**Leflunomide an Immunosuppressive Drug for Antiviral Purpose in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation**

Christophe Bazin

*Hôpital Européen Georges-Pompidou, Assistance Publique – Hôpitaux de Paris France* 

#### **1. Introduction**

#### **1.1 BK Virus**

34 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

Tajunisah, I.; Reddy, S. C.; Tan, L. H. (2009). Acute retinal necrosis by cytomegalovirus in an

Tam, P. M.; Hooper, C. Y.; Lightman, S. (2010). Antiviral selection in the management of

Theil, D.; Derfuss, T.; Paripovic, I.; Herberger, S.; Meinl, E.; Schueler, O.; Strupp, M.; Arbusow,

Tibbetts, M. D.; Shah, C. P.; Young, L. H.; Duker, J. S.; Maguire, J. I.; Morley, M. G. (2010). Treatment of acute retinal necrosis. *Ophthalmology,* Vol. 117, No. 4, pp. 818-824 Tran, T. H.; Rozenberg, F.; Cassoux, N.; Rao, N. A.; LeHoang, P.; Bodaghi, B. (2003a).

Tran, T. H.; Stanescu, D.; Caspers-Velu, L.; Rozenberg, F.; Liesnard, C.; Gaudric, A.; Leho-

Urayama, A.; Yamada, N.; Sasaki, T. (1971). Unilateral acute uveitis with periarteritis and detachment. *Japanese Journal of Clinical Ophthalmology,* Vol. 25, pp. 607-619 Usui, Y.; Takeuchi, M.; Goto, H.; Mori, H.; Kezuka, T.; Sakai, J.; Usui, M. (2008). Acute retinal

Vandercam, T.; Hintzen, R. Q.; de Boer, J. H.; van der Lelij, A. (2008). Herpetic encephalitis is a risk factor for acute retinal necrosis. *Neurology,* Vol. 71, No. 16, pp. 1268-1274 van Gelder, R. N.; Willig, J. L.; Holland, G. N.; Kaplan, H. J. (2001). Herpes simplex virus

Velez, G.; Roy, C. E.; Whitcup, S. M.; Chan, C. C.; Robinson, M. R. (2001). High-dose intra-

Voros, G. M.; Pandit, R.; Snow, M.; Griffiths, P. G. (2006). Unilateral recurrent acute retinal

Winterhalter, S.; Adams, O.; Althaus, Ch.; Stammen, J.; Schöler, E. M.; Joussen, A. M. (2007).

Yamamoto, S.; Nakao, T.; Kajiyama, K. (2007). Acute retinal necrosis following herpes sim-

Yamamoto, S.; Sugita, S.; Sugamoto, Y.; Shimizu, N.; Morio, T.; Mochizuki, M. (2008). Quan-

Zambarakji, H. J.; Obi, A. A.; Mitchell, S. M. (2002). Successful treatment of varicella-zoster

acute retinal necrosis. Clinical Ophthalmology, Vol.4, pp. 11-20

retinitis. *British Journal of Ophthalmology,* Vol. 87, No. 1, pp. 79-83.

*American Journal of Ophthalmology,* Vol. 137, No. 5, pp. 872-879

necrosis in Japan. *Ophthalmology,* Vol. 115, No. 9, pp. 1632-1633

*European Journal of Ophthalmology,* Vol. 16, No. 3, pp. 484-486

plex encephalitis. *Archives of Neurology,* Vol. 64, No. 2, pp. 283

*Immunology and Inflammation,* Vol. 10, No. 1, pp. 41-46

No. 7, pp. 928-932

*thalmology,* Vol. 29, No. 2, pp. 85-90

*mology,* Vol. 108, No. 5, pp. 869-876

*gy,* Vol. 62, No. 9, pp. 581-590

pp. 567-574

*Ophthalmology,* Vol. 131, No. 3, pp. 396-397

me in ocular fluids of patients with uveitis. *British Journal of Ophthalmology,* Vol. 92,

immunocompetent adult: case report and review of the literature. *International Oph-*

V.; Brandt, T. (2003). Latent herpesvirus infection in human trigeminal ganglia causes chronic immune response. *American Journal of Pathology,* Vol. 163, No. 6, pp. 2179-2184

Polymerase chain reaction analysis of aqueous humour samples in necrotising

ang, P.; Bodaghi, B. (2003b). Clinical characteristics of acute HSV-2 retinal necrosis.

type 2 as a cause of acute retinal necrosis syndrome in young patients. *Ophthal-*

vitreal ganciclovir and foscarnet for cytomegalovirus retinitis. *American Journal of* 

necrosis syndrome caused by cytomegalovirus in an immune-competent adult.

Acute retinal necrosis. *Klinische Monatsblätter für Augenheilkunde,* Vol. 224, No. 7,

titative PCR for the detection of genomic DNA of Epstein-Barr virus in ocular fluids of patients with uveitis. *Japanese Journal of Ophthalmology,* Vol. 52, No. 6, pp. 463-467 Young, N. J.; Bird, A. C. (1978). Bilateral acute retinal necrosis. *British Journal of Ophthalmolo-*

virus retinitis with aggressive intravitreal and systemic antiviral therapy. *Ocular* 

BK virus is a polyomavirus belonging to the *papovaviridae* branch. In addition to BK, the human polyomavirus family includes John Cunningham virus (JCV), Washington University virus (WUV), Karolinska Institute virus (KIV) and Merkel cell viruses (Boothpur et al. 2010). BK virus is a virus without a shell and it has a double-stranded circular nonenveloped DNA. It was first discovered and isolated in 1971 just like JC virus, responsible for Progressive Multifocal Leukoencephalopathy (PML). Contamination usually occurs during early childhood through the airway without clinical symptoms. BK virus seroprevalence in general population is around 60%. The main latency areas are the kidney and the urothelium. Asymptomatic BK virus infection is often acquired in childhood and the virus persists in a dormant state in urothelium and kidneys of healthy and immunocompetent individuals, where it can be reactivated under immunosuppression (Nickeleit et al. 2000a; Brocker et al. 2011).

#### **1.2 Prevalence and incidence**

Urinary viral prevalence for BK virus is between 0.3% and 6% in general population, and increases in functions of immunosuppression degree; between 10% and 45% in patients after renal transplant, 30% in patients after bone marrow graft and 25% in patients with Human immunodeficiency virus. In patients with renal graft, the annual incidence of the nephropathy is between 3% and 5% (Randhawa et al. 2000; Pavlakis et al. 2006).

#### **1.3 Risk factors**

BK virus-associated nephropathy seems to be promoted by the concurrent presence of several risk factors. The immunosuppressive regimen strength, with high level blood concentrations, is the first factor involved. Most patients affected by BK virus-associated nephropathy previously had an intensification of immunosuppressive regimen due to a rejection event or a treatment including tacrolimus and/or mycophenolate mofetil

Leflunomide an Immunosuppressive Drug for Antiviral Purpose

(Chen et al. 2001; Randhawa et al. 2002).

**1.6 Histology** 

(Nickeleit et al. 2000a).

**1.7 Interstitial inflammation** 

highly unlikely (Nickeleit et al. 2000a).

in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation 37

importance. Some authors have brought to light emerging mutations which could explain the renal physiopathologic effects of these viruses (Chen et al. 2001). Virus selection in patients with renal graft results in rearrangements in the T antigen region, mutations in the non coding regulatory zone, and above all variations in VP1 protein (Smith et al. 1998; Baksh et al. 2001; Randhawa et al. 2002). Heterogeneity and genetic instability in a same patient seem to favor renal damage and the risk of escaping immunologic surveillance

BK virus is usually associated with changes in the kidney and sometimes haemorrhagic cystitis and urethral stenosis. The virus affects tubular epithelial cells that show characteristic intranuclear inclusion bodies. Diagnosis relies upon urinary cytology, detection of viral DNA in fluids and renal biopsy. The nephropathy diagnosis can only be made histologically in a graft biopsy. Intranuclear viral inclusions are exclusively seen in epithelial cells and tubular cells reveal focal necrosis. Four different variants of intranuclear inclusion bodies can be seen throughout the entire nephron. Type 1 is the most frequently observed; it is an amorphous basophilic ground-glass variant. Type 2 is an eosinophilic granular type, halo surrounded. Type 3 is a finely granular form lacking a halo. And finally type 4 is a vesicular variant presenting markedly enlarged nuclei and irregular chromatin. Infected cells which are rounded-up and extruded from the epithelial cell layer into tubular lumens are frequently observed. Viral replication often causes tubular epithelial cell necrosis with denudation of basement membranes. Although cytopathic signs can be seen along the entire nephron, they are mostly abundant in distal tubular parts and collecting ducts

Interstitial inflammation in BK virus-associated nephropathy still remains controversial and needs to be fully explained. The major outcome is to distinguish between virally induced interstitial nephritis and cellular rejection. As lowering immunosuppression is the first option which can be chosen in the treatment, this choice requires two conditions, first the absence of rejection and second the BK virus should not trigger rejection. BK virus is frequently accompanied by an heterogeneous inflammatory reaction (Drachenberg et al. 1999). This inflammation can be minimal or absent in up to 17% of biopsies (Nickeleit et al. 2000a). When inflammation is encountered, the inflammatory cell infiltrate is composed of lymphocytes, macrophages and occasional plasma cells. Polymorphonuclear leukocytes can be seen in response to markedly damaged tubules with urinary leakage (Drachenberg et al. 1999). About 50% of biopsies performed during persistent BK virus-associated nephropathy show evidence of cellular rejection as conventionally defined with abundant tubulitis and transplant endarteritis in about 25%. Typically, mononuclear cell infiltrates and tubulitis are pronounced in areas without viral inclusions making virally induced interstitial nephritis

The upregulation of MCH-class II (HLA-DR) and ICAM-1 on tubular epithelial cells is a typical finding in graft biopsies with cellular rejection and can serve as an adjunct diagnostic tool (Seron et al. 1989; Nickeleit et al. 1998). HLA-DR expression can stimulate an allogenic

combined with monoclonal or polyclonal antibodies (De Luca et al. 2000; Nickeleit et al. 2000b; Randhawa et al. 2000; Hirsch et al. 2001). Conversely, no cases have been reported in patients treated with cyclosporine and corticosteroids (Binet et al. 1999; Mengel et al. 2003).

The other risk factors identified comprise donor characteristics, such as female gender, deceased donation, ischemia-reperfusion injury, high BK virus specific antibody titres, HLA mismatch and African-American ethnicity. The recipient characteristics in cause are older age, male gender, white race, diabetes, obesity, retransplantation, lack of HLA C-7, low or absent BK virus specific T-cell activity. Lastly, in addition to high immunosuppressive drug levels and tacrolimus based combinations, other post-transplant factors can be mentioned as acute rejection and antirejection treatment, cumulative steroid exposure and lymphocyte depleting antibodies (Gupta & Gupta, 2011).

Although immunosuppression increases the probability of latent BK virus reactivation, clinical manifestation of disease is rare. When symptoms occur, on the clinical point of view, a progressive decline of the renal functions can be observed up to 45 % of patients, usually 9 to 12 months after the renal transplant (Nickeleit et al. 2000a; Randhawa et al. 2000). The most serious form of the infection turns out to be the interstitial nephritis; although the BK virus was discovered in the 70's, this serious complication has first been seen in 1995. This fact can probably be explained by the commercialization of two drugs in 1995 and 1996, tacrolimus and mycophenolate mofetil.

Interestingly, BK virus-associated nephropathy happens hardly only in patients with renal graft. Some explanations could be found, such as the role of vesico-urethral reflux, quite usual in renal transplantation, with the systemic pathway of collecting tubes in peritubular capillary and the tubular localization of the infection. Some authors evoked easiness in the viral antigens presentation in a context of allograft, cold ischemia, tubular necrosis and graft rejection. For that matter BK virus-associated nephropathy is generally related to rejection, both events being linked in time; most cases of nephropathy are falsely tagged and treated just like a rejection. This confusion suggests that rejection is a risk factor on its own. Viral antigens could probably lead to rejection and conversely a rejection event could reactivate viral replication. In mice, Atencio et al proved an inductive effect of tubular damage upon BK virus linked interstitial nephritis (Atencio et al. 1993).

#### **1.4 Clinical aspects**

BK virus infection may lead to encephalitis, retinitis, pneumonitis, damage of the kidneys, bleeding of the bladder, and blockage of urine passageways. Minor infections are most of the time asymptomatic and can lead to urethral stenosis. This infection occurs 1 to 45 months (average 12.5 months) after the graft. It is linked to the conjunction of multiple factors, including an intense immunosuppressive regimen, viral reactivation, existence of an immune-allogenic conditions, and a suffering tubular due to ischemia or rejection (Nickeleit et al. 2003).

#### **1.5 Genotypes**

BK virus comes in the form of 4 different genotypes, type I being the most common seen. The coding regions for non structural proteins T and t antigens (pathogenic viral power), viral capsid proteins (cellular tropism) and a regulatory non coding zone have a vital importance. Some authors have brought to light emerging mutations which could explain the renal physiopathologic effects of these viruses (Chen et al. 2001). Virus selection in patients with renal graft results in rearrangements in the T antigen region, mutations in the non coding regulatory zone, and above all variations in VP1 protein (Smith et al. 1998; Baksh et al. 2001; Randhawa et al. 2002). Heterogeneity and genetic instability in a same patient seem to favor renal damage and the risk of escaping immunologic surveillance (Chen et al. 2001; Randhawa et al. 2002).

#### **1.6 Histology**

36 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

combined with monoclonal or polyclonal antibodies (De Luca et al. 2000; Nickeleit et al. 2000b; Randhawa et al. 2000; Hirsch et al. 2001). Conversely, no cases have been reported in patients treated with cyclosporine and corticosteroids (Binet et al. 1999; Mengel et al. 2003). The other risk factors identified comprise donor characteristics, such as female gender, deceased donation, ischemia-reperfusion injury, high BK virus specific antibody titres, HLA mismatch and African-American ethnicity. The recipient characteristics in cause are older age, male gender, white race, diabetes, obesity, retransplantation, lack of HLA C-7, low or absent BK virus specific T-cell activity. Lastly, in addition to high immunosuppressive drug levels and tacrolimus based combinations, other post-transplant factors can be mentioned as acute rejection and antirejection treatment, cumulative steroid exposure and lymphocyte

Although immunosuppression increases the probability of latent BK virus reactivation, clinical manifestation of disease is rare. When symptoms occur, on the clinical point of view, a progressive decline of the renal functions can be observed up to 45 % of patients, usually 9 to 12 months after the renal transplant (Nickeleit et al. 2000a; Randhawa et al. 2000). The most serious form of the infection turns out to be the interstitial nephritis; although the BK virus was discovered in the 70's, this serious complication has first been seen in 1995. This fact can probably be explained by the commercialization of two drugs in 1995 and 1996,

Interestingly, BK virus-associated nephropathy happens hardly only in patients with renal graft. Some explanations could be found, such as the role of vesico-urethral reflux, quite usual in renal transplantation, with the systemic pathway of collecting tubes in peritubular capillary and the tubular localization of the infection. Some authors evoked easiness in the viral antigens presentation in a context of allograft, cold ischemia, tubular necrosis and graft rejection. For that matter BK virus-associated nephropathy is generally related to rejection, both events being linked in time; most cases of nephropathy are falsely tagged and treated just like a rejection. This confusion suggests that rejection is a risk factor on its own. Viral antigens could probably lead to rejection and conversely a rejection event could reactivate viral replication. In mice, Atencio et al proved an inductive effect of tubular damage upon

BK virus infection may lead to encephalitis, retinitis, pneumonitis, damage of the kidneys, bleeding of the bladder, and blockage of urine passageways. Minor infections are most of the time asymptomatic and can lead to urethral stenosis. This infection occurs 1 to 45 months (average 12.5 months) after the graft. It is linked to the conjunction of multiple factors, including an intense immunosuppressive regimen, viral reactivation, existence of an immune-allogenic conditions, and a suffering tubular due to ischemia or rejection (Nickeleit

BK virus comes in the form of 4 different genotypes, type I being the most common seen. The coding regions for non structural proteins T and t antigens (pathogenic viral power), viral capsid proteins (cellular tropism) and a regulatory non coding zone have a vital

depleting antibodies (Gupta & Gupta, 2011).

tacrolimus and mycophenolate mofetil.

**1.4 Clinical aspects** 

et al. 2003).

**1.5 Genotypes** 

BK virus linked interstitial nephritis (Atencio et al. 1993).

BK virus is usually associated with changes in the kidney and sometimes haemorrhagic cystitis and urethral stenosis. The virus affects tubular epithelial cells that show characteristic intranuclear inclusion bodies. Diagnosis relies upon urinary cytology, detection of viral DNA in fluids and renal biopsy. The nephropathy diagnosis can only be made histologically in a graft biopsy. Intranuclear viral inclusions are exclusively seen in epithelial cells and tubular cells reveal focal necrosis. Four different variants of intranuclear inclusion bodies can be seen throughout the entire nephron. Type 1 is the most frequently observed; it is an amorphous basophilic ground-glass variant. Type 2 is an eosinophilic granular type, halo surrounded. Type 3 is a finely granular form lacking a halo. And finally type 4 is a vesicular variant presenting markedly enlarged nuclei and irregular chromatin. Infected cells which are rounded-up and extruded from the epithelial cell layer into tubular lumens are frequently observed. Viral replication often causes tubular epithelial cell necrosis with denudation of basement membranes. Although cytopathic signs can be seen along the entire nephron, they are mostly abundant in distal tubular parts and collecting ducts (Nickeleit et al. 2000a).

#### **1.7 Interstitial inflammation**

Interstitial inflammation in BK virus-associated nephropathy still remains controversial and needs to be fully explained. The major outcome is to distinguish between virally induced interstitial nephritis and cellular rejection. As lowering immunosuppression is the first option which can be chosen in the treatment, this choice requires two conditions, first the absence of rejection and second the BK virus should not trigger rejection. BK virus is frequently accompanied by an heterogeneous inflammatory reaction (Drachenberg et al. 1999). This inflammation can be minimal or absent in up to 17% of biopsies (Nickeleit et al. 2000a). When inflammation is encountered, the inflammatory cell infiltrate is composed of lymphocytes, macrophages and occasional plasma cells. Polymorphonuclear leukocytes can be seen in response to markedly damaged tubules with urinary leakage (Drachenberg et al. 1999). About 50% of biopsies performed during persistent BK virus-associated nephropathy show evidence of cellular rejection as conventionally defined with abundant tubulitis and transplant endarteritis in about 25%. Typically, mononuclear cell infiltrates and tubulitis are pronounced in areas without viral inclusions making virally induced interstitial nephritis highly unlikely (Nickeleit et al. 2000a).

The upregulation of MCH-class II (HLA-DR) and ICAM-1 on tubular epithelial cells is a typical finding in graft biopsies with cellular rejection and can serve as an adjunct diagnostic tool (Seron et al. 1989; Nickeleit et al. 1998). HLA-DR expression can stimulate an allogenic

Leflunomide an Immunosuppressive Drug for Antiviral Purpose

lowering immunosuppression (Dheir et al. 2011).

CNI, sirolimus, prednisone at diagnosis CNI, MMF, prednisone at diagnosis

CNI withdrawal AP withdrawal Dose reduction within 1 mo of diagnosis

Ancillary therapy

**2.2 Cidofovir** 

Cidofovir Intravenous Ig Leflunomide Acute rejection after diagnosis

CsA, cyclosporine A (Weiss et al. 2008).

Median serum creatinine at diagnosis (mg/dl) Agent withdrawal within 1 mo of diagnosis

> CNI reduction, AP reduction < 50% CNI reduction, AP reduction ≥ 50% Tac to CsA switch, AP reduction < 50%

tacrolimus and 100 ng/mL for cyclosporine (Gupta & Gupta, 2011).

in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation 39

In a previous study, mycophenolate mofetil was stopped the day leflunomide treatment was initiated; tacrolimus and everolimus were respectively reduced of 50% and 12.5%. Therapeutic drug monitoring target for tacrolimus was lowered to 4 - 6 ng/mL on immunoenzymatic techniques on whole blood. Corticosteroids were kept with average dosage of 5 to 10 mg per day (Bazin et al. 2009). Other authors recommend even lower targets with 3 ng/mL for

Besides, lowering immunosuppressive regimens together with a specific treatment for BK virus-associated nephropathy recently turns out to be effective to prolong graft survival, and moreover a safe treatment with acute rejection rates not increased significantly after

Two different therapeutic strategies have been evaluated: the immunosuppression withdrawal (3-drug to 2-drug immunosuppression) within the first month versus reduction of immunosuppression. The regimen modifications and results are presented in table 1 and figure 1. The Withdrawal cohort had significantly better graft survival at 1 year compared with the Reduction cohort (1-year graft survival 87.8% versus 56.2%, P = 0.03) (Weiss et al. 2008).

Table 1. Immunosuppression modifications comparing immunosuppression withdrawal *versus* immunosuppression reduction after diagnosis of BK virus-associated nephropathy. CNI, calcineurin inhibitor; MMF, mycophenolate; AP, antiproliferative; Tac, tacrolimus;

Cidofovir (Vistide®) is an injectable antiviral drug. It belongs to nucleoside analogues. It is used in infections due to human Cytomegalovirus (CMV) in adults suffering of AIDS (Acquired immune deficiency syndrome) without renal insufficiency, and it should only be used when other treatments are considered as inappropriate. Cidofovir counters CMV replication thanks to a selective inhibition of viral DNA polymerase in *herpesviridae* viruses

Withdrawal cohort (*n =* 17)

> 12 5 2.5

14 3

> - - -

Reduction cohort (*n* = 18)

> 11 7 2.2

> > - -

8 7 3

p

0.56 0.56 0.30

> - -

> - - -

0.40 0.88 1.0 1.0

lymphocytic reaction and also enhance T cell mediated lysis (Rosenberg et al. 1992). Consequently, BK virus could probably trigger rejection episodes by inducing HLA-DR upregulation as previously proposed for CMV (von Willebrand et al. 1986). However, no association could be found between BK virus infection and tubular HLA-DR expression based on immunofluorescence double labeling staining techniques. It is only in biopsies showing characteristic morphological evidence of rejection with marked tubulitis that typical upregulation of HLA-DR and ICAM-1 could be observed (Nickeleit et al. 2000a). Therefore, BK virus does not stimulate HLA-DR expression. Consequently no significant difference can be found between the prevalence of rejection in tissue samples taken during persistent BK virus-associated nephropathy and time matched controls without BK virus nephropathy. Thus, BK virus does not seem to provoke a constant and pronounced interstitial inflammatory reaction and should probably not be considered as associated with an increased prevalence of rejection episodes (Nickeleit et al. 2000a).

#### **1.8 PCR**

BK-virus DNA in the plasma and the urine, which can be detected by PCR (Polymerase Chain Reaction), is closely associated with nephropathy. Quantitative PCR can be used to follow the disease evolution and the treatment efficiency (Randhawa et al. 2004).

As for BK virus infection, this technique has proven a 100% sensivity, a 88% specificity and above all a negative predictive test of 100%. Hirsch et al. have even shown a correlation between viral load and nephropathy and proposed a cut-off above which the risk of nephropathy is significant: all patients with more than 7700 copies/mL in plasma had typical BK virus-associated nephropathy lesions on the biopsy (Hirsch et al. 2002).

The nephropathy evolution is very poor with a cytopathogenic effect persistent in up to 70% of patients, a graft loss in 45% of cases; and major sequel fibrosis in 75% of cases, even if viremia can be controlled (Nickeleit et al. 2000a; Randhawa et al. 2000; Mylonakis et al. 2001; Mengel et al. 2003).

#### **2. Classical treatments for BKV nephropathy**

Therapeutic alternatives are quite few in number. Despite the absence of randomized clinical trials, the current approach generally includes reduction of immunosuppression (Brennan et al. 2005; Hardinger et al. 2010). The rational is to allow host immune function to combat the virus, with the risk to increase acute and subclinical rejection. Lowering immunosuppression with smaller dosage and/or less drugs is partially efficient and seems to be the first thing to do. Except from lowering immunosuppression, to date no treatment seem to be efficient enough to be recommended to all patients, and new research have to be performed because of the poor evidence in small series of patients (Johnston et al. 2010).

#### **2.1 Lowering immunosuppression**

Reduction of immunosuppression is to date the only consensus regarding the treatment of BK virus-associated nephropathy. Lowering tacrolimus dosage of 41% and mycophenolate mofetil dosage of 44% allowed to eradicate 24 patients' viremia in 6 months (Saad et al. 2008).

In a previous study, mycophenolate mofetil was stopped the day leflunomide treatment was initiated; tacrolimus and everolimus were respectively reduced of 50% and 12.5%. Therapeutic drug monitoring target for tacrolimus was lowered to 4 - 6 ng/mL on immunoenzymatic techniques on whole blood. Corticosteroids were kept with average dosage of 5 to 10 mg per day (Bazin et al. 2009). Other authors recommend even lower targets with 3 ng/mL for tacrolimus and 100 ng/mL for cyclosporine (Gupta & Gupta, 2011).

Besides, lowering immunosuppressive regimens together with a specific treatment for BK virus-associated nephropathy recently turns out to be effective to prolong graft survival, and moreover a safe treatment with acute rejection rates not increased significantly after lowering immunosuppression (Dheir et al. 2011).

Two different therapeutic strategies have been evaluated: the immunosuppression withdrawal (3-drug to 2-drug immunosuppression) within the first month versus reduction of immunosuppression. The regimen modifications and results are presented in table 1 and figure 1. The Withdrawal cohort had significantly better graft survival at 1 year compared with the Reduction cohort (1-year graft survival 87.8% versus 56.2%, P = 0.03) (Weiss et al. 2008).


Table 1. Immunosuppression modifications comparing immunosuppression withdrawal *versus* immunosuppression reduction after diagnosis of BK virus-associated nephropathy. CNI, calcineurin inhibitor; MMF, mycophenolate; AP, antiproliferative; Tac, tacrolimus; CsA, cyclosporine A (Weiss et al. 2008).

#### **2.2 Cidofovir**

38 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

lymphocytic reaction and also enhance T cell mediated lysis (Rosenberg et al. 1992). Consequently, BK virus could probably trigger rejection episodes by inducing HLA-DR upregulation as previously proposed for CMV (von Willebrand et al. 1986). However, no association could be found between BK virus infection and tubular HLA-DR expression based on immunofluorescence double labeling staining techniques. It is only in biopsies showing characteristic morphological evidence of rejection with marked tubulitis that typical upregulation of HLA-DR and ICAM-1 could be observed (Nickeleit et al. 2000a). Therefore, BK virus does not stimulate HLA-DR expression. Consequently no significant difference can be found between the prevalence of rejection in tissue samples taken during persistent BK virus-associated nephropathy and time matched controls without BK virus nephropathy. Thus, BK virus does not seem to provoke a constant and pronounced interstitial inflammatory reaction and should probably not be considered as associated with

BK-virus DNA in the plasma and the urine, which can be detected by PCR (Polymerase Chain Reaction), is closely associated with nephropathy. Quantitative PCR can be used to

As for BK virus infection, this technique has proven a 100% sensivity, a 88% specificity and above all a negative predictive test of 100%. Hirsch et al. have even shown a correlation between viral load and nephropathy and proposed a cut-off above which the risk of nephropathy is significant: all patients with more than 7700 copies/mL in plasma had

The nephropathy evolution is very poor with a cytopathogenic effect persistent in up to 70% of patients, a graft loss in 45% of cases; and major sequel fibrosis in 75% of cases, even if viremia can be controlled (Nickeleit et al. 2000a; Randhawa et al. 2000; Mylonakis et al. 2001;

Therapeutic alternatives are quite few in number. Despite the absence of randomized clinical trials, the current approach generally includes reduction of immunosuppression (Brennan et al. 2005; Hardinger et al. 2010). The rational is to allow host immune function to combat the virus, with the risk to increase acute and subclinical rejection. Lowering immunosuppression with smaller dosage and/or less drugs is partially efficient and seems to be the first thing to do. Except from lowering immunosuppression, to date no treatment seem to be efficient enough to be recommended to all patients, and new research have to be performed because of the poor evidence in small series of patients (Johnston et al. 2010).

Reduction of immunosuppression is to date the only consensus regarding the treatment of BK virus-associated nephropathy. Lowering tacrolimus dosage of 41% and mycophenolate mofetil dosage of 44% allowed to eradicate 24 patients' viremia in 6 months (Saad et al.

follow the disease evolution and the treatment efficiency (Randhawa et al. 2004).

typical BK virus-associated nephropathy lesions on the biopsy (Hirsch et al. 2002).

an increased prevalence of rejection episodes (Nickeleit et al. 2000a).

**2. Classical treatments for BKV nephropathy** 

**2.1 Lowering immunosuppression** 

**1.8 PCR** 

Mengel et al. 2003).

2008).

Cidofovir (Vistide®) is an injectable antiviral drug. It belongs to nucleoside analogues. It is used in infections due to human Cytomegalovirus (CMV) in adults suffering of AIDS (Acquired immune deficiency syndrome) without renal insufficiency, and it should only be used when other treatments are considered as inappropriate. Cidofovir counters CMV replication thanks to a selective inhibition of viral DNA polymerase in *herpesviridae* viruses

Leflunomide an Immunosuppressive Drug for Antiviral Purpose

anti-proliferative action (Williamson et al. 1995; Fox et al. 1999).

responsible for the activity and side effects of leflunomide.

respiratory syncytial virus (RSV) *in vitro* and *in vivo* (Dunn et al. 2011).

Aventis 2009).

**3.2 Pharmacodynamy** 

**3.3 Pharmacokinetics** 

maintenance dose per day.

**3.4 Predictive efficiency** 

in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation 41

line to treat severe and active forms of psoriatic arthritis (Maddison et al. 2005; Sanofi-

Its immunosuppressive action lies in the dihydroorotate dehydrogenase (DHOH) inhibition, an enzyme necessary for de novo synthesis of pyrimidic bases in lymphocytes. It also has an

Besides, leflunomide has proven abilities to reduce the viral proliferation for Human Cytomegalovirus (CMV), Herpes Simplex Viruses (HSV) *in vitro* (Knight et al. 2001) and

After per os administration, leflunomide is promptly and almost fully metabolized into its active form, terflunomide or A77 1726. This metabolism happens during first pass and consists in a carbon cycle opening in the intestinal wall and the liver. 95% of leflunomide is turned into A77 1726 this way, the remains into minor metabolites. Terflunomide is the drug

Leflunomide bioavailability is about 82% in healthy volunteers (Sanofi-Aventis 2009). Elimination plasma half-life of A77 1726 is quite considerable, with some 15 days in average. Patients are so compelled to take a 100 mg charging dose for 3 days before a 10 to 20 mg

After a unique charging dose, A77 1726 Tmax is comprised between 6 and 12 hours, with a high inter-individual variability in patients with rheumatoid arthritis (Rozman 2002).

The volume of distribution (Vd) is quite low, with about 12.7 L (6 to 30.8 L), which is logical with its high affinity and linkage to albumin (99.4% in healthy volunteers) (Rozman 2002). Elimination of A77 1726 is slow, it is characterized by an apparent clearance of 0.051 L/h (Rozman 2002). This elimination is mostly renal (43%) and biliary (48%), as a consequence renal insufficiency alone does not significantly impair A77 1726 plasma concentrations (Beaman et al. 2002). Furthermore haemodialysis does not modify concentrations or clearance of A77 1726, which allows the patients to be on a dialysis without any dose adjustment. In vitro studies showed that cytochroms P450, in particular cytochroms 1A2, 2C19, 3A4 and 3A5 were involved in leflunomide metabolism (Kalgutkar et al. 2003). A pharmacogenetic study also showed the link between a polymorphism of cytochrom 1A2 and a risk of toxicity for patients with rheumatoid arthritis (Bohanec Grabar et al. 2008).

In rheumatoid arthritis, plasma concentrations above 13 µg/mL seem to be efficient. These concentrations are usually reached with 20 mg per day dosage (van Roon et al. 2005). Some authors tried to establish a relation between plasma concentrations and efficiency in patients with BK virus nephropathy, showing a tendency but with no absolute proof. Finally to date, no link between plasma concentrations and side effects has been shown (Bazin et al. 2009). Yet *in vitro* studies seem to show a predictive correlation between concentrations and the

Fig. 1. Immunosuppression withdrawal preserves graft function compared with reduction (Weiss et al. 2008).

 (Gilead 2010). Cidofovir has also demonstrated *in vitro* activity against murine and simian polyomavirus strains and appears to have activity against JC virus *in vivo* (De Luca et al. 2000). Pharmacokinetic studies have demonstrated that cidofovir is highly concentrated in urine and renal tissue which are the primary sites of BK virus infection (Kadambi et al. 2003)*.* This fact highlights the possibility that low doses might be sufficient for treating an infectious process, such as BK virus-associated nephropathy, that appears to be largely localized to the kidney and genitourinary tract.

The treatment consists in a low-dose treatment, 0.25 mg/kg/day intravenous during 2 weeks, associated to a prior hydration of 1 litre of saline solution. Cidofovir seems to be efficient in BK virus as well, but it tends to concentrate itself inside the kidney and can be responsible of a nephrotoxicity mostly for tubular cells leading to renal insufficiency.

Few cases have been described in literature and no conclusion can be given on the real efficacy of cidofovir. Indeed, despite viremia control, viruria remains detectable and the treatment is not able to avoid the evolution towards fibrosis and renal insufficiency (Kadambi et al. 2003; Kuypers et al. 2005).

In some cases, cidofovir may also become deleterious (Pallet et al. 2010; Talmon et al. 2010).

#### **3. Leflunomide**

#### **3.1 Drug generalities**

Leflunomide (Arava®) is a disease-modifying antirheumatic drug (DMARD) used in adult patients with methotrexate intolerance, failure or loss of efficiency; it is also used in a second line to treat severe and active forms of psoriatic arthritis (Maddison et al. 2005; Sanofi-Aventis 2009).

#### **3.2 Pharmacodynamy**

40 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

Fig. 1. Immunosuppression withdrawal preserves graft function compared with reduction

 (Gilead 2010). Cidofovir has also demonstrated *in vitro* activity against murine and simian polyomavirus strains and appears to have activity against JC virus *in vivo* (De Luca et al. 2000). Pharmacokinetic studies have demonstrated that cidofovir is highly concentrated in urine and renal tissue which are the primary sites of BK virus infection (Kadambi et al. 2003)*.* This fact highlights the possibility that low doses might be sufficient for treating an infectious process, such as BK virus-associated nephropathy, that appears to be largely

The treatment consists in a low-dose treatment, 0.25 mg/kg/day intravenous during 2 weeks, associated to a prior hydration of 1 litre of saline solution. Cidofovir seems to be efficient in BK virus as well, but it tends to concentrate itself inside the kidney and can be

Few cases have been described in literature and no conclusion can be given on the real efficacy of cidofovir. Indeed, despite viremia control, viruria remains detectable and the treatment is not able to avoid the evolution towards fibrosis and renal insufficiency

In some cases, cidofovir may also become deleterious (Pallet et al. 2010; Talmon et al. 2010).

Leflunomide (Arava®) is a disease-modifying antirheumatic drug (DMARD) used in adult patients with methotrexate intolerance, failure or loss of efficiency; it is also used in a second

responsible of a nephrotoxicity mostly for tubular cells leading to renal insufficiency.

(Weiss et al. 2008).

**3. Leflunomide** 

**3.1 Drug generalities** 

localized to the kidney and genitourinary tract.

(Kadambi et al. 2003; Kuypers et al. 2005).

Its immunosuppressive action lies in the dihydroorotate dehydrogenase (DHOH) inhibition, an enzyme necessary for de novo synthesis of pyrimidic bases in lymphocytes. It also has an anti-proliferative action (Williamson et al. 1995; Fox et al. 1999).

Besides, leflunomide has proven abilities to reduce the viral proliferation for Human Cytomegalovirus (CMV), Herpes Simplex Viruses (HSV) *in vitro* (Knight et al. 2001) and respiratory syncytial virus (RSV) *in vitro* and *in vivo* (Dunn et al. 2011).

#### **3.3 Pharmacokinetics**

After per os administration, leflunomide is promptly and almost fully metabolized into its active form, terflunomide or A77 1726. This metabolism happens during first pass and consists in a carbon cycle opening in the intestinal wall and the liver. 95% of leflunomide is turned into A77 1726 this way, the remains into minor metabolites. Terflunomide is the drug responsible for the activity and side effects of leflunomide.

Leflunomide bioavailability is about 82% in healthy volunteers (Sanofi-Aventis 2009). Elimination plasma half-life of A77 1726 is quite considerable, with some 15 days in average. Patients are so compelled to take a 100 mg charging dose for 3 days before a 10 to 20 mg maintenance dose per day.

After a unique charging dose, A77 1726 Tmax is comprised between 6 and 12 hours, with a high inter-individual variability in patients with rheumatoid arthritis (Rozman 2002).

The volume of distribution (Vd) is quite low, with about 12.7 L (6 to 30.8 L), which is logical with its high affinity and linkage to albumin (99.4% in healthy volunteers) (Rozman 2002). Elimination of A77 1726 is slow, it is characterized by an apparent clearance of 0.051 L/h (Rozman 2002). This elimination is mostly renal (43%) and biliary (48%), as a consequence renal insufficiency alone does not significantly impair A77 1726 plasma concentrations (Beaman et al. 2002). Furthermore haemodialysis does not modify concentrations or clearance of A77 1726, which allows the patients to be on a dialysis without any dose adjustment. In vitro studies showed that cytochroms P450, in particular cytochroms 1A2, 2C19, 3A4 and 3A5 were involved in leflunomide metabolism (Kalgutkar et al. 2003). A pharmacogenetic study also showed the link between a polymorphism of cytochrom 1A2 and a risk of toxicity for patients with rheumatoid arthritis (Bohanec Grabar et al. 2008).

#### **3.4 Predictive efficiency**

In rheumatoid arthritis, plasma concentrations above 13 µg/mL seem to be efficient. These concentrations are usually reached with 20 mg per day dosage (van Roon et al. 2005). Some authors tried to establish a relation between plasma concentrations and efficiency in patients with BK virus nephropathy, showing a tendency but with no absolute proof. Finally to date, no link between plasma concentrations and side effects has been shown (Bazin et al. 2009). Yet *in vitro* studies seem to show a predictive correlation between concentrations and the

Leflunomide an Immunosuppressive Drug for Antiviral Purpose

in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation 43

Fig. 3. Interactions between BK virus and inhibitors, sirolimus and leflunomide (Liacini et al.

2010)

viral inhibitory effect: 10 µg/mL reduced the extracellular BKV load by 90% (IC90) but with significant host cytostatic effects (see figure 2) (Bernhoff et al. 2010).

Fig. 2. Effect of Terflunomide on BK Virus load *in vitro* (Bernhoff et al. 2010)

#### **3.5 Mechanism of action**

Researches about the mechanism of leflunomide have recently been brightened. Leflunomide has two mechanisms of action: inhibition of dihydroorotate dehydrogenase, a key enzyme in the pyrimidine synthesis pathway, and tyrosine kinase inhibition. Dihydroorotate dehydrogenase inhibition is the primary mechanism involved in rheumatoid arthritis treatment. Interactions between the BK virus and the cellular protein kinase AKt / mammalian target of rapamycin (mTOR) pathway have been discovered (Liacini et al. 2010). These interactions are described in figure 3.

Akt (protein kinase B) is a serine/threonine kinase activated by growth factors, cytokines and mitogens (Fayard et al. 2010). The mTOR pathway which controls protein synthesis is located downstream of Akt. Akt indirectly activates mTOR. Two mTOR complexes have yet been described, mTOR complex 1 (mTORC1) which controls translation initiation, and mTOR complex 2 (mTORC2) which controls cytoskeletal changes and is also a 3' phosphoinositide-dependent kinase-2 (PDK2), phosphorylating Akt, which may alter its substrate specificity (Bhaskar et al. 2007). Liacini et al showed that BK virus infecting renal tubular epithelial cells was able to activate the Akt/mTOR pathway; that leflunomide active metabolite, A77 1726 could inhibit PDK1 and Akt phosphorylation in a dose-dependent manner and in this way to reduce BK large T antigen expression and DNA replication. The combination of serine/threonine kinase inhibition of mTOR and tyrosine kinase inhibition significantly reduce the ability of the virus to survive and to produce new virions. More interesting though seems to be the combination of leflunomide and sirolimus targeting the Akt/mTOR pathway on different sites. Because both leflunomide and sirolimus possess immunosuppressive activity, this combination may allow treatment of BK virus-associated nephropathy without reduction of immunosuppression (Liacini et al. 2010).

viral inhibitory effect: 10 µg/mL reduced the extracellular BKV load by 90% (IC90) but with

significant host cytostatic effects (see figure 2) (Bernhoff et al. 2010).

Fig. 2. Effect of Terflunomide on BK Virus load *in vitro* (Bernhoff et al. 2010)

nephropathy without reduction of immunosuppression (Liacini et al. 2010).

(Liacini et al. 2010). These interactions are described in figure 3.

Researches about the mechanism of leflunomide have recently been brightened. Leflunomide has two mechanisms of action: inhibition of dihydroorotate dehydrogenase, a key enzyme in the pyrimidine synthesis pathway, and tyrosine kinase inhibition. Dihydroorotate dehydrogenase inhibition is the primary mechanism involved in rheumatoid arthritis treatment. Interactions between the BK virus and the cellular protein kinase AKt / mammalian target of rapamycin (mTOR) pathway have been discovered

Akt (protein kinase B) is a serine/threonine kinase activated by growth factors, cytokines and mitogens (Fayard et al. 2010). The mTOR pathway which controls protein synthesis is located downstream of Akt. Akt indirectly activates mTOR. Two mTOR complexes have yet been described, mTOR complex 1 (mTORC1) which controls translation initiation, and mTOR complex 2 (mTORC2) which controls cytoskeletal changes and is also a 3' phosphoinositide-dependent kinase-2 (PDK2), phosphorylating Akt, which may alter its substrate specificity (Bhaskar et al. 2007). Liacini et al showed that BK virus infecting renal tubular epithelial cells was able to activate the Akt/mTOR pathway; that leflunomide active metabolite, A77 1726 could inhibit PDK1 and Akt phosphorylation in a dose-dependent manner and in this way to reduce BK large T antigen expression and DNA replication. The combination of serine/threonine kinase inhibition of mTOR and tyrosine kinase inhibition significantly reduce the ability of the virus to survive and to produce new virions. More interesting though seems to be the combination of leflunomide and sirolimus targeting the Akt/mTOR pathway on different sites. Because both leflunomide and sirolimus possess immunosuppressive activity, this combination may allow treatment of BK virus-associated

**3.5 Mechanism of action** 

Fig. 3. Interactions between BK virus and inhibitors, sirolimus and leflunomide (Liacini et al. 2010)

Leflunomide an Immunosuppressive Drug for Antiviral Purpose

cidofovir (Avery et al. 2004).

**5. Conclusion** 

in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation 45

least as efficient as ganciclovir in CMV infections and does not seem to be affected by resistant viruses (John et al. 2004). Leflunomide has even be successfully used in a patient with bone marrow graft and infected by a resistant virus to ganciclovir, foscarnet and

The studies where leflunomide is used as an immunosuppressive drug in renal and hepatic graft are more and more, because leflunomide allows to reduce anti-calcineurin drugs which have the major inconvenient of nephrotoxicity, and potentially protects aside from CMV, HSV and BK virus infections (Hardinger et al. 2002; Williams et al. 2002). Moreover leflunomide seems to be an interesting alternative in BK virus-associated nephropathy in renal transplant by eradicating detectable viremia in some patients. Leflunomide also allows avoiding rejection in most cases in spite of classical immunosuppressive drugs dosage reduction. Besides one of leflunomide's main asset is its absence of renal toxicity, contrary to cidofovir (Williams et al. 2005; Josephson et al. 2006; Teschner et al. 2006; Faguer et al. 2007). Thanks to the encountered success in renal transplant, leflunomide is now used to treat hemorrhagic cystitis linked to BK virus in bone marrow transplant (Dropulic et al. 2008). However, due to the absence of randomized clinical trials with a sufficient number of patients,

some authors consider its use in a first-line drug not recommended (Chon et al. 2011).

Leflunomide pharmacokinetics is characterized by a great inter-individual variability with terflunomide concentrations from 15 to 130 µg/mL obtained with the same dosage (Bazin et al. 2009). In BK virus infection, terflunomide concentrations between 15-30 µg/mL and 35- 100 µg/mL are sufficient to suppress respectively 50% and 90% of the replication for CMV and BK virus *in vitro*. That is why a therapeutic margin between 50 and 100 µg/mL has been proposed in this indication (Josephson et al. 2006). However, current strategic therapy so as to limit BK virus incidence tends to manage an early reduction of immunosuppressive regimen to avoid the apparition of a nephropathy. A prospective study with a significant number of patients would be probably necessary to definitely conclude about this relation between plasma concentrations and efficacy or in terms of rapidity of viral load eradication.

Leflunomide appears to be an alternative treatment in nephropathy due to BK virus in addition to lower immunosuppression regimen. In case of leflunomide use, a major standard seems to be high plasma terflunomide concentrations so as to obtain rapid virus eradication. Concentrations comprised between 15 and 60 µg/mL appear to be pertinent; these concentrations are usually reached with 20-40 mg per day. In patients with insufficient concentrations, further studies should be carried out to determine whether exists a benefit to use higher dosage up to 60 or 80 mg a day. Even if tolerance is quite satisfying, it will

Despite the small number of studies and the weak number of patients in each of them, a correlation seems to exist between plasma terflunomide concentrations and the treatment

Due to its great inter-individual variability and alongside classical virological and clinical follow-up therapeutic drug monitoring appears to be an important step to take into care

probably be the most important parameter in such high-dose treatments.

efficacy. This relation has not yet been proved with tolerance.

patients with BK virus related nephropathy.

Results in terms of biological evolution in patients speak for itself. In a mean monitoring time of 16 months (12-24 months), viral load with leflunomide can be reduced about up to 50%, and even be brought to undetectable, in blood like in urine. But more important is the lowering of renal failure and graft rejection thanks to this treatment. Creatinine clearance (Cockroft-Gault) can be stabilized and even improved (Bazin et al. 2009).

#### **3.6 Therapeutic drug monitoring**

Initial dosage for leflunomide is 20 mg once a day, and can be raised to 30 or 40 mg for patients with viral loads remaining important. Plasma concentrations in therapeutic drug monitoring fluctuate between 15 and 135 µg/mL. These concentrations set out low intraindividual but high inter-individual variability, and moreover without apparent correlation with prescribed dosage. It is interesting to notice that these concentrations were outside usual targets used in most studies - 50 to 100 µg/mL - which are supposed to offer the best efficiency and to limit the hepatotoxicity risk which can be lethal. Besides, the patient with the highest concentration - 135 µg/mL - had its viremia turned undetectable after only a two months treatment and showed no side effect of any kind. This result suggests that higher concentrations lead to higher efficacy and vice versa (Bazin et al. 2009).

#### **3.7 Tolerance**

Concerning tolerance, very few patients suffer from serious side effects. Loss of taste or lethargy can be observed but without any correlation with plasma concentrations. These side effects can prompt the treatment to be stopped, but in most cases viremia tends to increase strongly (Bazin et al. 2009).

#### **4. Discussion**

The main risk factor for BK virus-associated nephropathy is undeniably the immunosuppressive regimen intensity, in particular an intensification due to an acute rejection event (Binet et al. 1999; Nickeleit et al. 2000a; Barri et al. 2001; Nickeleit et al. 2003). Drugs in cause for these events seem to be the combination of tacrolimus, mycophenolate mofetil and monoclonal or polyclonal antibodies (Binet et al. 1999; Nickeleit et al. 1999; Nickeleit et al. 2000a; Nickeleit et al. 2003; Benavides et al. 2007). The organ and the graft type also play a role. For instance, Benavides et al. showed that the incidence of BK virusassociated nephropathy is higher in patients with kidney and pancreas rather than kidney alone; and that an alive donor would had a protective effect, probably explained by a lighter immunosuppressive regimen (Benavides et al. 2007).

Other risk factors have been evoked, like age and sex: nephropathy incidence seem to be greater for aged men (Ramos et al. 2002).

Furthermore many patients improve their symptoms at a distance of the surgery with the lowering of immunosuppression. We already have at our disposal a few experimental studies testing leflunomide on chronic or acute graft rejection (Williams et al. 1994; Xiao et al. 1995; Shen et al. 1998). More recently the inhibitory effects of leflunomide upon HSV, CMV and BK virus have been proved *in vitro* like *in vivo* (Waldman et al. 1999; Waldman et al. 1999; Knight et al. 2001; Farasati et al. 2005). Indeed a study suggests leflunomide is at least as efficient as ganciclovir in CMV infections and does not seem to be affected by resistant viruses (John et al. 2004). Leflunomide has even be successfully used in a patient with bone marrow graft and infected by a resistant virus to ganciclovir, foscarnet and cidofovir (Avery et al. 2004).

The studies where leflunomide is used as an immunosuppressive drug in renal and hepatic graft are more and more, because leflunomide allows to reduce anti-calcineurin drugs which have the major inconvenient of nephrotoxicity, and potentially protects aside from CMV, HSV and BK virus infections (Hardinger et al. 2002; Williams et al. 2002). Moreover leflunomide seems to be an interesting alternative in BK virus-associated nephropathy in renal transplant by eradicating detectable viremia in some patients. Leflunomide also allows avoiding rejection in most cases in spite of classical immunosuppressive drugs dosage reduction. Besides one of leflunomide's main asset is its absence of renal toxicity, contrary to cidofovir (Williams et al. 2005; Josephson et al. 2006; Teschner et al. 2006; Faguer et al. 2007).

Thanks to the encountered success in renal transplant, leflunomide is now used to treat hemorrhagic cystitis linked to BK virus in bone marrow transplant (Dropulic et al. 2008). However, due to the absence of randomized clinical trials with a sufficient number of patients, some authors consider its use in a first-line drug not recommended (Chon et al. 2011).

Leflunomide pharmacokinetics is characterized by a great inter-individual variability with terflunomide concentrations from 15 to 130 µg/mL obtained with the same dosage (Bazin et al. 2009). In BK virus infection, terflunomide concentrations between 15-30 µg/mL and 35- 100 µg/mL are sufficient to suppress respectively 50% and 90% of the replication for CMV and BK virus *in vitro*. That is why a therapeutic margin between 50 and 100 µg/mL has been proposed in this indication (Josephson et al. 2006). However, current strategic therapy so as to limit BK virus incidence tends to manage an early reduction of immunosuppressive regimen to avoid the apparition of a nephropathy. A prospective study with a significant number of patients would be probably necessary to definitely conclude about this relation between plasma concentrations and efficacy or in terms of rapidity of viral load eradication.

#### **5. Conclusion**

44 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

Results in terms of biological evolution in patients speak for itself. In a mean monitoring time of 16 months (12-24 months), viral load with leflunomide can be reduced about up to 50%, and even be brought to undetectable, in blood like in urine. But more important is the lowering of renal failure and graft rejection thanks to this treatment. Creatinine clearance

Initial dosage for leflunomide is 20 mg once a day, and can be raised to 30 or 40 mg for patients with viral loads remaining important. Plasma concentrations in therapeutic drug monitoring fluctuate between 15 and 135 µg/mL. These concentrations set out low intraindividual but high inter-individual variability, and moreover without apparent correlation with prescribed dosage. It is interesting to notice that these concentrations were outside usual targets used in most studies - 50 to 100 µg/mL - which are supposed to offer the best efficiency and to limit the hepatotoxicity risk which can be lethal. Besides, the patient with the highest concentration - 135 µg/mL - had its viremia turned undetectable after only a two months treatment and showed no side effect of any kind. This result suggests that higher

Concerning tolerance, very few patients suffer from serious side effects. Loss of taste or lethargy can be observed but without any correlation with plasma concentrations. These side effects can prompt the treatment to be stopped, but in most cases viremia tends to

The main risk factor for BK virus-associated nephropathy is undeniably the immunosuppressive regimen intensity, in particular an intensification due to an acute rejection event (Binet et al. 1999; Nickeleit et al. 2000a; Barri et al. 2001; Nickeleit et al. 2003). Drugs in cause for these events seem to be the combination of tacrolimus, mycophenolate mofetil and monoclonal or polyclonal antibodies (Binet et al. 1999; Nickeleit et al. 1999; Nickeleit et al. 2000a; Nickeleit et al. 2003; Benavides et al. 2007). The organ and the graft type also play a role. For instance, Benavides et al. showed that the incidence of BK virusassociated nephropathy is higher in patients with kidney and pancreas rather than kidney alone; and that an alive donor would had a protective effect, probably explained by a lighter

Other risk factors have been evoked, like age and sex: nephropathy incidence seem to be

Furthermore many patients improve their symptoms at a distance of the surgery with the lowering of immunosuppression. We already have at our disposal a few experimental studies testing leflunomide on chronic or acute graft rejection (Williams et al. 1994; Xiao et al. 1995; Shen et al. 1998). More recently the inhibitory effects of leflunomide upon HSV, CMV and BK virus have been proved *in vitro* like *in vivo* (Waldman et al. 1999; Waldman et al. 1999; Knight et al. 2001; Farasati et al. 2005). Indeed a study suggests leflunomide is at

(Cockroft-Gault) can be stabilized and even improved (Bazin et al. 2009).

concentrations lead to higher efficacy and vice versa (Bazin et al. 2009).

**3.6 Therapeutic drug monitoring** 

increase strongly (Bazin et al. 2009).

immunosuppressive regimen (Benavides et al. 2007).

greater for aged men (Ramos et al. 2002).

**3.7 Tolerance** 

**4. Discussion** 

Leflunomide appears to be an alternative treatment in nephropathy due to BK virus in addition to lower immunosuppression regimen. In case of leflunomide use, a major standard seems to be high plasma terflunomide concentrations so as to obtain rapid virus eradication. Concentrations comprised between 15 and 60 µg/mL appear to be pertinent; these concentrations are usually reached with 20-40 mg per day. In patients with insufficient concentrations, further studies should be carried out to determine whether exists a benefit to use higher dosage up to 60 or 80 mg a day. Even if tolerance is quite satisfying, it will probably be the most important parameter in such high-dose treatments.

Despite the small number of studies and the weak number of patients in each of them, a correlation seems to exist between plasma terflunomide concentrations and the treatment efficacy. This relation has not yet been proved with tolerance.

Due to its great inter-individual variability and alongside classical virological and clinical follow-up therapeutic drug monitoring appears to be an important step to take into care patients with BK virus related nephropathy.

Leflunomide an Immunosuppressive Drug for Antiviral Purpose

transplantation.]." *Pathologe*.

*Clin Immunol* 7(3): 273-81.

*Ther* 16(3): 309-17.

79(1): 116-8.

*Immunol* 346: 31-56.

*Transplant* 10(2): 407-15.

*Am J Transplant* 2(9): 867-71.

in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation 47

Brocker, V., Schwarz, A. and Becker, J. U. (2011). "[BK virus nephropathy after kidney

Chen, C. H., Wen, M. C., Wang, M., Lian, J. D., Wu, M. J., Cheng, C. H., Shu, K. H. and

De Luca, A., Giancola, M. L., Ammassari, A., Grisetti, S., Cingolani, A., Paglia, M. G.,

Dheir, H., Sahin, S., Uyar, M., Gurkan, A., Turunc, V., Kacar, S., Bayirli Turan, D. and

Dropulic, L. K. and Jones, R. J. (2008). "Polyomavirus BK infection in blood and marrow

Dunn, M. C., Knight, D. A. and Waldman, W. J. (2011). "Inhibition of respiratory syncytial

Faguer, S., Hirsch, H. H., Kamar, N., Guilbeau-Frugier, C., Ribes, D., Guitard, J., Esposito, L.,

Farasati, N. A., Shapiro, R., Vats, A. and Randhawa, P. (2005). "Effect of leflunomide and

Fayard, E., Xue, G., Parcellier, A., Bozulic, L. and Hemmings, B. A. (2010). "Protein kinase B

Fox, R. I., Herrmann, M. L., Frangou, C. G., Wahl, G. M., Morris, R. E., Strand, V. and

Gupta, A. and Gupta, P. (2011). "BK virus associated nephropathy in renal transplantation:

Hardinger, K. L., Koch, M. J., Bohl, D. J., Storch, G. A. and Brennan, D. C. (2010). "BK-virus

Hardinger, K. L., Wang, C. D., Schnitzler, M. A., Miller, B. W., Jendrisak, M. D., Shenoy, S.,

preferable approach for graft surveillance." *Transplant Proc* 43(3): 867-70. Drachenberg, C. B., Beskow, C. O., Cangro, C. B., Bourquin, P. M., Simsir, A., Fink, J., Weir,

progressive multifocal leukoencephalopathy." *Aids* 14(14): F117-21.

with urine cytology." *Hum Pathol* 30(8): 970-7.

transplantation." *Transpl Int* 20(11): 962-9.

arthritis." *Clin Immunol* 93(3): 198-208.

Gilead (2010). "Vistide®." *Résumé des Caractéristiques du Produit*.

where do we stand." *Minerva Urol Nefrol* 63(2): 155-67.

transplant recipients." *Bone Marrow Transplant* 41(1): 11-8.

Chang, D. (2001). "A regulatory region rearranged BK virus is associated with tubulointerstitial nephritis in a rejected renal allograft." *J Med Virol* 64(1): 82-8. Chon, W. J. and Josephson, M. A. (2011). "Leflunomide in renal transplantation." *Expert Rev* 

Govoni, A., Murri, R., Testa, L., Monforte, A. D. and Antinori, A. (2000). "Cidofovir added to HAART improves virological and clinical outcome in AIDS-associated

Basdemir, G. (2011). "Intensive polyoma virus nephropathy treatment as a

M. R., Klassen, D. K., Bartlett, S. T. and Papadimitriou, J. C. (1999). "Human polyoma virus in renal allograft biopsies: morphological findings and correlation

virus in vitro and in vivo by the immunosuppressive agent leflunomide." *Antivir* 

Cointault, O., Modesto, A., Lavit, M., Mengelle, C. and Rostaing, L. (2007). "Leflunomide treatment for polyomavirus BK-associated nephropathy after kidney

cidofovir on replication of BK virus in an in vitro culture system." *Transplantation*

(PKB/Akt), a key mediator of the PI3K signaling pathway." *Curr Top Microbiol* 

Kirschbaum, B. J. (1999). "Mechanism of action for leflunomide in rheumatoid

and the impact of pre-emptive immunosuppression reduction: 5-year results." *Am J* 

Lowell, J. A. and Brennan, D. C. (2002). "Prospective, pilot, open-label, short-term study of conversion to leflunomide reverses chronic renal allograft dysfunction."

#### **6. Acknowledgment**

The author would like to acknowledge the staff of Henri-Mondor and Bicetre University Hospital Pharmacology laboratories and especially Anne Hulin, PharmD, PhD, Valerie Furlan, PharmD, PhD and Caroline Barau, PharmD.

#### **7. References**


The author would like to acknowledge the staff of Henri-Mondor and Bicetre University Hospital Pharmacology laboratories and especially Anne Hulin, PharmD, PhD, Valerie

Atencio, I. A., Shadan, F. F., Zhou, X. J., Vaziri, N. D. and Villarreal, L. P. (1993). "Adult mouse

Baksh, F. K., Finkelstein, S. D., Swalsky, P. A., Stoner, G. L., Ryschkewitsch, C. F. and

Barri, Y. M., Ahmad, I., Ketel, B. L., Barone, G. W., Walker, P. D., Bonsib, S. M. and Abul-

Bazin, C., Barau, C., Grimbert, P., François, H., Blanchet, B., Astier, A., Furlan, V. and Hulin,

Beaman, J. M., Hackett, L. P., Luxton, G. and Illett, K. F. (2002). "Effect of hemodialysis on

Benavides, C. A., Pollard, V. B., Mauiyyedi, S., Podder, H., Knight, R. and Kahan, B. D.

Binet, I., Nickeleit, V., Hirsch, H. H., Prince, O., Dalquen, P., Gudat, F., Mihatsch, M. J. and

cause of renal graft dysfunction and graft loss." *Transplantation* 67(6): 918-22. Bohanec Grabar, P., Rozman, B., Tomsic, M., Suput, D., Logar, D. and Dolzan, V. (2008).

Boothpur, R. and Brennan, D. C. (2010). "Human polyoma viruses and disease with

Brennan, D. C., Agha, I., Bohl, D. L., Schnitzler, M. A., Hardinger, K. L., Lockwood, M.,

preemptive immunosuppression reduction." *Am J Transplant* 5(3): 582-94.

leflunomide plasma concentrations." *Ann Pharmacother* 36(1): 75-7.

Bhaskar, P. T. and Hay, N. (2007). "The two TORCs and Akt." *Dev Cell* 12(4): 487-502.

rheumatoid arthritis patients." *Eur J Clin Pharmacol* 64(9): 871-6.

emphasis on clinical BK and JC." *J Clin Virol* 47(4): 306-12.

immunosuppressive therapy." *Clin Transplant* 15(4): 240-6.

kidneys become permissive to acute polyomavirus infection and reactivate persistent infections in response to cellular damage and regeneration." *J Virol* 67(3): 1424-32. Avery, R. K., Bolwell, B. J., Yen-Lieberman, B., Lurain, N., Waldman, W. J., Longworth, D.

L., Taege, A. J., Mossad, S. B., Kohn, D., Long, J. R., Curtis, J., Kalaycio, M., Pohlman, B. and Williams, J. W. (2004). "Use of leflunomide in an allogeneic bone marrow transplant recipient with refractory cytomegalovirus infection." *Bone* 

Randhawa, P. (2001). "Molecular genotyping of BK and JC viruses in human polyomavirus-associated interstitial nephritis after renal transplantation." *Am J* 

Ezz, S. R. (2001). "Polyoma viral infection in renal transplantation: the role of

A. (2009). "Etude pilote du suivi thérapeutique du Léflunomide en transplantation

(2007). "BK virus-associated nephropathy in sirolimus-treated renal transplant patients: incidence, course, and clinical outcomes." *Transplantation* 84(1): 83-8. Bernhoff, E., Tylden, G. D., Kjerpeseth, L. J., Gutteberg, T. J., Hirsch, H. H. and Rinaldo, C.

H. (2010). "Leflunomide inhibition of BK virus replication in renal tubular epithelial

Thiel, G. (1999). "Polyomavirus disease under new immunosuppressive drugs: a

"Genetic polymorphism of CYP1A2 and the toxicity of leflunomide treatment in

Torrence, S., Schuessler, R., Roby, T., Gaudreault-Keener, M. and Storch, G. A. (2005). "Incidence of BK with tacrolimus versus cyclosporine and impact of

**6. Acknowledgment** 

**7. References** 

Furlan, PharmD, PhD and Caroline Barau, PharmD.

*Marrow Transplant* 34(12): 1071-5.

rénale." *J Pharm Clin* 28(1): 21-6.

cells." *J Virol* 84(4): 2150-6.

*Kidney Dis* 38(2): 354-65.


Leflunomide an Immunosuppressive Drug for Antiviral Purpose

transplantation." *Adv Exp Med Biol* 577: 185-9.

transplant patients." *J Am Soc Nephrol* 13(8): 2145-51.

Sanofi-Aventis (2009). "Arava®." *Résumé des Caractéristiques du Produit*.

allogeneic heart transplantation." *Transplantation* 66(6): 692-8.

proliferating cells." *Transplantation* 47(5): 811-6.

*Hypertens* 12(6): 599-605.

*Clin Microbiol* 42(3): 1176-80.

*Engl J Med* 342(18): 1361-3.

230-2.

in Treatment for BK Virus-Associated Nephropathy After Kidney Transplantation 49

Nickeleit, V., Hirsch, H. H., Zeiler, M., Gudat, F., Prince, O., Thiel, G. and Mihatsch, M. J.

Nickeleit, V., Zeiler, M., Gudat, F., Thiel, G. and Mihatsch, M. J. (1998). "Histological

Pallet, N., Burgard, M., Quamouss, O., Rabant, M., Bererhi, L., Martinez, F., Thervet, E.,

Ramos, E., Drachenberg, C. B., Papadimitriou, J. C., Hamze, O., Fink, J. C., Klassen, D. K.,

Randhawa, P., Ho, A., Shapiro, R., Vats, A., Swalsky, P., Finkelstein, S., Uhrmacher, J. and

Randhawa, P. S. and Demetris, A. J. (2000). "Nephropathy due to polyomavirus type BK." *N* 

Randhawa, P. S., Khaleel-Ur-Rehman, K., Swalsky, P. A., Vats, A., Scantlebury, V., Shapiro,

in patients with BK virus interstitial nephritis." *Transplantation* 73(7): 1090-4. Rosenberg, A. S. and Singer, A. (1992). "Cellular basis of skin allograft rejection: an in vivo model of immune-mediated tissue destruction." *Annu Rev Immunol* 10: 333-58. Rozman, B. (2002). "Clinical pharmacokinetics of leflunomide." *Clin Pharmacokinet* 41(6): 421-30. Saad, E. R., Bresnahan, B. A., Cohen, E. P., Lu, N., Orentas, R. J., Vasudev, B. and Hariharan,

immunosuppression without antiviral therapy." *Transplantation* 85(6): 850-4.

Seron, D., Alexopoulos, E., Raftery, M. J., Hartley, R. B. and Cameron, J. S. (1989). "Diagnosis

Shen, J., Chong, A. S., Xiao, F., Liu, W., Huang, W., Blinder, L., Foster, P., Sankary, H.,

Smith, R. D., Galla, J. H., Skahan, K., Anderson, P., Linnemann, C. C., Jr., Ault, G. S.,

(2000). "BK-virus nephropathy in renal transplants-tubular necrosis, MHC-class II expression and rejection in a puzzling game." *Nephrol Dial Transplant* 15(3): 324-32. Nickeleit, V., Klimkait, T., Binet, I. F., Dalquen, P., Del Zenero, V., Thiel, G., Mihatsch, M. J. and

Hirsch, H. H. (2000). "Testing for polyomavirus type BK DNA in plasma to identify renal-allograft recipients with viral nephropathy." *N Engl J Med* 342(18): 1309-15. Nickeleit, V., Singh, H. K. and Mihatsch, M. J. (2003). "Polyomavirus nephropathy:

morphology, pathophysiology, and clinical management." *Curr Opin Nephrol* 

characteristics of interstitial renal allograft rejection." *Kidney Blood Press Res* 21(2-4):

Anglicheau, D., Noel, L. H., Rouzioux, C. and Legendre, C. (2010). "Cidofovir may be deleterious in BK virus-associated nephropathy." *Transplantation* 89(12): 1542-4. Pavlakis, M., Haririan, A. and Klassen, D. K. (2006). "BK virus infection after non-renal

Drachenberg, R. C., Wiland, A., Wali, R., Cangro, C. B., Schweitzer, E., Bartlett, S. T. and Weir, M. R. (2002). "Clinical course of polyoma virus nephropathy in 67 renal

Weck, K. (2004). "Correlates of quantitative measurement of BK polyomavirus (BKV) DNA with clinical course of BKV infection in renal transplant patients." *J* 

R. and Finkelstein, S. (2002). "DNA sequencing of viral capsid protein VP-1 region

S. (2008). "Successful treatment of BK viremia using reduction in

of rejection in renal allograft biopsies using the presence of activated and

Jensik, S., McChesney, L., Mital, D. and Williams, J. W. (1998). "Histological characterization and pharmacological control of chronic rejection in xenogeneic and

Ryschkewitsch, C. F. and Stoner, G. L. (1998). "Tubulointerstitial nephritis due to a


Hirsch, H. H., Knowles, W., Dickenmann, M., Passweg, J., Klimkait, T., Mihatsch, M. J. and

Johnston, O., Jaswal, D., Gill, J. S., Doucette, S., Fergusson, D. A. and Knoll, G. A. (2010).

Josephson, M. A., Gillen, D., Javaid, B., Kadambi, P., Meehan, S., Foster, P., Harland, R.,

Kadambi, P. V., Josephson, M. A., Williams, J., Corey, L., Jerome, K. R., Meehan, S. M. and

Kalgutkar, A. S., Nguyen, H. T., Vaz, A. D., Doan, A., Dalvie, D. K., McLeod, D. G. and

Knight, D. A., Hejmanowski, A. Q., Dierksheide, J. E., Williams, J. W., Chong, A. S. and

Maddison, P., Kiely, P., Kirkham, B., Lawson, T., Moots, R., Proudfoot, D., Reece, R., Scott,

Mengel, M., Marwedel, M., Radermacher, J., Eden, G., Schwarz, A., Haller, H. and Kreipe,

modern immunosuppressive drugs." *Nephrol Dial Transplant* 18(6): 1190-6. Mylonakis, E., Goes, N., Rubin, R. H., Cosimi, A. B., Colvin, R. B. and Fishman, J. A. (2001).

Nickeleit, V., Hirsch, H. H., Binet, I. F., Gudat, F., Prince, O., Dalquen, P., Thiel, G. and

latent infection to manifest disease." *J Am Soc Nephrol* 10(5): 1080-9.

therapy for BK virus infection." *Transplantation* 90(12): 1450-7.

nephropathy in renal-transplant recipients." *N Engl J Med* 347(7): 488-96. Hirsch, H. H., Mohaupt, M. and Klimkait, T. (2001). "Prospective monitoring of BK virus

virus nephropathy." *J Infect Dis* 184(11): 1494-5; author reply 1495-6. John, G. T., Manivannan, J., Chandy, S., Peter, S. and Jacob, C. K. (2004). "Leflunomide

77(9): 1460-1.

review." *Transplantation* 89(9): 1057-70.

with leflunomide." *Transplantation* 81(5): 704-10.

of aldoximes." *Drug Metab Dispos* 31(10): 1240-50.

*Rheumatology (Oxford)* 44(3): 280-6.

*Transplantation* 72(10): 1587-92.

with cidofovir." *Am J Transplant* 3(2): 186-91.

Steiger, J. (2002). "Prospective study of polyomavirus type BK replication and

load after discontinuing sirolimus treatment in a renal transplant patient with BK

therapy for cytomegalovirus disease in renal allograft recepients." *Transplantation*

"Treatment of polyomavirus infection in kidney transplant recipients: a systematic

Thistlethwaite, R. J., Garfinkel, M., Atwood, W., Jordan, J., Sadhu, M., Millis, M. J. and Williams, J. (2006). "Treatment of renal allograft polyoma BK virus infection

Limaye, A. P. (2003). "Treatment of refractory BK virus-associated nephropathy

Murray, J. C. (2003). "In vitro metabolism studies on the isoxazole ring scission in the anti-inflammatory agent lefluonomide to its active alpha-cyanoenol metabolite A771726: mechanistic similarities with the cytochrome P450-catalyzed dehydration

Waldman, W. J. (2001). "Inhibition of herpes simplex virus type 1 by the experimental immunosuppressive agent leflunomide." *Transplantation* 71(1): 170-4. Kuypers, D. R., Vandooren, A. K., Lerut, E., Evenepoel, P., Claes, K., Snoeck, R., Naesens, L. and

Vanrenterghem, Y. (2005). "Adjuvant low-dose cidofovir therapy for BK polyomavirus interstitial nephritis in renal transplant recipients." *Am J Transplant* 5(8): 1997-2004. Liacini, A., Seamone, M. E., Muruve, D. A. and Tibbles, L. A. (2010). "Anti-BK virus

mechanisms of sirolimus and leflunomide alone and in combination: toward a new

D., Sword, R., Taggart, A., Thwaites, C. and Williams, E. (2005). "Leflunomide in rheumatoid arthritis: recommendations through a process of consensus."

H. (2003). "Incidence of polyomavirus-nephropathy in renal allografts: influence of

"BK virus in solid organ transplant recipients: an emerging syndrome."

Mihatsch, M. J. (1999). "Polyomavirus infection of renal allograft recipients: from


**4** 

*USA* 

Yangxin Huang

*University of South Florida, Tampa, FL,* 

**Modeling Virologic Response in HIV-1 Infected** 

Although the advent of highly active antiretroviral therapy (HAART), including potent protease inhibitors (PIs), has profoundly reduced human immunodeficiency virus (HIV) mortality and morbidity (Palella et al., 1998; CDC, 2009)**,** these combination regimens are not a cure for HIV infection and therapy may be life long. While many patients benefit from HAART treatment, others do not benefit or only experience a temporary benefit. There are several reasons why treatment fails, with poor patient adherence to HAART a leading contributing factor (Ickovics & Meisler, 1997; Paterson, 2000). Thus, assessment of medication adherence within AIDS clinical trials is a critical component of the successful evaluation of therapy outcomes. Maintaining adherence may be particularly difficult when the drug regimen is complex or side-effects are common, as is often the case for current HIV therapy especially in highly treatment experienced patients (Ickovics & Meisler, 1997).

The measurement of adherence remains problematic; a standard definition of optimal adherence and completely reliable measures of adherence are lacking. Nevertheless, there has been substantial progress in both of these areas in the past few years. First, it appears that higher levels of adherence are needed for HIV disease than other diseases to achieve the desired therapeutic benefit. Using questionnaires (patient self-reporting and/or face-to-face interview) and electronic compliance monitoring caps (Medication Event Monitoring System [MEMS]), viral suppression is common with at least 54%–100% mean adherence level to antiviral regimens (Bangsberg, 2006). Second, a better appreciation of the value and limitations of different adherence measurements has been addressed (Berg & Arnsten, 2006; Bova et al., 2005). In AIDS clinical trials, adherence to a medication regimen is currently measured by two major methods: by use of questionnaires and by use of MEMS. The MEMS is considered an objective adherence measure. It consists of a microprocessor in the cap of a medication bottle which records the date and time of bottle opening. The results are downloaded to a computer for analysis. Results demonstrate that medication-taking patterns are highly variable among patients (Kastrissios et al., 1998) and that they often give a more precise measure of adherence than self-report (Arnsten et al., 2001a). However, MEMS data are also subject to error and are not widely available in the clinical setting. Adherence assessment by self-report is usually evaluated by a patient's ability to recall their medication dosing during a specific time interval. Often self-reported measures tend to overestimate HIV medication adherence compared to other methods (Arnsten et al., 2001b,

**1. Introduction** 

**Patients to Assess Medication Adherence** 

*Department of Epidemiology and Biostatistics, College of Public Health,* 

mutant polyomavirus BK virus strain, BKV(Cin), causing end-stage renal disease." *J Clin Microbiol* 36(6): 1660-5.


## **Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence**

#### Yangxin Huang

*Department of Epidemiology and Biostatistics, College of Public Health, University of South Florida, Tampa, FL, USA* 

#### **1. Introduction**

50 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

Talmon, G., Cornell, L. D. and Lager, D. J. (2010). "Mitochondrial changes in cidofovir

Teschner, S., Geyer, M., Wilpert, J., Schwertfeger, E., Schenk, T., Walz, G. and Donauer, J.

van Roon, E. N., Jansen, T. L., van de Laar, M. A., Janssen, M., Yska, J. P., Keuper, R.,

Waldman, W. J., Knight, D. A., Blinder, L., Shen, J., Lurain, N. S., Miller, D. M., Sedmak, D.

Waldman, W. J., Knight, D. A., Lurain, N. S., Miller, D. M., Sedmak, D. D., Williams, J. W.

Weiss, A. S., Gralla, J., Chan, L., Klem, P. and Wiseman, A. C. (2008). "Aggressive

Williams, J. W., Javaid, B., Kadambi, P. V., Gillen, D., Harland, R., Thistlewaite, J. R.,

Williams, J. W., Mital, D., Chong, A., Kottayil, A., Millis, M., Longstreth, J., Huang, W.,

Williams, J. W., Xiao, F., Foster, P., Clardy, C., McChesney, L., Sankary, H. and Chong, A. S.

Williamson, R. A., Yea, C. M., Robson, P. A., Curnock, A. P., Gadher, S., Hambleton, A. B.,

Xiao, F., Chong, A., Shen, J., Yang, J., Short, J., Foster, P., Sankary, H., Jensik, S., Mital, D.,

immunomodulatory compound." *J Biol Chem* 270(38): 22467-72.

transplant rejection." *Transplantation* 60(10): 1065-72.

therapy for BK virus nephropathy." *Transplant Proc* 42(5): 1713-5.

dose leflunomide." *Nephrol Dial Transplant* 21(7): 2039-40.

*Clin Microbiol* 36(6): 1660-5.

*Intervirology* 42(5-6): 412-8.

*Soc Nephrol* 3(6): 1812-9.

transplantation." *Transplantation* 73(3): 358-66.

cyclosporine." *Transplantation* 57(8): 1223-31.

352(11): 1157-8.

mutant polyomavirus BK virus strain, BKV(Cin), causing end-stage renal disease." *J* 

(2006). "Remission of polyomavirus-induced graft nephropathy treated with low-

Houtman, P. M. and Brouwers, J. R. (2005). "Therapeutic drug monitoring of A77 1726, the active metabolite of leflunomide: serum concentrations predict response to treatment in patients with rheumatoid arthritis." *Ann Rheum Dis* 64(4): 569-74. von Willebrand, E., Pettersson, E., Ahonen, J. and Hayry, P. (1986). "CMV infection, class II

antigen expression, and human kidney allograft rejection." *Transplantation* 42(4): 364-7.

D., Williams, J. W. and Chong, A. S. (1999). "Inhibition of cytomegalovirus in vitro and in vivo by the experimental immunosuppressive agent leflunomide."

and Chong, A. S. (1999). "Novel mechanism of inhibition of cytomegalovirus by the experimental immunosuppressive agent leflunomide." *Transplantation* 68(6): 814-25.

immunosuppression minimization reduces graft loss following diagnosis of BK virus-associated nephropathy: a comparison of two reduction strategies." *Clin J Am* 

Garfinkel, M., Foster, P., Atwood, W., Millis, J. M., Meehan, S. M. and Josephson, M. A. (2005). "Leflunomide for polyomavirus type BK nephropathy." *N Engl J Med*

Brady, L. and Jensik, S. (2002). "Experiences with leflunomide in solid organ

(1994). "Leflunomide in experimental transplantation. Control of rejection and alloantibody production, reversal of acute rejection, and interaction with

Woodward, K., Bruneau, J. M., Hambleton, P., Moss, D., Thomson, T. A., Spinella-Jaegle, S., Morand, P., Courtin, O., Sautes, C., Westwood, R., Hercend, T., Kuo, E. A. and Ruuth, E. (1995). "Dihydroorotate dehydrogenase is a high affinity binding protein for A77 1726 and mediator of a range of biological effects of the

McChesney, L. and et al. (1995). "Pharmacologically induced regression of chronic

Although the advent of highly active antiretroviral therapy (HAART), including potent protease inhibitors (PIs), has profoundly reduced human immunodeficiency virus (HIV) mortality and morbidity (Palella et al., 1998; CDC, 2009)**,** these combination regimens are not a cure for HIV infection and therapy may be life long. While many patients benefit from HAART treatment, others do not benefit or only experience a temporary benefit. There are several reasons why treatment fails, with poor patient adherence to HAART a leading contributing factor (Ickovics & Meisler, 1997; Paterson, 2000). Thus, assessment of medication adherence within AIDS clinical trials is a critical component of the successful evaluation of therapy outcomes. Maintaining adherence may be particularly difficult when the drug regimen is complex or side-effects are common, as is often the case for current HIV therapy especially in highly treatment experienced patients (Ickovics & Meisler, 1997).

The measurement of adherence remains problematic; a standard definition of optimal adherence and completely reliable measures of adherence are lacking. Nevertheless, there has been substantial progress in both of these areas in the past few years. First, it appears that higher levels of adherence are needed for HIV disease than other diseases to achieve the desired therapeutic benefit. Using questionnaires (patient self-reporting and/or face-to-face interview) and electronic compliance monitoring caps (Medication Event Monitoring System [MEMS]), viral suppression is common with at least 54%–100% mean adherence level to antiviral regimens (Bangsberg, 2006). Second, a better appreciation of the value and limitations of different adherence measurements has been addressed (Berg & Arnsten, 2006; Bova et al., 2005). In AIDS clinical trials, adherence to a medication regimen is currently measured by two major methods: by use of questionnaires and by use of MEMS. The MEMS is considered an objective adherence measure. It consists of a microprocessor in the cap of a medication bottle which records the date and time of bottle opening. The results are downloaded to a computer for analysis. Results demonstrate that medication-taking patterns are highly variable among patients (Kastrissios et al., 1998) and that they often give a more precise measure of adherence than self-report (Arnsten et al., 2001a). However, MEMS data are also subject to error and are not widely available in the clinical setting. Adherence assessment by self-report is usually evaluated by a patient's ability to recall their medication dosing during a specific time interval. Often self-reported measures tend to overestimate HIV medication adherence compared to other methods (Arnsten et al., 2001b,

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 53

study of amprenavir (APV) as part of several dual protease inhibitor (PI) regimens in subjects with HIV infection in whom initial PI therapy had failed. One of objectives of the ACTG 398 study was to evaluate the genotypic and phenotypic resistance profiles that emerge on treatment and their relationship to the plasma HIV-1 RNA and CD4 cell count responses, and to determine the relationship between drug exposure measured from combined PK and adherence data to the degree and duration of viral response. Subjects in all arms received APV (1200 mg twice a day [q12h]), efavirenz (EFV, 600 mg once a day [qd]), abacavir (300 mg q12h) and adefovir dipivoxil (60 mg qd). A total of 481 subjects were randomized to four treatment arms and received a second PI or placebo: Arm A (*n*=116) saquinavir (1600 mg q12h); Arm B (*n*=69) indinavir (1200 mg q12h); Arm C (*n*=139) nelfinavir (NFV, 1250 mg q12h); and Arm D (*n*=157) received a placebo matched for one of these three PIs. Assignment of subjects to treatment arms depended on past PI exposure in the arm. Subjects were scheduled for follow-up visits at study (day 0); at weeks 2, 4, 8, 16 and every 8 weeks thereafter until week 72; and at the time of confirmed virologic failure. More detailed descriptions of this study and study results are given by Hammer et al. (2002) and Pfister et al. (2003). Because phenotype sensitivity testing was performed only on a subset of randomly selected subjects, the number of subjects available for our analysis was greatly reduced. We chose to consider only the subjects within Arm C for our analysis because this arm afforded the greatest number of subjects (*n*=31) with phenotypic drug susceptibility data on the two PIs (APV and NFV) and had available adherence data, as required for our model. Among these 31 subjects, 13 had phenotypic drug susceptibility

**RNA viral load:** RNA viral load was measured in copies/mL at study weeks 0, 2, 4, 8 and every 8 weeks thereafter until week 72 by the ultrasensitive reverse transcriptasepolymerase chain reaction HIV-1 RNA assay. Only measurements taken while on protocoldefined treatment were used in the analysis. All viral load values were log (base 10) transformed. Although, the lower limit of assay quantification was 200 copies/mL, when lower values (<200 copies/mL) were detected, these values were used in the analysis. The exact day of viral load measurement (not predefined study week) was used to compute

**Medication adherence:** Medication adherence was measured by two methods-- by the use of questionnaires and by the use of electronic monitoring caps. Subjects completed an adherence questionnaire (AACTG study 398 questionnaire QL0702) at study weeks 4, 8, 12, 16, and every 8 weeks thereafter. The questionnaire was completed by the study participant and/or by a face-to-face interview with study personnel. The subject was asked to specify the number of prescribed doses of each drug that he or she had failed to take on each of the preceding 4 days. Questionnaire adherence rates for APV and NFV were determined at each visit as the number of prescribed doses taken divided by the number prescribed doses during the preceding 4 day interval. For electronically monitored adherence, an MEMS cap (Medication Event Monitoring Systems, Aprex Corp., Menlo CA) was used to monitor APV and EFV compliance only. Subjects were asked to bring their medication bottles and caps to the clinic at each study visit (weeks 2, 4, 8, 12, 16, and every 8 weeks thereafter), where cap data were downloaded to computer files and stored for later analysis. Since APV was

data at the time of protocol-defined virologic failure.

**2.2 Observed measurements** 

study day in our analysis.

Bangsberg et al. 2000; Levine et al., 2006; Liu et al., 2001). Finally, it is important to note that the measurement of viral load levels is of special utility as an indirect measure of adherence in HIV therapeutics. It has been argued that this is not a good adherence measure because other factors may influence viral load (pharmacokinetics, drug resistance etc.). However, there is a tight correlation between viral load and adherence (Haubrich et al., 1999; Paterson et al., 2000), but results vary by adherence method and summary adherence statistic (Vrijens & Goetghebeur, 1997; Vrijens et al., 2005) . Several recent papers explore the methodological and operational issues when evaluating electronic drug monitoring adherence on viral load (Arnsten et al., 2001b; Fennie et al., 2006; Fletcher et al., 2005; Llabre et al., 2006; Liu et al. , 2006; Liu et al. , 2007; Pearson et al. , 2007; Vrijens et al., 2005). Most importantly, a favorable change in viral load is the desired therapeutic outcome of adherence to HAART.

In this paper, we propose using a viral dynamic model with consideration of long-term medication adherence and drug susceptibility to explore the relationship between adherence to two protease inhibitors, as part of an HAART regimen, and long-term virologic response. In particular, we will use different adherence measures from an AIDS clinical trial study-- ACTG398 (Hammer et al., 2002) and compare their performance for predicting virologic response. The dynamic modeling approach (Huang et al., 2006; 2010) allows us to appropriately capture the sophisticated nonlinear relationships and interactions among important factors and virologic response. The complete HIV-1 RNA (viral load) trajectories serve as the virologic response index, which is more informative and sensitive to clinical and drug factors. Thus, this method is more powerful to detect the effect of a clinical or drug factor on the response. Using a Bayesian method (Huang et al., 2006), we fit a long-term viral dynamic model to data from the AIDS clinical trial study to explore the association between adherence and viral load in HIV-infected patients with adjustment of the potential confounding factor, drug susceptibility. In this study, we employed the proposed mechanism-based dynamic model to assess how to efficiently use the adherence data based on questionnaires and the MEMS to predict virologic response. In particular, we intend to address the questions (i) how to summarize the MEMS adherence data for efficient prediction of the virologic response, and (ii) which adherence assessment method, questionnaire or MEMS, is more efficient in predicting the virologic response after accounting for the potential confounding factor, drug resistance. We expect that viral dynamic modeling not only provides a powerful tool to evaluate the effect of adherence on long-term virologic responses, but also can be used to predict antiviral responses for various scenarios that may help with understanding the role of different adherence measure statistics in antiviral activities and assist clinicians in treatment decisions.

#### **2. Materials**

In this section, we describe the subject population to be studied and observed data to be used in this research. These measurements include RNA viral load, phenotypic drug sensitivity and medication adherence. We also discuss how to evaluate assessment interval lengths and time frames (delay effect of timing) for the MEMS adherence data.

#### **2.1 Subject population**

The subject sample in our analysis was drawn from the AIDS Clinical Trials Group (ACTG) 398 study (Hammer et al., 2002), a randomized, double-blind, placebo-controlled phase II study of amprenavir (APV) as part of several dual protease inhibitor (PI) regimens in subjects with HIV infection in whom initial PI therapy had failed. One of objectives of the ACTG 398 study was to evaluate the genotypic and phenotypic resistance profiles that emerge on treatment and their relationship to the plasma HIV-1 RNA and CD4 cell count responses, and to determine the relationship between drug exposure measured from combined PK and adherence data to the degree and duration of viral response. Subjects in all arms received APV (1200 mg twice a day [q12h]), efavirenz (EFV, 600 mg once a day [qd]), abacavir (300 mg q12h) and adefovir dipivoxil (60 mg qd). A total of 481 subjects were randomized to four treatment arms and received a second PI or placebo: Arm A (*n*=116) saquinavir (1600 mg q12h); Arm B (*n*=69) indinavir (1200 mg q12h); Arm C (*n*=139) nelfinavir (NFV, 1250 mg q12h); and Arm D (*n*=157) received a placebo matched for one of these three PIs. Assignment of subjects to treatment arms depended on past PI exposure in the arm. Subjects were scheduled for follow-up visits at study (day 0); at weeks 2, 4, 8, 16 and every 8 weeks thereafter until week 72; and at the time of confirmed virologic failure. More detailed descriptions of this study and study results are given by Hammer et al. (2002) and Pfister et al. (2003). Because phenotype sensitivity testing was performed only on a subset of randomly selected subjects, the number of subjects available for our analysis was greatly reduced. We chose to consider only the subjects within Arm C for our analysis because this arm afforded the greatest number of subjects (*n*=31) with phenotypic drug susceptibility data on the two PIs (APV and NFV) and had available adherence data, as required for our model. Among these 31 subjects, 13 had phenotypic drug susceptibility data at the time of protocol-defined virologic failure.

#### **2.2 Observed measurements**

52 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

Bangsberg et al. 2000; Levine et al., 2006; Liu et al., 2001). Finally, it is important to note that the measurement of viral load levels is of special utility as an indirect measure of adherence in HIV therapeutics. It has been argued that this is not a good adherence measure because other factors may influence viral load (pharmacokinetics, drug resistance etc.). However, there is a tight correlation between viral load and adherence (Haubrich et al., 1999; Paterson et al., 2000), but results vary by adherence method and summary adherence statistic (Vrijens & Goetghebeur, 1997; Vrijens et al., 2005) . Several recent papers explore the methodological and operational issues when evaluating electronic drug monitoring adherence on viral load (Arnsten et al., 2001b; Fennie et al., 2006; Fletcher et al., 2005; Llabre et al., 2006; Liu et al. , 2006; Liu et al. , 2007; Pearson et al. , 2007; Vrijens et al., 2005). Most importantly, a favorable

In this paper, we propose using a viral dynamic model with consideration of long-term medication adherence and drug susceptibility to explore the relationship between adherence to two protease inhibitors, as part of an HAART regimen, and long-term virologic response. In particular, we will use different adherence measures from an AIDS clinical trial study-- ACTG398 (Hammer et al., 2002) and compare their performance for predicting virologic response. The dynamic modeling approach (Huang et al., 2006; 2010) allows us to appropriately capture the sophisticated nonlinear relationships and interactions among important factors and virologic response. The complete HIV-1 RNA (viral load) trajectories serve as the virologic response index, which is more informative and sensitive to clinical and drug factors. Thus, this method is more powerful to detect the effect of a clinical or drug factor on the response. Using a Bayesian method (Huang et al., 2006), we fit a long-term viral dynamic model to data from the AIDS clinical trial study to explore the association between adherence and viral load in HIV-infected patients with adjustment of the potential confounding factor, drug susceptibility. In this study, we employed the proposed mechanism-based dynamic model to assess how to efficiently use the adherence data based on questionnaires and the MEMS to predict virologic response. In particular, we intend to address the questions (i) how to summarize the MEMS adherence data for efficient prediction of the virologic response, and (ii) which adherence assessment method, questionnaire or MEMS, is more efficient in predicting the virologic response after accounting for the potential confounding factor, drug resistance. We expect that viral dynamic modeling not only provides a powerful tool to evaluate the effect of adherence on long-term virologic responses, but also can be used to predict antiviral responses for various scenarios that may help with understanding the role of different adherence measure

change in viral load is the desired therapeutic outcome of adherence to HAART.

statistics in antiviral activities and assist clinicians in treatment decisions.

lengths and time frames (delay effect of timing) for the MEMS adherence data.

In this section, we describe the subject population to be studied and observed data to be used in this research. These measurements include RNA viral load, phenotypic drug sensitivity and medication adherence. We also discuss how to evaluate assessment interval

The subject sample in our analysis was drawn from the AIDS Clinical Trials Group (ACTG) 398 study (Hammer et al., 2002), a randomized, double-blind, placebo-controlled phase II

**2. Materials** 

**2.1 Subject population** 

**RNA viral load:** RNA viral load was measured in copies/mL at study weeks 0, 2, 4, 8 and every 8 weeks thereafter until week 72 by the ultrasensitive reverse transcriptasepolymerase chain reaction HIV-1 RNA assay. Only measurements taken while on protocoldefined treatment were used in the analysis. All viral load values were log (base 10) transformed. Although, the lower limit of assay quantification was 200 copies/mL, when lower values (<200 copies/mL) were detected, these values were used in the analysis. The exact day of viral load measurement (not predefined study week) was used to compute study day in our analysis.

**Medication adherence:** Medication adherence was measured by two methods-- by the use of questionnaires and by the use of electronic monitoring caps. Subjects completed an adherence questionnaire (AACTG study 398 questionnaire QL0702) at study weeks 4, 8, 12, 16, and every 8 weeks thereafter. The questionnaire was completed by the study participant and/or by a face-to-face interview with study personnel. The subject was asked to specify the number of prescribed doses of each drug that he or she had failed to take on each of the preceding 4 days. Questionnaire adherence rates for APV and NFV were determined at each visit as the number of prescribed doses taken divided by the number prescribed doses during the preceding 4 day interval. For electronically monitored adherence, an MEMS cap (Medication Event Monitoring Systems, Aprex Corp., Menlo CA) was used to monitor APV and EFV compliance only. Subjects were asked to bring their medication bottles and caps to the clinic at each study visit (weeks 2, 4, 8, 12, 16, and every 8 weeks thereafter), where cap data were downloaded to computer files and stored for later analysis. Since APV was

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 55

concentration (IC50) (Molla et al., 1996). All 31 subjects used in our analysis had baseline APV and NFV IC50 values, of which 13 subjects had follow-up APV and NFV IC50 values at

We fit the dynamic model to the viral load data from 31 subjects with the following considerations. (i) In the model we incorporate the two clinical factors, drug adherence (questionnaire or MEMS) and drug susceptibility (phenotype IC50 values), into a function of treatment efficacy. (ii) We only consider the PI drug effects in the drug efficacy model because the effect of RTI drugs is considered less important compared to the PI drugs and would require a different efficacy model. (iii) We assume that NFV has the same compliance rate as determined for APV by the MEMS method. Details of the mathematical models and statistical methods are described in Huang et al. (2006) and Wu et al. (2005). For

As Molla et al*.* (1996) suggested, the phenotype marker, median inhibitory concentration (IC50), can be used to quantify agent-specific drug susceptibility. We use the following model to approximate the within-host changes over time in IC50 (Huang et al., 2003; Huang

0

*I I I t tt*

*r*

Where 0*I* and *rI* are respective values of 50 *IC t*( ) at baseline and time point *rt* at which resistant mutations dominate. In our study, *rt* is the time of virologic failure. For subjects

Poor adherence to a treatment regimen is one of the major causes of treatment failure. (Ickovics abd Meisler, 1997). The following model is used to represent adherence for a time

1 if all doses are taken in ( ] ( ) if 100 doses are taken in ( ]

where 0 1 *Rk* , with *Rk* indicating the adherence rate computed for each assessment interval (*T Tk k*<sup>1</sup> ] based on the questionnaire or MEMS data; *Tk* denotes the adherence

In most viral dynamic studies, investigators assumed that either drug efficacy was constant over treatment time (Perelson and Nelson, 1999; Wu and Ding, 1999; Ding and Wu, 2001) or

*A t R R% T T*

*k k k k*

for 0

*<sup>r</sup> <sup>r</sup>*

*r r*

*I t t*

for

1 1

 

*k k*

*T T*

(1)

(2)

completeness, a brief summary of the models and methods is given as follows.

0

*IC t t*

50

( )

without a failure time IC50, baseline IC50 was held constant over time.

the time of virologic failure.

**3.1 Drug resistance model** 

et al., 2006; Wu et al., 2005).

**3.2 Medication adherence model** 

assessment time at the *k*th clinical visit.

interval*T tT k k* <sup>1</sup> ,

**3.3 Drug efficacy model** 

**3. Mathematical models and statistical methods** 

prescribed twice daily, a prescribed AM and PM dosing period was defined for each subject. If a subject opened the bottle at least once during a dosing period, then the subject was recorded as having a positive event (*x*=1), otherwise (*x*=0). The MEMS adherence rate for APV was determined as the sum of positive dosing events divided by the sum of prescribed dosing events during the specified time interval. A positive dosing event assumes a presumptive dose. If the MEMS cap was recorded as not in use, the MEMS dosing event was set to missing. In our analysis, we assumed that NFV had the same MEMS adherence rate as APV.


Table 1. Summary of the MEMS assessment interval notation and definitions

To determine the best summary metric of the MEMS adherence rate, we evaluated different assessment interval lengths (averaging adherence dosing events over 1, 2, or 3 week intervals) and different assessment time frames (fixing the assessment interval times to end either immediately or 1, 2 or 3 weeks prior to the next measured viral load). Table 1 summarizes the MEMS assessment interval notation and definitions for the 13 models. As an example, M2.2 in Table 1 denotes an MEMS adherence interval length of 2 weeks fixed to end 2 weeks prior to the next viral load measurement; for instance, the MEMS adherence rate for a subject at study week 8 (day 56) was calculated as the number of nominal dosing events divided by the number of prescribed dosing events over study days 29 - 42. The case M serves as a reference and averages all the available MEMS data between viral load measurements.

**Phenotypic drug susceptibility:** Retrospectively, 200 subjects were randomly selected from the entire ACTG 398 study population for phenotypic sensitivity testing. Of these 139 subjects were tested at baseline based on receiving study treatment for at least 8 weeks and having an available sample. Among these subjects, 59 subjects experienced protocol-defined virologic failure and phenotypic sensitivity testing was performed at the time of failure (Hammer et al., 2002). Phenotypic drug susceptibility was determined by a recombinant virus assay (PhenoSense, ViroLogic, Inc) and values were expressed as the 50% inhibitory concentration (IC50) (Molla et al., 1996). All 31 subjects used in our analysis had baseline APV and NFV IC50 values, of which 13 subjects had follow-up APV and NFV IC50 values at the time of virologic failure.

#### **3. Mathematical models and statistical methods**

We fit the dynamic model to the viral load data from 31 subjects with the following considerations. (i) In the model we incorporate the two clinical factors, drug adherence (questionnaire or MEMS) and drug susceptibility (phenotype IC50 values), into a function of treatment efficacy. (ii) We only consider the PI drug effects in the drug efficacy model because the effect of RTI drugs is considered less important compared to the PI drugs and would require a different efficacy model. (iii) We assume that NFV has the same compliance rate as determined for APV by the MEMS method. Details of the mathematical models and statistical methods are described in Huang et al. (2006) and Wu et al. (2005). For completeness, a brief summary of the models and methods is given as follows.

#### **3.1 Drug resistance model**

54 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

prescribed twice daily, a prescribed AM and PM dosing period was defined for each subject. If a subject opened the bottle at least once during a dosing period, then the subject was recorded as having a positive event (*x*=1), otherwise (*x*=0). The MEMS adherence rate for APV was determined as the sum of positive dosing events divided by the sum of prescribed dosing events during the specified time interval. A positive dosing event assumes a presumptive dose. If the MEMS cap was recorded as not in use, the MEMS dosing event was set to missing. In our analysis, we assumed that NFV had the same MEMS adherence rate as

> Interval length

Table 1. Summary of the MEMS assessment interval notation and definitions

To determine the best summary metric of the MEMS adherence rate, we evaluated different assessment interval lengths (averaging adherence dosing events over 1, 2, or 3 week intervals) and different assessment time frames (fixing the assessment interval times to end either immediately or 1, 2 or 3 weeks prior to the next measured viral load). Table 1 summarizes the MEMS assessment interval notation and definitions for the 13 models. As an example, M2.2 in Table 1 denotes an MEMS adherence interval length of 2 weeks fixed to end 2 weeks prior to the next viral load measurement; for instance, the MEMS adherence rate for a subject at study week 8 (day 56) was calculated as the number of nominal dosing events divided by the number of prescribed dosing events over study days 29 - 42. The case M serves as a reference and averages all the available MEMS data between viral load

**Phenotypic drug susceptibility:** Retrospectively, 200 subjects were randomly selected from the entire ACTG 398 study population for phenotypic sensitivity testing. Of these 139 subjects were tested at baseline based on receiving study treatment for at least 8 weeks and having an available sample. Among these subjects, 59 subjects experienced protocol-defined virologic failure and phenotypic sensitivity testing was performed at the time of failure (Hammer et al., 2002). Phenotypic drug susceptibility was determined by a recombinant virus assay (PhenoSense, ViroLogic, Inc) and values were expressed as the 50% inhibitory

1 M visit time 0 Days 28 – 55 2 M0.1 1 week 0 Days 49 - 55 3 M0.2 2 weeks 0 Days 42 – 55 4 M0.3 3 weeks 0 Days 35 – 55 5 M1.1 1 week 1 week Days 43 – 49 6 M1.2 2 weeks 1 week Days 36 – 49 7 M1.3 3 weeks 1 week Days 29 – 49 8 M2.1 1 week 2 weeks Days 36 – 42 9 M2.2 2 weeks 2 weeks Days 29 – 42 10 M2.3 3 weeks 2 weeks Days 22 – 42 11 M3.1 1 week 3 weeks Days 29 – 35 12 M3.2 2 weeks 3 weeks Days 22 – 35 13 M3.3 3 weeks 3 weeks Days 15 - 35

Adherence assessment definition

Time frame length (weeks to RNA measurement)

Example for day 56, adherence computed over

APV.

measurements.

Case MEMS adherence

interval notation

As Molla et al*.* (1996) suggested, the phenotype marker, median inhibitory concentration (IC50), can be used to quantify agent-specific drug susceptibility. We use the following model to approximate the within-host changes over time in IC50 (Huang et al., 2003; Huang et al., 2006; Wu et al., 2005).

$$IC\_{50}(t) = \begin{cases} I\_0 + \frac{I\_r - I\_0}{t\_r}t & \text{for } 0 < t < t\_r, \\\ I\_r & \text{for } t \ge t\_r, \end{cases} \tag{1}$$

Where 0*I* and *rI* are respective values of 50 *IC t*( ) at baseline and time point *rt* at which resistant mutations dominate. In our study, *rt* is the time of virologic failure. For subjects without a failure time IC50, baseline IC50 was held constant over time.

#### **3.2 Medication adherence model**

Poor adherence to a treatment regimen is one of the major causes of treatment failure. (Ickovics abd Meisler, 1997). The following model is used to represent adherence for a time interval*T tT k k* <sup>1</sup> ,

$$A(t) = \begin{cases} 1 & \text{if all doses are taken in ( $T\_k, T\_{k+1}$ )}, \\ R\_k & \text{if } 100R\_k\% \text{ doses are taken in ( $T\_k, T\_{k+1}$ )}. \end{cases} \tag{2}$$

where 0 1 *Rk* , with *Rk* indicating the adherence rate computed for each assessment interval (*T Tk k*<sup>1</sup> ] based on the questionnaire or MEMS data; *Tk* denotes the adherence assessment time at the *k*th clinical visit.

#### **3.3 Drug efficacy model**

In most viral dynamic studies, investigators assumed that either drug efficacy was constant over treatment time (Perelson and Nelson, 1999; Wu and Ding, 1999; Ding and Wu, 2001) or

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 57

numerical solution of the differential equations (4) for the *i*th subject at time *<sup>j</sup> t* . Let ( ) *ij y t* and ( ) *<sup>i</sup> <sup>j</sup> e t* denote the repeated measurements of common logarithmic viral load and a measurement error with mean zero, respectively. The Bayesian nonlinear mixed-effects model can be written as the following three stages (Davidian and Giltinan, 1995; Huang et

2 2 ( ) (0 )*<sup>i</sup>* **yf e e** *ii i i*

 *i i* 

2 1

*Ga a b N* () ( ) ( )

 

where the mutually independent Gamma ( *Ga* ), Normal ( *N* ) and Wishart ( *Wi* ) prior distributions are chosen to facilitate computations (Davidian and Giltinan, 1995). The hyper-

(Perelson and Nelson, 1999; Ho et al., 1995; Perelson et al., 1996, 1997; Nowak and May, 2000). See Huang et al. (2006) for a detailed discussion of the Bayesian modeling approach, including the choice of the hyper-parameters, and the implementation of the Markov chain

Of the 31 subjects used in our analysis, the mean age was 40 years (SD=7); 94% were men; and 65% were white, 23% black, 10% Hispanic and 3% Asian. At baseline, 58% had prior nonnucleoside reverse transcriptase inhibitor (NNRTI) experience. Median baseline CD4 cell count was 196 cells/uL (interquartile range=120-308 cells/uL) and median baseline viral load was 38,019 copies/mL (interquartile range=19,498-181,970 copies/mL). Median time to the last viral load measurement while on protocol-defined treatment was 227 days (interquartile range=168-321 days). Median baseline IC50 values were 21.2 ng/mL and 38.9 ng/mL for APV and NFV, respectively. Among the 13 subjects with IC50 values at failure time, the median time to virologic failure was 157 days. Overall mean questionnaire adherence rate was 0.95 and 0.96 respectively for APV and NFV and the MEMS adherence rate for APV was 0.80. Fig. 1 shows the viral load (log10 transformed) and adherence rates over time based on questionnaire data for APV and NFV drugs and APV MEMS data (13

 

*<sup>T</sup>* **f***i i*

*c d Nk* , { } { } *<sup>i</sup> <sup>l</sup>*

 

 *ii i ft f t im m* , <sup>1</sup> ( ( ) ( )) *<sup>i</sup>*

(0 )

(7)

were determined from previous studies and the literature

*Wi*

**b b** *i i N* (6)

 *t Vt* , where ( ) *V t <sup>i</sup>*

*N* **I***m* (5)

*<sup>T</sup>* **e***i i im et et* .

*l i* and

*<sup>i</sup> <sup>j</sup>* denotes the

*i i i i i Ti i i*

 

*i n* , (ln ln ln ln ln ln ln )*<sup>T</sup>*

*<sup>T</sup>* **<sup>y</sup>***i i yt y t im m* , 1 1 ( ) ( ( ) ( )) *i i*

Monte Carlo (MCMC) procedures (Gamerman, 1997; Wakefield, 1996).

 *<sup>i</sup>* 

 and 

summary metrics) for one representative subject.

{1 1} *ij <sup>i</sup>* **Y** *y i n j m* . Let 10 ( ) log ( ( )) *ij j i i <sup>i</sup> <sup>j</sup> f*

*Stage* 1. Within-subject variation:

where 1 1 ( ( ) ( )) *i i*

*Stage* 2. Between-subject variation:

*Stage* 3. Hyperprior distributions:

**4.1 Subject characteristics** 

parameters *a b*

**4. Results** 

{ 1} *<sup>i</sup>* 

al., 2006).

antiviral regimens had perfect effect in blocking viral replication (Ho et al., 1995; Perelson et al., 1996, 1997). However, the drug efficacy may change as concentrations of antiretroviral drugs and other factors (*e.g.* drug resistance) vary during treatment (Dixit et al., 2004). We employ the following modified *Emax* model **(**Sheiner, 1985) to represent the time-varying drug efficacy ( see Wu et al.(2005) for more discussion about the drug effect *Emax* model) for two antiretroviral agents within a class (for example, the two PI drugs APV and NFV),

$$\gamma(t) = \frac{A\_1(t) / I\mathbb{C}\_{50}^1(t) + A\_2(t) / I\mathbb{C}\_{50}^2(t)}{\phi + A\_1(t) / I\mathbb{C}\_{50}^1(t) + A\_2(t) / I\mathbb{C}\_{50}^2(t)},\tag{3}$$

where 50( ) *<sup>k</sup> IC t* ( 1 2) *<sup>k</sup>* are median inhibitory concentration change over time for the two agents; ( ) *A t <sup>k</sup>* ( 1 2) *k* are adherence profiles of the two drugs measured by questionnaire or the MEMS method. Parameter can be regarded as a conversion factor between *in vitro* and *in vivo* IC50 s and will be estimated from the data. Note that ( )*t* ranges from 0 to 1.

#### **3.4 Antiviral response model**

We consider a simplified HIV dynamic model with antiviral treatment as follows. (Huang et al., 2006; Wu et al., 2005).

$$\begin{aligned} \frac{d}{dt}T &= \quad \lambda - d\_T T - [1 - \gamma(t)]kTV, \\ \frac{d}{dt}T^\* &= \quad [1 - \gamma(t)]kTV - \delta T^\*, \\ \frac{d}{dt}V &= \quad \quad N\delta T^\* - cV, \end{aligned} \tag{4}$$

where the three differential equations represent three compartments: target uninfected cells (*T* ), infected cells (*T* ) and free virions (*V* ). Parameter represents the rate at which new T cells are generated from sources within the body, such as the thymus, *Td* is the death rate of T cells, *k* is the infection rate without treatment, is the death rate of infected cells, *N* is the number of new virions produced from each of infected cell during its life-time, and *c* is the clearance rate of free virions. The time-varying parameter ( )*t* is the antiviral drug efficacy at treatment time *t* .

#### **3.5 Bayesian modeling approach**

Although a number of studies investigated various statistical methods, including Bayesian approaches, of fitting viral dynamic models to predicting virologic responses using shortterm viral load data (Wu and Ding, 1999; Ho et al., 1995; Perelson et al., 1996, 1997; Wu et al., 1999; Notermans et al., 1998; Markowitz et al., 2003; Han et al., 2002), little work has been undertaken to investigate long-term virologic responses. In this paper, we used a hierarchical Bayesian modeling approach (Huang et al., 2006) to estimate the dynamic parameters.

We denote the number of subjects by *n* and the number of measurements on the *i*th subject by *mi* . For notational convenience, let (ln ln ln ln ln ln ln )*<sup>T</sup> T c d Nk* ,

{ 1} *<sup>i</sup> i n* , (ln ln ln ln ln ln ln )*<sup>T</sup> i i i i i Ti i i c d Nk* , { } { } *<sup>i</sup> <sup>l</sup> l i* and {1 1} *ij <sup>i</sup>* **Y** *y i n j m* . Let 10 ( ) log ( ( )) *ij j i i <sup>i</sup> <sup>j</sup> f t Vt* , where ( ) *V t <sup>i</sup> <sup>i</sup> <sup>j</sup>* denotes the numerical solution of the differential equations (4) for the *i*th subject at time *<sup>j</sup> t* . Let ( ) *ij y t* and ( ) *<sup>i</sup> <sup>j</sup> e t* denote the repeated measurements of common logarithmic viral load and a measurement error with mean zero, respectively. The Bayesian nonlinear mixed-effects model can be written as the following three stages (Davidian and Giltinan, 1995; Huang et al., 2006).

*Stage* 1. Within-subject variation:

56 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

antiviral regimens had perfect effect in blocking viral replication (Ho et al., 1995; Perelson et al., 1996, 1997). However, the drug efficacy may change as concentrations of antiretroviral drugs and other factors (*e.g.* drug resistance) vary during treatment (Dixit et al., 2004). We employ the following modified *Emax* model **(**Sheiner, 1985) to represent the time-varying drug efficacy ( see Wu et al.(2005) for more discussion about the drug effect *Emax* model) for two antiretroviral agents within a class (for example, the two PI drugs APV and NFV),

> 1 2 1 50 2 50

*A t IC t A t IC t*

where 50( ) *<sup>k</sup> IC t* ( 1 2) *<sup>k</sup>* are median inhibitory concentration change over time for the two agents; ( ) *A t <sup>k</sup>* ( 1 2) *k* are adherence profiles of the two drugs measured by questionnaire or

We consider a simplified HIV dynamic model with antiviral treatment as follows. (Huang et

[1 ( )]

where the three differential equations represent three compartments: target uninfected cells

T cells are generated from sources within the body, such as the thymus, *Td* is the death rate

is the number of new virions produced from each of infected cell during its life-time, and *c*

Although a number of studies investigated various statistical methods, including Bayesian approaches, of fitting viral dynamic models to predicting virologic responses using shortterm viral load data (Wu and Ding, 1999; Ho et al., 1995; Perelson et al., 1996, 1997; Wu et al., 1999; Notermans et al., 1998; Markowitz et al., 2003; Han et al., 2002), little work has been undertaken to investigate long-term virologic responses. In this paper, we used a hierarchical Bayesian modeling approach (Huang et al., 2006) to estimate the dynamic

We denote the number of subjects by *n* and the number of measurements on the *i*th subject by *mi* . For notational convenience, let (ln ln ln ln ln ln ln )*<sup>T</sup>*

 

 *c d Nk* ,

*<sup>d</sup> <sup>T</sup> t kTV T*

*<sup>d</sup> <sup>V</sup> N T cV*

*T <sup>d</sup> T d T t kTV*

the MEMS method. Parameter

**3.4 Antiviral response model** 

al., 2006; Wu et al., 2005).

efficacy at treatment time *t* .

parameters.

**3.5 Bayesian modeling approach** 

*dt*

*dt*

*dt*

is the clearance rate of free virions. The time-varying parameter

(*T* ), infected cells (*T* ) and free virions (*V* ). Parameter

of T cells, *k* is the infection rate without treatment,

*in vivo* IC50 s and will be estimated from the data. Note that

1 2 1 50 2 50 () () () () ( ) () () () () *A t IC t A t IC t <sup>t</sup>*

[1 ( )]

 

 

(3)

can be regarded as a conversion factor between *in vitro* and

( )*t* ranges from 0 to 1.

(4)

represents the rate at which new

( )*t* is the antiviral drug

*T*

is the death rate of infected cells, *N*

$$\mathbf{y}\_{i} = \mathbf{f}\_{i}(\boldsymbol{\rho}\_{i}) + \mathbf{e}\_{i}, \quad \mathbf{e}\_{i} \mid \boldsymbol{\sigma}^{2}, \boldsymbol{\rho}\_{i} \sim \mathrm{N}(\mathbf{0}, \boldsymbol{\sigma}^{2} \mathbf{I}\_{m\_{i}}) \tag{5}$$

where 1 1 ( ( ) ( )) *i i <sup>T</sup>* **<sup>y</sup>***i i yt y t im m* , 1 1 ( ) ( ( ) ( )) *i i <sup>T</sup>* **f***i i ii i ft f t im m* , <sup>1</sup> ( ( ) ( )) *<sup>i</sup> <sup>T</sup>* **e***i i im et et* . *Stage* 2. Between-subject variation:

$$\mathbf{b}\_{i}\theta\_{i} = \mu + \mathbf{b}\_{i}, \quad \mathbf{b}\_{i} \mid \Sigma - N(\mathbf{0}, \Sigma) \tag{6}$$

*Stage* 3. Hyperprior distributions:

$$
\sigma^{-2} \rightharpoonup \text{Ca}(a, b), \quad \mu \rightharpoonup \text{N}(\eta, \Lambda), \quad \Sigma^{-1} \rightharpoonup \text{Wi}(\Omega, \nu) \tag{7}
$$

where the mutually independent Gamma ( *Ga* ), Normal ( *N* ) and Wishart ( *Wi* ) prior distributions are chosen to facilitate computations (Davidian and Giltinan, 1995). The hyperparameters *a b* and were determined from previous studies and the literature (Perelson and Nelson, 1999; Ho et al., 1995; Perelson et al., 1996, 1997; Nowak and May, 2000). See Huang et al. (2006) for a detailed discussion of the Bayesian modeling approach, including the choice of the hyper-parameters, and the implementation of the Markov chain Monte Carlo (MCMC) procedures (Gamerman, 1997; Wakefield, 1996).

#### **4. Results**

#### **4.1 Subject characteristics**

Of the 31 subjects used in our analysis, the mean age was 40 years (SD=7); 94% were men; and 65% were white, 23% black, 10% Hispanic and 3% Asian. At baseline, 58% had prior nonnucleoside reverse transcriptase inhibitor (NNRTI) experience. Median baseline CD4 cell count was 196 cells/uL (interquartile range=120-308 cells/uL) and median baseline viral load was 38,019 copies/mL (interquartile range=19,498-181,970 copies/mL). Median time to the last viral load measurement while on protocol-defined treatment was 227 days (interquartile range=168-321 days). Median baseline IC50 values were 21.2 ng/mL and 38.9 ng/mL for APV and NFV, respectively. Among the 13 subjects with IC50 values at failure time, the median time to virologic failure was 157 days. Overall mean questionnaire adherence rate was 0.95 and 0.96 respectively for APV and NFV and the MEMS adherence rate for APV was 0.80. Fig. 1 shows the viral load (log10 transformed) and adherence rates over time based on questionnaire data for APV and NFV drugs and APV MEMS data (13 summary metrics) for one representative subject.

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 59

are specified based on the combination of drug susceptibility (*IC50*) data and 14 different adherence summary metrics (1 questionnaire and 13 MEMS summary metrics listed in Table 1). Note that the abbreviation IM2.2, for example, denotes the model incorporating the data of drug resistance (I) and MEMS adherence rate (M2.2) summarized as an interval length of 2 weeks fixed to end 2 weeks prior to the next viral load measurement. For example, the MEMS adherence rate for a subject at study week 8 (day 56) was calculated over a 14 day interval from study days 29-42 and this value was used to represent adherence from the previous

In order to assess how adherence rates, determined from 14 different scenarios, interact with drug susceptibility to contribute to virologic response, we fitted the models to all 14 scenarios as well as the control model and compared the fitting results. We found that, overall, the model with adherence rate determined from MEMS dosing events averaged over a 2 week assessment interval either 1 week prior to a viral load measurement (IM1.2) or 2 weeks prior to a viral load measurement (IM2.2) provided the best fits to the observed data, compared to the other 13 models for most subjects; the control model, lacking factors for subject-specific drug adherence and susceptibility, failed to fit viral load rebounds and

Fig. 2. The estimated viral trajectory for three representative subjects from the model fitting: (i) Control model (solid curves), (ii) IM1.2 (dotted curves) and (iii) IM2.2 (dashed curves).

fluctuations and provided the worst fitting results for the majority of subjects. For the purpose of illustration, the model fitting curves for three representative subjects from the control model (solid curves), the IM1.2 model (dotted curves), and the IM2.2 model (dashed

Table 2 presents the results of estimated dynamic parameters for individual subjects and the sample summary statistics (minimum, median, mean, maximum, standard deviation (SD) and coefficient of variation (CV) for the model IM2.2 that provided the best fit to the observed data. We can see from Table 2 a relatively large between-subject variation in the

 ( ) 2 ( 2) *t* 

. Other 14 models

corresponds to setting *A*(*t*) and *IC*50(*t*) to be 1 in Eq. (3), i.e.,

study visit to the study visit at day 56 for modeling fitting.

The observed values are indicated by circles.

**4.3 Individual dynamics parameter estimates** 

curves) are displayed in Fig. 2.

**4.2 Model fitting** 

Fig. 1. The trajectories of viral load on log10 scale (solid curves) and adherence rates (stairsteps) over time based on questionnaire data for APV and NFV drugs (upper-left panel) and MEMS data summarized by the 13 models for APV drug (other panels) for one subject

We fit the viral dynamic model to the data from 31 subjects described previously using the proposed Bayesian approach. We incorporated the two clinical factors, drug adherence (questionnaire or MEMS) and drug susceptibility (phenotype *IC*50 values), into a function of treatment efficacy (3). For model fitting and the purpose of comparisons, we set up a control model as the one without using any adherence and drug susceptibility data which corresponds to setting *A*(*t*) and *IC*50(*t*) to be 1 in Eq. (3), i.e., ( ) 2 ( 2) *t* . Other 14 models are specified based on the combination of drug susceptibility (*IC50*) data and 14 different adherence summary metrics (1 questionnaire and 13 MEMS summary metrics listed in Table 1). Note that the abbreviation IM2.2, for example, denotes the model incorporating the data of drug resistance (I) and MEMS adherence rate (M2.2) summarized as an interval length of 2 weeks fixed to end 2 weeks prior to the next viral load measurement. For example, the MEMS adherence rate for a subject at study week 8 (day 56) was calculated over a 14 day interval from study days 29-42 and this value was used to represent adherence from the previous study visit to the study visit at day 56 for modeling fitting.

#### **4.2 Model fitting**

58 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

Fig. 1. The trajectories of viral load on log10 scale (solid curves) and adherence rates (stairsteps) over time based on questionnaire data for APV and NFV drugs (upper-left panel) and MEMS data summarized by the 13 models for APV drug (other panels) for one

We fit the viral dynamic model to the data from 31 subjects described previously using the proposed Bayesian approach. We incorporated the two clinical factors, drug adherence (questionnaire or MEMS) and drug susceptibility (phenotype *IC*50 values), into a function of treatment efficacy (3). For model fitting and the purpose of comparisons, we set up a control model as the one without using any adherence and drug susceptibility data which

subject

In order to assess how adherence rates, determined from 14 different scenarios, interact with drug susceptibility to contribute to virologic response, we fitted the models to all 14 scenarios as well as the control model and compared the fitting results. We found that, overall, the model with adherence rate determined from MEMS dosing events averaged over a 2 week assessment interval either 1 week prior to a viral load measurement (IM1.2) or 2 weeks prior to a viral load measurement (IM2.2) provided the best fits to the observed data, compared to the other 13 models for most subjects; the control model, lacking factors for subject-specific drug adherence and susceptibility, failed to fit viral load rebounds and

Fig. 2. The estimated viral trajectory for three representative subjects from the model fitting: (i) Control model (solid curves), (ii) IM1.2 (dotted curves) and (iii) IM2.2 (dashed curves). The observed values are indicated by circles.

fluctuations and provided the worst fitting results for the majority of subjects. For the purpose of illustration, the model fitting curves for three representative subjects from the control model (solid curves), the IM1.2 model (dotted curves), and the IM2.2 model (dashed curves) are displayed in Fig. 2.

#### **4.3 Individual dynamics parameter estimates**

Table 2 presents the results of estimated dynamic parameters for individual subjects and the sample summary statistics (minimum, median, mean, maximum, standard deviation (SD) and coefficient of variation (CV) for the model IM2.2 that provided the best fit to the observed data. We can see from Table 2 a relatively large between-subject variation in the

60 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 61

seven viral dynamic parameters was observed (CV ranges from 20% to 148%) among the 31 subjects. Generally speaking, the virologically successful subjects (maintaining plasma HIV-1 RNA levels of less than 200 copies/mL) have higher clearance rates of free virions (*c*), but

results show the similar patterns to those displayed in Figure 4 studied by Wu et al. (2005) The individual parameter estimates from both the IM2.2 and IM1.2 models are significantly correlated for all seven parameters, while the individual parameter estimates for the control model appear significantly different from those for the model IM2.2 for most of the seven

In order to assess how different adherence rates measured by questionnaires and MEMS contribute to the virologic response, we compared the fitting results of models with all 14 adherence scenarios and the control model. The mean of the sum of the squared deviations

SSDs are plotted for all the models in Fig. 3, with the best fitting models having a smaller

), and lower death rates of infected cells (

*ij y* are the observed and predicted values, respectively. The mean

1 ( ) ˆ *mi <sup>j</sup> ij ij <sup>y</sup> <sup>y</sup>* for each

); these

**4.4 Effects of adherence rate determined by questionnaire vs. MEMS data** 

(SSD) was used to assess model fit and the SSD was calculated by <sup>2</sup>

mean SSD, and sign test *p*-values from pairwise comparisons are reported in Table 3.

Fig. 3. Comparison of mean of SSDs for models from the 14 different determinants of adherence with drug resistance and the control model. The three horizontal lines represent

The pattern in Fig. 3 shows that when the time frame for MEMS assessment is fixed, models with a 2 week MEMS assessment interval length generally outperform models with an

mean of the SSDs for control, IA and IM models, respectively.

**4.5 What MEMS assessment interval length is best?** 

assessment interval length of 1 or 3 weeks.

smaller efficacy parameter estimates (

parameters (data not shown here).

subject, where *ij y* and ˆ


Table 2. The estimated dynamic parameters from the IM2.2 model for individual subjects, where Min, Med, Max, SD and CV=SD/Mean denote the minimum, median, maximum, standard deviation and coefficient of variation, respectively.

1 0.002 3.144 0.105 68.830 0.041 126.353 2.083 2 0.002 3.366 0.203 193.493 0.021 763.481 0.800 3 0.002 3.065 0.228 103.261 0.051 516.932 1.363 4 0.001 3.963 0.294 67.549 0.110 709.988 0.738 5 0.002 2.499 1.238 171.962 0.142 9384.339 0.868 6 0.002 3.219 0.224 108.737 0.056 495.182 1.526 7 0.002 3.631 0.078 111.508 0.017 109.594 1.549 8 0.001 4.103 0.157 73.195 0.050 354.894 1.124 9 0.002 2.286 0.809 177.204 0.097 4070.423 1.112 10 0.002 2.620 1.058 208.394 0.091 6031.472 1.033 11 0.002 3.213 0.219 83.127 0.058 479.131 1.316 12 0.002 2.804 0.315 96.976 0.071 806.579 1.622 13 0.001 3.705 0.164 88.307 0.056 581.903 0.727 14 0.002 2.877 0.132 130.303 0.026 341.015 1.347 15 0.003 2.316 0.434 207.091 0.043 2197.731 1.064 16 0.003 1.753 0.924 160.891 0.120 4224.349 1.867 17 0.001 4.041 0.211 127.465 0.023 765.045 0.692 18 0.002 3.367 0.216 85.716 0.051 508.451 1.148 19 0.002 3.955 0.114 74.292 0.033 164.046 1.498 20 0.001 3.938 0.116 117.911 0.023 356.109 0.801 21 0.001 2.887 0.314 306.351 0.019 1636.315 0.486 22 0.003 2.003 0.569 100.966 0.067 1186.222 0.654 23 0.001 4.273 0.260 43.015 0.136 474.269 1.258 24 0.001 3.506 0.131 157.873 0.028 643.514 0.847 25 0.002 2.277 0.839 174.871 0.105 4912.339 1.189 26 0.001 3.847 0.340 54.983 0.135 714.569 1.116 27 0.003 2.730 0.218 103.326 0.042 411.427 1.690 28 0.002 3.510 0.073 133.204 0.015 106.627 1.366 29 0.002 3.751 0.186 108.477 0.026 435.276 1.111 30 0.002 3.760 0.162 96.575 0.031 354.927 1.192 31 0.002 3.415 0.144 112.288 0.034 392.298 1.140 Min 0.001 1.753 0.073 43.015 0.015 106.627 0.486 Med 0.002 3.367 0.218 108.737 0.050 516.932 1.140 Max 0.003 4.273 1.238 306.351 0.142 9384.339 2.083 Mean 0.0017 3.285 0.328 124.134 0.059 1427.574 1.172 SD 0.0006 0.647 0.305 55.726 0.039 2118.271 0.373 CV (%) 33.037 19.686 93.066 44.892 66.245 148.383 31.801 Table 2. The estimated dynamic parameters from the IM2.2 model for individual subjects, where Min, Med, Max, SD and CV=SD/Mean denote the minimum, median, maximum,

*<sup>i</sup> ci* δ*i* λ*i dTi Ni ki* x 104

Subject

standard deviation and coefficient of variation, respectively.

seven viral dynamic parameters was observed (CV ranges from 20% to 148%) among the 31 subjects. Generally speaking, the virologically successful subjects (maintaining plasma HIV-1 RNA levels of less than 200 copies/mL) have higher clearance rates of free virions (*c*), but smaller efficacy parameter estimates (), and lower death rates of infected cells (); these results show the similar patterns to those displayed in Figure 4 studied by Wu et al. (2005) The individual parameter estimates from both the IM2.2 and IM1.2 models are significantly correlated for all seven parameters, while the individual parameter estimates for the control model appear significantly different from those for the model IM2.2 for most of the seven parameters (data not shown here).

#### **4.4 Effects of adherence rate determined by questionnaire vs. MEMS data**

In order to assess how different adherence rates measured by questionnaires and MEMS contribute to the virologic response, we compared the fitting results of models with all 14 adherence scenarios and the control model. The mean of the sum of the squared deviations (SSD) was used to assess model fit and the SSD was calculated by <sup>2</sup> 1 ( ) ˆ *mi <sup>j</sup> ij ij <sup>y</sup> <sup>y</sup>* for each subject, where *ij y* and ˆ *ij y* are the observed and predicted values, respectively. The mean SSDs are plotted for all the models in Fig. 3, with the best fitting models having a smaller mean SSD, and sign test *p*-values from pairwise comparisons are reported in Table 3.

Fig. 3. Comparison of mean of SSDs for models from the 14 different determinants of adherence with drug resistance and the control model. The three horizontal lines represent mean of the SSDs for control, IA and IM models, respectively.

#### **4.5 What MEMS assessment interval length is best?**

The pattern in Fig. 3 shows that when the time frame for MEMS assessment is fixed, models with a 2 week MEMS assessment interval length generally outperform models with an assessment interval length of 1 or 3 weeks.

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 63

better predictor of virologic response than adherence assessed by questionnaires, MEMS

Fig. 4. Comparison of SSDs for the models from 9 different determinants of adherence

Models Control I A M M1.2 M2.2 IA IM IM1.2 IM2.2

Several studies investigated the association between virologic responses and adherence assessed by MEMS data only without considering other confounding factors such as drug resistance using standard modeling methods including Poisson regression (Knafl et al., 2004), logistic regression (Vrijens et al., 2005) and linear mixed-effects model (Liu et al., 2007). In this article, we developed a mechanism-based nonlinear time-varying differential equation model for long-term dynamics to (i) establish the relationship of virologic response (viral load trajectory) with drug adherence and drug resistance, (ii) to describe both suppression and resurgence of virus, (iii) to directly incorporate observed drug adherence

and/or drug resistance as well as the control model

M 0.007 0.106 0.858 M1.2 <0.001 0.106 0.590 0.858 M2.2 <0.001 0.209 0.858 0.858 0.369 IA <0.001 0.209 0.209 0.590 0.209 0.590 IM <0.001 0.590 0.209 0.048 0.048 0.209 0.106 IM1.2 <0.001 0.048 0.002 0.002 0.020 0.048 0.048 0.026 IM2.2 <0.001 0.048 0.002 0.002 0.007 0.048 0.007 0.020 0.590 SSD Mean 11.30 6.09 6.68 6.69 6.68 6.45 6.06 5.27 4.75 4.11 ±SD 9.28 6.84 6.02 6.87 6.79 6.33 5.66 5.52 5.38 4.18 Table 4. Pairwise comparisons of sum of squared deviations (SSD) from individual subjects

for 10 models. The *p*-values were obtained using the sign test.

I <0.001

**5. Conclusion and discussion** 

A 0.007 0.106

*p*

alone or two-factor combinations.


Table 3. Pairwise comparisons of sum of squared deviations (SSD) from individual subjects for 15 models. The *p*-values were obtained using the sign test and MSSD is the mean of SSD

#### **4.6 What MEMS assessment time frame (delay effect of timing) is best?**

As seen in Fig. 3, regardless of the assessment interval length, models which assess compliance 2 weeks prior to viral load generally outperform models which assess compliance immediately before viral load, 1 week before or 3 weeks before viral load measurement. Overall, the model with a MEMS assessment interval length of 2 weeks measured from 4 to 2 weeks prior to viral load measurement (IM2.2) was significantly a better predicator of viral load over time than any other models, with the exception of the IM1.2 model.

From Table 3, the means and standard deviations of the SSDs for the models based on IM1.2 ( 4 75 5 38 ) and IM2.2 ( 4 11 4 18 ) were significantly less than those of the other 13 models. We can see that that the IM1.2 and IM2.2 models were significantly better than the models based on the other 13 models ( *p* 0 001 0 048 ), but they were not significantly different each other (*p*=0.590). The control model was significantly worse than those based on all other models ( *p* 0 001 0 020 ) except for the 2 models (IM3.1 and IM3.2: *p* =0.106).

#### **4.7 What adherence assessment method (questionnaire vs MEMS) is best?**

Further, we compared the model fittings with all possible combinations of IC50 and the four determinants of adherence (A, M, M1.2 and M2.2). The mean of SSD for all the 10 models is plotted in Figure 4, and sign test *p* ─values from pairwise comparisons are reported in Table 4. The results indicate that (i) the control model was significantly worse than those based on all other 9 models (*p ≤*0.001~0.007); (ii) the models IM1.2 and IM2.2 were significantly better than the other eight models (*p*≤0.001 0.048); (iii) the models I, A, M, M1.2, M2.2, IA and IM do not provide significantly different results (*p*=0.048 0.858) except for two marginally significant results. In particular, the models IA and IM are not better than the model I (*p*=0.209, 0.590), and the models IM1.2 and IM2.2 are significantly better than the models I, M1.2 and M2.2 (*p*=0.007 0.048). Overall, adherence assessed by an optimal summary MEMS metric with the confounding resistance factor combinations (IM1.2 and IM2.2) was a

Model Control IA IM IM0.1 IM0.2 IM0.3 IM1.1 IM1.2 IM1.3 IM2.1 IM2.2 IM2.3 IM3.1 IM3.2

**4.6 What MEMS assessment time frame (delay effect of timing) is best?** 

As seen in Fig. 3, regardless of the assessment interval length, models which assess compliance 2 weeks prior to viral load generally outperform models which assess compliance immediately before viral load, 1 week before or 3 weeks before viral load measurement. Overall, the model with a MEMS assessment interval length of 2 weeks measured from 4 to 2 weeks prior to viral load measurement (IM2.2) was significantly a better predicator of viral load over time than any other models, with the exception of the

From Table 3, the means and standard deviations of the SSDs for the models based on IM1.2 ( 4 75 5 38 ) and IM2.2 ( 4 11 4 18 ) were significantly less than those of the other 13 models. We can see that that the IM1.2 and IM2.2 models were significantly better than the models based on the other 13 models ( *p* 0 001 0 048 ), but they were not significantly different each other (*p*=0.590). The control model was significantly worse than those based on all other models ( *p* 0 001 0 020 ) except for the 2 models (IM3.1 and IM3.2: *p* =0.106).

Further, we compared the model fittings with all possible combinations of IC50 and the four determinants of adherence (A, M, M1.2 and M2.2). The mean of SSD for all the 10 models is plotted in Figure 4, and sign test *p* ─values from pairwise comparisons are reported in Table 4. The results indicate that (i) the control model was significantly worse than those based on all other 9 models (*p ≤*0.001~0.007); (ii) the models IM1.2 and IM2.2 were significantly better than the other eight models (*p*≤0.001 0.048); (iii) the models I, A, M, M1.2, M2.2, IA and IM do not provide significantly different results (*p*=0.048 0.858) except for two marginally significant results. In particular, the models IA and IM are not better than the model I (*p*=0.209, 0.590), and the models IM1.2 and IM2.2 are significantly better than the models I, M1.2 and M2.2 (*p*=0.007 0.048). Overall, adherence assessed by an optimal summary MEMS metric with the confounding resistance factor combinations (IM1.2 and IM2.2) was a

**4.7 What adherence assessment method (questionnaire vs MEMS) is best?** 

IA <0.001 IM <0.001 0.106 IM0.1 0.106 0.007 0.002 IM0.2 0.019 0.209 0.007 0.020 IM0.3 0.048 0.048 0.020 0.106 0.369 IM1.1 0.001 0.590 0.020 0.106 0.209 0.858 IM1.2 <0.001 0.048 0.029 0.002 0.007 0.007 0.007 IM1.3 0.001 0.858 0.858 0.048 0.048 0.048 0.209 0.048 IM2.1 <0.001 0.106 0.209 0.020 0.007 0.048 0.048 0.048 0.209 IM2.2 <0.001 0.007 0.020 <0.001 0.001 <0.001 0.048 0.590 0.048 0.020 IM2.3 <0.001 0.369 0.209 0.011 0.002 0.002 0.209 0.106 0.858 0.590 0.048 IM3.1 0.106 0.007 <0.001 0.369 0.590 0.048 0.007 <0.001 0.007 0.001 0.002 0.002 IM3.2 0.106 0.209 0.002 0.858 0.858 0.007 0.007 0.001 0.002 0.007 0.007 0.002 0.048 IM3.3 0.001 0.858 0.020 0.048 0.020 0.369 0.209 0.007 0.369 0.369 0.001 0.020 0.002 0.020 MSSD 11.30 6.06 5.27 9.77 7.77 8.27 8.09 4.75 6.00 6.73 4.11 5.16 11.12 11.56 Table 3. Pairwise comparisons of sum of squared deviations (SSD) from individual subjects for 15 models. The *p*-values were obtained using the sign test and MSSD is the mean of SSD

IM1.2 model.

better predictor of virologic response than adherence assessed by questionnaires, MEMS alone or two-factor combinations.

Fig. 4. Comparison of SSDs for the models from 9 different determinants of adherence and/or drug resistance as well as the control model


Table 4. Pairwise comparisons of sum of squared deviations (SSD) from individual subjects for 10 models. The *p*-values were obtained using the sign test.

#### **5. Conclusion and discussion**

Several studies investigated the association between virologic responses and adherence assessed by MEMS data only without considering other confounding factors such as drug resistance using standard modeling methods including Poisson regression (Knafl et al., 2004), logistic regression (Vrijens et al., 2005) and linear mixed-effects model (Liu et al., 2007). In this article, we developed a mechanism-based nonlinear time-varying differential equation model for long-term dynamics to (i) establish the relationship of virologic response (viral load trajectory) with drug adherence and drug resistance, (ii) to describe both suppression and resurgence of virus, (iii) to directly incorporate observed drug adherence

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 65

exist for the proposed modeling method. Firstly, our model is a simplified model and there are many possible variations (Perelson and Nelson, 1999; Nowak and May, 2000; Callaway and Perelson, 2002). We did not separately consider the compartments of short-lived productively infected cells, long-lived and latently infected cells.(Perelson et al., 1997). Instead we examined a pooled productively infected cell population. The virus compartment was not further decomposed into infectious virions and non-infectious virions as in the paper by Perelson *et al*. (1996). Thus, different mechanisms of RTI and PI drug effects were not modeled. In fact, we only considered PI drug effects in the drug efficacy model (3) since the RTI drugs have a different adherence-resistance relationship. Further studies will be conducted in considering both PI and RTI drug effects in the models. Secondly, the availability of IC50 data was limited to baseline and failure time, as is typical in clinical trials. Thus, we extrapolated the IC50 data linearly to the whole treatment period in our modeling. The linear extrapolation is the best approximation that we can get from the sparse *IC*50 data (Wu et al., 2005). The linear assumption might have some influence on the estimation results since the *IC*50 might have jumped to a higher level earlier before the failure time when we obtained the sample for drug resistance test. However, we expect that this assumption had little effect on the prediction of virologic response since we had relatively frequent monitoring (monthly in the later stage) of virologic failure in this study. Thirdly, a more complete model of antiretroviral treatment efficacy would ideally also consider the time-varying function of concentrations of drug in plasma (Huang et al., 2003). Unfortunately, the limited availability of drug concentration data prohibited our inclusion of PK parameters in our model. Lastly, as measurements of adherence may not reflect actual adherence profiles for individual patients, the data quality would affect our estimation results for viral dynamic parameters. For example, adherence data measured by questionnaires may not be accurate. More accurate measurements of the MEMS adherence data were used in this paper and it was found that the MEMS adherence data can provide a better prediction of virologic response compared with the questionnaire adherence data, when the MEMS data are summarized in an appropriate way. Further studies on these issues are definitely needed. Nevertheless, these limitations would not offset the major findings from our

modeling approach, although further improvement may be warranted.

response with drug exposure and drug susceptibility.

(MEMS). 8th CROI. Chicago, USA

and MSP/NSA grant H98230-09-1-0053.

**6. Acknowledgment** 

**7. References** 

In summary, MEMS adherence data may not be correlated better to virologic response compared to questionnaire adherence data unless the MEMS cap data are summarized in an appropriate way where adherence was assessed over 2 week interval measured from 1 or 2 weeks prior to a RNA measurement in our case. Our study also shows that the mechanismbased dynamic model is powerful and effective to establish a relationship of antiviral

The author would like to thank the ACTG398 study team for allowing me to use the clinical data from their study. This research was partially supported by NIAID/NIH grant AI080338

Arnsten, J.; Demas, P. & Gourvetch, M.; et al. (February, 2001a). Adherence and viral load in

HIV infected drug users: Comparison of self-report and medication event monitors

and susceptibility into a function of treatment efficacy and (iv) to use a hierarchical Bayesian mixed-effects modeling approach that can not only combine prior information with current clinical data for estimating dynamic parameters, but also characterize inter-subject variability. Our modeling approach allows us to estimate time-varying antiretroviral efficacy during the entire course of a treatment regimen by incorporating the information of drug exposure and drug susceptibility. Thus, the results of estimated dynamic parameters based on this model should be more reliable and reasonable to interpret long-term HIV dynamics. Our models are simplified with the main goals of retaining crucial features of HIV dynamics and, at the same time, guaranteeing their applicability to typical clinical data, in particular, long-term viral load measurements.

We employed the proposed mechanism-based dynamic model to assess how to efficiently use adherence rates based on questionnaires and MEMS dosing events to predict virologic response. In particular, we intended to address the questions (i) how to summarize the MEMS adherence data for efficient prediction of virologic response, and (ii) which adherence assessment method, questionnaire or MEMS, is a more efficient predictor of virologic response after accounting for potential confounding factors such as drug resistance between subjects.

For the MEMS data, we found that the best summary metric for prediction of virologic response in terms of model fitting residuals (prediction error) is the adherence rate determined from MEMS dosing events averaged over a 2 week assessment interval, 1 week or 2 weeks prior to the next measured RNA observation (denoted by IM1.2 or IM2.2). The model fitting residuals from both models (IM2.2 and IM1.2) are significantly smaller than any other 13 models (*p*≤0.001 0.048), but they were not significantly different each other (*p*=0.590).

The model which used all available MEMS data between study visits to determine the adherence rate (the standard analysis) did not perform significantly better in terms of prediction of virologic response compared to the model with questionnaire adherence data (*p*=0.106).

We also compared the model fittings with all possible combinations of IC50 and the four determinants of adherence data (see Fig. 4). The results indicate that (i) the control model was significantly worse than those based on all other 9 models (*p*≤0.001 0.007); (ii) the models IM1.2 and IM2.2 were significantly better than the eight other models (*p*≤0.001 0.048); (iii) the models I, A, M, M1.2, M2.2, IA and IM do not provide significantly different results (*p*=0.048 0.858) except for two marginally significant results. In particular, the models IA and IM did not improve upon the model I, which indicates that adherence measured by questionnaire and MEMS dosing events averaged over study visit interval did not provide any additional information to drug susceptibility in predicting virologic response. However, the models IM1.2 and IM2.2 did outperform the models I, M1.2 and M2.2, which indicates that the combination of drug susceptibility and adherence assessed over 2 week interval measured from 1 or 2 weeks prior to a RNA measurement provided significant additional information compared to either drug susceptibility or adherence alone in predicting virologic response.

Although the analysis presented here used a simplified model, which appeared to perform well in capturing and explaining the observed patterns, and characterizing the biological mechanisms of HIV infection under relatively complex clinical situations, some limitations exist for the proposed modeling method. Firstly, our model is a simplified model and there are many possible variations (Perelson and Nelson, 1999; Nowak and May, 2000; Callaway and Perelson, 2002). We did not separately consider the compartments of short-lived productively infected cells, long-lived and latently infected cells.(Perelson et al., 1997). Instead we examined a pooled productively infected cell population. The virus compartment was not further decomposed into infectious virions and non-infectious virions as in the paper by Perelson *et al*. (1996). Thus, different mechanisms of RTI and PI drug effects were not modeled. In fact, we only considered PI drug effects in the drug efficacy model (3) since the RTI drugs have a different adherence-resistance relationship. Further studies will be conducted in considering both PI and RTI drug effects in the models. Secondly, the availability of IC50 data was limited to baseline and failure time, as is typical in clinical trials. Thus, we extrapolated the IC50 data linearly to the whole treatment period in our modeling. The linear extrapolation is the best approximation that we can get from the sparse *IC*50 data (Wu et al., 2005). The linear assumption might have some influence on the estimation results since the *IC*50 might have jumped to a higher level earlier before the failure time when we obtained the sample for drug resistance test. However, we expect that this assumption had little effect on the prediction of virologic response since we had relatively frequent monitoring (monthly in the later stage) of virologic failure in this study. Thirdly, a more complete model of antiretroviral treatment efficacy would ideally also consider the time-varying function of concentrations of drug in plasma (Huang et al., 2003). Unfortunately, the limited availability of drug concentration data prohibited our inclusion of PK parameters in our model. Lastly, as measurements of adherence may not reflect actual adherence profiles for individual patients, the data quality would affect our estimation results for viral dynamic parameters. For example, adherence data measured by questionnaires may not be accurate. More accurate measurements of the MEMS adherence data were used in this paper and it was found that the MEMS adherence data can provide a better prediction of virologic response compared with the questionnaire adherence data, when the MEMS data are summarized in an appropriate way. Further studies on these issues are definitely needed. Nevertheless, these limitations would not offset the major findings from our modeling approach, although further improvement may be warranted.

In summary, MEMS adherence data may not be correlated better to virologic response compared to questionnaire adherence data unless the MEMS cap data are summarized in an appropriate way where adherence was assessed over 2 week interval measured from 1 or 2 weeks prior to a RNA measurement in our case. Our study also shows that the mechanismbased dynamic model is powerful and effective to establish a relationship of antiviral response with drug exposure and drug susceptibility.

#### **6. Acknowledgment**

The author would like to thank the ACTG398 study team for allowing me to use the clinical data from their study. This research was partially supported by NIAID/NIH grant AI080338 and MSP/NSA grant H98230-09-1-0053.

#### **7. References**

64 Antiviral Drugs – Aspects of Clinical Use and Recent Advances

and susceptibility into a function of treatment efficacy and (iv) to use a hierarchical Bayesian mixed-effects modeling approach that can not only combine prior information with current clinical data for estimating dynamic parameters, but also characterize inter-subject variability. Our modeling approach allows us to estimate time-varying antiretroviral efficacy during the entire course of a treatment regimen by incorporating the information of drug exposure and drug susceptibility. Thus, the results of estimated dynamic parameters based on this model should be more reliable and reasonable to interpret long-term HIV dynamics. Our models are simplified with the main goals of retaining crucial features of HIV dynamics and, at the same time, guaranteeing their applicability to typical clinical data,

We employed the proposed mechanism-based dynamic model to assess how to efficiently use adherence rates based on questionnaires and MEMS dosing events to predict virologic response. In particular, we intended to address the questions (i) how to summarize the MEMS adherence data for efficient prediction of virologic response, and (ii) which adherence assessment method, questionnaire or MEMS, is a more efficient predictor of virologic response after accounting for potential confounding factors such as drug resistance

For the MEMS data, we found that the best summary metric for prediction of virologic response in terms of model fitting residuals (prediction error) is the adherence rate determined from MEMS dosing events averaged over a 2 week assessment interval, 1 week or 2 weeks prior to the next measured RNA observation (denoted by IM1.2 or IM2.2). The model fitting residuals from both models (IM2.2 and IM1.2) are significantly smaller than any other 13

The model which used all available MEMS data between study visits to determine the adherence rate (the standard analysis) did not perform significantly better in terms of prediction of virologic response compared to the model with questionnaire adherence data

We also compared the model fittings with all possible combinations of IC50 and the four determinants of adherence data (see Fig. 4). The results indicate that (i) the control model was significantly worse than those based on all other 9 models (*p*≤0.001 0.007); (ii) the models IM1.2 and IM2.2 were significantly better than the eight other models (*p*≤0.001 0.048); (iii) the models I, A, M, M1.2, M2.2, IA and IM do not provide significantly different results (*p*=0.048 0.858) except for two marginally significant results. In particular, the models IA and IM did not improve upon the model I, which indicates that adherence measured by questionnaire and MEMS dosing events averaged over study visit interval did not provide any additional information to drug susceptibility in predicting virologic response. However, the models IM1.2 and IM2.2 did outperform the models I, M1.2 and M2.2, which indicates that the combination of drug susceptibility and adherence assessed over 2 week interval measured from 1 or 2 weeks prior to a RNA measurement provided significant additional information compared to either drug susceptibility or adherence alone

Although the analysis presented here used a simplified model, which appeared to perform well in capturing and explaining the observed patterns, and characterizing the biological mechanisms of HIV infection under relatively complex clinical situations, some limitations

models (*p*≤0.001 0.048), but they were not significantly different each other (*p*=0.590).

in particular, long-term viral load measurements.

between subjects.

(*p*=0.106).

in predicting virologic response.

Arnsten, J.; Demas, P. & Gourvetch, M.; et al. (February, 2001a). Adherence and viral load in HIV infected drug users: Comparison of self-report and medication event monitors (MEMS). 8th CROI. Chicago, USA

Modeling Virologic Response in HIV-1 Infected Patients to Assess Medication Adherence 67

Huang, Y.; Rosenkranz, S. L. & Wu, H. (2003). Modeling HIV dynamics and antiviral

Huang, Y.; Liu, D. & Wu, H. (2006). Hierarchical Bayesian methods for estimation of parameters in a longitudinal HIV dynamic system. *Biometrics*, Vol.62, pp.413-423 Huang, Y., Wu, H. & Acosta, P. E. (2010). Hierarchical Bayesian inference for HIV dynamic

Ickovics, J. R. & Meisler, A. W. (1997). Adherence in AIDS clinical trial: a framework for

Knafl, G. J.; Fennie, K. P.; Bova, C.; et al. (2004). Electronic monitoring device event

Levine, A. J.; Hinkin, C. H.; Marion, S.; et al. (2006). Adherence to antiretroviral medications

Llabre, M. M.; Weaver, K. E.; Duran, R. E.; et al.(2006). A measurement model of medication

Liu, H.; Golin, C. E.; Miller, L. G.; et al. (2001). A comparison study of multiple measures of adherence to HIV protease inhibitors. *Ann Intern Med*. , Vol.134, pp.968-77 Liu, H.; Miller, L. G.; Hays, R. D.; et al. (2006). Repeated measures longitudinal analyses of

Markowitz, M.; Louie, M.; Hurley, A.; et al. (2003). A novel antiviral intervention results in

dynamics and T-cell decay in vivo. *Journal of Virology*, Vol.77, pp.5037–5038 Molla, A.; Korneyeva, M.; Gao, Q.; et al. (1996). Ordered accumulation of mutations in HIV

Nowak, M. A. & May, R.M. (Eds). (2000). *Virus Dynamics: Mathematical Principles of* 

Notermans, D. W.; Goudsmit, J.; Danner, S.A.; et al. (1998). Rate of HIV-1 decline following

Palella, F. J.; Delaney, K. M; Moorman, A. C.; et al. (1998). Declining morbidity and mortality

*England Journal of Medicine*, Vol.338, No.13, pp. 853-860, ISSN 0028-4793 Paterson, D. L.; Swindells, S.; Mohr, J.; et al. (2000). Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. *J Intern Med.*, Vol.133, pp.21-30

protease confers resistance to ritonavir. *Nat Med*., Vol.2, pp.760-766

*Immunology and Virology*. Oxford University Press, Oxford

*Immune Defic Syndr*., Vol.12, pp.1483-1490

HIV-positive adults. *AIDS Patient Care STDS*., Vol.20, pp.701-711

clinical research and clinical care. *J Clin Epidemiol.*, Vol., pp.385–391 Kastrissios, H.; Suarez, J. R.; Katzenstein, D.; et al. (1998). Characterizing patterns of drug-

adherence. *Math. Biosci*., Vol.184, pp.165–86

*Journal,* Vol.52, No.4, pp.470–486

*Statistics in Medicine*, Vol.23, pp.783-801

*Health Psychol*. , Vol.25, pp.329-35

*in Medicine*, Vol.26, pp.991-1007

Vol.12, pp.2295–2303

responses with consideration of time-varying drug exposures, sensitivities and

differential equation models incorporating multiple treatment factors. *Biometrical* 

taking behavior with a multiple drug regimen in an AIDS clinical trial. *AIDS*,

modeling on an individual-subject basis using adaptive Poisson regression.

in HIV: differences in data collected via self-report and electronic monitoring.

adherence to highly active antiretroviral therapy and its relation to viral load in

HIV virologic response as a function of percent adherence, dose timing, genotypic sensitivity, and other factors. *J Acquir Immune Defic Syndr*. , Vol.41, pp.315-322 Liu, H.; Miller, L. G.; Golin, C.E.; et al. (2007). Repeated measures analyses of dose timing of

antiretroviral medication and its relationship to HIV virologic outcomes. *Statistics* 

more accurate assessment of human immunodeficiency virus type 1 replication

antiretroviral therapy is related to viral load at baseline and drug regimen. *J Acquir* 

among patients with advanced human immunodeficiency virus infection. *The New* 


Arnsten, J. H.; Demas, P. A.; Farzadegan, H.; et al. (2001b). Antiretroviral therapy adherence

Bangsberg, D. R. (2006). Less than 95% adherence to nonnucleoside reverse-transcriptase

Berg, K. M. & Arnsten, J. H. (2006). Practical and conceptual challenges in measuring antiretroviral adherence. *J Acquir Immune Defic Syndr*., Vol.43, pp.S79-S87 Bova, C. A.; Fennie, K. P.; Knafl, G. J.; et al. (2005). Use of electronic monitoring devices to

Callaway, D. S. & Perelson, A. S. (2002). HIV-1 infection and low steady state viral loads.

Centers for Disease Control and Prevention (CDC). (2009). *HIV/AIDS Surveillance Report*,

Davidian, M. & Giltinan, D. M. (Eds). (1005). *Nonlinear Models for Repeated Measurement Data*.

Ding, A. A. & Wu, H. (2001). Assessing antiviral potency of anti-HIV therapies *in vivo* by comparing viral decay rates in viral dynamic models. *Biostatistics*. Vol.2, pp.13-29 Dixit, N. M.; Markowitz, M.; Ho DD; et al. (2004). Estimates of intracellular delay and

Fennie, K. P.; Bova, C. A. & Williams, A. B. (2006). Adjusting and censoring electronic

Fletcher, C. V.; Testa, M. A.; Brundage, R. C.; et al. (2005). Four measures of antiretroviral

Gamerman, D. (1997). *Markov Chain Monte Carlo: Stochastic Simulation for Bayesian Inference*.

Hammer, S. M.; Vaida, F.; Bennett, K. K.; et al. (2002). Dual vs single protease inhibitor

Han, C.; Chaloner, K. & Perelson, A. S. (2002). Bayesian analysis of a population HIV

Haubrich, R. H.; Little, S. J.; Currier, J. S.; et al. (1999). The value of patient-reported

Ho, D. D.; Neumann, A. U.; Perelson, A. S.; et al. (1995). Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 infection. *Nature*, Vol.373, pp.123–126

antiretroviral therapy. *Antiviral Therapy*. Vol.9, pp.237-246

359. *J Acquir Immune Defic Syndr*. , Vol.40, pp.301-306

average drug efficacy from viral load data of HIV-infected individuals under

monitoring device data. Implications for study outcomes. *J Acquir Immune Defic* 

medication adherence and virologic response in AIDS clinical trials group study

therapy following antiretroviral treatment failure: a randomized trial. *JAMA*,

dynamic model. *Case Studies in Bayesian Statistics, Vol. 6*. Springer-Verlag, New

adherence to antiretroviral therapy in predicting virologic and immunologic

electronic monitoring. *Clinical Infectious Diseases*, Vol.33, pp.1417-1423 Bangsberg, D. R.; Hecht, F. M.; Charlebois, E. D.; et al. (March, 2000). Adherence to protease

population. *AIDS,* Vol.14, No.4, pp.357-366

pp.939–941

Vol.9, pp.103-110

*Bull. Math. Biol*., Vol.64, pp.29–64

pp.1-44, Atlanta, GA, USA

Chapman *&* Hall, London

*Syndr.*, Vol.43, pp.S88-S95

Chapman *&* Hall, London

response. *AIDS*, Vol.13, pp.1099-1107

Vol.288, pp.169–180

York

and viral suppression in HIV-infected drug users: comparison of self-report and

inhibitors, HIV-1 viral load, and development of drug resistance in an indigent

inhibitor therapy can lead to viral suppression. *Clinical Infectious Diseases*, Vol.43,

measure antiretroviral adherence: practical considerations. *AIDS and Behavior*,


**Part 2** 

**Developing New Antivirals** 

