Determining a healthy reference range and factors potentially influencing PRO-C3 – A biomarker of liver fibrosis

Background & Aims Progressive fibrosis has been identified as the major predictor of mortality in patients with non-alcoholic fatty liver disease (NAFLD). Several biomarkers are currently being evaluated for their ability to substitute the liver biopsy as the reference standard. Recent clinical studies in NAFLD/NASH patients support the utility of PRO-C3, a marker of type III collagen formation, as a marker for the degree of fibrosis, disease activity, and effect of treatment. Here we establish the healthy reference range, optimal sample handling conditions for both short- and long-term serum storage, and robustness for the PRO-C3 assay. Methods PRO-C3 was measured in 269 healthy volunteers and in 222 NAFLD patients. Robustness of the PRO-C3 assay was measured according to Clinical and Laboratory Standards Institute standards and included validation of interference, precision, and reagent stability, whilst sample stability was defined for storage at different temperatures and for 3 freeze-thaw cycles. Fibrosis scoring was based on histological assessments and used as a reference for the diagnostic ability of PRO-C3 to discriminate between patients with different levels of fibrosis. Results Robustness of the PRO-C3 analysis validated by interference, precision, and reagent stability was found to be within the predefined acceptance criteria. The healthy reference range was determined to be 6.1–14.7 ng/ml. Levels of PRO-C3 were not affected by sex, age, BMI, or ethnicity. Levels of PRO-C3 were able to identify patients with clinically significant fibrosis and advanced fibrosis (AUC = 0.83 (95% CI [0.77–0.88], p <0.0001), and AUC = 0.79 (95% CI [0.73–0.85], p <0.0001), respectively). Conclusions The assay proved to be robust and sample stability was found to comply with hospital sample handling requirements. PRO-C3 measured in samples from patients with NAFLD/NASH was diagnostic for significant and advanced liver fibrosis. Lay summary We showed that PRO-C3 levels were stable under conditions conforming with hospital sample-handling requirements. We determined a healthy reference range and showed that PRO-C3 levels were not associated with sex, age, BMI, or ethnicity. Finally, we provide further evidence of an association of PRO-C3 with increasing liver fibrosis.

Determining a healthy reference range and factors potentially influencing PRO-C3 -A biomarker of liver fibrosis Introduction Non-alcoholic fatty liver disease (NAFLD) is a common progressive disorder often associated with the clinical features of metabolic syndrome. NAFLD starts with an excessive deposition of fat within hepatocytes for reasons other than excessive alcohol intake, especially obesity and type 2 diabetes mellitus (T2DM). 1,2 The condition encompasses a wide spectrum ranging from isolated steatosis and non-alcoholic steatohepatitis (NASH), a condition with additional inflammation and hepatocyte injury. NASH can progress through varying degrees of fibrosis to cirrhosis, liver decompensation, and potentially hepatocellular carcinoma. 3,4 In parallel with the worldwide increase in obesity and T2DM, NAFLD prevalence is also increasing. [5][6][7] With a global prevalence of 25%, NAFLD is now the leading cause of chronic liver disease in the USA and Europe. 8,9 Although simple steatosis does not correlate with increased short-term morbidity or mortality, NAFLD patients can progress to include liver cell damage (NASH) and may also have fibrosis. 10,11 Because of the dependence of histological features for a diagnosis of NASH and the necessity of invasive liver-biopsy confirmation, the exact prevalence of NASH in the general population is currently unknown. However, studies have shown that between 7 and 30% of NAFLD patients undergoing voluntary liver biopsies had NASH, and that the overall prevalence of NASH was between 1.5 and 6.45%. 6 Despite several efforts to identify a non-invasive diagnostic method, liver biopsy remains the established, but imperfect, reference standard for definitive diagnosis of the spectrum of NAFLD disease. 12,13 It is invasive, resource intensive, prone to sampling error and carries a small but significant risk of complications such as bleeding and pain. 14 Participants in ongoing NASH drug trials are required to be staged by liver biopsy to qualify for enrolment. 15,16 During the enrolment many patients undergoing liver biopsy do not meet the prespecified histological requirements and fail the screening process, having undergone an invasive procedure. Consequently, there is an urgent need for regulatory approved diagnostic biomarkers to facilitate the evaluation of new drugs. PRO-C3 is a type III collagen marker reflecting levels of the N-terminal pro-peptide released by A Disintegrin and Metalloproteases (ADAM)-TS2 during collagen maturation. 17 The assay specifically detects the neo-epitope when the pro-peptide is released. In clinical NAFLD studies, PRO-C3 has been shown to be correlated with the fibrosis stage and additionally to be related to disease activity. [18][19][20] Furthermore, the anti-fibrotic efficacy of a drug in patients responding to therapy has been shown to correlate to serologically assessed PRO-C3. 21 In the current study we established a healthy reference range, tested the optimal sample handling conditions for short-and long-term serum storage, and investigated the diagnostic ability of PRO-C3 to detect the histological fibrosis stage.

Study groups
All samples used to determine the reference range and the technical validations included in the analysis for assay robustness were collected from apparently healthy individuals (healthy volunteers) and NAFLD patients. All individuals gave their written informed consent. The study protocols were approved by appropriate ethics committees, assigned for the individual hospitals (Hvidovre hospital) or clinics (Discovery Life Sciences and Reprocell). The NAFLD/NASH diagnosis was based on a liver biopsy. Liver biopsy was not available for healthy adult volunteers with samples available (obtained from Discovery Life Sciences and Reprocell).
Serum samples from 269 healthy adult volunteers representative of the US population in age, race, and sex were obtained from Discovery Life Sciences. All tested negative for chronic hepatitis. Of these, 101 were considered overweight (BMI 25.0-29.5 kg/m 2 ) and 85 considered obese (BMI above 30). All volunteers were determined to be healthy after annual health checks. Volunteers with a history of diabetes or liver diseases or abnormal values of alanine aminotransferase (ALT; >50 IU/L), aspartate aminotransferase (AST; >50 IU/L), alkaline phosphatase (ALP; >129 IU/L), or bilirubin (>22 lmol/L) were excluded from the reference range estimation.
For the evaluation of PRO-C3 levels in NAFLD patients, a total of 222 patients with NAFLD were included from Nottingham University Hospitals NHS Trust (n = 83; Cohort 1) and the University Medical Center Mainz (n = 139; Cohort 2). 22,23 All patients were informed about the rationale and possible risks of the study and provided their informed consent. The study protocols were approved by the Nottingham University Hospitals NHS Trust and the Ethikkommision of the Landesärztekammer Rheinland-Pfalz (No. 837.199'10 [7208]). Based on a liver biopsy taken within 3 months of obtaining blood samples, histological activities of inflammation and fibrosis were assessed for all patients. The grade of NASH was assessed by an experienced histopathologist using the NAFLD activity score (NAS) scored from 0 to 8, incorporating scores of steatoses (0-3), ballooning (0-2), and lobular inflammation (0-3). The NASH Clinical Research Network (NASH-CRN) grading and staging system was used for quantification of fibrosis. 13 Biochemical measurements PRO-C3 was assessed using a competitive ELISA developed and produced at Nordic Bioscience (Herlev, Denmark). The assay procedure was as previously described. 17 The PRO-C3 assay used a monoclonal antibody detecting the sequence CPTGPQNYSP (Nordic Bioscience), corresponding to the cleavage site of the pro-peptide from the mature collagen in position 153 of the collagen type III a1 chain. Each analytical run was calibrated by an 8-point standard curve measured in duplicates and 3 quality controls were included in duplicates per analytical run. Data below the limit of quantification were reported as the LLOQ value.
Biochemical measurements such as ALT, AST, ALP, and other variables were measured on validated platforms in central laboratories at the respective clinical centres: Nottingham University Hospitals NHS and the University Medical Center Mainz. The biochemical measurements of ALT, AST, ALP, and bilirubin in all healthy volunteers included in the reference range study were measured on validated platforms in a Clinical Laboratory Improvement Amendments accredited laboratory.

Technical validation
To determine robustness of the PRO-C3 assay, we tested the assay according to available standards. 24-28 The PRO-C3 assay was tested in the following categories: analyte stability (storage and freeze-thaw), reagent stability, interference from known endogenous and the most relevant exogenous compounds used in treating patients with NAFLD/NASH, and assay precision. The test conditions, number of lots, operators, samples, and minimal acceptance criteria for each of the tests are summarised in Table 1.

Statistics
To test for differences between 2 groups, we used a Mann-Whitney test. For multiple groups, we used a Kruskal-Wallis test followed by Dunn's test with correction for multiple comparisons.
The reference range for PRO-C3 was determined according to the Clinical and Laboratory Standards Institute standard EP28-A3c. 29 The data distribution was evaluated by density plots and boxplots, both for the reference range and the NAFLD/NASH cohort. Outliers were detected on log-transformed data overall or within strata of sex, age, race, or obesity using Tukey's criterion, and excluded before the reference range estimation. The reference range was estimated using the robust method on logtransformed data, and back transformed to the original scale of PRO-C3. Ninety percent CI estimates for the lower and the upper limit of the estimated reference were calculated using 5,000 bootstrap replicates of the data.
Outliers for the reference intervals of PRO-C3 in patients with NAFLD/NASH were detected on log-transformed data within strata of fibrosis stage, NASH, or fibrotic NASH using Tukey's criterion, and excluded before the reference interval estimation. Area under receiving operator characteristic (AUROC) analyses were used to explore the diagnostic accuracy of PRO-C3 with respect to fibrosis staging in the NAFLD/NASH cohort. Ninety-five percent CI estimates for the estimated AUCs were calculated by the method of DeLong et al. 30 Optimal cut-offs were estimated by maximising the Youden index.
The analyses were performed in R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria). 31 Reference ranges were estimated using the referenceIntervals package, 32 and AUROC analyses were performed using the pROC package. 33

Technical validation
The serum PRO-C3 analyte was found to be stable for 8 days when stored at 8 C (Fig. S1A), for up to 48 h when stored at 25 C for 24 h (Fig. S1B), and remained stable up to the maximum time tested when stored at -80 C (Fig. S1C). PRO-C3 remained stable for all 3 freeze-thaw cycles tested (Fig. S1D). The PRO-C3 ELISA passed all predefined acceptance criteria for precision (intra-and interassay variation, lot variation, and interobserver variations (Table S1).
To test interference of potential endogenous and exogenous compounds (medication), the highest expected concentrations to be observed in patients' circulation were spiked into serum samples from patients with NASH that either contained low or high PRO-C3 levels (Table S2). No interference was detected from the endogenous and exogenous compounds that were deemed most relevant for patients with NAFLD (Table S1).

Reference ranges
Reference ranges for healthy volunteers were based on 269 apparently healthy individuals with no prior history of diabetes, liver disease, and with levels of ALT < − 50 IU/L, AST < − 50 IU/L, ALP < − 129 IU/L, or bilirubin < − 22 lmol/L were considered to be normal.
The reference intervals for PRO-C3 in healthy individuals stratified by sex, age, BMI categories, and race are listed in Table 2. When comparing levels of PRO-C3 in the different stratifications, we saw that there was no significant difference in PRO-C3 levels between healthy males and females (Fig. 1A), based on age (Fig. 1B), BMI categories (Fig. 1C) or ethnicity (Fig. 1D).

Discussion
Several biomarkers have been developed for NAFLD in recent years, including surrogate risk scores, individual blood-based markers, complex panels, and new imaging modalities. 12,35 PRO-C3 is a relatively new biomarker and is detected by an ELISA that specifically targets the neo-epitope of the N-terminal pro-peptide of type III collagen, which is released during deposition of the fibrillar collagen in the extracellular matrix. In liver fibrosis, type III collagen production is markedly upregulated. During its formation and accumulation pro-peptides are released from type III pro-collagen by ADAM-TS2-mediated cleavage and released into the circulation. 36 Consequently, the PRO-C3 ELISA measures a fragment reflecting active fibrogenesis. Several studies have explored PRO-C3 within clinical trials. 15,18,19,21,23,[36][37][38][39][40][41][42] A recent meta-analysis included studies that provided data on biopsy-proven NAFLD, PRO-C3 test, and data on true positive, false positive, true negative, and false negative results, or data that allowed a reconstruction of the classification table, were eligible for inclusion in the metaanalysis (submitted). Five studies consisting of 1,312 patients  Estimated AUCs and cut-off values for the identification of patients with F > − 2, F > − 3, and fibrotic NASH (NASH with F > − 2) are listed along with sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Curves were generated using area under the receiver operating characteristics curve analysis. The 95% CIs for the AUC and cut-offs were estimated by bootstrapping. Reference intervals were estimated by the robust method according to the recommended approach (CLSI C28-A3). PRO-C3 levels below the lower limit of quantification were assigned the lowest acceptable concentration. NASH was defined as lobular inflammation > − 1, ballooning > − 1, and NAS > − 4. Fibrotic NASH was defined as NASH with fibrosis stage > − 2. Outliers were detected on log-transformed data using test Tukey's criterion, and excluded. Skewness estimates are based on the log-transformed data. Statistical significance: *p <0.05. CLSI, Clinical and Laboratory Standards Institute; NAS, non-alcoholic fatty liver disease activity score; NASH, non-alcoholic steatohepatitis.
reported data on significant fibrosis with thresholds ranging from 13.2 to 21.3 ng/ml. 15,18,23,38,39 Two of the studies included data on PRO-C3 in detecting (fibrotic) NAFLD/NASH and found thresholds of 14.5 and 14.7 ng/ml, respectively. 15,39 For cirrhosis related to NAFLD/NASH thresholds of 15.6 and 16.5 ng/ml have previously been reported. 15,23 Our findings seem to be in line with previous findings.
To support further clinical investigations and use of PRO-C3 it is critical to establish robustness of the analysis, optimal sample handling, conditions for short-and long-term serum storage, and a healthy reference range. In this article we have proven the assay to be robust, including precision within expected limits, shown that the analyte and reagents are stable for periods that are in alignment with clinical use, and not influenced by potential endogenous and exogenous (medications) which are highly relevant for the NAFLD population. However, as the NAFLD population includes patients who often suffer from accompanying diseases, it may be beneficial to test whether medications of selected co-morbidities may influence PRO-C3 levels.
Based on our findings in healthy individuals, we showed that PRO-C3 levels were not influenced by sex, age, or race. This finding contrasts with the Enhanced Liver Fibrosis (ELF TM ) test which has been found to be significantly dependent on age and sex, leading the authors to stress that this score needs to be interpreted with caution. 43 Importantly, being overweight or obesea major risk factor for NAFLD did not impact PRO-C3. Serum PRO-C3 values were not associated with BMI, and levels of PRO-C3 were similar in patients classified as normal weight (BMI 18.5-24.9), overweight (BMI 25.0-29.5), and obese (BMI > − 30).
At current, regulatory agencies (FDA and EMA) require phase IIb and phase III clinical trials to report NASH-resolution or fibrosis regression based on liver biopsy results in order to consider conditional approval for a drug. This is a major limitation for drug development as recruitment rates are impacted and thereby major delays in the development timelines for a specific compound emerge. Large-scale efforts are currently ongoing by 2 major consortia, namely the European consortium LITMUS (Liver Investigation: Testing Marker Utility in Steatohepatitis) 44 and the US consortium NIMBLE (Non-Invasive Biomarkers of Metabolic Liver Disease), 45 to define what the FDA and EMA requires for a biomarker qualification (e.g. level of technical, clinical, statistical analysis) in different contexts of uses in NAFLD/NASH. A qualification of a diagnostic biomarker could ultimately amend or even eliminate the need for a liver biopsy for patients entering clinical trials, if robust data are available.
We next evaluated levels of PRO-C3 in a cohort of NAFLD patients with histological assessments available. The NAFLD patient group consisted of cohorts from 2 centres in Germany and the UK. We showed that levels of PRO-C3 were able to discriminate patients with no or mild fibrosis from patients with significant fibrosis (> − F2). We also tested the performance of PRO-C3 for specifically identifying advanced fibrosis, as well as fibrotic NASH. The performance of PRO-C3 to discriminate these patients proved to be slightly lower than the discrimination of patients with significant fibrosis; however, the AUROC was only slightly lower. The lower AUROC may be attributable to a relatively low number of patients in these 2 groups. Overall, the performance of PRO-C3 in our paper is in line with findings from a recent review and metanalysis of PRO-C3 in previous studies of patients with NAFLD/NASH (submitted).
The current data support the use of PRO-C3 as a diagnostic screening marker. Such a marker could be used to de-select patients that would be unlikely to meet histological enrolment criteria, and thereby avoid an unnecessary procedure with the potential for complications.
Another option for use of PRO-C3 is as a prognostic enrichment biomarker which could enrich a trial with patients who are more likely to develop adverse outcomes. However, for the EMA and FDA to qualify a prognostic biomarker for clinical events, they are likely to require real-time data on clinical endpoints, which will only be generated in large scale clinical studies exceeding to 5-8 years in duration considering the (in most cases) slowly progressive nature of fibrosis in NAFLD.
In recent studies, a combination of PRO-C3 with other clinical markers was shown to be a good diagnostic screening tool. 20,23 The studies investigated the ability of a combination of different markers (PRO-C3 and other laboratory markers 23 ) to identify patients meeting the recommended biopsy requirements by the FDA and EMA for phase III trial enrolment (i.e. NAS > − 4 with at least 1 point each in inflammation and ballooning along with a NASH-CRN >1 but < stage 4). 16,34 The combination was suggested to be a possible means to decrease the number of patients to be biopsied, as patients with a low score would have a low probability of meeting the biopsy enrolment criteria and would undergo an unnecessary biopsy.
Biomarkers are needed in the management of NAFLD patients to: (i) determine severity of disease (staging), (ii) develop new treatments, and (iii) monitor patients to identify patients at risk of disease progression and adverse outcomes. In relation to determining disease severity, such markers may reduce the number of unnecessary biopsies performed for clinical trial enrolment.
Several clinical studies have evaluated and defined usage of PRO-C3 in the NAFLD/NASH population. This includes positive data on; the ability of PRO-C3 to determine severity of disease, 15,18,23,37,42 its ability as a prognostic biomarker to enrich clinical trials to monitor treatment effect, 19,21,38,41,42,46 and for determining disease activity. 20 The fact that PRO-C3 is a marker that directly reflects fibrosis and fibrosis activity makes it suitable for use in clinical practice for the general practitioner to support/take decisions on referral of patients to specialised liver centres, and in specialised liver centres for support in decisions on patient management and prognosispotentially even without the need for a biopsy. The advantage of using PRO-C3 for initial evaluation of liver fibrosis compared with fibrosis-4 (FIB-4) and NAFLD fibrosis score is that PRO-C3 is a direct marker of fibrosis in contrast to FIB-4 and the NAFLD fibrosis score, which are indirect composite scores of fibrosis that rather reflect liver disease. The markers included in these composites scores are clinical characteristics, which mostly are markers of liver function, thus their origin is not the liver scar tissue. The PRO-C3 and ELF score composite scores are direct markers of liver fibrosis, as they assess proteins which originate from scar tissue of the liver. Furthermore, in a study of more than 400 patients with NAFLD, it was shown that PRO-C3, as a single marker or as a composite marker within the ADAPT score (age, diabetes, platelets, and PRO-C3), was superior to APRI (AST to platelet ratio index), FIB-4, and NAFLD Fibrosis score for identification of advanced liver fibrosis. 23 A similar analysis was performed in more than 300 patients supporting that ADAPT is superior. 47 The ELF score requires the assessment of 3 fibrosisrelated markers (PIIINP, HA, TIMP-1) in contrast to PRO-C3 which is a single marker. Nevertheless, it has been indicated that PRO-C3 is a more dynamic marker of fibrogenesis compared to ELF, shown in several clinical trial studies of NASH or primary sclerosing cholangitis (PSC), where reduction of PRO-C3 were greater compared with ELF and also the PIIINP component alone, during treatment of patients with potential anti-fibrotic therapies. 40,48,49 This may suggest that PRO-C3 is a superior marker for liver fibrosis because of its dynamics and origin within liver fibrosis. Future studies will determine whether PRO-C3 also has a place in the general clinical management of NAFLD patients. However, it has already been shown that the serial combination of ADAPT with liver stiffness measurement has high diagnostic accuracy with a low requirement for liver biopsy. 47 The proposed algorithm would help stratify those who need biopsies and narrow down those patients who would need to be referred to specialty clinics.
In apparently healthy volunteers resembling the US population we found that PRO-C3 levels were not associated with sex, age, BMI, or race. The assay was robust as measured in relation to interference, precision, reagent, and sample stability. Furthermore, we provided further evidence that PRO-C3 is a promising marker for pre-screening of patients for clinical trials to avoid unnecessary biopsies.

Financial support
The biomarker validation studies were funded by Bristol Myers-Squibb, Princeton, NJ, USA. The study was supported by the Danish Research Foundation.