A Reduced Pancreatic Polypeptide Response is Associated With New-onset Pancreatogenic Diabetes Versus Type 2 Diabetes

Abstract

Purpose

Pancreatogenic diabetes refers to diabetes mellitus (DM) that develops in the setting of a disease of the exocrine pancreas, including pancreatic ductal adenocarcinoma (PDAC) and chronic pancreatitis (CP). We sought to evaluate whether a blunted nutrient response of pancreatic polypeptide (PP) can differentiate these DM subtypes from type 2 DM (T2DM).

Methods

Subjects with new-onset DM (<3 years’ duration) in the setting of PDAC (PDAC-DM, n = 28), CP (CP-DM, n = 38), or T2DM (n = 99) completed a standardized mixed meal tolerance test, then serum PP concentrations were subsequently measured at a central laboratory. Two-way comparisons of PP concentrations between groups were performed using Wilcoxon rank-sum test and analysis of covariance while adjusting for age, sex, and body mass index.

Results

The fasting PP concentration was lower in both the PDAC-DM and CP-DM groups than in the T2DM group (P = 0.03 and <0.01, respectively). The fold change in PP at 15 minutes following meal stimulation was significantly lower in the PDAC-DM (median, 1.869) and CP-DM (1.813) groups compared with T2DM (3.283; P < 0.01 for both comparisons). The area under the curve of PP concentration was significantly lower in both the PDAC-DM and CP-DM groups than in T2DM regardless of the interval used for calculation and remained significant after adjustments.

Conclusions

Fasting PP concentrations and the response to meal stimulation are reduced in new-onset DM associated with PDAC or CP compared with T2DM. These findings support further investigations into the use of PP concentrations to characterize pancreatogenic DM and to understand the pathophysiological role in exocrine pancreatic diseases (NCT03460769).

Pancreatogenic diabetes mellitus (DM) refers to DM that is a consequence of a disease of the exocrine pancreas, including pancreatic ductal adenocarcinoma (PDAC) and chronic pancreatitis (CP) (1). Although there are no specific treatments for pancreatogenic DM (also referred to as type 3c DM, pancreoprivic DM, or diabetes of the exocrine pancreas), there are multiple important reasons to differentiate pancreatogenic DM from the much more prevalent type 2 DM (T2DM) (2). For example, in patients with PDAC, DM often precedes the onset of other cancer-related symptoms and may represent an opportunity for early detection (3, 4). In patients with CP, additional comparisons and contrasts with T2DM are needed to understand the underlying physiological changes, which will help to identify therapies that are more effective and/or associated with lower morbidity (such as treatment-related hypoglycemia). Although criteria have been proposed for diagnosis of pancreatogenic DM, they have not been validated, and this remains an important clinical challenge (5).

A multidisciplinary group of experts previously recommended that a blunted response in pancreatic polypeptide (PP) hormone levels to mixed meal stimulation is a method of distinguishing pancreatogenic DM from T2DM (6). Prior studies examining fasting PP levels to separate groups have yielded mixed results. A small pilot study (n = 18) demonstrated that the PP response was significantly lower in subjects with new-onset DM associated with PDAC than T2DM (7). This effect was particularly evident when the tumor was in the proximal portion of the pancreas. Considering the persistent, unmet clinical need to differentiate these various types of DM, we designed the DETECT study to evaluate these preliminary findings in a larger multicenter study (8). In the current analysis, we investigated differences in the PP response in new-onset pancreatogenic DM associated with pancreatic ductal adenocarcinoma (PDAC-DM) or chronic pancreatitis (CP-DM) from T2DM.

Materials and Methods

The current analysis includes subjects with new-onset DM who participated in the DETECT study from September 2018 to August 2021 at one of 9 clinical centers from the Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer (NCT03460769). The study was approved by each center's institutional review board, then subsequently transitioned to a single institutional review board in 2021. Clinical center teams, including gastroenterologists, endocrinologists, radiologists, and research staff, collected clinical data using standardized case report forms, which were electronically entered into the Consortium's central database. Missing or discordant data were reconciled by the Coordination and Data Management Center.

Eligibility Criteria

Patients with new-onset diabetes in the setting of T2DM without structural disease, or with new-onset diabetes in the setting of PDAC or CP, were invited for participation, as previously described (8). For the purposes of this study, “new onset” was defined as DM that was diagnosed within 36 months of the mixed meal tolerance test (MMTT) (or from clinical diagnosis for the PDAC-DM group). Definitions for CP and PDAC were adopted from those used in other studies from the Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer (CPDPC) (9, 10). Patients with pancreatic cancer other than ductal adenocarcinoma were not included. The age at enrollment ranged from 40 to 85 years. Patients with previous pancreatic surgery, gastric surgery, or treatment for PDAC (including chemotherapy) were excluded from participation.

Mixed Meal Test Protocol and Sample Processing

All subjects underwent a standardized 2-hour MMTT, as previously described (8). All antidiabetic medications and pancreatic enzyme replacement therapy were held on the morning of the MMTT (8). Blood was collected at −5, 0, 15, 45, 60, 90, and 120 minutes following ingestion of a standard mixed meal stimulus (12 ounces of original Boost drink). Results from the samples collected at −5 and 0 minutes were averaged to represent the fasting baseline value. Blood collection, processing, and storage were performed according to specific standard operating procedures (11). Blood samples used in this analysis were collected at each time point without the addition of enzyme inhibitors and allowed to clot for 30 to 60 minutes before centrifugation. Aliquots were subsequently frozen and then transferred for storage at a central biorepository. Serum samples were batch analyzed by the Translation Core in the Center for Diabetes and Metabolic Diseases at Indiana University. Glucose was quantitated using a Randox Daytona clinical analyzer and PP concentrations using the Meso Scale Diagnostics V-PLEX platform. The lower limit of detection for this PP assay was 0.41 pg/mL and within-assay coefficient of variation was 2.4%. All samples for each subject were run in duplicate on the same plate. Nonspecific binding to numerous hormones, including peptide YY isoforms, is below the lower limit of quantification as tested by the manufacturer. Hemoglobin A1c (HbA1c) for each subject was determined by the pathology laboratory at each study site using the gold standard HPLC method.

Power Calculation

Based on preliminary data examining PP responses in PDAC and T2DM, we planned to enroll up to 136 subjects with PDAC-DM, 136 subjects with CP-DM, and 100 subjects with new-onset T2DM to achieve our specified parameters of diagnostic performance, specifically, 25% (null hypothesis) vs 44% specificity assuming a sensitivity of 90%, with 80% power at a 2-sided significance level of 5% each (7, 8). The initial analysis reported here was performed after 3 years of enrollment, which was operationally challenging because of the ongoing COVID-19 pandemic and slower than projected enrollment. Because of the sample size at the time of the current analysis, receiver operating characteristic analyses would be unreliable, and we sought to study the hypothesis that the fold change in PP concentration following meal stimulation is significantly lower in the groups with DM associated with PDAC or CP than T2DM. Fold change was calculated by division of the PP concentration at the postprandial timepoints and the baseline value.

Statistical Analysis

Baseline characteristics were compared using 2-sample t tests/Wilcoxon rank-sum tests, and χ²/Fisher exact tests, as appropriate, to provide between-group comparisons of interest (ie, PDAC-DM vs T2DM and CP-DM vs T2DM). The raw data distribution of PP concentrations at baseline and postprandial time points demonstrated nonnormal distributions within groups. Therefore, 2-way comparisons of PP response (including fold change and area under the curve [AUC]) between groups were performed using Wilcoxon rank-sum test. The AUC was calculated using the trapezoidal method and was assessed by considering the increase from baseline (incremental AUC). Analysis of covariance was performed on log10-transformed data to assess for persistent between-group differences after adjusting for age, sex, and body mass index (BMI). Linear regressions were used to evaluate the effects of potential confounders (including baseline HbA1c, fasting glucose, age, BMI, and tumor size [for the PDAC-DM group]) for the between-group differences in PP response. A P value <0.05 was considered statistically significant.

Results

Study Population

A total of 165 subjects with either new-onset PDAC-DM (n = 28), CP-DM (n = 38), or T2DM (n = 99) completed the DETECT study during the study period. Compared with T2DM controls, the mean age of participants was older in the PDAC-DM group and younger in the CP-DM group (Table 1). The proportion of women was significantly lower in the CP-DM group compared with the T2DM group, and the mean BMI was lower in both PDAC-DM and CP-DM groups. The duration of DM was similar between groups, as were fasting blood glucose levels on the day of the MMTT. However, the median HbA1c values were significantly higher in the PDAC-DM (7.1%) and CP-DM (6.7%) groups than in the T2DM group (6.4%; < 0.01 and 0.01, respectively). There were differences in treatment of diabetes before the study visit with higher proportions using insulin in the PDAC-DM and CP-DM groups. Approximately two-thirds (68%) of PDAC-DM subjects had an early cancer stage (ie, I or II) at the time of enrollment.

Table 1.

Baseline characteristics of subjects with new onset PDAC-DM, CP-DM, and T2DM

	T2DM, n = 99	PDAC-DM, n = 28	CP-DM, n = 38	P value (PDAC-DM vs T2DM)	P value (CP-DM vs T2DM)
Demographics
ȃMean age (SD), y	60.4 (9.1)	69.0 (7.1)	56.0 (11.2)	<0.0001	0.019
ȃFemale sex	54 (54.4%)	13 (46.4%)	9 (23.7%)	0.52	0.0012
ȃMedian BMI (kg/m2), IQR	33.2 (28.8-37.5)	27.0 (25.4-30.2)	24.4 (21.2-29.7)	0.0005	<0.0001
Race
ȃȃWhite	78 (79.6%)	26 (92.9%)	35 (92.1%)	0.049	0.22
ȃȃBlack	15 (15.3%)	0 (0.0%)	3 (7.9%)
ȃȃOther	5 (5.1%)	2 (7.4%)	0 (0.0%)
Ethnicity
ȃȃHispanic	3 (3.1%)	0 (0.0%)	1 (2.6%)	>0.99	>0.99
ȃȃNon-Hispanic	94 (96.9%)	28 (100.0%)	37 (97.4%)
Diabetes characteristics
ȃMedian duration of DM, month, IQR	11.2 (2.60-18.9)	7.0 (0.13-15.0)	14.8 (3.92-21.8)	0.47	0.57
ȃFamily history of DM in a first- degree relative	63 (63.6%)	17 (60.7%)	15 (39.5%)	0.83	0.013
ȃMedian fasting blood glucose (IQR), mg/dL	135 (116-151)	138 (126-167)	144 (125-171)	0.06	0.12
ȃMedian hemoglobin A1c (IQR), %	6.4 (5.9-7.1)	7.1 (6.1-8.3)	6.7 (6.2-7.6)	0.0035	0.012
ȃUse of oral antidiabetic medication(s)	70 (70.7%)	15 (53.6%)	16 (42.1%)	0.11	0.0029
ȃUse of insulin	6 (6.1%)	11 (39.3%)	11 (28.9%)	<0.0001	0.0007
ȃNo oral medication or insulin	27 (27.3%)	6 (21.4%)	15 (39.5%)	0.63	0.21
Characteristics of pancreatic disease
Use of pancreatic enzyme replacement therapy
ȃYes	0 (0.0%)	2 (7.1%)	25 (65.8%)	—	—
ȃNo	99 (100.0%)	26 (92.9%)	13 (34.2%)
Tumor location
ȃProximal	N/A	22 (78.6%)	N/A	—	—
ȃDistal	N/A	6 (21.4%)	N/A
AJCC cancer stage
ȃI/II	N/A	17 (68.0%)	N/A	—	—
ȃIII/IV	N/A	8 (32.0%)	N/A
ȃMedian tumor size (IQR), cm	N/A	3.50 (3.00-4.00)	N/A	—	—
ȃPresence of calcification(s)	N/A	N/A	32 (84.2%)	—	—

Abbreviations: AJCC, American Joint Committee on Cancer; BMI, body mass index; CP, chronic pancreatitis; DM, diabetes mellitus; IQR, interquartile range; N/A, not available; PDAC, pancreatic ductal adenocarcinoma; T2DM, type 2 diabetes mellitus.

Assessment of Pancreatic Polypeptide Concentrations

The median fasting PP concentration was significantly lower in the PDAC-DM (45.3 pg/mL [24.6-110.8]) and CP-DM (23.9 pg/mL [12.1-80.7]) groups when compared with T2DM (88.8 pg/mL [52.0-141.0]) (P = 0.01 and <0.01, respectively; Fig. 1). These differences remained significant when adjusting for age, sex, and BMI (P = 0.00012 [PDAC-DM vs T2DM] and P < 0.0001 [CP-DM vs T2DM]). The PP concentrations were significantly lower in the CP-DM group at each of the subsequent time points compared with T2DM (Fig. 2) and remained significant after adjustments for the covariates (data not shown). Similar differences were observed when comparing PDAC-DM and T2DM, with the exception that the difference at 120 minutes did not maintain statistical significance (P = 0.06). The fold change in PP at 15 minutes following meal stimulation was significantly lower in the PDAC-DM (median, 1.869) and CP-DM (1.813) groups compared with T2DM (3.283, P < 0.01 for both comparisons).

Figure 1.

Box plots depicting the fasting (basal) serum concentrations of pancreatic polypeptide (PP) in participants with new-onset PDAC-DM, CP-DM, and T2DM. Lines represent medians and boxes illustrate interquartile ranges.

Figure 2.

Line graph depicting the pancreatic polypeptide (PP) concentration (pg/mL) during a 120-minute mixed meal tolerance test in subjects with new-onset PDAC-DM, CP-DM, and T2DM. Dots indicated medians and bars represent interquartile ranges.

Postprandial Pancreatic Polypeptide Response Between Groups

The fold change in PP following meal stimulation was significantly reduced in the PDAC-DM group at 15 minutes (median, 1.869 vs 3.283 in T2DM, P = 0.00058), but not at 30 minutes (P = 0.07; Fig. 3). The change in relative PP was reduced in the CP-DM group at both 15 and 30 minutes (1.813 at 15 minutes, 2.375 at 30 minutes; P = 0.0002 and P = 0.0067 compared with T2DM, respectively). There were no differences observed in either group compared with T2DM controls beyond 30 minutes. After adjusting for age, sex, and BMI, the differences remained significant for CP-DM compared with T2DM, but not for the comparison with the PDAC-DM group (Supplementary Table 1) (12). Last, we did not identify a linear correlation between the fold change in PP (at 15, 30, and 60 minutes) with age, BMI, tumor size, fasting blood glucose, and baseline HbA1c values for the groups, with the exception of a negative correlation with advancing age in the T2DM controls (representative data from the fold change at 15 minutes are shown in Supplementary Fig. 1) (12).

Figure 3.

Box plot depicting the fold change in pancreatic polypeptide (PP) concentrations values during a 120-minute mixed meal in participants with new onset PDAC-DM, CP-DM, and T2DM. Lines represent medians and bars indicate the interquartile ranges.

We also sought to quantify differences in PP hormone exposure in the PDAC-DM and CP-DM groups using AUC. The AUC of PP concentrations was significantly lower at all time points in both the PDAC-DM and CP-DM groups compared with T2DM (Table 2). Importantly, these differences persisted for both comparisons for log10-transformed AUC across all time points after adjusting for age, sex, and BMI. However, although significant differences were observed for both between-group comparisons at all time points (except at 120 minutes for the PDAC-DM vs control comparison) for incremental AUC based on log10-transformed PP concentration, only the comparisons between the CP-DM and control groups at all time points remained significant after controlling for age, sex, and BMI, but not for any comparison between PDAC-DM and control groups (Supplementary Table 2) (12).

Table 2.

Area under the curve of postprandial pancreatic polypeptide concentrations in participants with new-onset PDAC-DM, CP-DM, and T2DM. Values represent medians and interquartile ranges (pg min/mL). Between-group differences remained significant after adjusting for age, sex, and BMI

	T2DM, n = 99	PDAC-DM, n = 28	CP-DM, n = 38	Unadjusted P value (PDAC-DM vs T2DM)a	Adjusted P value (PDAC-DM vs T2DM)b	Unadjusted P value (CP-DM vs T2DM)a	Adjusted P value (CP-DM vs T2DM)b
AUC0-15	2969.7 (1642.0-4913.3)	1184.6 (453.0-3319.9)	610.7 (306.9-2009.8)	<0.0018	<0.0001	<0.0001	<0.0001
AUC0-30	7824.2 (4554.7-12 278.5)	3119.2 (1030.7-8823.2)	1439.1 (656.1-5119.5)	<0.0031	<0.0001	<0.0001	<0.0001
AUC0-45	12 483.4 (7547.2-19 154.5)	5625.0 (1652.7-14 758.1)	2583.0 (1056.6-8316.3)	<0.0062	0.00012	<0.0001	<0.0001
AUC0-60	16 404.2 (10 387.5-25 108.4)	8116.6 (2295.4-20 109.6)	3541.2 (1522.2-10 966.3)	<0.0093	0.00021	<0.0001	<0.0001
AUC0-90	22 263.5 (14 711.6-36 047.1)	12 432.4 (3661.1-30 236.8)	549 801 (2374.3-15 567.6)	0.013	0.00035	<0.0001	<0.0001
AUC0-120	27 332.9 (18 176.4-45 180.4)	16 007.0 (5083.9-39 594.2)	7143.8 (3158.0-18 966.0)	0.015	0.00058	<0.0001	<0.0001

Abbreviations: AUC, area under the curve; BMI, body mass index; CP, chronic pancreatitis; DM, diabetes mellitus; PDAC, pancreatic ductal adenocarcinoma; PP, pancreatic polypeptide; T2DM, type 2 diabetes mellitus.

Unadjusted P values are based on Wilcoxon rank-sum tests for PP concentration.

Adjusted P values are based on analysis of covariance adjusting for age, sex, and BMI for log10-transformed values (ie, log10(pg/mL) min).

Subgroup Analysis

Because of the variable distribution of PP producing cells in the pancreas, a subgroup analysis was performed for all comparisons with the PDAC-DM group restricted to the 22 subjects with a tumor located in the proximal portion of the gland. In summary, the fasting and PP responses (both fold change of PP and AUC of PP concentrations) in the PDAC-DM proximal group were further reduced leading to larger between-group differences compared with T2DM controls (Supplementary Figs. 2, 3, 4; Supplementary Tables 3, 4) (12). The comparisons of incremental AUC based on log10-transformed PP concentration between the PDAC (proximal)-DM and control groups at 15 and 30 minutes remained significant after controlling for age, sex, and BMI. New significant differences were identified in the fold change of PP at 30 and 45 minutes but were not observed after adjustments.

Conclusions

We demonstrate that basal PP levels and the PP response to meal stimulation are reduced in new-onset DM associated with either PDAC or CP compared with T2DM. Among various methods of quantifying PP response, the AUC of serum PP concentrations manifested the largest differences between groups, with significantly reduced responses in both forms of pancreatogenic DM (compared with T2DM) that persisted after adjustments for age, sex, and BMI at all time points. Fasting and postprandial PP concentrations were further reduced in a subgroup analysis of participants with a tumor located in the proximal gland. Additional investigations are needed to confirm our observations in a larger study population, clarify the potential role of a blunted PP response in the diagnostic approach to pancreatogenic DM in the setting of PDAC or CP, and to further compare the underlying pathogenic mechanisms of pancreatogenic DM.

The blunted PP response (compared with T2DM) in our subects with new-onset DM in the setting of PDAC is congruent with results from a pilot study and corroborates the findings of a lower fasting PP value reported by others (7, 13). The previous observation of further reduction in PP response when the pancreatic tumor was in the proximal pancreas (ie, uncinate or head) was also replicated. This observation is not surprising because the majority of PP producing islet cells are found in this area of the pancreas. Our results also support prior studies suggesting a reduced PP response in CP subjects with DM. In fact, the reduction in subjects with CP-DM was even more robust than those seen in the PDAC-DM group. Although fasting PP concentrations were also lower in both groups with pancreatogenic DM compared with T2DM, this observation has not been consistent across all studies and may benefit from further studies to determine to what degree these changes may reflect technical variations (eg, different assays) or differences in the study population (14).

A key observation from our study is the difference in baseline characteristics of age, sex, and BMI between the groups. These findings were not unexpected because attempts were not made to match participant characteristics across the 3 groups. Although the current study design does not allow us to directly provide a conclusion, a potential explanation is that these differences reflect true biological characteristics associated with these types of DM. This hypothesis is supported by other, recent clinical studies illustrating that advancing age and lower body weight at time of diagnosis are associated with increasing risk for PDAC (15–17). Additionally, although the sex distribution for T2DM and PDAC (with or without DM) is typically equal, there is a male predominance in CP. Previous studies have demonstrated that age, sex, and BMI are associated with differences in fasting PP and/or PP responses (18, 19). Therefore, to evaluate for potential confounding, we assessed for correlation between age, BMI, and PP response and did not observe a correlation. Nevertheless, we ultimately elected to evaluate for residual effects in all PP assessments following adjustment for age, sex, and BMI to eliminate potential confounding and believe this is an important consideration for future studies comparing these and other types of pancreatogenic DM.

The current study generates important, unanswered questions. First, “Why are basal and postprandial PP responses lower in these forms of pancreatogenic DM?” Although the current study was not designed to address this mechanistic question, our data do provide some clarity to the pathophysiology of new-onset DM in adults with PDAC. An important finding relates to the larger between-group differences observed in both fasting PP concentration and PP responses when only considering proximal cancers (the location of more than two-thirds of all pancreatic cancers and the portion of the gland with the highest concentration of PP secreting islets). Importantly, though, there was no correlation between tumor size and PP response in these subjects with PDAC, suggesting this change is not simply related to islet destruction. An alternative hypothesis based on observations in cystic fibrosis-related DM (another form of pancreatogenic DM) is that islets experience denervation due to local changes from intra-islet inflammation because of the tumor, potentially comediated by pancreatic stellate cells (20). Last, there is accumulating evidence supporting the broader theory that PDAC-DM represents a paraneoplastic syndrome because of tumor-secreted mediators in a subset of patients with PDAC (4). Our study provides evidence to suggest this mechanism directly alters the PP responses. Speculations regarding an explanation for the blunted PP response in subjects with CP are somewhat similar, with the exception that the extent of glandular fibrosis is likely more extensive and the potential contribution from pancreatic stellate cells is even stronger. Additional studies are needed to better understand if the reduction in PP secretion is due to the fibro-inflammatory changes of CP (therefore preceding and contributing to CP-DM) or is a consequence of CP-DM (and potentially a biomarker of CP-DM).

Another intuitive question regarding these data is, “Can a lower PP response be used to diagnose new-onset pancreatogenic DM (specifically secondary to PDAC or CP)?” This remains an important but unanswered question that has clinical implications. Primarily, an abnormal PP response (or other diagnostic biomarkers for pancreatogenic DM) would indicate a clinical need to pursue additional diagnostic testing for underlying PDAC or CP. The identification of subclinical PDAC in adults with new-onset DM is viewed as a key opportunity for early detection, which is associated with improved long-term survival (3). In regard to CP, this differentiation would identify patients who are at potentially increased risk for diabetic complications and/or warrant further consideration of medical management (2). Although the current study is the largest to date of these types of pancreatogenic DM cases who completed an MMTT, the sample sizes of these groups remain small. Although we demonstrated significant differences in both fasting and postprandial PP concentrations, substantial overlap in the current data set does not definitively establish the ability to discriminate between the types of diabetes. Of note, the estimates of variance are imprecise because of the current sample size. Therefore, it would be premature to dismiss fasting PP or PP response (estimated with AUC or any other measurement) as a suboptimal diagnostic marker based on this study. We expect a larger sample size will provide a more precise assessment of the diagnostic performance for between-group classification. Other opportunities in a larger study population may also include combining PP values with clinical features, physiological indices (eg, insulin secretion), and/or other biomarkers. Increasing the sample size is conceptually straightforward, but it should be noted that identification of patients with treatment-naïve PDAC or CP who meet stringent eligibility criteria and are willing to participate is challenging and will require careful planning. Considerations could include an abbreviated duration or reduced sampling during the mixed meal tolerance test to lower participant burden.

There are multiple strengths in the current study. First, this study population represents the largest to date of subjects with pancreatogenic DM, including CP and pancreatic cancer, who prospectively completed an MMTT. This larger sample size allowed us to use multiple statistical tests (including adjusting for key covariates) and approaches to assess change in PP concentration to demonstrate the robustness of our findings that were not possible in prior studies. Lastly, the verification of the blunted PP response in new-onset diabetes associated with PDAC within a multicenter framework provides additional confidence in the results.

There are also limitations that must be considered to accurately interpret these data. First, even though the study was relatively large, the group sizes remain statistically small. As a result, we were unable to assess for differential responses related to disease characteristics (eg, cancer stage, etiology of CP). We believe the heterogeneity between sites is low because of the extensive efforts that were made to optimize the rigor of the study design and execution (eg, use of a standardized study protocol and case report forms frequent interactions between study teams, uniform methods of collecting, processing, and storing blood samples, analyses conducted in a central laboratory), but this cannot be completely eliminated in any multicenter study. There are recent data demonstrating that PP can be degraded in vitro into peptide fragments under various conditions, including exposure to dipeptidyl peptidase IV (21). Samples used for the current analysis were not pretreated with dipeptidyl peptidase IV inhibitors or other enzyme inhibitors; thus, we cannot rule out the possibility that some degradation of PP may have occurred during the clotting time, contributing to variability. A limitation with studies involving pancreatogenic diabetes relates to the challenge of performing group classification with high certainty. Despite eligibility criteria used to minimize this uncertainty, it remains possible that a few individuals in the PDAC-DM or CP-DM category had underlying T2DM, which may have contributed to the overlap in PP values. Last, although we systematically asked subjects to hold diabetes medications on the day of the MMTT, we cannot rule out a potential influence from residual effects of those medications on PP responses.

In summary, we demonstrate fasting PP concentrations and the PP response to mixed meal stimulation are reduced in subjects with new onset DM (<3 years of onset) in the setting of PDAC or CP compared with T2DM. Although we are unable to firmly conclude that a blunted PP response is a valid diagnostic method for these (and other) subtypes of pancreatogenic DM, these data clearly support additional investigations into the use of PP response in isolation or combination with other clinical factors for this indication. Confirmation of our findings in larger studies would suggest that this could potentially be used in a screening approach for patients at increased risk for pancreatic cancer. Further research is needed to understand the physiological relevance of these findings to the pathogenesis of diabetes and the underlying pancreatic diseases to ultimately provide guidance to prioritize the study of therapies that are more effective and/or associated with lower morbidity in this patient population.

Abbreviations

AUC

area under the curve

BMI

body mass index

CP

chronic pancreatitis

CP-DM

chronic pancreatitis-related diabetes

DM

diabetes mellitus

HbA1c

hemoglobin A1c

MMTT

mixed meal tolerance test

PDAC

pancreatic ductal adenocarcinoma

PDAC-DM

pancreatic ductal adenocarcinoma-related diabetes

PDAC-DM-proximal

diabetes associated with tumor in the proximal pancreas

PP

pancreatic polypeptide

T2DM

type 2 diabetes mellitus

Funding

Research reported in this publication was supported by the National Cancer Institute and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) under award numbers: U01DK108288 (Mayo Clinic), U01DK108320 (University of Florida), U01DK108323 (Indiana University), U01DK108326 (Baylor College of Medicine), U01DK108327 (The Ohio State University), U01DK108328 (University of Texas—MD Anderson Cancer Center), U01DK108300 (Stanford University), U01DK108306 (University of Pittsburgh), U01DK108314 (Cedars-Sinai Medical Center), U01DK126300 (University of Minnesota), and 2P30DK097512 (Indiana University). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Author Contributions

P.H., Y.K., D.Y., D.A., M.B., D.B., K.C., W.F., K.M., W.P., Z.S., R.C., S.G., J.R., J.S., and M.G. were involved with the conception and design of the study. P.H., Y.K., D.Y., D.A., Y.L., F.T., and M.G. developed the initial draft of the manuscript. All authors were involved with the conduct of the study, data collection, edited, reviewed, and approved the draft and final versions of the manuscript. P.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Data Availability

Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.

Clinical Trial Information

Clinical Trials.gov identifier: NCT03460769.