**Results on objective 3:**

### **Practical guidelines for the safe and effective long-term vancomycin administration**

The financial and health burden for managing ARF due to Van-AKI is currently unknown and clearly cannot be determined by our retrospective studies, especially with such a small series and observed over such a short time window. Furthermore, we have no good information on the true incidence of Van-AKI. However, it suffices to note that for our first 5 patients, it took between 20 and 70 days of inpatient care and management (a mean of 35 ± 9.6 days after the diagnosis of ARF) before regaining partial renal function for discharge to outpatient follow-up. The clinical impact of Van-AKI was also substantial to affected patients since we found significant residual losses of renal function ~ 37 % (by CrCl) even 31 days after diagnosing ARF or after stopping the drug, using the lowest serum creatinine in the recovery. For these two and other additional reasons, the prevention and amelioration of Van-AKI should be a top priority. Drawing upon our own experience and analyses plus that published in the literature, we would like to turn the lessons and insights (from results on Objective II) into the following guidelines.


The financial and health burden for managing ARF due to Van-AKI is currently unknown and clearly cannot be determined by our retrospective studies, especially with such a small series and observed over such a short time window. Furthermore, we have no good information on the true incidence of Van-AKI. However, it suffices to note that for our first 5 patients, it took between 20 and 70 days of inpatient care and management (a mean of 35 ± 9.6 days after the diagnosis of ARF) before regaining partial renal function for discharge to outpatient follow-up. The clinical impact of Van-AKI was also substantial to affected patients since we found significant residual losses of renal function ~ 37 % (by CrCl) even 31 days after diagnosing ARF or after stopping the drug, using the lowest serum creatinine in the recovery. For these two and other additional reasons, the prevention and amelioration of Van-AKI should be a top priority. Drawing upon our own experience and analyses plus that published in the literature, we would like to turn the lessons and insights (from results

1. We propose that vancomycin be viewed as nephrotoxic till proven otherwise, just like aminoglycosides, cis-platinum, amphotericin B, and radio-contrast dyes. If there are no compelling indications, as was the case for 4 of our 6 patients (cases 1, 2, 3, and 5; Table 6 A), vancomycin should not be used, in deference to other safer suitable alternatives

2. If it must be used, the index of suspicion for Van-AKI should be high because even the slightest degree of renal injury (generally undetected by the meager increase in serum creatinine from its normal baseline) will impair excretion, predispose to drug accumulation and excess levels, which in turn inflicts more tissue damage and further compromises elimination, setting up a viscous cycle. Such a rapid buildup of vancomycin with steeply rising serum creatinine was amply illustrated by our patients 4 and 6, the former precipitously over 48 h and the latter in 13 days, but both were mediated by the same mechanism. There had been case reports of similarly steep functional decline

3. Drug levels and serial renal function should be closely monitored continually throughout treatment, daily the first week, preferably thrice weekly but no fewer than twice weekly thereafter. For example, three of our 6 patients did not have drug levels and/or creatinine measured for 8-21 days immediately prior to their ARF and/or development of excessive vancomycin levels (cases 2, 5 and 6). In at least six subsets of patients who are particularly vulnerable to AKI, these preventive measures should be

a. Those treated in the outpatient setting, nursing homes, or long-term care facilities where physician involvement and supervision are inherently minimal, indirect, and less than immediate, like our patients 2, 5 and 6. Lab results must be received, reviewed and acted upon in a timely fashion (within 23 h if dosed daily or within 11 h if dosed every 12 h) by professionals trained to monitor for nephrotoxicity and supervised by physicians experienced in this issue. Timely dose adjustments or stoppage must be

b. Critically ill and complicated patients, like those in the ICU, who are at increased risks for ARF due to other potential nephrotoxic insults or hemodynamic instability (Colomo

caused by sharply increasing vancomycin levels (Shah-Khan et al, 2011).

**Practical guidelines for the safe and effective long-term vancomycin administration** 

**Results on objective 3:** 

on Objective II) into the following guidelines.

(see suggestions under Discussion).

feasible and reliable to prevent AKI.

mandatory.

et al (2010)).

f. ARF or those with fluctuating serum creatinine.

Pre-emptive renal consultation at the earliest signs of potential nephrotoxicity may prevent costly AKI and y hospitalization.


In cases 1, 4, and 6, for instance, toxic trough vancomycin levels were created and found because of the rapid escalation without attaining a relative steady state at each incremental step. Our experience amply confirmed the virtually identical experience reported in the one case by Barraclough et al in 2007. We would support and re-emphasize their caution regarding the need not only to monitor very tightly and frequently during rapid dose escalation but also ordering one dose at a time.

By routinely refraining from a multi-day scheduled order, physicians could use the latest serum creatinine and vancomycin data to adjust the next dose to avoid further damage which otherwise could easily happen with a standing order. The danger of the latter approach was statistically and pictorially shown in our six patients by the delayed stoppage despite theoretically prior knowledge of vancomycin levels of 70 mg/L and creatinine elevation to 4-fold of the baseline 2 days earlier (Fig 7, Tables 6 B & C). The rationale and justification are identical to writing daily coumadin orders in similar non-steady states like dose titration or escalation. The trade off for the inconvenience and extra though manageable work load would be a far lower incidence of AKI [(and perhaps fewer cases of chronic kidney disease (CKD)] and reduced health care expenses.


We urge serious considerations for an immediate stoppage if any 2 of these 3 criteria are present, at least temporarily withholding vancomycin until additional tests show stable or improved renal function and significant decline in vancomycin levels.

6. We would re-emphasize the observation from classical renal physiology that serum creatinine is a very insensitive index of renal function in terms of detecting early decline in GFR. It is also grossly inaccurate in quantifying the loss of renal function, especially when the absolute values are below 1.5 mg/dl or when the changes occur between 0.5 and 2 mg/dl. Sole reliance on the increases of serum creatinine or the absolute values as

Vancomycin-Induced Nephrotoxicity 213

On scientific ground, we would discourage if not deplore the practice of "random" vancomycin levels. It condones uncertainties and fosters the culture and attitude of making and accepting subjective arbitrary interpretations. We would therefore endorse getting only a true trough level like 10-12 h (after the last dose if given at q 12 h frequency), or 24- or 48-

Parenthetically, though with undefined clinical impact, the AUC per unit time is smaller (thus the nephrotoxic risk lower) if dosed once q 12 h versus dosing q 24 or q 48 h even when the trough levels are identical, say, at 15 mg/L for all three regimens. This is because AUC (or the total drug exposure by time and concentration) has been shown to play a role in Van-AKI (Bailie et al 1988). We would therefore favor and recommend the q 12 h (or at the longest, < q 24 h) dosing schedule over the q 48 h regimen and accordingly suggest measuring the 12-h trough levels unless logistically impossible. If q 24-h dosing is necessary, experience has shown comparable safety compared to q 12 h dosing if the 24-h rough levels

8. We propose changing our default mode of ordering vancomycin "*to give* the *next dose only if the trough level falls below the therapeutic target*", as opposed to the current default mode of "*keep giving to sustain the trough level above the target range*". In practice, presently most physicians would re-dose even if the trough level was as high as 20-25 mg/L (or even 30), for fear that if we withhold, the level might drop precipitously below the therapeutic range regardless of the prevailing serum creatinine. Consequently, the actual trough

Our recommendation of a conservative dosing is based on two considerations. First, by definition all levels prior to the trough would have exceeded 15-20 mg/L, which were shown to pose greater risks for nephrotoxicity (Hidayat et al (2006); Pritchard et al (2008); our series of six patients). Second, there is no published evidence that trough levels of 10-15 mg/L, for example, are necessarily associated with poorer clinical cure or response than levels of 15-20 mg/L if the MIC against a "sensitive" MRSA is supposed to be < 1 mg/L or

9. Although the initial loading dose of vancomycin (typically 15 mg/kg) is the same regardless of the level of renal function, the maintenance dose must be reduced in preexisting renal insufficiency, newly developed ARF, and/or deteriorating function. This basic safety principle was forgotten or ignored in virtually all the reported cases including our 6 patients. We recommend using the nomogram (15 mg x GFR in ml/min) for daily maintenance dose (in mg per day) first suggested by Moellering et al (1981) for renal impairment. This rough guideline has stood the test of time and provides a good though crude first approximation, allowing us to make later and

levels are always substantially if not markedly higher than 20-25 mg/L.

at the worst < 2 mg/L (Hermsen, Hanson et al. 2010; Chan, Pham et al. 2011).

continual adjustments based on subsequent trough levels.

h troughs (if dosed at 24 to 48 h frequency for whatever reasons).

were kept below 10 mg/L (Cohen, Dadashev et al. 2002).

very difficult.

details of the last administration or verifying this assumption. Thus frequently if not invariably, high or "toxic" values are simply attributed to sampling within a few hours of the infusion when in fact they may actually be a 10-12 h trough level. A high peak value (if indeed verified to be peak) may not necessarily require dose reduction or discontinuation although even this assumption may not be correct or safe. But a true trough but high level should mandate immediate consideration of stopping vancomycin (or at least until nephrotoxicity is excluded). Reconstruction of the timing of a "random" level relative to the last infused dose is tedious, time-consuming and prohibitive. They render making sound clinical decisions on proper dosage adjustments

indicator of ARF will delay detection and recognition of AKI. Decline in GFR is not linearly related to the rise in serum creatinine. An initial small rise in serum creatinine from a perfectly normal baseline actually represents a marked fall in GFR whereas a marked rise in advanced CKD represents only modest drops. Thus, even small increments from the normal should raise concerns of AKI, especially in the early phase. The 500 ml of IV fluids typically used to deliver the vancomycin q 12 h could easily mask a genuine increase in serum creatinine, making detection of AKI in the early phase even harder unless the index of suspicion is high.

The emaciated 40-kg patient described by Barraclough et al (2007) illustrated this point well because his baseline serum creatinine was only 0.3 mg/dl. Although it went up precipitously to 0.5 mg/dl by day 4 day (already 40% loss of GFR) and then to 1.4 mg/dl (already 79 % loss of GFR) by day 8 day of therapy and with a vancomycin level 66 mg/L, the standing order of 1 g twice daily was not reduced until day 9, when creatinine finally peaked at 1.9 mg/dl (84% loss of GFR).

We thus recommend using the reciprocal (times a convenient constant like 100) as a simple, reliable and accurate estimate of CrCl and its changes reflect *relative changes* in GFR for a given patient. This approach will enhance the sensitivity and detection of early kidney injury, at a time when timely and appropriate dosage reduction or cessation should be made to prevent further nephrotoxicity.

The contrast between using serum creatinine and using 100/serum creatinine is best illustrated in 4 of our patients (cases 1, 3, 4, and 6). Two of them (cases 1 & 3) lost 62 to 76 % of their CrCl or GFR in 1 day (from 132 to 43 ml/min in case 1 and from 106 to 39 ml/min in case 3) if renal function is evaluated by using CrCl (Fig 1 B and Fig 3 B). In contrast, superficially there appeared to be quite "minimal" or "manageable" loss by serum creatinine over the same one day [(0.8 to 2.3 mg/dl in case 1 (Fig 1A) and 0.9 to 2.6 mg/dl in case 3 (Fig 3A)]. Similarly, case 4 lost 62 % of the GFR in 2 days when judged by CrCl (Fig 4 B), contrary to the "modest" rise in serum creatinine from 0.9 to 2.6 mg/dl (Fig 4 A). Likewise, patient # 6 suffered 36% loss of CrCl in 2 days (Fig 6B) as the corresponding serum creatinine went up by a "meager" delta of 0.5 mg/dl (from 0.8 to 1.3; Fig 6A) over the same 2 days.

We therefore recommend quantifying relative GFR loss by the decrements in CrCl, as estimated by 100 /serum creatinine. Specifically, we suggest that 20-30 % drop in GFR estimated by this serum creatinine reciprocal method would provide a better and earlier warning signal for possible nephrotoxicity than the thresholds of doubling of serum creatinine or values ≥ 1.5 mg/dl. As recommended later, a renal consult can be requested to assist with such a less conventional approach of evaluating GFR.

7. As shown in our patients individually and collectively, the current practice of ordering and measuring "random" vancomycin levels, without regard to the timing of the last administered dose, will continue to confuse and confound us. Random levels are essentially un-interpretable, often misleading and unreliable, generally inaccurate as an index of the area under the curve (AUC) relating drug concentration against time, and at times simply useless if not dangerous. For instance, to the best of our knowledge there is no published "normal" range to define what to expect for levels between 4 to 8 hours post-dosing. Such grey-zone times create unnecessary ambiguity and obligate extrapolation and speculation. In addition, there is a general tendency (and thus a common problem) for busy clinicians working as a team to assume a high value or "toxic levels" as "peak" previously ordered by a colleague without always checking

The emaciated 40-kg patient described by Barraclough et al (2007) illustrated this point well because his baseline serum creatinine was only 0.3 mg/dl. Although it went up precipitously to 0.5 mg/dl by day 4 day (already 40% loss of GFR) and then to 1.4 mg/dl (already 79 % loss of GFR) by day 8 day of therapy and with a vancomycin level 66 mg/L, the standing order of 1 g twice daily was not reduced until day 9, when creatinine finally

We thus recommend using the reciprocal (times a convenient constant like 100) as a simple, reliable and accurate estimate of CrCl and its changes reflect *relative changes* in GFR for a given patient. This approach will enhance the sensitivity and detection of early kidney injury, at a time when timely and appropriate dosage reduction or cessation should be made

The contrast between using serum creatinine and using 100/serum creatinine is best illustrated in 4 of our patients (cases 1, 3, 4, and 6). Two of them (cases 1 & 3) lost 62 to 76 % of their CrCl or GFR in 1 day (from 132 to 43 ml/min in case 1 and from 106 to 39 ml/min in case 3) if renal function is evaluated by using CrCl (Fig 1 B and Fig 3 B). In contrast, superficially there appeared to be quite "minimal" or "manageable" loss by serum creatinine over the same one day [(0.8 to 2.3 mg/dl in case 1 (Fig 1A) and 0.9 to 2.6 mg/dl in case 3 (Fig 3A)]. Similarly, case 4 lost 62 % of the GFR in 2 days when judged by CrCl (Fig 4 B), contrary to the "modest" rise in serum creatinine from 0.9 to 2.6 mg/dl (Fig 4 A). Likewise, patient # 6 suffered 36% loss of CrCl in 2 days (Fig 6B) as the corresponding serum creatinine went up by a "meager" delta of 0.5 mg/dl (from 0.8 to 1.3; Fig 6A) over the

We therefore recommend quantifying relative GFR loss by the decrements in CrCl, as estimated by 100 /serum creatinine. Specifically, we suggest that 20-30 % drop in GFR estimated by this serum creatinine reciprocal method would provide a better and earlier warning signal for possible nephrotoxicity than the thresholds of doubling of serum creatinine or values ≥ 1.5 mg/dl. As recommended later, a renal consult can be requested to

7. As shown in our patients individually and collectively, the current practice of ordering and measuring "random" vancomycin levels, without regard to the timing of the last administered dose, will continue to confuse and confound us. Random levels are essentially un-interpretable, often misleading and unreliable, generally inaccurate as an index of the area under the curve (AUC) relating drug concentration against time, and at times simply useless if not dangerous. For instance, to the best of our knowledge there is no published "normal" range to define what to expect for levels between 4 to 8 hours post-dosing. Such grey-zone times create unnecessary ambiguity and obligate extrapolation and speculation. In addition, there is a general tendency (and thus a common problem) for busy clinicians working as a team to assume a high value or "toxic levels" as "peak" previously ordered by a colleague without always checking

phase even harder unless the index of suspicion is high.

assist with such a less conventional approach of evaluating GFR.

peaked at 1.9 mg/dl (84% loss of GFR).

to prevent further nephrotoxicity.

same 2 days.

indicator of ARF will delay detection and recognition of AKI. Decline in GFR is not linearly related to the rise in serum creatinine. An initial small rise in serum creatinine from a perfectly normal baseline actually represents a marked fall in GFR whereas a marked rise in advanced CKD represents only modest drops. Thus, even small increments from the normal should raise concerns of AKI, especially in the early phase. The 500 ml of IV fluids typically used to deliver the vancomycin q 12 h could easily mask a genuine increase in serum creatinine, making detection of AKI in the early details of the last administration or verifying this assumption. Thus frequently if not invariably, high or "toxic" values are simply attributed to sampling within a few hours of the infusion when in fact they may actually be a 10-12 h trough level. A high peak value (if indeed verified to be peak) may not necessarily require dose reduction or discontinuation although even this assumption may not be correct or safe. But a true trough but high level should mandate immediate consideration of stopping vancomycin (or at least until nephrotoxicity is excluded). Reconstruction of the timing of a "random" level relative to the last infused dose is tedious, time-consuming and prohibitive. They render making sound clinical decisions on proper dosage adjustments very difficult.

On scientific ground, we would discourage if not deplore the practice of "random" vancomycin levels. It condones uncertainties and fosters the culture and attitude of making and accepting subjective arbitrary interpretations. We would therefore endorse getting only a true trough level like 10-12 h (after the last dose if given at q 12 h frequency), or 24- or 48 h troughs (if dosed at 24 to 48 h frequency for whatever reasons).

Parenthetically, though with undefined clinical impact, the AUC per unit time is smaller (thus the nephrotoxic risk lower) if dosed once q 12 h versus dosing q 24 or q 48 h even when the trough levels are identical, say, at 15 mg/L for all three regimens. This is because AUC (or the total drug exposure by time and concentration) has been shown to play a role in Van-AKI (Bailie et al 1988). We would therefore favor and recommend the q 12 h (or at the longest, < q 24 h) dosing schedule over the q 48 h regimen and accordingly suggest measuring the 12-h trough levels unless logistically impossible. If q 24-h dosing is necessary, experience has shown comparable safety compared to q 12 h dosing if the 24-h rough levels were kept below 10 mg/L (Cohen, Dadashev et al. 2002).

8. We propose changing our default mode of ordering vancomycin "*to give* the *next dose only if the trough level falls below the therapeutic target*", as opposed to the current default mode of "*keep giving to sustain the trough level above the target range*". In practice, presently most physicians would re-dose even if the trough level was as high as 20-25 mg/L (or even 30), for fear that if we withhold, the level might drop precipitously below the therapeutic range regardless of the prevailing serum creatinine. Consequently, the actual trough levels are always substantially if not markedly higher than 20-25 mg/L.

Our recommendation of a conservative dosing is based on two considerations. First, by definition all levels prior to the trough would have exceeded 15-20 mg/L, which were shown to pose greater risks for nephrotoxicity (Hidayat et al (2006); Pritchard et al (2008); our series of six patients). Second, there is no published evidence that trough levels of 10-15 mg/L, for example, are necessarily associated with poorer clinical cure or response than levels of 15-20 mg/L if the MIC against a "sensitive" MRSA is supposed to be < 1 mg/L or at the worst < 2 mg/L (Hermsen, Hanson et al. 2010; Chan, Pham et al. 2011).

9. Although the initial loading dose of vancomycin (typically 15 mg/kg) is the same regardless of the level of renal function, the maintenance dose must be reduced in preexisting renal insufficiency, newly developed ARF, and/or deteriorating function. This basic safety principle was forgotten or ignored in virtually all the reported cases including our 6 patients. We recommend using the nomogram (15 mg x GFR in ml/min) for daily maintenance dose (in mg per day) first suggested by Moellering et al (1981) for renal impairment. This rough guideline has stood the test of time and provides a good though crude first approximation, allowing us to make later and continual adjustments based on subsequent trough levels.

Vancomycin-Induced Nephrotoxicity 215

substantial body of indirect evidence to support the existence of Van-AKI, mainly based on the close correlations between increased blood levels and/or increased dosage on the one hand and increased incidence on the other hand (Rybak et al, 2009) (Table 1). Typically, there was observed a very low incidence of Van-AKI with low trough vancomycin levels like < 10 mg/L (Sorrel et al, 1985), but increased incidence with higher trough levels like >14 (Pritchard et al, 2008) or >15-20 mg/L (Hidayat et al, 2006), or with a high steady-state level > 28 mg/L (Ingram et al, 2008), and a 3-fold higher incidence

Additional support was provided by the observations of synergism in nephrotoxicity between vancomycin and aminoglycoside (Farber et al, 1983; Sorrel et al, 1985; Rybak et al, 1990; Goetz & Sayer, 1993), the increased risks of Van-AKI with prolonged administration (Goetz & Sayer, 1993; Hidayat et al, 2006; Pritchard et al 2008), and the enhanced risks of nephrotoxicity (Lodise et al, 2008) or poorer renal outcome with vancomycin (Rodriguez

The second line of evidence was obtained from the 2 dozen cases of ARF associated with vancomycin administration (Table 2). Many were somewhat equivocal in terms of a clear cut etiology for the ARF, especially when no vancomycin levels were given and/or other common etiologies had been or could be vigorously excluded. There however remained about half a dozen well documented and unambiguous cases of Van-AKI, as evidenced by toxic drug levels and the absence of any other contributing factors or confounding variables for the ARF (Frimat et al, 1995; Barraclough et al, 2007; Ladino et al, 2008 [2 of 5 convincing

The third and perhaps the strongest line of evidence is derived from our own experience, which includes 6 cases we have encountered and treated in the course of a month of renal consultation. There are probably two reasons for the relative ease with which these 6 patients with Van-AKI were discovered. One is the changing microbiology and characteristics of modern era patients and our obligated responses to these changes and adoption of current dosing practices. Two is the unique patient cohorts treated with vancomycin nowadays compared to the invariably septic or bacteremic patients with shock

First, there has been an apparent increase in the incidence of AKI during vancomycin therapy, largely due to three factors. One, the incidence of infections by documented MRSA and MRSE is growing rapidly. Two, there is an exponential increase in the use of vancomycin not only for sensitive and documented pathogens, but also for HCAP and osteomyelitis (especially in diabetics) in whom MRSA must be considered and/or covered, typically by vancomycin. After all, it is inexpensive, time-honored, tried, true, and proven to be effective against MSRA, the most prevalent and the deadliest bacteria. Three due to the widespread use (if not abuse) of vancomycin, there is a steady emergence of organisms sensitive only to rather high MIC, leading to the ID recommendation of trough levels > 15- 20 mg/L (Rybak, Lomaestro et al. 2009). These three factors have combined to contribute to

Second, as opposed to the older cases where sepsis, bacteremia, hemodynamic instability, concurrently administered aminoglycosides, amphotericin B or contrast dyes could not be definitely excluded as etiologic factors for the ARF, none of these risk factors could have contributed to the AKI in our 6 patients (3 with HCAP and 3 with osteomyelitis, and none bacteremic or hypotensive) (Table 6 A). By providing and correlating serial vancomycin levels before, during, and after the ARF with the corresponding changes in renal function

Colomo et al, 2010) compared to linezolid in treating similar patient cohorts.

and pancytopenia in earlier decades. We shall elaborate on these two points.

when daily dose >4 g (Lodise et al, 2008).

cases]; Shah-Khan et al, 2011; Table 2).

a significant upsurge of Van-AKI in our view.

In practice, if serum creatinine is relatively stable, GFR can be estimated by the equation of Cockcroft-Gault for CrCl (in ml/min) [= (140 – age in years) x (lean body mass in kg) / (serum creatinine in mg/dl x 72)] (Cockcroft and Gault (1976)). For instance, the maintenance dose will be ~ 1.5 g /d (=15 mg/d x 100) for a CrCl of 100 ml/min. Likewise, it will be ~ 450 mg/d (=15 mg/d x 30) for a CrCl of 30 ml/min. We should note that even with a steady state creatinine, this equation is known to over-estimate CrCl in the (a) elderly, (b) emaciated, (c) edematous, (d) obese, and (e) paralysis or amputees. An even smaller dose must be considered in these situations.

For patients with changing serum creatinine, it is advisable not only to measure creatinine and vancomycin more frequently due to the non-steady state, but also obtain renal consultation. These patients are at increased risks created by the predictable positive feedback loop between falling GFR (as denoted by steadily rising serum creatinine) and increasing kidney vancomycin exposure (as reflected by rising vancomycin levels). For patients functionally anuric or anephric, 2 mg/kg/d is a reasonable initial dose. In these patients and those with established end-stage renal disease or dialysis dependency, nephrology should be consulted even though they fall outside of the scope of cohorts to be considered in this Chapter (Van-AKI).

10. Although unproven by randomized controlled trial, there are theoretical reasons and some anecdotal evidence to support the consideration of prompt and significant removal of vancomycin by hemodialysis in patients with Van-AKI and burdened with sustained toxic levels and severe renal failure. We would therefore recommend earliest possible referral to nephrology for assistance and support for such a therapeutic option.

Though without personal or literature data to address this issue, we would submit that it is an unresolved theory as to the scientific basis and/or the clinical superiority of targeting trough vancomycin levels between 15 and 20 mg/L for those MRSA with MIC > 1 but < 2 mg/L (Hermsen, Hanson et al. 2010; Chan, Pham et al. 2011).

We would urge exercising circumspection in accepting this recommendation and showing discretion and flexibility in applying the same if the goal is to achieve the bacterial killing without renal toxicity.

#### **5. Discussion**

For nearly half a century, vancomycin has been used successfully to treat infections caused by gram positive bacteria, notably MRSA, from various sources and in various organs. The issue of Van-AKI has been controversial due to the difficulty in establishing a cause-and-effect relationship between vancomycin and the alleged ARF. This is true among the affected patients reported in large epidemiologic surveys or drug toxicity monitoring studies because they generally provide little details on individual patients for an objective review or independent determination (Table 1). Similarly, among the two dozen or so reported cases of Van-AKI (Table 2), fewer than 10 had unequivocally excluded the usual confounding variables like sepsis, bacteremia, hemodynamic factors and concurrent nephrotoxins. Many also failed to provide serial vancomycin levels to show the temporal evolution with the ARF. Thus, to date, the existence of Van-AKI has been intensely debated and at times categorically dismissed. Our first objective was to more firmly establish this clinical entity by performing a

vigorous and comprehensive review of the existing literature and by reporting our own experience. We have obtained and presented three lines of evidence to argue for the entity of Van-AKI. First, the drug toxicity monitoring studies in the aggregate have offered a

In practice, if serum creatinine is relatively stable, GFR can be estimated by the equation of Cockcroft-Gault for CrCl (in ml/min) [= (140 – age in years) x (lean body mass in kg) / (serum creatinine in mg/dl x 72)] (Cockcroft and Gault (1976)). For instance, the maintenance dose will be ~ 1.5 g /d (=15 mg/d x 100) for a CrCl of 100 ml/min. Likewise, it will be ~ 450 mg/d (=15 mg/d x 30) for a CrCl of 30 ml/min. We should note that even with a steady state creatinine, this equation is known to over-estimate CrCl in the (a) elderly, (b) emaciated, (c) edematous, (d) obese, and (e) paralysis or amputees. An even smaller dose

For patients with changing serum creatinine, it is advisable not only to measure creatinine and vancomycin more frequently due to the non-steady state, but also obtain renal consultation. These patients are at increased risks created by the predictable positive feedback loop between falling GFR (as denoted by steadily rising serum creatinine) and increasing kidney vancomycin exposure (as reflected by rising vancomycin levels). For patients functionally anuric or anephric, 2 mg/kg/d is a reasonable initial dose. In these patients and those with established end-stage renal disease or dialysis dependency, nephrology should be consulted even though they fall outside of the scope of cohorts to be

10. Although unproven by randomized controlled trial, there are theoretical reasons and some anecdotal evidence to support the consideration of prompt and significant removal of vancomycin by hemodialysis in patients with Van-AKI and burdened with sustained toxic levels and severe renal failure. We would therefore recommend earliest possible referral to nephrology for assistance and support for such a therapeutic option. Though without personal or literature data to address this issue, we would submit that it is an unresolved theory as to the scientific basis and/or the clinical superiority of targeting trough vancomycin levels between 15 and 20 mg/L for those MRSA with MIC > 1 but < 2

We would urge exercising circumspection in accepting this recommendation and showing discretion and flexibility in applying the same if the goal is to achieve the bacterial killing

For nearly half a century, vancomycin has been used successfully to treat infections caused by gram positive bacteria, notably MRSA, from various sources and in various organs. The issue of Van-AKI has been controversial due to the difficulty in establishing a cause-and-effect relationship between vancomycin and the alleged ARF. This is true among the affected patients reported in large epidemiologic surveys or drug toxicity monitoring studies because they generally provide little details on individual patients for an objective review or independent determination (Table 1). Similarly, among the two dozen or so reported cases of Van-AKI (Table 2), fewer than 10 had unequivocally excluded the usual confounding variables like sepsis, bacteremia, hemodynamic factors and concurrent nephrotoxins. Many also failed to provide serial vancomycin levels to show the temporal evolution with the ARF. Thus, to date, the existence of Van-AKI has been intensely debated and at times categorically dismissed. Our first objective was to more firmly establish this clinical entity by performing a vigorous and comprehensive review of the existing literature and by reporting our own experience. We have obtained and presented three lines of evidence to argue for the entity of Van-AKI. First, the drug toxicity monitoring studies in the aggregate have offered a

must be considered in these situations.

considered in this Chapter (Van-AKI).

without renal toxicity.

**5. Discussion** 

mg/L (Hermsen, Hanson et al. 2010; Chan, Pham et al. 2011).

substantial body of indirect evidence to support the existence of Van-AKI, mainly based on the close correlations between increased blood levels and/or increased dosage on the one hand and increased incidence on the other hand (Rybak et al, 2009) (Table 1). Typically, there was observed a very low incidence of Van-AKI with low trough vancomycin levels like < 10 mg/L (Sorrel et al, 1985), but increased incidence with higher trough levels like >14 (Pritchard et al, 2008) or >15-20 mg/L (Hidayat et al, 2006), or with a high steady-state level > 28 mg/L (Ingram et al, 2008), and a 3-fold higher incidence when daily dose >4 g (Lodise et al, 2008).

Additional support was provided by the observations of synergism in nephrotoxicity between vancomycin and aminoglycoside (Farber et al, 1983; Sorrel et al, 1985; Rybak et al, 1990; Goetz & Sayer, 1993), the increased risks of Van-AKI with prolonged administration (Goetz & Sayer, 1993; Hidayat et al, 2006; Pritchard et al 2008), and the enhanced risks of nephrotoxicity (Lodise et al, 2008) or poorer renal outcome with vancomycin (Rodriguez Colomo et al, 2010) compared to linezolid in treating similar patient cohorts.

The second line of evidence was obtained from the 2 dozen cases of ARF associated with vancomycin administration (Table 2). Many were somewhat equivocal in terms of a clear cut etiology for the ARF, especially when no vancomycin levels were given and/or other common etiologies had been or could be vigorously excluded. There however remained about half a dozen well documented and unambiguous cases of Van-AKI, as evidenced by toxic drug levels and the absence of any other contributing factors or confounding variables for the ARF (Frimat et al, 1995; Barraclough et al, 2007; Ladino et al, 2008 [2 of 5 convincing cases]; Shah-Khan et al, 2011; Table 2).

The third and perhaps the strongest line of evidence is derived from our own experience, which includes 6 cases we have encountered and treated in the course of a month of renal consultation. There are probably two reasons for the relative ease with which these 6 patients with Van-AKI were discovered. One is the changing microbiology and characteristics of modern era patients and our obligated responses to these changes and adoption of current dosing practices. Two is the unique patient cohorts treated with vancomycin nowadays compared to the invariably septic or bacteremic patients with shock and pancytopenia in earlier decades. We shall elaborate on these two points.

First, there has been an apparent increase in the incidence of AKI during vancomycin therapy, largely due to three factors. One, the incidence of infections by documented MRSA and MRSE is growing rapidly. Two, there is an exponential increase in the use of vancomycin not only for sensitive and documented pathogens, but also for HCAP and osteomyelitis (especially in diabetics) in whom MRSA must be considered and/or covered, typically by vancomycin. After all, it is inexpensive, time-honored, tried, true, and proven to be effective against MSRA, the most prevalent and the deadliest bacteria. Three due to the widespread use (if not abuse) of vancomycin, there is a steady emergence of organisms sensitive only to rather high MIC, leading to the ID recommendation of trough levels > 15- 20 mg/L (Rybak, Lomaestro et al. 2009). These three factors have combined to contribute to a significant upsurge of Van-AKI in our view.

Second, as opposed to the older cases where sepsis, bacteremia, hemodynamic instability, concurrently administered aminoglycosides, amphotericin B or contrast dyes could not be definitely excluded as etiologic factors for the ARF, none of these risk factors could have contributed to the AKI in our 6 patients (3 with HCAP and 3 with osteomyelitis, and none bacteremic or hypotensive) (Table 6 A). By providing and correlating serial vancomycin levels before, during, and after the ARF with the corresponding changes in renal function

Vancomycin-Induced Nephrotoxicity 217

**Characteristics (N= 6 patients) (Mean ± SE) (Units)** 

**Age (years) 55.7 ± 6.0 years Body weights 85.5 ± 2.8 kg Pre-existing chronic kidney disease N=2 (33 %) % Hypertension N=3 (50 %) % Diabetes mellitus N=3 (50 %) % Congestive heart failure or coronary disease 0 % History of liver disease or hepatic dysfunction N=2 (33 %) % Signs of volume depletion N=1 (16 %) % Positive blood or urine cultures 0 %** 

**(but both temporally unrelated to ARF) N=2 (33 %) % Indication for vancomycin: - Pneumonia N=3 (50%) (1 MRSA) Osteomyelitis N=3 (50%) (1 MRSA)** 

**shock 0 % Fever (temperature > 38 degrees C or 100.5 F) N=2 (33%) % Baseline WBC 10.6 ± 1.9 v Baseline absolute neutrophils 8.6 ± 1.8 K/mm3 Baseline absolute eosinophils 74 ± 32 per mm3 Urine eosinophils 0 % Ultrasound evidence for obstructive uropathy 0 %** 

**of vancomycin in the AKI 85 ± 1 %** 

We therefore attempted to identify independent common risk factors resulting in Van-AKI. In this pursuit, we have confirmed but extended the three previously reported risk factors for Van-AKI (a) High blood vancomycin levels (Rybak et al, 1990; Hidayat et al, 2006; Pritchard et al, 2008; Ingram et al, 2008; Lodise et al, 2008). In all 6 of our patients, the clinical intent was to dose to achieve a target trough level >12-20 mg/L, but during the execution, toxic levels had developed. (b) Prolonged duration of administration (Goetz & Sayer, 1993; Hidayat et al, 2006; Pritchard et al 2008). Two of our patients (#2 & # 6) had received vancomycin for 56 and 78 days, primarily in the outpatient setting where the monitoring mechanism and dose adjustment and response time were suboptimal. (c) Rapid dose escalation without achieving steady state (Barraclough et al, 2007). Three of our patients (#1, #4, and #6) suffered as a result of the desire to achieve the higher 15-20 mg/L target levels because of the apparent failure to await a steady state between dosage increments. In other two patients (# 3 & # 5), the intent to attain therapeutic levels within the first few days of administration resulted in excessive levels

In our opinion, therefore, the most important but recurrent lesson to learn from all these cases and the literature would be meticulous avoidance of excess vancomycin levels since low levels were rarely reported to induce Van-AKI. Toxic levels typically develop during the non-steady state of either the initial days of a fixed dose schedule or the early phase of rapid dose escalation. While we have no data to base any proposed recommendation, it is

Table 6. A. Demographics and baseline clinical characteristics (N= 6)

**Gender Male : Female 6 : 0** 

**Exposure to radio-contrast agents** 

**Hypotension (SBP <95 x >4 h) or signs of** 

**Overall assessment of the pathogenic role** 

also due to non-steady state kinetics.

during the evolution phase and recovery period of the AKI, we believe we have vigorously documented the existence of Van-AKI in these six patients in whom we have complete access to and full review of all their clinical and laboratory data (Fig 1-7).

Collectively, we believe these three independent lines of evidence firmly establish the fact that vancomycin is unquestionably nephrotoxic, no different than aminoglycoside, cisplatinum, and radio-contrast dyes. The degree of renal failure was severe enough to initiate dialysis in one patient though he got less than one week of vancomycin. The other five patients had various degrees of residual renal impairment even a month after the last dose (Table 6 B, Fig 7). We therefore submit that the issue is no longer whether Van-AKI exists, but how to prevent or ameliorate it. To generate some practical guidelines towards this goal (our third objective), we took an intermediate step by pursuing the next objective.

Our second objective was to statistically analyze our 6 patients to generate a clinical pattern and to define a typical profile of Van-AKI, with the intent to abstract some insights and derive some lessons which can eventually help us formulate preventive strategies. Generally speaking, we note that Van-AKI is a real and common complication of vancomycin treatment, especially during rapid dose escalation and/or prolonged infusion of fixed doses without frequent monitoring of drug levels and serum creatinine. Van-AKI could be costly both financially and clinically since significant irreversible functional loss can ensue (Table 6 B, Fig 7). In the detailed analysis of our 6 cases, we found that, in retrospect if not prospectively, most cases of Van-AKI could have been prevented or ameliorated if only the returned results on levels and serum creatinine were carefully examined and interpreted within the clinical context and if only timely and appropriate corrective responses were made.

Fig. 7. Renal functional profile and changes in serum Vancomycin levels as averages of the 6 patients with AKI plotted against time since the initiation of vancomycin.

during the evolution phase and recovery period of the AKI, we believe we have vigorously documented the existence of Van-AKI in these six patients in whom we have complete

Collectively, we believe these three independent lines of evidence firmly establish the fact that vancomycin is unquestionably nephrotoxic, no different than aminoglycoside, cisplatinum, and radio-contrast dyes. The degree of renal failure was severe enough to initiate dialysis in one patient though he got less than one week of vancomycin. The other five patients had various degrees of residual renal impairment even a month after the last dose (Table 6 B, Fig 7). We therefore submit that the issue is no longer whether Van-AKI exists, but how to prevent or ameliorate it. To generate some practical guidelines towards this goal (our third objective), we took an intermediate step by pursuing the next objective. Our second objective was to statistically analyze our 6 patients to generate a clinical pattern and to define a typical profile of Van-AKI, with the intent to abstract some insights and derive some lessons which can eventually help us formulate preventive strategies. Generally speaking, we note that Van-AKI is a real and common complication of vancomycin treatment, especially during rapid dose escalation and/or prolonged infusion of fixed doses without frequent monitoring of drug levels and serum creatinine. Van-AKI could be costly both financially and clinically since significant irreversible functional loss can ensue (Table 6 B, Fig 7). In the detailed analysis of our 6 cases, we found that, in retrospect if not prospectively, most cases of Van-AKI could have been prevented or ameliorated if only the returned results on levels and serum creatinine were carefully examined and interpreted within the clinical

access to and full review of all their clinical and laboratory data (Fig 1-7).

context and if only timely and appropriate corrective responses were made.

Fig. 7. Renal functional profile and changes in serum Vancomycin levels as averages of the 6

patients with AKI plotted against time since the initiation of vancomycin.


Table 6. A. Demographics and baseline clinical characteristics (N= 6)

We therefore attempted to identify independent common risk factors resulting in Van-AKI. In this pursuit, we have confirmed but extended the three previously reported risk factors for Van-AKI (a) High blood vancomycin levels (Rybak et al, 1990; Hidayat et al, 2006; Pritchard et al, 2008; Ingram et al, 2008; Lodise et al, 2008). In all 6 of our patients, the clinical intent was to dose to achieve a target trough level >12-20 mg/L, but during the execution, toxic levels had developed. (b) Prolonged duration of administration (Goetz & Sayer, 1993; Hidayat et al, 2006; Pritchard et al 2008). Two of our patients (#2 & # 6) had received vancomycin for 56 and 78 days, primarily in the outpatient setting where the monitoring mechanism and dose adjustment and response time were suboptimal. (c) Rapid dose escalation without achieving steady state (Barraclough et al, 2007). Three of our patients (#1, #4, and #6) suffered as a result of the desire to achieve the higher 15-20 mg/L target levels because of the apparent failure to await a steady state between dosage increments. In other two patients (# 3 & # 5), the intent to attain therapeutic levels within the first few days of administration resulted in excessive levels also due to non-steady state kinetics.

In our opinion, therefore, the most important but recurrent lesson to learn from all these cases and the literature would be meticulous avoidance of excess vancomycin levels since low levels were rarely reported to induce Van-AKI. Toxic levels typically develop during the non-steady state of either the initial days of a fixed dose schedule or the early phase of rapid dose escalation. While we have no data to base any proposed recommendation, it is

Vancomycin-Induced Nephrotoxicity 219

On the other hand, to pursue the other equally important goal of prevention of Van-AKI, a more appropriate default mode of ordering vancomycin, we propose, would be to *infuse only if trough levels fall below certain target ranges,* as long as the attained trough levels are sufficiently high to achieve bacterial killing. We submit that these two goals are not mutually exclusive but in fact achievable in the same patient at the same time. At least two retrospective studies could be cited to support this notion. In the treatment of deep-seated MRSA infections, a retrospective cohort study failed to find any difference in clinical outcome between those with measurably high (>15-20 mg/L) and those with demonstrably lower trough levels (Hermsen, Hanson et al. 2010). In another retrospective study on vancomycin in the treatment of MRSA ventilator-associated pneumonia, the authors did not find any significant difference in survival or clinical cure in patients with trough level < 15

**Cumulative dose gram 59.3 23.6 Duration of vancomycin treatment days 28.2 12.7 Average daily dose g/day 2.4 0.6 Cumulative dose per unit body weight mg/kg 679 260 Average daily dose per unit body weight mg/kg/day 28 8** 

**("pre-peak" day) mg/L 30.2 9.8 Time from the first dose to the day of "pre-peak" vancomycin level days 22 10 Time from the day of "pre-peak" level to the day of peak vancomycin days 4.2 2.0** 

**patient during the entire course irrespective of when it was given) mg/L 70.0 9.8 Time from the first dose to the day of peak vancomycin level days 26.2 11.1 Time lag from the last dose to the appearance of peak vancomycin hours 9.5 4.9 Vancomycin levels just before discontinuation mg/L 65.1 12.5 Time when the last vancomycin dose given since the day of initiation days 28.2 12.7 Amount of vancomycin given in the last dose g 1.3 0.2 Nadir vancomycin level measured and recorded during recovery mg/L 17.5 7.5 Time from the last dose to nadir vancomycin level in recovery days 8.0 2.5 Time from the first dose to the day of peak serum creatinine days 27.2 10.7**

**creatinine hours 68.7 32.4 Time from the last dose to nadir serum creatinine (days to nadir) days 30.8 10.4 Recovery time as a ratio of vancomycin exposure time ratio 3.2 1.7**  Table 6. C. Summary of Vancomcyin dosage and serum levels during the course of AKI and

Additionally, in the treatment of MRSA bacteremia, it has been shown that high trough vancomycin levels of 15 to 20 mg/L per se might not be a good determinant or predictor for therapeutic success, at least not in those with pneumonia or MRSA endocarditis (Walraven, North et al. 2011). Therefore, until prospective randomized control studies comparing certain trough vancomycin level ranges are done to provide hard evidence to prove the importance of trough levels in excess of >15-20 mg/L, we propose physicians exercise appropriate caution, some circumspection, and some discretion in individual patients and

**serum creatinine Units Mean S.E.**

mg/L and those with trough levels > 15 mg/L (Chan, Pham et al. 2011).

**Vancomycin Dosages and Serum Levels vs. the time course of changing** 

**Vancomycin level on the day just before peak level** 

**Peak vancomycin levels (defined as the highest value for a given** 

**Interval between the last dose and the appearance of peak serum** 

its recovery (N=6)

fair to state that the ID recommendation of targeting 15-20 mg/L represents the consensus opinion of a panel of experienced experts in this field. The primary goal of combating infections is of course complete bacterial eradication. Viewed from this perspective, it is understandable and reasonable that the default mode of ordering vancomycin is to *keep giving to sustain trough levels >15-20 mg/L.* In practice though, the empirically observed trough levels would almost always exceed 15-20 mg/L, sometimes even up to 25-30, due to lack of foolproof dosing formula and due to invariably changing renal functions. Thus these drug levels seemed to be constantly at the threshold of flirting with nephrotoxicity. This could occur not only at the time of the documented trough levels but also most certainly during all the preceding hours when levels (though typically not measured) could be expected to be significantly if not markedly elevated.


Table 6. B. Summary of Serial Renal Function data during the Evolution of AKI & during its Recovery (N=6) (Mean ± SE)

fair to state that the ID recommendation of targeting 15-20 mg/L represents the consensus opinion of a panel of experienced experts in this field. The primary goal of combating infections is of course complete bacterial eradication. Viewed from this perspective, it is understandable and reasonable that the default mode of ordering vancomycin is to *keep giving to sustain trough levels >15-20 mg/L.* In practice though, the empirically observed trough levels would almost always exceed 15-20 mg/L, sometimes even up to 25-30, due to lack of foolproof dosing formula and due to invariably changing renal functions. Thus these drug levels seemed to be constantly at the threshold of flirting with nephrotoxicity. This could occur not only at the time of the documented trough levels but also most certainly during all the preceding hours when levels (though typically not measured) could be

expected to be significantly if not markedly elevated.

**Serum creatinine on the day just before vancomycin level** 

**Rise in serum creatinine on the day of "pre-peak"** 

**Rise in serum creatinine from the day of "pre-peak"** 

**Further rise in serum creatinine from the day of peak** 

**Further fall in CrCl from the day of peak vancomycin** 

**CrCl at the time of nadir serum creatinine (maximal** 

**Time from the first dose to nadir serum creatinine** 

Recovery (N=6) (Mean ± SE)

**Rise in serum creatinine from baseline** 

**Renal Function by serum creatinine or Creatinine Clearance estimated by** 

**100/serum creatinine (absolute values or changes)** 

**reached the peak (day of "pre-peak" vancomycin) (mg/dl) 1.85 0.48 n.s. vs.** 

**vancomycin vs. baseline (mg/dl) 0.89 0.40 n.s. vs.** 

**Drop in Crcl on the day of "pre-peak" vancomycin (ml/min) - 35 14 p = 0.06** 

**to the day of peak vancomycin (mg/dl) 3.14 0.67 p < 0.005 Fall in CrCl from baseline to the day of peak vancomycin (ml/min) - 84.7 11.2 p < 0.001**

**vancomycin to the day of peak vancomycin (mg/dl) 2.24 0.74 p < 0.04 Drop in CrCl from "pre-peak" to peak vancomycin (ml/min) - 49.7 19.3 p < 0.05**

**Increase in serum creatinine vs. baseline (mg/dl) 4.97 1.12 p < 0.01**

**Fall in CrCl at peak serum creatinine (worst decline) (ml/min) - 92.6 13.6 p < 0.005**

**vancomycin level to the day of peak creatinine level (mg/dl) 1.83 0.73 p < 0.05**

**level to the day of peak creatinine level (ml/min) - 7.9 3.0 p < 0.05**

**Drop in serum creatinine between peak & nadir values (mg/dl) 4.36 1.21 p < 0.02 Best CrCl recovery (maximal CrCl – worst CrCl) (ml/min) 50.7 13 p < 0.01 Irreversible increase in serum creatinine vs. baseline (mg/dl) 0.61 0.2 p < 0.04 Residual decline in CrCl despite maximal recovery (ml/min) - 41.9 12.9 p < 0.03** Table 6. B. Summary of Serial Renal Function data during the Evolution of AKI & during its

**Baseline serum creatinine (mg/dl) 0.96 0.13 Creatinine Clearance (CrCl) (ml/min) 114 15** 

**CrCl on the day of "pre-peak" vancomycin (ml/min) 79.2 21.1** 

**Serum creatinine on the day of peak vancomycin (mg/dl) 4.10 0.78 CrCl on the day of peak vancomycin (ml/min) 29.5 5.5** 

**Peak serum creatinine (mg/dl) 5.93 1.23** 

**CrCl at peak serum creatinine (worst CrCl) (ml/min) 21.6 5.1** 

**Interval between the first dose and peak serum creatinine (days) 28.8 9.9** 

**Nadir serum creatinine during recovery (mg/dl) 1.57 0.22** 

**recovery) (ml/min) 72.3 12.4** 

**(recovery time) (days) 59 15.7** 

**Units Mean SE P VALUES**

**baseline** 

**baseline** 

On the other hand, to pursue the other equally important goal of prevention of Van-AKI, a more appropriate default mode of ordering vancomycin, we propose, would be to *infuse only if trough levels fall below certain target ranges,* as long as the attained trough levels are sufficiently high to achieve bacterial killing. We submit that these two goals are not mutually exclusive but in fact achievable in the same patient at the same time. At least two retrospective studies could be cited to support this notion. In the treatment of deep-seated MRSA infections, a retrospective cohort study failed to find any difference in clinical outcome between those with measurably high (>15-20 mg/L) and those with demonstrably lower trough levels (Hermsen, Hanson et al. 2010). In another retrospective study on vancomycin in the treatment of MRSA ventilator-associated pneumonia, the authors did not find any significant difference in survival or clinical cure in patients with trough level < 15 mg/L and those with trough levels > 15 mg/L (Chan, Pham et al. 2011).


Table 6. C. Summary of Vancomcyin dosage and serum levels during the course of AKI and its recovery (N=6)

Additionally, in the treatment of MRSA bacteremia, it has been shown that high trough vancomycin levels of 15 to 20 mg/L per se might not be a good determinant or predictor for therapeutic success, at least not in those with pneumonia or MRSA endocarditis (Walraven, North et al. 2011). Therefore, until prospective randomized control studies comparing certain trough vancomycin level ranges are done to provide hard evidence to prove the importance of trough levels in excess of >15-20 mg/L, we propose physicians exercise appropriate caution, some circumspection, and some discretion in individual patients and

Vancomycin-Induced Nephrotoxicity 221

at 24-h trough levels below 10 mg/L to achieve similar safety margins as q 12 h dosing

The cornerstone to avoiding Van-AKI is abstinence, if not absolutely indicated as for 4 of our 6 patients, and if suitable safer alternatives are available. Despite the vast and positive overall clinical experience with vancomycin as an anti-MRSA antibiotic, several newer, less nephrotoxic or non-nephrotoxic alternatives have emerged, some even proven in clinical trials to confer comparable efficacy in certain bacterial infections. A few of these studies merit our comments and considerations as alternative agents because they demonstrate noninferiority or comparable efficacy to that of vancomycin, at least for certain organ infections. Thus, in MRSA ventilator-associated pneumonia, linezolid has been found in one retrospective study to produce similar survival rates but a trend towards higher cure rates than vancomycin (Chan, Pham et al. 2011). Clinical and microbiological outcomes in the treatment of nosocomial pneumonia were also found in one prospective randomized control trial to be comparable

In patients with SA bacteremia and endocarditis, daptomycin has been shown to produce similar clinical responses as standard vancomycin therapy (Fowler, Boucher et al. 2006) and the reported success rates favored daptomycin over vancomycin among those patients

In skin and soft tissue infections, a prospective single-blinded multicenter study reported similar efficacy between daptomycin and vancomycin (Pertel et al, 2009). Similarly, teicoplanin (Van Laethem et al. 1988) and telavancin (Wilson et al. 2009) have been found to yield comparable cure rates as vancomycin for skin and soft tissue infections. It should be noted that teicoplanin is a glycopeptide with similar spectrum of anti-bacterial activities as vancomycin but with one third lower nephrotoxic risks, based on a recent Cochrane review

Finally, two 5th generation cephalosporin prodrugs (ceftaroline fosamil and ceftobiprole medocaril) have been found to possess anti-MRSA activities. Ceftaroline has been shown to produce similar clinical cure rates as vancomycin in complicated skin and skin structure infections (Iizawa, Nagai et al. 2004; Ge, Biek et al. 2008), whereas ceftobiprole was found to show similar efficacy as vancomycin in suspected gram positive infections, diabetic foot and mixed bacterial complicated skin and skin structure infectons (Noel, Bush et al. 2008 a; Noel,

In summary, several newer antibiotics have been shown to provide a potential equally effective but less nephrotoxic alternative to vancomycin for deep-seated MRSA infections. It is beyond the scope and our goal to comment on the advisability of deploying such

Our third and final objective was to use the lessons and insights from the literature and our case series to generate and recommend some simple practical guidelines targeted to the prevention and amelioration of Van-AKI. We will present these recommendations in a

In conclusion, the era of vancomycin administration has spanned over half a century. Due to the widespread use of antibiotics whether indicated or not, there has been a growing emergence of microorganisms increasingly resistant to the existing antibiotics. MRSA has dictated the greater reliance on vancomycin. This in turn breeds the development of strains relatively insensitive to vancomycin, forces physicians to target higher drug levels and

without sacrificing efficacy (Cohen, Dadashev et al. 2002).

between linezolid and vancomycin (Rubinstein, Cammarata et al. 2001).

of 24 studies involving 2,400 patients (Cavalcanti, Goncalves et al. 2010).

Strauss et al. 2008; Noel, Strauss et al. 2008 b).

summary form in Table 7 below (Section VII).

**6. Conclusions and recommendations** 

alternatives other than updating their availabilities.

infected with MRSA.

base final dosing decisions on the entire clinical contexts, including the prevailing renal function.

Statistical analyses of our group data have yielded some new perhaps noteworthy insights. One, in 4 of our 6 patients (#1, #2, # 3, and # 5), the indication for vancomycin was not compelling, at least in retrospect, since only two had documented MRSA. Thus similar to some of the reported cases, in 2/3 of our patients, Van-AKI could have been avoided. Two, the failure to closely monitor drug levels or renal function had definitely contributed to the unexpectedly toxic levels and to Van-AKI in 3 of our patients (#2, #5, and # 6). In them, levels had not been checked for 8 to 21 days. Two of them (#2 and # 6) were under a designed 10-weeks treatment plan as an outpatient.

Three, there was no appropriate response to the discovery of excessive vancomycin levels (e. g. 30 mg/L) on day 22 of therapy despite a doubling of serum creatinine from the normal baseline (1.85 vs. 0.96 mg/dl) (Tables 6 B and C). Four days had been allowed to elapse, letting the steady climb of vancomycin level to its highest value on day 26, when little to nothing was done to reduce or withhold the dose during this interim. The same concern could be stated for the last dose given on day 28 since it should have been stopped or drastically reduced, as opposed to the 1.3 g dose actually used despite a vancomycin level of 70 mg/L and a serum creatinine of 4.1 mg/dl already noted 1-2 days earlier.

Four, when this highest level of 70 mg/L was finally reported on day 26 of therapy, at a time when serum creatinine (4.1 mg/dl) was already increased 4 fold, there was a 1-2 day time delay before the drug was stopped (Fig 7). Five, there was no evidence for any systematic dose adjustments for the known renal impairment in 3 of our 6 patients (#3, # 4, and # 6), either because of the absence of renal consultation or the lack of familiarity with the nomogram by Moellering et al (1981) [(15 x GFR in ml/min) for daily maintenance dose (in mg per day)]. This general equation has been found to be quite useful as it provides the first though crude approximation for dosages as a function of the residual renal function, permitting later finer adjustments based on subsequent trough levels. In practice, if serum creatinine is relatively stable, GFR can be estimated by using the equation of Cockcroft-Gault for CrCl (in ml/min) [= (140 – age in years) x (lean body mass in kg) / (serum creatinine in mg/dl x 72)] (Cockcroft and Gault, 1976).

The common basic issue among what appeared to have been judgment or logistic errors is either the lack of adequate monitoring or the lack of appropriate timely responses to typically already known warning signals for Van-AKI. Perhaps one additional source of problem or lesson to learn is the ordering, trusting, using and interpreting "random" vancomycin levels, here crudely defined any non-peak or non-trough levels, obtained at times totally without regard to the last administered dose. Such "random" levels are basically un-decipherable, generally misleading and unreliable, typically inaccurate as a surrogate of the AUC relating drug levels vs. time, and often simply useless if not hazardous. There is no published "normal" range statistically derived to define what one can expect for levels 4 to 8 hours post-dosing. This vacuum of information leaves plenty of doubts and much room for inaccurate extrapolations and erroneous speculations, making sound clinical decisions on proper dosage adjustments impossible.

We believe the practice of "random" levels should be abandoned, replaced by true 10-12 h trough levels. It should be noted that the AUC per unit time is smaller (thus nephrotoxic risk lower) if dosed once q 12 h vs. q 24 or q 48 h for identical trough levels for all three schedules. If a q 24-h dosing must be used, published experience would recommend aiming

base final dosing decisions on the entire clinical contexts, including the prevailing renal

Statistical analyses of our group data have yielded some new perhaps noteworthy insights. One, in 4 of our 6 patients (#1, #2, # 3, and # 5), the indication for vancomycin was not compelling, at least in retrospect, since only two had documented MRSA. Thus similar to some of the reported cases, in 2/3 of our patients, Van-AKI could have been avoided. Two, the failure to closely monitor drug levels or renal function had definitely contributed to the unexpectedly toxic levels and to Van-AKI in 3 of our patients (#2, #5, and # 6). In them, levels had not been checked for 8 to 21 days. Two of them (#2 and # 6) were under a

Three, there was no appropriate response to the discovery of excessive vancomycin levels (e. g. 30 mg/L) on day 22 of therapy despite a doubling of serum creatinine from the normal baseline (1.85 vs. 0.96 mg/dl) (Tables 6 B and C). Four days had been allowed to elapse, letting the steady climb of vancomycin level to its highest value on day 26, when little to nothing was done to reduce or withhold the dose during this interim. The same concern could be stated for the last dose given on day 28 since it should have been stopped or drastically reduced, as opposed to the 1.3 g dose actually used despite a vancomycin level of

Four, when this highest level of 70 mg/L was finally reported on day 26 of therapy, at a time when serum creatinine (4.1 mg/dl) was already increased 4 fold, there was a 1-2 day time delay before the drug was stopped (Fig 7). Five, there was no evidence for any systematic dose adjustments for the known renal impairment in 3 of our 6 patients (#3, # 4, and # 6), either because of the absence of renal consultation or the lack of familiarity with the nomogram by Moellering et al (1981) [(15 x GFR in ml/min) for daily maintenance dose (in mg per day)]. This general equation has been found to be quite useful as it provides the first though crude approximation for dosages as a function of the residual renal function, permitting later finer adjustments based on subsequent trough levels. In practice, if serum creatinine is relatively stable, GFR can be estimated by using the equation of Cockcroft-Gault for CrCl (in ml/min) [= (140 – age in years) x (lean body mass in kg) / (serum

The common basic issue among what appeared to have been judgment or logistic errors is either the lack of adequate monitoring or the lack of appropriate timely responses to typically already known warning signals for Van-AKI. Perhaps one additional source of problem or lesson to learn is the ordering, trusting, using and interpreting "random" vancomycin levels, here crudely defined any non-peak or non-trough levels, obtained at times totally without regard to the last administered dose. Such "random" levels are basically un-decipherable, generally misleading and unreliable, typically inaccurate as a surrogate of the AUC relating drug levels vs. time, and often simply useless if not hazardous. There is no published "normal" range statistically derived to define what one can expect for levels 4 to 8 hours post-dosing. This vacuum of information leaves plenty of doubts and much room for inaccurate extrapolations and erroneous speculations, making

We believe the practice of "random" levels should be abandoned, replaced by true 10-12 h trough levels. It should be noted that the AUC per unit time is smaller (thus nephrotoxic risk lower) if dosed once q 12 h vs. q 24 or q 48 h for identical trough levels for all three schedules. If a q 24-h dosing must be used, published experience would recommend aiming

70 mg/L and a serum creatinine of 4.1 mg/dl already noted 1-2 days earlier.

designed 10-weeks treatment plan as an outpatient.

creatinine in mg/dl x 72)] (Cockcroft and Gault, 1976).

sound clinical decisions on proper dosage adjustments impossible.

function.

at 24-h trough levels below 10 mg/L to achieve similar safety margins as q 12 h dosing without sacrificing efficacy (Cohen, Dadashev et al. 2002).

The cornerstone to avoiding Van-AKI is abstinence, if not absolutely indicated as for 4 of our 6 patients, and if suitable safer alternatives are available. Despite the vast and positive overall clinical experience with vancomycin as an anti-MRSA antibiotic, several newer, less nephrotoxic or non-nephrotoxic alternatives have emerged, some even proven in clinical trials to confer comparable efficacy in certain bacterial infections. A few of these studies merit our comments and considerations as alternative agents because they demonstrate noninferiority or comparable efficacy to that of vancomycin, at least for certain organ infections.

Thus, in MRSA ventilator-associated pneumonia, linezolid has been found in one retrospective study to produce similar survival rates but a trend towards higher cure rates than vancomycin (Chan, Pham et al. 2011). Clinical and microbiological outcomes in the treatment of nosocomial pneumonia were also found in one prospective randomized control trial to be comparable between linezolid and vancomycin (Rubinstein, Cammarata et al. 2001).

In patients with SA bacteremia and endocarditis, daptomycin has been shown to produce similar clinical responses as standard vancomycin therapy (Fowler, Boucher et al. 2006) and the reported success rates favored daptomycin over vancomycin among those patients infected with MRSA.

In skin and soft tissue infections, a prospective single-blinded multicenter study reported similar efficacy between daptomycin and vancomycin (Pertel et al, 2009). Similarly, teicoplanin (Van Laethem et al. 1988) and telavancin (Wilson et al. 2009) have been found to yield comparable cure rates as vancomycin for skin and soft tissue infections. It should be noted that teicoplanin is a glycopeptide with similar spectrum of anti-bacterial activities as vancomycin but with one third lower nephrotoxic risks, based on a recent Cochrane review of 24 studies involving 2,400 patients (Cavalcanti, Goncalves et al. 2010).

Finally, two 5th generation cephalosporin prodrugs (ceftaroline fosamil and ceftobiprole medocaril) have been found to possess anti-MRSA activities. Ceftaroline has been shown to produce similar clinical cure rates as vancomycin in complicated skin and skin structure infections (Iizawa, Nagai et al. 2004; Ge, Biek et al. 2008), whereas ceftobiprole was found to show similar efficacy as vancomycin in suspected gram positive infections, diabetic foot and mixed bacterial complicated skin and skin structure infectons (Noel, Bush et al. 2008 a; Noel, Strauss et al. 2008; Noel, Strauss et al. 2008 b).

In summary, several newer antibiotics have been shown to provide a potential equally effective but less nephrotoxic alternative to vancomycin for deep-seated MRSA infections. It is beyond the scope and our goal to comment on the advisability of deploying such alternatives other than updating their availabilities.

Our third and final objective was to use the lessons and insights from the literature and our case series to generate and recommend some simple practical guidelines targeted to the prevention and amelioration of Van-AKI. We will present these recommendations in a summary form in Table 7 below (Section VII).
