**19. Biomarkers of disease severity and how they predict prognosis**

### **19.1. Sputum and bronchoalveolar lavage fluid biomarkers**

The need for useful and clinically relevant biomarkers is vital, and their use as predictors of prognosis as well as surrogates of clinical response to treatment is increasingly important. Numerous different sputum biomarkers have been investigated and used to correlate with clinical condition, including correlating with exacerbation frequency, pulmonary function, microbiological colonisation and overall prognosis. The use of sputum biomarkers is ideal in CF as patients produce significant amounts of sputum, hence samples are easy to obtain and samples are routinely collected to assess bacterial colonisation as part of best practice care.

Several studies have demonstrated how neutrophil elastase (NE) activity is significantly increased in the CF lung and a recent study by Sly et al. revealed how NE activity is an early biomarker for the development of bronchiectasis in CF [101]. The levels of NE in CF bron‐ choalveolar lavage fluid (BALF) correlate inversely with FEV1 [10] and furthermore the higher the level of NE detectable the more rapid the decline in lung function [12], neutrophil counts also correlate well with these measures. These results support the use of NE levels and activity as a useful biomarker available from BALF or sputum, and many other potential biomarkers are benchmarked against this measure as well as clinical correlation. Several inflammatory cytokines which are known to be elevated in the CF lung have potential to be useful biomarkers also, particularly interleukin-8 (IL-8) and tumour necrosis alpha (TNF-α). IL-8 is a neutrophil chemo-attractant and elevated levels are seen in CF BALF [102] and levels measured in sputum correlate inversely with FEV1 [10]. IL-8 and TNF- α levels are elevated at the time of an acute pulmonary exacerbation, and have been shown to significantly decrease in response to antibiotic treatment in an inverse pattern compared to lung function measurements [103]. Furthermore, the levels of IL-8 found in the lungs of CF patients correlates with microbial colonisation, with higher levels identifiable in patients colonised with *Pseudomonas Aerugino‐ sa* (PA) and *Staphylococcus Aureus* (SA) compared to non-colonised CF patients and healthy controls [104]. Further support for the usefulness of these cytokines as markers is how they correlate with symptoms of deterioration as reported by patients [105], hence IL-8 and TNFα are markers of degree of lung damage and bacterial colonisation. There are limitations to the use of sputum and BALF biomarkers such as these, as they do not always correlate with all clinical measurements of disease; in one study IL-8 did not correlate [106] with modified Bhalla scores [107], or other reversible changes on high resolution computed tomography (HRCT) that did correlate with lung function measurements.

Glycosaminoglycans (GAGs) are involved in the modulation of IL-8 activity in BALF and increased expression of GAGs is related to the sustained inflammation in the CF lung, hence the levels of GAGs may be a potential biomarker of disease progression in CF as their expres‐ sion is related to the neutrophil chemotaxis in the lung [108]. Other cytokines including IL-10 and IL-4 are also significantly elevated in CF BALF compared to healthy controls [109]. However, fewer studies have been done on these cytokines.

Several other sputum and BALF biomarkers have been studied and some have more potential future utility than others. Some of the early sputum biomarkers investigated were nitrites, with sputum NO2/NO3 shown to be significantly higher in CF patients with acute exacerbations [110] and correlated with neutrophil counts and inversely with FEV1 [111]. However, the levels did not return to normal despite intensive antibiotic treatment, hence negating their potential use as a marker of response to treatment [110]. Biomarkers that correlate to PA colonisation also include club cell secretory protein (CCSP), which is inversely related to the NE concen‐ tration in CF sputum and patients with pseudomonas aeruginosa have significantly lower CCSP than non-colonised individuals [112]. Furthermore, levels of leukotriene B4 (LTB4) correlate with PA colonisation in CF compared to other organisms [104].

To date, the most accurate sputum and BALF biomarker in CF remains NE activity and levels, hence several other biomarkers have been compared to this to demonstrate their usefulness. Other proteases such as matrix metalloproteinases 2 (MMP-2) and MMP-9 correlate well with increased levels noted in BALF which possessed increased NE activity [109], while cathepsins do not correlate with PA colonisation or exacerbations [113]. Biomarkers specific to pulmonary exacerbations are ideal as defining the response to treatment is very difficult in CF,; these include mucins MUC5AB and MUC5AC which are degraded at the time of exacerbation and demonstrate increased sialylation compared to controls [114]. Granulocyte-macrophage colony-stimulating factor (GM-CSF) levels at the time of pulmonary exacerbation predicts a larger acute decline in FEV1[115], which is highly predictive of poor prognosis [13], hence this may be a useful marker of future disease progression. Finally, calprotectin, a neutrophil derived protein, may also be a useful biomarker, where levels both in sputum and serum have been demonstrated to decrease significantly following treatment of an acute exacerbation and predicted time to next exacerbation [116], this is extremely useful as both exacerbations and lung function decline are the most sensitive markers of disease progression in CF.

### **19.2. Serum biomarkers as markers of systemic inflammation**

chemo-attractant and elevated levels are seen in CF BALF [102] and levels measured in sputum correlate inversely with FEV1 [10]. IL-8 and TNF- α levels are elevated at the time of an acute pulmonary exacerbation, and have been shown to significantly decrease in response to antibiotic treatment in an inverse pattern compared to lung function measurements [103]. Furthermore, the levels of IL-8 found in the lungs of CF patients correlates with microbial colonisation, with higher levels identifiable in patients colonised with *Pseudomonas Aerugino‐ sa* (PA) and *Staphylococcus Aureus* (SA) compared to non-colonised CF patients and healthy controls [104]. Further support for the usefulness of these cytokines as markers is how they correlate with symptoms of deterioration as reported by patients [105], hence IL-8 and TNFα are markers of degree of lung damage and bacterial colonisation. There are limitations to the use of sputum and BALF biomarkers such as these, as they do not always correlate with all clinical measurements of disease; in one study IL-8 did not correlate [106] with modified Bhalla scores [107], or other reversible changes on high resolution computed tomography (HRCT)

Glycosaminoglycans (GAGs) are involved in the modulation of IL-8 activity in BALF and increased expression of GAGs is related to the sustained inflammation in the CF lung, hence the levels of GAGs may be a potential biomarker of disease progression in CF as their expres‐ sion is related to the neutrophil chemotaxis in the lung [108]. Other cytokines including IL-10 and IL-4 are also significantly elevated in CF BALF compared to healthy controls [109].

Several other sputum and BALF biomarkers have been studied and some have more potential future utility than others. Some of the early sputum biomarkers investigated were nitrites, with sputum NO2/NO3 shown to be significantly higher in CF patients with acute exacerbations [110] and correlated with neutrophil counts and inversely with FEV1 [111]. However, the levels did not return to normal despite intensive antibiotic treatment, hence negating their potential use as a marker of response to treatment [110]. Biomarkers that correlate to PA colonisation also include club cell secretory protein (CCSP), which is inversely related to the NE concen‐ tration in CF sputum and patients with pseudomonas aeruginosa have significantly lower CCSP than non-colonised individuals [112]. Furthermore, levels of leukotriene B4 (LTB4)

To date, the most accurate sputum and BALF biomarker in CF remains NE activity and levels, hence several other biomarkers have been compared to this to demonstrate their usefulness. Other proteases such as matrix metalloproteinases 2 (MMP-2) and MMP-9 correlate well with increased levels noted in BALF which possessed increased NE activity [109], while cathepsins do not correlate with PA colonisation or exacerbations [113]. Biomarkers specific to pulmonary exacerbations are ideal as defining the response to treatment is very difficult in CF,; these include mucins MUC5AB and MUC5AC which are degraded at the time of exacerbation and demonstrate increased sialylation compared to controls [114]. Granulocyte-macrophage colony-stimulating factor (GM-CSF) levels at the time of pulmonary exacerbation predicts a larger acute decline in FEV1[115], which is highly predictive of poor prognosis [13], hence this may be a useful marker of future disease progression. Finally, calprotectin, a neutrophil derived protein, may also be a useful biomarker, where levels both in sputum and serum have

that did correlate with lung function measurements.

16 Cystic Fibrosis in the Light of New Research

However, fewer studies have been done on these cytokines.

correlate with PA colonisation in CF compared to other organisms [104].

Similar to sputum biomarkers, the use of serum biomarkers is highly appealing to monitor disease progression and also predict outcomes. Several studies have investigated in CF an array of potentially useful peptides and cytokines measurable in serum that correlate with exacerbation severity, length and recurrence as well as lung function and microbiological colonisation. While there are many studies of biomarkers of inflammation in CF and it further strengthens our understanding of the disease, the use of biomarkers has not become routine clinical practice, hence the need for accurate measurements of both pulmonary and systemic inflammation, which makes serum levels a key factor in this process as they may also be useful in measuring the response to treatment and predicting prognosis especially as new therapeutic options emerge.

As discussed previously, markers of neutrophil activity and elastase levels and activity in the lung correlate excellently with clinical condition, hence the use of neutrophil markers from serum would be ideal as biomarkers. One such biomarker is calprotectin, a neutrophil derived protein that is released during neutrophil activation, which has been shown to correlate with the resolution of exacerbations [116], as well as correlating with radiological scores and pulmonary symptoms [117]. CD16b is a receptor expressed by human neutrophils which binds immunoglobulins and is involved in the inflammatory response in CF. Levels of alpha-1 antitrypsin (AAT) complexed to CD16b (AAT:CD16b complex) have been shown to be significantly higher in CF patients than healthy controls, and correlate to other pro-inflam‐ matory cytokines including IL-8 and TNF-α [11]. Plasma concentrations of AAT:CD16b complex at the time of acute pulmonary exacerbation responded to antibiotic treatment and demonstrated a significant correlation to FEV1 improvement. These results suggest this may be a potentially useful biomarker of acute exacerbations and hence a useful prognostic tool.

The use of cytokines, similarly to sputum, is often viewed as an ideal surrogate of disease state, and many have been studied. Immunoglobulin G (IgG) and IL-6 have been shown previously to correlate with mortality, however, when adjusted for confounding factors, in particular FEV1, this correlation did not remain independent [118]. There is significant cross correlation between inflammatory markers such as TNF-α, IL-8 and cathepsins as well as IL-6 and IgG. However, none have shown any significant correlation with mortality. As survival becomes longer in this disease, the need increases for biomarkers that correlate together with other indices of disease such as pulmonary function, bacterial colonisation, exacerbation frequency, and radiological changes.

Leucocyte ribonucleic acid (RNA) measurements may be one such biomarker for potential future use. In identifying the presence of airway infection, whole blood levels of leucocyte RNA is an accurate measure and is more sensitive than FEV1 or c-reactive protein (CRP), and together with FEV1 is highly sensitive [119]. Other novel biomarkers include anti-neutrophil cytoplasmic antibodies specific for bactericidal/permeability-increasing protein (BPI-ANCA). Bactericidal/permeability-increasing protein (BPI) is one of the most potent endogenously produced antibacterials secreted by neutrophils, and autoantibodies against this protein are present in up to 90% of CF patients [120]. Levels of BPI-ANCA correlate strongly with markers of lung disease and it may indicate a pathogenic role in CF [121] and a recent study has shown that the presence of BPI-ANCA correlated significantly with a poor outcome, being death or requiring a lung transplantation. Furthermore, in patients colonised with PA the outcome was significantly worse if associated with this biomarker [122].

While no single biomarker of disease severity exists from either sputum or serum, there are several potential biomarkers that need further study and more robust results to support their routine use in clinical practice.

One biomarker that has recently re-emerged as a useful index of disease severity is the measurement of CFTR function. This may be measured by nasal potential difference (NPD) and sweat chloride concentration. It is important as a biomarker as it has been shown that CFTR and sweat chloride concentration at diagnosis predicts long term prognosis [123]. NPD is now reproducible in a more robust manor than previously, as it was difficult to accurately to so in the past [124]. The use of this biomarker has been further established in the recent clinical trials of CFTR modulators and correctors such as ivacaftor, with a change in NPD and sweat chloride seen as marker of improvement and correlating with improvements in lung function and BMI, whether this will be a long-term prognostic indictor remains to be seen [4].
