Risk Factors Associated With Progression Toward Endocrine Insufficiency in Chronic Pancreatitis.

Table 1 shows patient and disease characteristics for the 325 included in the analyses. The primary causes of pancreatitis were genetic/familial (42%), idiopathic (28%), pancreas divisum (12%), Sphincter of Oddi dysfunction (11%), alcohol (3%), and other causes (3%). The median interval between HbA1c tests was 86.5 days (IQR, 56–146; n = 316); the median interval between MMTT tests was 104 days (IQR, 78–162; n = 183). Because of this study used clinical data, the time intervals between repeated HbA1c and MMTT tests varied. To normalize for interpatient differences in time interval, we estimated the rate of change in each outcome over a 6-month time period. Glycated hemoglbin increased significantly by 0.062% ± standard error (SE), 0.029% ( P = 0.037) per 6 months; the ratio of AUC C-peptide to AUC C-glucose (“ratio”) decreased 0.0028 ± SE, 0011 per 6 months ( P = 0.014). Overall this suggests a modest increasing trend in average glucose and decreasing trend in prandial C-peptide secretion relative to prandial glucose in this cohort of patients with pancreatitis. However, on average per 6 months, AUC C-peptide (−12.62 ± SE, 14.94; P = 0.40), AUC glucose (+313.9 ± SE, 160.9, P = 0.053), and fasting glucose (+1.087 ± SE, 1.21; P = 0.37) did not change significantly. For each of the metabolic measures we considered, we tested the interaction of each of several dichotomous clinical features with the change per 6 months, specifically, these clinical features: sex, age (adult vs pediatric), history of surgery for CP, history of endoscopic retrograde pancreatography, history of tobacco smoking, pancreatic calcifications on imaging (present or absent), history of diabetes in a first-degree relative, clinical diagnosis of endocrine insufficiency, and any pre-existing diagnosis of diabetes. Table 2 shows the clinical subgroupings for which the rate of change in glycemic control parameters per 6 months has P < 0.05 for one or both subgroups, or for which the two subgroups differ in their respective rates of change per 6 month. Change in HbA1c differed significantly for those with pancreatic calcifications versus those without, with a more rapid increase in HbA1c in the group with calcific pancreatitis ( P = 0.003 for calcific vs non-calcific disease). We found a statistical significant increase over time in HbA1c level for the subgroups of patients who had pancreatic calcifications (n = 68, change over 6 months +0.22% ± SE, 0.06%, P = 0.0003) and the subgroup of patients without prior surgery for CP (n = 269, change over 6 months +0.064% ± SE, 0.32%, P = 0.047). Glycated hemoglobin also increased significantly per 6 months in non-smokers but not smokers. Similarly, progression in fasting glucose per 6 months was more pronounced for non-smokers than smokers ( P = 0.028), with a modest but non-significant trend towards increasing fasting glucose in non-smokers. Area under the curve glucose (from MMTT) increased more rapidly per 6 month for those with prior surgical drainage history compared to those without ( P = 0.0011); AUC glucose increased significantly in the group with prior surgical drainage and also in patients who were non-smokers, had no family history of diabetes, and had not yet been diagnosed with exocrine insufficiency. Table 3 shows the clinical subgroupings for which the rate of change in beta cell function parameters per 6 months has P < 0.05 for one or both subgroups, or for which the two subgroups differ in their respective rates of change per 6 month. We also measured C-peptide response to mixed meal tolerance testing and calculated (1) AUC C-peptide and (2) the ratio (of AUC C-peptide to AUC glucose) as markers of beta cell function. Here, history of tobacco smoking was associated with a statistically significant decline in AUC C-peptide (n = 68, rate of change per 6 months −66.5 ± SE, 30.3; P = 0.029), and smokers progressed more rapidly in loss of AUC C-peptide than non-smokers ( P = 0.04). To determine if a dose-dependent effect of smoking might be present, we examined the association between pack-years tobacco smoking and rate of change in AUC C-peptide and ratio in smokers. Greater pack-years was associated with more rapid decline in AUC C-peptide (rate of change per 6 months −3.02 ± SE, 1.32; P = 0.0273) and ratio (rate of change per 6 months −0.00022 ± SE, 0.000074; P = 0.0042). Notably, the ratio declined over time in multiple subgroups, suggesting an overall loss of beta cell function over time across the cohort. Statistically significant declines in ratio were observed for female patients, smokers, and those without pancreatic calcifications, those without family history of diabetes, and those without personal history of diabetes. The ratio also declined significantly in patients with no history of prior surgery for CP but also showed a trend to decline in those with a history of any CP surgery. For drainage procedures, specifically (Puestow, Frey), however, the decline in the ratio was greater per 6 months than for those with no prior drainage ( P = 0.03). Because of potential associations between patient and disease risk factors, we further evaluated change over time for each of our 5 metabolic measures in a multivariate model. For each metabolic measure previously described, we tested the adjusted effect of one of these seven risk factors on progression: sex, age, smoking history, family history of diabetes, personal history of diabetes, pancreatic drainage surgery, and pancreatic calcifications, after adjusting for the remaining six risk factors. Table 4 shows results from multivariate models for measures of glycemic control and beta cell function. Analyses of AUC glucose showed no significant results. Patients with calcific pancreatitis had a greater rise in HbA1c than those without calcific disease ( P = 0.002). Fasting glucose declined in smokers, those without diabetes mellitus history, and those with drainage, differing from their counterpart subgroups. The multivariate analyses for AUC C-peptide and ratio showed similar results to the unadjusted univariate analyses. Area under the curve C-peptide declined more rapidly in females vs. males ( P = 0.04) and in smokers vs. non-smokers ( P = 0.016). The decline in ratio was similarly more rapid in females ( P = 0.03) and smokers ( P = 0.016). The decline in ratio per 6 months remained significant in those with prior drainage surgery, but fell short of significance in those without drainage ( P = 0.069). As an exploratory analysis, because the patients included in this study were undergoing TPIAT, we evaluated whether the rate of change in any glycemic or islet function measure could predict the results of the islet isolation procedure in islet equivalents (IEQ) and islet equivalents per kilogram (IEQ/kg). There was no significant association between unit of change in any of the metabolic measures and IEQ or IEQ/kg ( P > 0.05 for all).

We reviewed clinical characteristics and laboratory results from before TPIAT for 325 patients undergoing TPIAT for CP and/or acute recurrent pancreatitis (ARP) at the University of Minnesota Medical Center. All patients were enrolled in a single-center longitudinal study of TPIAT over a 10-year interval from October 2008 to May 2018. Informed consent or parental consent, as appropriate, was obtained from all patients in the study. The study protocol was reviewed and approved by the University of Minnesota Institutional Review Board. All patients were reviewed by a multi-disciplinary team for TPIAT and were included only if diagnostic criteria for intractable CP or ARP were met as previously described. 14 Briefly, patients were required to have at least one of these criteria supporting diagnosis: (1) 2 or more episodes of acute pancreatitis (amylase/lipase > 3 times upper limit of normal) with disability or intractable pain; (2) in non-calcific CP, at least 2 abnormal studies suggestive of CP from magnetic resonance cholangiopancreatography, endoscopic ultrasound (≥4 criteria for CP), and pancreatic function tests; (3) pancreatic calcifications on CT scan; or (4) histopathologic confirmation of CP. Participants were included in the study if they had testing repeated on 2 different days for: (1) HbA1c level; (2) MMTT with 0, 60, and 120 minute samples for glucose and C-peptide levels; or (3) both HbA1c and MMTT. As part of their TPIAT evaluation, all patients underwent MMTT during the initial consultation visit. The MMTT was repeated immediately before TPIAT, when able, if there had been a >3-month interval between the consult and surgery date or if the patient was a participant in a study that required MMTT at baseline study. For the MMTT, glucose and C-peptide were drawn in the morning following an overnight fast, 6 mL/kg Boost HP (maximum dose of 360 mL, Nestle Health Science, Bridgewater, NJ) was ingested over 5–10 minutes, and glucose and C-peptide were drawn at 60 and 120 minutes following the Boost HP. The area under the curve for C-peptide (AUC C-peptide) and glucose (AUC glucose) were calculated using the trapezoid method. The ratio of AUC C-peptide to AUC glucose (“ratio”) was calculated. HbA1c level was also obtained. For this study’s purposes, participants were included in the analyses only if they had two MMTTs (n = 183) or two Hba1c levels (n = 316) successfully completed. If HbA1c or AUC glucose and fasting glucose increased over time, these were considered suggestive of worsening of glycemic control, while decreasing AUC C-peptide and/or ratio were considered suggestive of worsening islet function. Based on the literature, we identified clinical features that might be risk factors for worsening glycemic control or islet function. The potential risk factors for progression of endocrine insufficiency included in univariate analyses were: age at TPIAT (analyzed as a continuous measure and also dichotomized as adult vs pediatric [<18 years]), family history of diabetes (one or more first-degree relatives), pre-existing diagnosis of diabetes before TPIAT, history of tobacco smoking (former or current), exocrine insufficiency (defined by low fecal elastase-1 or a physician-reported diagnosis of pancreatic exocrine insufficiency in the electronic medical record), pancreatic parenchymal calcifications or ductal stones on imaging, prior surgical procedures (lateral pancreatojejunostomy, Whipple, distal pancreatectomy, or open sphincterotomy), and prior endoscopic retrograde cholangiopancreatography. Among smokers, we collected tobacco smoking dose as pack-years from the medical record where available (n = 65). Islet mass isolated as islet equivalent (IEQ) and IEQ per kg of body weight was also abstracted from the islet lab records. Patient and disease characteristics are reported as median (and interquartile range [IQR]) for continuous characteristics and count (and percentage) for categorical characteristics. Changes in metabolic measures over time were tested using linear mixed models with time as the fixed effect and person-specific slopes and intercepts as the random effects, distributed as bivariate normal with a general covariance matrix. We also tested changes in metabolic measures over time in different pairs of subgroups, together with associations between corresponding dichotomous clinical features and changes in metabolic measures over time, by using the above linear mixed models with added fixed effects of one (unadjusted model) or seven (adjusted multivariate model) dichotomous clinical features and their interaction(s) with time. Associations between TPIAT islet yield and change over time in metabolic measures were tested using simple linear regression models with the predictor being the unit of change in metabolic measures, which was defined as the ratio {follow-up measure minus baseline measure} divided by the follow-up time. We used R (version: 3.5.1, R Core Team 2018, Vienna, Austria) for the above analyses. 15 We implemented the linear mixed models using the function “lme” in R package “nlme”, and the function “emtrends” in R package “emmeans” to calculate the effects for subgroups in the adjusted models. We used the “lm” function in R package “stats” for the simple linear regression models.

Diabetes is a well-known late complication of recurrent acute and chronic pancreatitis, but the literature has little data describing progressive, longitudinal changes in metabolic measures in patients with chronic pancreatitis. We had a unique opportunity to access laboratory data from a large cohort of patients with pancreatitis who had at least two HbA1c and MMTT, with a primary study aim to assess the rate of change in glycemic and insulin secretory measures derived from these tests. We observed a rise in HbA1c (0.062% ± SE, 0.029% per 6 months) and a decline in the ratio of AUC C-peptide to AUC glucose (0.0028 ± SE, 0.0011 per 6 months) in the overall cohort. These changes, although small, were detectable even in the relatively short interval between the repeat tests in this dataset. Most importantly, insulin production (C-peptide) relative to glucose level (the ratio) declined in nearly all subgroups. These results support the expected finding of a slow but significant attrition in endocrine pancreatic function over time in patients with CP, and suggest that this progression may be measurable with clinically available testing protocols. Our second aim was to identify clinical features of patients with significant decline in endocrine function. In this study, smoking appeared detrimental to insulin secretion and beta cell function in a dose dependent manner. When we adjusted for other clinical features, smoking remained associated with apparent disease progression, with smokers showing greater decline in AUC C-peptide and ratio compared to non-smoker. Within the cohort of smokers with available history of pack-years of smoking, decline was more rapid with heavier smoking history. Tobacco smoking in CP patients has been associated with greater levels of pro-inflammatory cytokines and fibrosis increased risk of pancreatic calcification, pancreatic islet destruction, and increased incidence of diabetes. 9 , 10 , 16 , 17 Therefore, one possible hypothesis is that additional damage to the pancreas induced by tobacco smoking could contribute to the accelerated decline in insulin secretion and beta cell function that we observed in current or past smokers. Likewise, cigarette smoking has been associated with increased risk of type 2 diabetes, possibly through alterations in body composition, insulin sensitivity, and beta cell function in smokers. 18 – 20 We also found that prior surgical drainage was associated with worsening glycemic control (AUC glucose) and decline in beta cell function (the ratio). When adjusted for other clinical features, this association was no longer statistically significant but it retained a strong (though non-significant) trend towards a greater decline in ratio in patients with drainage than in those without prior drainage. Surgical drainage procedures were defined as Peustow, Frey, or related variants performed in patients with CP and a dilated duct. 21 – 23 24 , 25 Whether surgical drainage slows diabetes progression in CP patients by preserving exocrine and endocrine tissue or increases the risk for subsequent diabetes has been unclear in prior literature. In the larger North American study, CP patients with a history of pancreatic surgery, including drainages, had increased odds of diabetes diagnosis. 7 In contrast, Malka et al. reported that except for distal pancreatectomy (which had increased risk for diabetes), elective pancreatic surgical procedures including drainage were not associated with increased risk of diabetes in CP patients. 8 Although our study had few patients with surgical drainage (only 13), we observed a rapid rate of decline in beta cell function in those with drainage compared to those without drainage after adjusting for other clinical features. Further investigation with a larger sample size and longer time interval would be needed to confirm an association between surgical drainage and accelerated decline in endocrine function. Pancreatic calcification has been previously identified as a major risk factor for development of DM in CP patients. 1 , 7 , 8 We also found people with pancreatic calcifications tended to have a faster rise in HbA1c compared to those without calcifications after adjusting for other clinical variables. Pancreatic calcifications are presumed to indicate more severe damage to the pancreatic parenchyma and have been associated with abnormal exocrine secretory function. 26 Our findings support the previous findings that calcifications indicate a risk for DM and may also be a risk marker for accelerated deterioration of glycemic control in CP, even before onset of frank DM. 27 In this study, females had a more rapid decline in ratio compared to males, adjusted for other risk factors. This may suggest an impact of sex on rate of progression of disease. Conversely, our female patients may differ from our male patients in this particular population, patients who present for consideration of TPIAT. One possibility is that female patients present at a different stage of disease than male patients, and thus have a different trajectory at the time that we studied them. In other subgroups, we observed a decline from baseline that was not necessarily predicted—i.e., those with no family history of diabetes, no exocrine insufficiency, and no personal history of diabetes had a more rapid deterioration in some metabolic measures. One possible explanation is that these subgroups have more pancreatic reserve, less scarring at time of initial assessment, so in essence they have ‘more to lose’ over a short interval of time. However, a more likely explanation is that these subgroups were large in size, and thus better powered to detect a change from their baseline. These groups did not differ significantly from their dichotomous counterpart—for example, although the decline in ratio over 6 months was significant in those with no family history of diabetes, the rate of decline in ratio did not differ significantly between those with and without diabetes. We observed mixed results using fasting glucose as a metabolic measure. The overall cohort of all patients showed no significant change in fasting glucose per 6-months. However, when looking at putative risk factors for decline in endocrine function, non-smokers appeared to fare worse than smokers, and those with a drainage procedure seemed to fare worse than those with no drainage. We suspect that fasting glucose is a less reliable measure than hemoglobin A1c, which assesses glucoses over a 3-month period, or AUC glucose, which incorporates both fasting and post-prandial glucoses. The small differences in fasting glucose that can occur from day to day may lead to imprecise estimates of rates of change in function. More fasting measures over a longer study interval may be necessary to determine the true progression in fasting glucose over time and risk factors for rise in fasting glucose. The strengths of our study are a prospective and longitudinal design, use of repeated glycemic control and beta cell function measurements, and a large, diverse group of CP patients. Inclusion of the MMTT has the advantage of greater assessment of insulin secretion and prandial glucose control, compared to using only fasting glucose and C-peptide (insulin) or HbA1c alone. The MMTT simulates a physiologically equivalent response from a normal human meal consumption as opposed to the pure glucose stimulus of an oral glucose tolerance test. 28 However, this study is limited by reliance on a clinical cohort for study. Despite the large size of the overall cohort, our power to find small differences between subgroups may also be limited by small sample sizes of certain risk subgroups. Because these tests were performed in the clinic setting, based on clinical need, the follow-up time interval varies between study patients, and the median time between the first and second test is relatively short. We also did not incorporate pancreatic polypeptide, glucagon, or incretin measures into the mixed meal because these measures were not available clinically, but alterations in these other pancreatic and enteroendocrine hormones may be important to fully understand endocrine dysfunction in CP. 29 – 32 Future studies to confirm these preliminary findings would ideally include longer follow up, at specified time intervals, in a research setting. In summary, we observed an overall decline in pancreatic endocrine function in a large population of patients with CP, with a significant increase in HbA1c and decline in ratio (of C-peptide to glucose in MMTT) over a 6 month interval of time. This decline in the glycemic control and beta cell functional appeared to be accelerated when tobacco smoking, surgical drainage, or pancreatic calcification were present, suggesting these subgroups may need to be followed more carefully for development of new onset diabetes. These clinical findings should be confirmed with prospectively designed and long-term studies.

Chronic pancreatitis (CP) is a fibroinflammatory disease of the pancreas associated with development of exocrine and endocrine insufficiency. Destruction of pancreatic islets from fibrosis leads to eventual pancreatogenic diabetes in 25–80% of patients affected with CP. 1 – 5 However, the time from pancreatitis diagnosis to diabetes diagnosis varies and subtle changes in glucose and C-peptide levels during mixed meal tolerance testing (MMTT) have been observed even in non-diabetic patients with CP, suggesting early endocrine insufficiency preceding frank diabetes. 6 Several risk factors for diabetes mellitus have been identified in patients with CP. Pancreatitis disease features that may increase risk for pancreatogenic diabetes include exocrine insufficiency, longer disease duration, older age at pancreatitis diagnosis, prior surgical treatment, and pancreatic calcifications. 1 , 7 , 8 Other clinical features associated with development of diabetes in those with CP include tobacco smoking, overweight or obese body mass index (BMI), and positive family history of diabetes. 7 , 9 , 10 Most research to date has focused on risk factors for the clinical diagnosis of diabetes. Stimulatory studies of insulin and C-peptide secretion have largely been cross-sectional, with little data on longitudinal changes. 11 – 13 We do not know how rapidly deficits in insulin secretion develop or how quickly glycemic control declines in those with CP, nor risk factors for changes in insulin/C-peptide and glucose measures. This study sought to (1) describe the rate of change of glycemic and insulin secretory measures from repeated glycated hemoglobin (HbA1c) and MMTT in patients being considered for total pancreatectomy with islet auto-transplantation (TPIAT), and (2) determine if clinical markers could identify those likely to have significant deterioration in glycemic control or stimulated C-peptide levels.