Statistically, COPPS significantly correlated with hospitalization outcomes, but the correlations were weaker than in previous studies, which may be related to the outpatient nature of the PROspective Evaluation of Chronic Pancreatitis for EpidEmiologic and Translational StuDies cohort and lower prevalence of high severity disease. Studies in other prospective cohorts are needed to understand the full utility of COPPS as a potential tool for clinical risk assessment and intervention.
Chronic pancreatitis (CP) is a fibroinflammatory syndrome of the pancreas in individuals with genetic, environmental, and/or other risk factors who develop persistent pathologic responses to parenchymal injury or stress (). CP is associated with debilitating symptoms that reduce physical and mental health severely impacting quality of life and leading to hospitalization(s) (). CP-related hospitalizations are increasing across the globe (). Over a 17-year period, the cost of hospitalizations associated with CP increased over 300% (). The estimated cost for pancreatitis-related hospitalizations in the United States is in excess of $2 billion (). In other chronic diseases, the development of tools that can predict prognosis has been incorporated into clinical decision-making. For example, the modified Model for End-Stage Liver Disease score is currently used to prioritize individuals on the liver transplant list. There are no tools or dynamic scoring systems available for gastroenterologists to predict the natural course of CP in a particular patient. To address this need, the Chronic Pancreatitis Prognosis Score (COPPS) was developed. COPPS, a composite score based on 5 selected parameters, was designed to predict the short-term course of pancreatic disease, enabling physicians to respond with disease-specific treatments (). Previous studies utilizing COPPS were conducted in Europe and Asia (India) (). While a recent meta-analysis suggests that CP in eastern and western populations share similar clinical profiles (), it remains unclear whether COPPS would perform similarly in other geographic regions. Thus, the purpose of this study was to evaluate the utility of COPPS in a larger adult CP population in the United States.
PROspective Evaluation of Chronic Pancreatitis for EpidEmiologic and Translational StuDies (PROCEED) is a prospective cohort study of adult subjects across the pancreatitis spectrum. A subset of PROCEED subjects with definite CP (N = 354) enrolled between June 2017 and October 2020 at 9 centers were eligible for this study (). Definite patients with CP include those who showed (i) presence of unequivocal CP by Cambridge grade 3 or 4, and/or parenchymal and/or ductal calcifications by computed tomography scan or magnetic resonance imaging/magnetic resonance cholangiopancreatography, or (ii) pancreatic histology diagnostic of CP including findings of fibrosis (Amman score ≥6), chronic inflammation, and acinar loss (). After excluding those who did not have complete data for all 5 COPPS parameters and those with <12-month follow-up, a total of 279 subjects were included in the primary analysis as the final analytical sample (Figure 1). Per the protocol of the PROCEED study, detailed patient and clinical data were collected on demographics, drinking and smoking status, quality of life, medication use, disease phenotype, laboratory testing, interventions, health outcomes such as number of hospitalizations or number of acute pancreatitis (AP) attacks and biospecimens at baseline and/or follow-up. This study utilized the comprehensive data collection from the PROCEED study for the validation of COPPS (Table 1). All protocols were approved by a central Institutional Review Board (MD Anderson Cancer Center).
COPPS was calculated by assigning point values to each of the 5 parameters: numerical pain rating scale, platelet count, hemoglobin A1c (HbA1c), body mass index (BMI), and c-reactive protein (CRP), then summing the points (see Supplementary Table 1, Supplementary Digital Content 1, https://links.lww.com/CTG/B183). COPPS values were also classified by severity as low (5-6 points), moderate (7-9 points), and high (10-15 points) (). Numerical pain rating scale, platelet count, HbA1c, and BMI data are routinely collected for the PROCEED study, as previously described.
Owing to the high number of missing values for CRP values from standard-of-care laboratory tests and variability between clinical laboratory assays, we systematically measured levels in baseline serum samples that were collected for research purposes. All assays were performed according to the manufacturers' instructions using an electrochemical immunoassay on the Meso Scale Discovery Platform. Positive and negative (diluent only) controls were included on each 96-well plate. All samples were run in duplicate. All research staff performing sample testing and data collection were blinded, and coded samples were randomized across plates.
The primary endpoints, the number and duration of hospitalizations for any reason, and the number and duration of pancreas-related hospitalizations were all assessed at 12 months after enrollment. Pancreas-related hospitalizations were identified as hospitalizations due to AP-related or pancreatitis-related complications. For a secondary analysis, imputation was done for missing values in the primary endpoints. If there were any data available beyond 12-month follow-up, they were used with adjustment in follow-up duration. For the subjects whose hospitalization data were not available either during 12-month follow-up or beyond, primary endpoints were imputed by multiple imputation and the imputed data were used with follow-up duration defined as 12 months. The secondary endpoints included continued usage of tobacco, alcohol, or opioids during the 12-month period following enrollment. Moreover, the presence of moderate to severe anxiety was assessed using the patient-reported outcomes measurement information system (PROMIS)-29 questionnaire. New interventions, including endoscopy and surgery, during the 12-month follow-up period were also recorded.
Clinical variables were collected during baseline and annual PROCEED study visits through completion of case report forms or by review of electronic medical records (). Diabetes mellitus (DM) was diagnosed based on the American Diabetes Association criteria: abnormal values on 2 of the following tests or 2 abnormal values of the same test: (i) fasting blood glucose ≥126 mg/dL, (ii) HbA1c ≥ 6.5%, (iii) random blood glucose ≥200 mg/dL or by the use of antidiabetic medications. Subjects with no documented diagnosis of DM at enrollment underwent a per protocol assessment of DM with blood glucose and HbA1c levels. Exocrine pancreas dysfunction (EPD) was defined at baseline by a documented diagnosis before enrollment (clinical history of steatorrhea or fecal elastase of <100 μg/g stool or quantitative fecal fat of >7 g/d on a 100 g fat diet) or a fecal elastase of <100 μg/g stool per protocol assessment in subjects with no known history of EPD. The DM or EPD status was labeled "not tested" for subjects with no documented diagnosis at enrollment and who did not undergo a per protocol assessment after enrollment for DM or EPD at their baseline visit. Anxiety was assessed using the PROMIS-29 questionnaire in which response options ranged from 1 to 5 (e.g., never/not at all to always/very much). Raw sum scores were calculated by adding together responses to all items within the anxiety domain. If there were any missing responses within the domain, prorated raw sum scores were calculated based on the number of items answered as suggested in the PROMIS-29 scoring manual. The raw sum scores were then converted to T scores where a T score ≥50 indicates anxiety worse than the general US population. T score >60 was used to define moderate-severe anxiety ().
All baseline characteristics were compared by COPPS severity categories using the Kruskal-Wallis test for continuous variables and either χ test or Fisher exact test for categorical variables. Correlation analysis was conducted to assess the linear relationship between COPPS and the primary endpoints using the Pearson correlation coefficient and Kendall tau rank correlation coefficient. The 95% confidence intervals of the estimated correlation coefficients were obtained through Fisher Z transformation (). Missing data in the primary endpoints were imputed by multiple imputations through the PROC MI procedure in SAS, and the statistical results were aggregated from the 5 imputed data sets through the PROC MIANALYZE procedure whenever feasible. The primary endpoints were also compared among the COPPS categories using the Kruskal-Wallis test. Owing to the high proportion of subjects without a hospitalization during follow-up, a subgroup analysis was conducted in the subgroup with at least 1 hospitalization. When using the imputed data, the Kruskal-Wallis test was conducted separately in each imputed data set, and an aggregated P value was calculated (). Finally, all secondary endpoints were compared across COPPS severity categories using the χ test or Fisher exact test as appropriate. All statistical tests used a significance level of 5%. All data management and analyses were conducted using SAS 9.4 version (SAS Institute, Cary, NC). All associations were investigated as independent outcomes; therefore, we did not perform familywise hypothesis testing (i.e., multiple comparison).
The median age in the study population was 53 years (IQR 45, 63) with a slight female predominance (51%) (Table 1). Almost half of subjects were current smokers, and 21% were current drinkers at enrollment. About half of the study population were opioid users (52%) and around one-third had moderate to severe anxiety at baseline. Of the final 279 subjects, the mean COPPS as a score was 8.4 (SD 1.6), and two-thirds were in the Moderate category (Figure 1). A summary of the values for individual parameters, COPPS as a score, and COPPS as severity category are provided in Table 2.
Age at enrollment, number of lifetime AP attacks, DM status, opioid use, and moderate to severe anxiety were significantly different between the COPPS categories (Table 1). Those in the low severity group showed a higher mean age than those in the moderate and high severity groups (P = 0.02). The likelihood of having experienced 2 or more AP attacks increased with increasing COPPS severity (P = 0.01). As expected, the low severity group showed the highest prevalence of DM (P < 0.001). Those in the moderate and high severity groups showed a higher prevalence of opioid use than those in the low severity group (P = 0.003). Similarly, the prevalence of moderate to severe anxiety was significantly higher in the moderate and high COPPS groups (P = 0.02).
The number of hospitalizations and the total duration of hospitalizations were summarized for the 12-month period following enrollment (Table 3). Overall, 39.4% of subjects had 1 or more hospitalization(s) for any reason; 30.8% had 1 or more pancreas-related hospitalization(s). The number of hospitalizations for any reason was estimated to have a mean of 1 (SD 1.9) per subject. Duration of hospitalizations for any reason was a mean of 4.5 days (SD 10.5) per subject for an overall total of 1,205 days of which 926 days were from hospitalizations due to pancreas-related reasons.
Correlation analysis including all subjects revealed that higher COPPS was positively correlated with increased number and duration of hospitalizations for any reason and for pancreas-related reasons (Figure 2). Correlation analysis using the imputed data showed similar results, although some of the correlations were not significant after imputation (see Supplementary Table 2, Supplementary Digital Content 1, https://links.lww.com/CTG/B183). Correlation analyses between individual COPPS parameters and hospitalizations suggest that in the PROCEED cohort, pain is the main driver of the association between COPPS and hospitalization because the other 4 parameters did not significantly correlate with hospitalization outcomes (see Supplementary Table 3, Supplementary Digital Content 1, https://links.lww.com/CTG/B183). Comparison between COPPS severity groups indicates that all hospitalization outcomes are significantly different (Table 3). On the other hand, excluding those with zero hospitalization, comparison of COPPS severity groups in all hospitalization outcomes did not show significant differences (Supplementary Table 4, Supplementary Digital Content 1, https://links.lww.com/CTG/B183). Data with and without imputation for missing values showed similar results, with the exception of duration of pancreas-related hospitalizations. In the analysis of data with imputation, P = 0.04, whereas in the analysis of imputed data, P > 0.05 (Supplementary Table 5, Supplementary Digital Content 1, https://links.lww.com/CTG/B183).
Of all the secondary outcomes analyzed, COPPS showed significant association with continued alcohol use and presence of moderate to severe anxiety (Table 4). There were no significant associations between COPPS and continued tobacco exposure, continued opioid use, new endoscopic therapy, new pancreatic surgery, cholecystectomy, or percutaneous celiac plexus block. We observed that subjects with moderate or high COPPS had a higher prevalence of anxiety at enrollment (P = 0.02) and at 12-month follow-up (P < 0.05). Subjects with low or moderate COPPS showed a higher prevalence of continued drinking at follow-up (P = 0.04). There was a trend toward higher prevalence of continued opioid use, which may be related to increased pain.
COPPS was developed as a tool to inform clinical decision-making related to the treatment of CP. Like previous studies, our data from a US cohort demonstrate that COPPS is significantly associated with short-term hospitalizations both for any reason and pancreas-related hospitalizations. Despite achieving statistical significance, the observed correlations between COPPS and hospitalization outcomes were weaker than in the previous studies. We also observed a significant association between high COPPS severity and anxiety and low COPPS severity and continued alcohol, neither of which were reported in other COPPS studies. Mood and other psychiatric disorders are frequent comorbidities in individuals with chronic pain syndromes, including CP (). Indeed, genetic loci for anxiety, major depression, and posttraumatic stress disorder are associated with worse pain experience (more severe, constant pain) in the US cohorts of individuals with CP (). More extensive studies elucidating the interactions between pain and mood disorders in the context of CP could inform better management of CP pain. For example, addition of neuromodulatory therapies could reduce the amount of opioids required in this patient population. In addition to psychiatric status, there were other differences between the PROCEED cohort and the previous cohorts related to participant characteristics and study design, which may influence the hospitalization rates and resulting correlations with COPPS severity.
There were differences in the participant characteristics of the PROCEED cohort when compared with other cohorts used to study COPPS that reflect geographic differences and may have contributed to the observed differences in the current analysis (Supplementary Table 6, Supplementary Digital Content 1, https://links.lww.com/CTG/B183). For instance, PROCEED had an even distribution of sex, while both the German and Indian cohorts had a male predominance. In addition, the Indian cohort had a younger age than the German and US cohorts. Approximately two-thirds of the Indian cohort were classified as idiopathic etiology, while only one-third of the PROCEED cohort was attributed to an idiopathic etiology. The proportion of subjects in PROCEED with etiology attributed to alcohol was in between the German and Indian cohorts. The Indian cohort had a larger proportion of subjects with DM. Race/ethnicity can also contribute to the severity of pancreatitis (); however, the racial/ethnic composition of the other cohorts may be more homogenous than PROCEED, but this information is not available for comparison. Geographic differences have also been reported in AP (). These differences are reflected in pancreatitis scoring systems. For instance, the pancreatitis activity scoring system used for AP was significantly different across 4 continents. Individual with mild AP had highest scores in North America at admission and 24 hours later (). The pancreatitis activity scoring system score incorporates opiate use in the calculation; however, granular data on morphine or morphine equivalent dose are not yet available for CP across the world. Once these data become available, adding it to the calculation may improve the predictive strength of COPPS.
Variations in the distribution of the COPPS parameters (and consequently the severity groups among cohorts) may be reflective of the differences in participant characteristics and study design, influencing the predictive ability of COPPS (see Supplementary Table 6, Supplementary Digital Content 1, https://links.lww.com/CTG/B183). Among the individual components of COPPS, the PROCEED cohort had similar mean values for BMI and platelet counts when compared with the previous cohorts. The mean level of CRP in the PROCEED cohort was lower than the German cohort, but higher than the Indian cohort. The German cohort participants were enrolled during hospitalization which may have inflated the CRP values (for those with acute inflammation). In the German cohort, all COPPS components except pain rating independently correlated with the primary outcomes of hospitalization (), whereas in the PROCEED cohort, pain rating was the only parameter that independently associated with hospitalizations (see Supplementary Table 4, Supplementary Digital Content 1, https://links.lww.com/CTG/B183). The overall COPPS severity of disease in the PROCEED cohort was lower despite the fact that the German cohort included subjects with Cambridge classifications 0 (10%), 1 (32%), and 2 (12%) (), whereas the eligibility criteria for PROCEED required more severe morphological changes.
The distributions of the primary outcomes were similar between the Indian and PROCEED cohorts, but the number of hospitalizations was nearly double in the German cohort, and duration of hospitalizations was 3 times higher. The PROCEED cohort had a much lower proportion of subjects with hospitalizations for any reason (39.4%) during 12-month follow-up compared with 58.2% of the German cohort. Similarly, only 30.8% of the PROCCED cohort had hospitalizations for pancreas-related reasons vs 52.7% of the German cohort. The observed differences in the primary (and secondary) outcomes may reflect key differences in the study design, particularly location of recruitment. The German cohort was primarily recruited as inpatients, which biases the cohort toward the high severity category and a greater probability of future hospitalizations. Studies have documented a high risk of subsequent hospitalization after an inpatient admission for pancreatitis (). There were also differences in the secondary outcomes that were analyzed in all 3 COPPS studies. Specifically, new therapeutic endoscopies only correlated with COPPS in the German cohort (). Variability in healthcare between countries can directly influence our primary and secondary outcomes, which complicates the interpretation of the differences between COPPS studies.
A key advantage of PROCEED is the prospective multicenter nature of the study and the recruitment of a larger sample size than previous COPPS studies. CRP levels were batch-analyzed in an unblinded manner to minimize heterogeneity between clinical laboratories. As the PROCEED study continues, data from multiple years will become available and may be incorporated to extend COPPS or another predictive tool that can be applied to long-term outcomes. However, the current data only provided the number of hospitalizations over a 1-year follow-up and did not include the specific dates of hospitalizations, preventing more detailed analyses. One important weakness of the study is most subjects were White and non-Hispanic, which is not representative of the diversity of the American population. Across countries and cultures, there are differences between healthcare systems, attitudes, and standards that impact likelihood of admission and discharge. These differences may impact primary outcomes about number or duration of hospitalizations independently of patient characteristics. Therefore, a standardized predictor of hospitalization is unlikely to perform uniformly across different countries; additional factors may need to be incorporated that adjust for populations originating in different countries and medical systems.
The COPPS system provided a score that was significantly correlated with short-term prognosis (likelihood and duration of hospitalization within 12 months) of patients with CP. However, the strength of the association was small, suggesting further optimization of a clinical prediction tool is needed before clinicians can implement this into patient care.
Guarantor of the article: Jami L. Saloman, PhD.
Specific author contributions: D.Y. and D.L.C. co-direct the PROCEED study. D.Y., D.L.C., and J.L.S. generated the concept for this study. J.L.S. initiated and coordinated the study. S.K.P. and J.L.S. drafted the initial manuscript. S.K.P., S.L., and L.L. analyzed the data. K.S. performed the CRP assay. P.A.H., E.L.F., W.E.F., C.E.F., S.J.P., W.G.P., M.T., S.S.V., and S.K.V. serve as PI for their respective centers on the PROCEED study. All authors gave input to and have approved the final manuscript and had final responsibility for the decision to submit for publication.
Financial support: This work was supported by the National Cancer Institute (NCI) and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) under award numbers: R21DK122293 and K01DK120737 (J.L.S), U01DK108288 (S.S.V. and M.T), U01DK108300 (W.G.P), U01DK108306 (D.Y), U01DK108314 (S.J.P), U01DK108323 (E.L.F), U01DK108326 (W.E.F), U01DK108327 (P.A.H. and D.L.C), U01DK108332 (S.K.V), U01DK108320 (C.E.F), and U01DK108328 (S.K.P., L.L). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.