Measurement of calprotectin (S100A8/S100A9) and S100A12 in serum: method development, analytical validation, and clinical application

Abstract

Background Faecal biomarkers of intestinal inflammation, in particular faecal calprotectin and to a lesser extent faecal S100A12, are used to discriminate between inflammatory bowel disease (IBD) and irritable bowel syndrome (IBS) and categorise active from inactive disease in established IBD. Faecal biomarkers have limitations including intra–individual and inter–individual variability, spot variability in the same sample and reluctance of some patients to provide stool samples. These issues may be overcome by using serum samples for the measurement of calprotectin and S100A12. This offers the prospect that serum calprotectin and serum S100A12 could replace or supplement faecal calprotectin and faecal S100A12 in the identification and assessment of IBD. The measurement in serum calprotectin and serum S100A12, however, requires method development, validation of assays for serum and evaluation of the validated assays for their diagnostic and prognostic utility in IBD. Method development switches between two processes. It may necessitate adapting an existing method to ensure its suitability for application in a new assay or devising a suitable method by integrating the expertise and experience of the personnel undertaking the task of method development. Assay validation process for commercially available serum immunoassay kits is necessary to underpin assay measurement in order to confirm accuracy of test results, cut costs of undertaking unnecessary and repeat testing procedures, reinforce analytical claim that assay measurement is devoid of uncertainty to justify 'fit for purpose' and confer additional benefit of good reputation to a clinical laboratory. Validation of an assay in serum confirms or disproves kit manufacturer’s analytical claim to robust assay performance characteristics that include accuracy, precision, dilution linearity/parallelism, recovery, sensitivity, interference and stability. Aim/Objectives This project was designed to (1) Develop and analytically validate a faecal S100A12 assay (ImmunodiagnostikTM AG, Stubenwald–Allee 8a, D–64625 Bensheim, Germany) for measurement of S100A12 in serum. Analytically validate serum calprotectin assays provided by Bühlmann (serum BMN®-Cp; Bϋhlmann Laboratories AG, Baselstrasse 55, CH – 4124 Schönenbuch, Switzerland) and ImmunodiagnostikTM (serum IDK®- Cp; ImmunodiagnostikTM AG, Stubenwald–Allee 8a, D–64625 Bensheim, Germany). (2) Assess whether serum BMN®-Cp, serum IDK®-Cp and serum S100A12 could replace or supplement faecal calprotectin and faecal S100A12 in excluding IBD in patients presenting with chronic diarrhoea. (3) Evaluate the utility of serum BMN®-Cp and serum IDK®-Cp in discriminating between active and inactive IBD. (4) Study the effect of the acute phase response (APR) on serum calprotectin determined with two different immunoassays kits (i.e., serum BMN®-Cp and serum IDK®-Cp), and to assess and compare the diagnostic performance of the two assays in APR. Methods (1) ELISA assays for faecal S100A12 were developed and optimised for measurement of S100A12 in serum. The serum BMN®-Cp, serum IDK®-Cp and serum IDK®-A12 were validated by determining analytical sensitivity, functional sensitivity, dilution linearity/parallelism, recovery, precision, and interference. (2) The diagnostic performances of the serum BMN®-Cp, serum IDK®-Cp and serum IDK®-A12 assays were compared against faecal calprotectin, as the diagnostic ‘gold standard’, in 40 patients with IBD and 5 control patients. (3) Serum BMN®-Cp and serum IDK®-Cp and other conventional inflammatory blood biomarkers (serum CRP and platelets) were compared to faecal calprotectin in discriminating between active and inactive disease in a cohort of 175 patients with IBD. (4) The effect of APR, as determined by serum CRP, and serum calprotectin was assessed by measuring serum BMN®-Cp and serum IDK®-Cp before and after elective knee or hip surgery in 30 patients. Results Analytical validation Analytical validation of the assays showed a dynamic working range in serum of 10 to 25000 ng/mL, good precision (%CV for intra– and inter–assay variability for the kits were < 10% respectively, for each assay) and good reproducibility. There was no interference from bilirubin, haemoglobin, or lipid in the assays. There was no significant carryover or cross–reactivity across the assays. Assay kits were stable over 12 months. Analytical sensitivity ranged from 0.673 to 577 ng/mL for limit of the blank (LoB), and 1.119 to 597 ng/mL for lower limit of detection (LLoD). Functional sensitivity or limit of quantitation (LoQ) ranged from 522 to 3615 ng/mL. Measured to Expected ratios for dilution linearity/parallelism and recovery for the kits ranged from 98.4% to 103.7%, and from 82.1% to 126.5% respectively. Method comparison showed 19% positive proportional bias of the BMN®-Cp assay compared to the IDK®-Cp assay. Serum BMN®-Cp, serum IDK®-Cp and serum S100A12 in identifying IBD Using faecal calprotectin as the ‘gold standard’ for identifying IBD, the AUC from ROC curves for serum IDK®-Cp (AUC = 0.793) was greater than that for serum BMN®-Cp (AUC = 0.771) and these were greater than that for serum S100A12 (AUC = 0.700). Faecal calprotectin correlated best with serum IDK®-Cp (r = 0.69), then serum BMN®-Cp (r = 0.66) and least with serum S100A12 (r = 0.44). Serum BMN®-Cp and serum IDK®-Cp in discriminating between active and inactive IBD. The cohort of 175 patients with IBD consisted of 101 (57.7%) patients with Crohns disease (CD), 71 (40.6%) with ulcerative colitis (UC) and 3 (1.7%) inflammatory bowel disease unclassified (IBDU). The clinical classification of disease activity was largely based on faecal calprotectin which indicated that the disease was quiescent in 99 (56.6%) patients, active in 73 (41.7%) patients and in 3 (1.7%) patients were IBDU. Faecal calprotectin was, therefore, higher (p < 0.0001) in active CD than in quiescent CD, and similarly higher (p < 0.0001) in active UC compared to quiescent UC. Serum BMN®-Cp and serum IDK®-Cp in 175 IBD patients were highly correlated (r = 0.97). Serum BMN®-Cp and serum IDK®-Cp were higher (p < 0.006) in active CD than in quiescent CD but were similar (p > 0.1) in active and quiescent UC. Serum CRP was higher (p = 0.0095) in active CD compared to quiescent CD but similar (p = 0.0638) in active and quiescent UC. Platelets were similar (p = 0.0579) in active and quiescent CD and similar (p = 0.8055) in active and quiescent UC. Serum BMN®-Cp and serum IDK®-Cp concentrations were higher (p < 0.05) in active CD than quiescent CD at the ileal and upper GI, and the colonic and ileo–colonic sites of the ileum. Serum BMN®-Cp and serum IDK®-Cp concentrations were similar (p > 0.05) in active UC and quiescent UC involving the rectum, distal colon and pancolon. Based on ROC curve analysis, the performance of serum CRP (AUC = 0.699) was marginally superior to that of serum BMN®-Cp (AUC = 0.662) and serum IDK®-Cp (AUC = 0.656), and these were superior to platelets (AUC = 0.547) in all patients with IBD. In patients with CD, none of the blood biomarkers performed well; serum CRP (AUC = 0.585), serum BMN®-Cp (AUC = 0.585), serum IDK®-Cp (AUC = 0.556) and platelets (AUC = 0.609). In patients with UC, the performance of serum CRP (AUC = 0.752) was superior to that of serum BMN®-Cp (AUC = 0.670) and serum IDK®-Cp (AUC = 0.660), and these were superior to platelets (AUC = 0.487). The effect of an APR on serum BMN®-Cp and serum IDK®-Cp Following elective knee and hip surgery in 30 patients, serum CRP, serum BMN®-Cp, serum IDK®-Cp and blood neutrophils increased (p < 0.0001); serum albumin and serum total protein decreased (p < 0.0001). The mean (SD) post–operative increase in serum BMN®-Cp (3.0 (1.9) fold) and serum IDK®-Cp (2.8 (1.8) fold) were similar (p = 0.6575) but these were both lower (p < 0.0001) than serum CRP (82.0 (60.8) fold). Logarithmically transformed serum CRP correlated positively with serum BMN®-Cp (r = 0.64), serum IDK®-Cp (r = 0.65) and neutrophil count (r = 0.66), and negatively with serum total protein (r = –0.43) and serum albumin (r = –0.70). Serum BMN®-Cp correlated positively with serum IDK®-Cp (r = 0.97) and neutrophil count (r = 0.68), and negatively with serum albumin (r = –0.54). Serum IDK®-Cp correlated positively with neutrophil count (r = 0.67), and negatively with serum albumin (r = –0.55; p < 0.0001). There was no correlation between serum total protein and either serum BMN®-Cp or serum IDK®-Cp. Conclusions The developed and optimised serum IDK®-Cp, serum BMN®-Cp and serum S100A12 assays have good analytical performance and compared favourably to manufacturer stated performance characteristics, where available. The large numerical difference between serum BMN®-Cp and serum IDK®-Cp values indicate that results and any derived cut–offs between assays are not directly inter–changeable. The serum BMN®-Cp and serum IDK®-Cp assays have acceptable diagnostic accuracy for the identification of IBD and these were superior to serum S100A12. Although serum BMN®-Cp results were 1.7–fold higher than matched serum IDK®-Cp results, for diagnostic purposes this was accounted for by their manufacturer provided cut–offs of >3900 ng/mL and >3000 ng/mL respectively. Serum BMN®-Cp and serum IDK®-Cp, however, are unlikely to replace faecal calprotectin but may have a role supplementing faecal calprotectin in the identification of IBD. An elevated serum calprotectin in patients with chronic diarrhoea would be an indication for endoscopy since it has a low false positive rate, but a normal serum calprotectin does not exclude IBD. In the cohort of 175 patients with IBD, serum BMN®-Cp and serum IDK®-Cp were significantly associated with disease activity in patients with CD irrespective of site of disease. There was, however, no significant association between serum BMN®-Cp and serum IDK®-Cp and disease activity in patients with UC. ROC curves analyses indicated that serum CRP performed better than serum BMN®-Cp and serum IDK®-Cp in discriminating between active and inactive disease in patients with CD and UC. In this patient cohort, serum calprotectin offers no advantages over serum CRP in discriminating between active and inactive IBD, particularly since serum CRP is easily available and less expensive. Serum calprotectin is a positive acute phase protein, and both the serum BMN®-Cp and serum IDK®-Cp assays perform equally well during an APR elicited by orthopaedic surgery. The increase in serum calprotectin elicited by trauma and previously reported increase in sepsis indicates that serum calprotectin is a non–specific biomarker of inflammation. At two days following an inflammatory insult, serum CRP may be a better discriminatory biomarker of the APR than serum calprotectin based on a much greater incremental response.