258 Care Volume 42, February 2019

Proinsulin Is a Emily K. Sims,1,2 Henry T. Bahnson,3 Julius Nyalwidhe,4 Leena Haataja,5 Persistent Feature of Type 1 Asa K. Davis,3 Cate Speake,3 Linda A. DiMeglio,1,2 Janice Blum,6 Diabetes Margaret A. Morris,7 Raghavendra G. Mirmira,1,2,8,9,10 7 10,11 Diabetes Care 2019;42:258–264 | https://doi.org/10.2337/dc17-2625 Jerry Nadler, Teresa L. Mastracci, Santica Marcovina,12 Wei-Jun Qian,13 Lian Yi,13 Adam C. Swensen,13 Michele Yip-Schneider,14 C. Max Schmidt,14 Robert V. Considine,9 Peter Arvan,5 Carla J. Greenbaum,3 Carmella Evans-Molina,2,8,9,10,15 and the T1D Exchange Residual C-peptide Study Group* OBJECTIVE Abnormally elevated secretion has been reported in type 2 and early when significant C-peptide is present. We questioned whether individuals with long-standing type 1 diabetes and low or absent C-peptide secretory capacity retained the ability to make proinsulin.

RESEARCH DESIGN AND METHODS C-peptide and proinsulin were measured in fasting and stimulated sera from 319 subjects with long-standing type 1 diabetes (‡3 years) and 12 control subjects without diabetes. We considered three categories of stimulated C-peptide: 1) 1Department of Pediatrics, Indiana University ‡ 2 School of Medicine, Indianapolis, IN C-peptide positive, with high stimulated values 0.2 nmol/L; ) C-peptide posi- 2 tive, with low stimulated values ‡0.017 but <0.2 nmol/L; and 3)C-peptide Center for Diabetes and Metabolic Diseases, Indiana University School of Medicine, Indian- <0.017 nmol/L. Longitudinal samples were analyzed from C-peptide–positive apolis, IN subjects with diabetes after 1, 2, and 4 years. 3Diabetes Clinical Research Program, Benaroya Research Institute, Seattle, WA RESULTS 4Department of Microbiology and Molecular Cell Biology, Eastern Virginia Medical School, Of individuals with long-standing type 1 diabetes, 95.9% had detectable serum Norfolk, VA proinsulin (>3.1 pmol/L), while 89.9% of participants with stimulated C-peptide 5Division of Metabolism, Endocrinology & Di- values below the limit of detection (<0.017 nmol/L; n = 99) had measurable abetes, University of Michigan Medical Center, proinsulin. Proinsulin levels remained stable over 4 years of follow-up, while Ann Arbor, MI 6Department of Microbiology and Immunology, C-peptide decreased slowly during longitudinal analysis. Correlations between Indiana University School of Medicine, Indian- proinsulin with C-peptide andmixed-meal stimulationof proinsulin were foundonly apolis, IN ‡ fi 7Department of Internal Medicine, Eastern Vir-

PATHOPHYSIOLOGY/COMPLICATIONS in subjects with high stimulated C-peptide values ( 0.2 nmol/L). Speci cally, increases in proinsulin with mixed-meal stimulation were present only in the ginia Medical School, Norfolk, VA 8Department of Cellular and Integrative Physi- group with high stimulated C-peptide values, with no increases observed among ology, Indiana University School of Medicine, subjects with low or undetectable (<0.017 nmol/L) residual C-peptide. Indianapolis, IN 9Department of Medicine, Indiana University CONCLUSIONS School of Medicine, Indianapolis, IN 10 In individuals with long-duration type 1 diabetes, the ability to secrete proinsulin Department of Biochemistry and Molecular Biology, Indiana University School of Medicine, persists, even in those with undetectable serum C-peptide. Indianapolis, IN 11Indiana Biosciences Research Institute, Indian- apolis, IN Type 1 diabetes results from autoimmune-mediated destruction of the pancreatic 12 Northwest Lipid Metabolism and Diabetes Re- b-cell, resulting in the need for exogenous treatment (1). The classic paradigm search Laboratories, University of Washington, that type 1 diabetes leads to a complete loss of b-cell mass and absolute insulin Seattle, WA deficiency has been challenged by recent data (1). Analysis of pancreatic sections from 13Biological Sciences Division, Pacific Northwest organ donors with diabetes indicates the presence of residual insulin-containing National Laboratory, Richland, WA 14Department of Surgery, Indiana University islets many years after disease onset (2). In addition, multiple groups have reported School of Medicine, Indianapolis, IN detectable levels of serum C-peptide in cohorts of individuals with long-duration T1D 15Richard L. Roudebush VA Medical Center, In- (3–9). These studies have included highly selected populations, such as the Joslin dianapolis, IN care.diabetesjournals.org Sims and Associates 259

Medalists, who were identified on the relative to insulin and C-peptide have received a or islet transplant) basis of long-term survival, as well as also been reported in whole pancreata (3). From this group, 319 subjects in groups more reflective of general diabe- from donors with long-standing type 1 whom mixed-meal tolerance tests tes populations, with estimates that up diabetes (18). However, whether de- (MMTTs) were performed were included to 80% of individuals with type 1 diabe- tectable proinsulin secretion is present in in the current analysis (3). In brief, this tes retain the ability to secrete small individuals with extended-duration type 1 included subjects with detectable random amounts of stimulated C-peptide (3–7). diabetes has not been adequately tested. nonfasting C-peptide and 10% of subjects For detection of residual b-cell mass To gain insight into the relationships with C-peptide ,0.017 nmol/L on a ran- and function, these studies have relied between C-peptide and proinsulin secretion dom nonfasting test (n = 99). For subjects nearly exclusively on the measurement in long-standing type 1 diabetes, we with serum C-peptide $0.017 nmol/L of C-peptide, which is generated from the analyzed longitudinal samples from a (n = 220), repeat MMTTs were performed processing of immature diverse cohort of subjects with estab- 1, 2, and 4 years after the initial MMTT. In molecules into insulin and C-peptide. lished disease ($3 years) who were addition, MMTT was performed in a small Preproinsulin processing begins with categorized based on the presence or group of healthy adult control subjects cleavage of the N-terminal absence of residual stimulated C-peptide locally at Indiana University School of to form proinsulin within the lumen of secretion. We sought to define patterns Medicine (n = 12). Subject characteristics the b-cell (ER) of change in C-peptide and proinsu- are described in Table 1. (10). Disulfide bond formation and ter- lin secretion over 4 years of follow-up minal protein folding occurs in the ER and to explore relationships between Measurements and Assays and Golgi, and proinsulin is eventu- meal-stimulated C-peptide and proinsulin Standard MMTTs were performed as ally cleaved into mature insulin and secretion. previously described with blood sam- C-peptide by the enzymes prohormone ples drawn at 210,0,30,60,90,and convertase 1/3, prohormone convertase RESEARCH DESIGN AND METHODS 120 min (3,19). Control samples drawn 2, and carboxypeptidase E within secre- Study Approval at Indiana University were immediately 2 tory granules (10). b-Cell dysfunction, Sample collections were performed processed and stored at 80 without such as that caused by inflammatory after institutional review board ap- freeze/thaw prior to testing. Serum or ER stress, results in the accumulation proval was obtained from T1D Ex- fromsubjects withdiabetes was shipped and release of incompletely processed change Network sites and Indiana to Northwest Lipid Metabolism and Di- proinsulin (11–13). Thus, measurement University. Written informed consent abetes Research Laboratories (NWRL, of C-peptide secretion alone could un- was obtained from participants prior University of Washington) on cold derestimate the ability of the b-cell to to study inclusion. packs the day of sample collections. initiate production, while in- Samples were immediately analyzed for creased proinsulin secretion may provide Subjects C-peptide. Otherwise, serum was stored insight into specific disease pathology. We previously reported nonfasting se- at 280°C without freeze/thaw until Studies in human cohorts have sug- rum C-peptide levels in 919 individuals proinsulin measurement. HLA-DRB1 gested that abnormalities in b-cell pro- with varying durations of disease (all $3 typing, HbA1c, fasting glucose, and insulin processing have clinical relevance years from disease onset) identified autoantibodies against GAD, islet an- to type 1 diabetes. Levels of circulating through the T1D Exchange clinic regis- tigen 2 (IA-2), and zinc transporter proinsulin relative to C-peptide (PI:C try who participated in the Residual 8 (ZnT8) were measured as previously ratios) are elevated at the time of C-peptide in Type 1 Diabetes Study described (19). type 1 diabetes onset and have been evaluating residual insulin secretion in C-peptide values were measured shown to be predictive of type 1 diabetes those with long-standing type 1 diabetes from each MMTT time point in a blinded development in autoantibody-positive (3). To be enrolled in this study, an fashion by NWRL using the Tosoh two- individuals (14–16). Similarly, the ratio individual must have had a clinical di- site immunoenzymometric assay (Tosoh of proinsulin–to–insulin-positive b-cell agnosis of type 1 diabetes made by an Bioscience), which had a sensitivity of area is increased in pancreatic sections endocrinologist on the basis of either 0.017 nmol/L at study start (3,19). from donors with recent-onset type 1 positive islet cell antibodies or insulin Though the TOSOH assay sensitivity diabetes and in autoantibody-positive therapy started around the time of changed to 0.007 nmol/L over the donors without diabetes (17). Persistent diagnosis and used continually thereafter course of the study, for consistency in elevations in islet proinsulin content (with the exception of individuals who our longitudinal analyses, the threshold

Corresponding author: Carmella Evans-Molina, This article contains Supplementary Data online © 2018 by the American Diabetes Association. [email protected], or Emily K. Sims, eksims@iu. at http://care.diabetesjournals.org/lookup/suppl/ Readers may use this article as long as the work is edu doi:10.2337/dc17-2625/-/DC1. properly cited, the use is educational and not for profit, and the work is not altered. More infor- Received 15 December 2017 and accepted 12 *A complete list of the T1D Exchange Residual October 2018 mation is available at http://www.diabetesjournals C-peptide Study Group can be found in .org/content/license. Supplementary Data online. See accompanying article, p. 183. 260 Persistent Serum Proinsulin in Type 1 Diabetes Diabetes Care Volume 42, February 2019

Table 1—Characteristics of the study population C-peptide ,0.017 C-peptide 0.017–0.2 C-peptide $0.2 Control subjects Variable nmol/L (n = 99) nmol/L (n = 117) nmol/L (n = 103) (n = 12) Age at initial MMTT (years)*** 34.0 (17.0, 56.0) 29.0 (17.0, 45.0) 41.0 (28.0, 48.8) 20.5 (18.0, 25.8) Sex (% male) 46.5 42.7 44.7 50 BMI (kg/m2) 25.5 (22.0, 29.0) 25.4 (22.1, 28.4) 26.1 (22.7, 29.2) 24.7 (23.1, 28.4) Race/ethnicity (% non-Hispanic white)* 93.9 87.2 86.7 66.7 T1D duration at consent (years)*** 16.0 (7.0, 33.0) 6.0 (4.0, 11.5) 7.0 (4.0, 13.0) n/a Age at diagnosis (years)*** 15.0 (7.0, 25.0) 19.0 (11.0, 45.0) 29.0 (21.5, 38.0) n/a

HbA1c (%)* 7.8 (5.3, 8.7) 7.7 (6.8, 9.3) 7.3 (6.6, 8.1) n/a GAD positive* 41.4 62.4 60 n/a IA2 positive** 35.4 44.4 22.9 n/a ZnT8 positive*** 14.1 40.17 25.7 n/a $1 high risk HLA DRB1 alleles present 83.2 76.2 75 n/a Data represent values obtained at initial visit and are median (interquartile range) or percent unless otherwise indicated. AUC, area under the curve; n/a, not applicable. *P , 0.05; **P , 0.01; ***P , 0.001 for differences among groups using Kruskal-Wallis test for continuous variables and x2 test for categorical variables.

of 0.017 nmol/L was applied through- MMTT 90-min values were used to of 319 individuals with diabetes of $3 out. Proinsulin levels were assayed at define stimulated C-peptide or proinsulin years’ duration (range 3–81). Demo- the 210 and 90 min time points in a secretion. For C-peptide–positive sub- graphics of the study cohort are blinded fashion by NWRL using a radio- jects, PI:C ratios were calculated as mo- reportedinTable1.MMTTswereper- immunoassay (catalog number HPI-15 K; lar ratios 3 100. formed at baseline and then repeated at Millipore), which detects 100% human 12-month, 2-year, and 4-year follow-up intact proinsulin, 95% human Des Statistics visits. As previously reported, we con- (31,32) proinsulin, and ,0.1% human A Kruskal-Wallis test with Dunn multiple sidered three categories of stimulated Des (64,65) proinsulin (20). The re- comparisons test was used to compare C-peptide: 1) C-peptide positive, with ported sensitivity of this assay is 3.1 continuous variables between groups, stimulated values $0.2 nmol/L; 2) x2 pmol/L. Intra- and interassay variation and a test was used to compare C-peptide positive, with stimulated val- determined by NWRL were 1.4–6.0% proportions. An intraclass correlation ues $0.017 but ,0.2 nmol/L; and 3) fi and 4.84–7.48%, respectively. Cross- coef cient was calculated to quantify C-peptide ,0.017 nmol/L (3). Median reactivity for both human C-peptide the degree of clustering for repeated fasting and stimulated C-peptide val- and human insulin are ,0.1%. Experi- longitudinal proinsulin values. Analyses ues from the initial visit for each ments performed to validate assay were done using SAS, version 9.4, and group are shown in Table 2, with in- variation, sensitivity, specificity, and GraphPad Prism 7.0. For all analyses, dividual stimulated values from each # cross-reactivity, including analysis of two-tailed P values of 0.05 were con- visit displayed in Fig. 1A. Although fi sera from patients before and after sidered signi cant. most subjects had small reductions pancreatectomy, analysis of insulin in stimulated C-peptide over 4 years antibody–positive serum, and vali- RESULTS of follow-up, most remained within dation of radioimmunoassay results To define patterns of C-peptide and their originally assigned C-peptide using targeted mass spectrometry proinsulin secretion in long-standing category. analysis of serum, are included in Supple- type 1 diabetes, we analyzed hormone To categorize residual proinsulin se- mentary Tables 1–6 and Supplementary secretion at fasting and in response cretion between the groups, we assayed Figs. 1–4. to mixed-meal stimulation in a cohort proinsulin levels from the MMTT fasting

Table 2—Fasting and stimulated C-peptide and proinsulin values C-peptide ,0.017 C-peptide 0.017–0.2 C-peptide $0.2 Control subjects Variable nmol/L (n = 99) nmol/L (n = 117) nmol/L (n = 103) (n = 12) Fasting C-peptide (nmol/L) n/a 0.023 (,0.017, 0.036)***# 0.156 (0.095, 0.30)* 0.601 (0.525, 0.800) Stimulated (90 min) C-peptide (nmol/L) n/a 0.079 (0.040, 0.111)***# 0.513 (0.344, 0.847)* 1.84 (1.08, 2.47) Fasting proinsulin (pmol/L) 13.5 (6.2, 23.8) 10.9 (7.3, 21.5) 15.2 (9.8, 23.8) 15.19 (11.8, 17.4) Stimulated (90 min) proinsulin (pmol/L) 11.2 (4.6, 19.7)***# 11.3 (7.3, 34.8)***# 22.5 (17.1, 34.8) 46.7 (34.3, 74.2) Fasting PI:C ratio n/a 37.7 (22.0, 58.5)***# 9.6 (5.9, 18.1)** 2.3 (2.0, 3.1) Stimulated PI:C ratio n/a 15.4 (8.4, 27.9)***# 4.9 (2.4, 6.9) 2.6 (2.4, 3.2) Data represent values obtained at initial visit and are expressed as median (interquartile range). n/a, not applicable. *P , 0.05; **P , 0.01; ***P , 0.001 compared with control subjects and #P , 0.001 compared with C-peptide $0.2 nmol/L group using Kruskal-Wallis test with Dunn multiple comparisons test. care.diabetesjournals.org Sims and Associates 261

Figure 1—Longitudinal C-peptide and proinsulin values in subjects with long-standing type 1 diabetes. Values are color coded based on stimulated C-peptide values (,0.017 nmol/L [teal], $0.017–0.2 nmol/L [light blue], or $0.2 nmol/L [royal blue]). A: Individual stimulated (90 min) serum C-peptide (C-pep) levels over 4 years of follow-up. B: Fasting proinsulin values during the initial MMTT, grouped by C-peptide category and type 1 diabetes (T1D) duration. C: Scatterplot of individual proinsulin increments from fasting to stimulated MMTT time points. D: Individual longitudinal proinsulin values for participants with stimulated C-peptide $0.017–0.2 nmol/L or $0.2 nmol/L. Subjects with undetectable stimulated C-peptide were not invited for longitudinal MMTTs per study protocol. For graphical depiction, values below the assay limit of detection were plotted as one-half the assay limit of detection. Bars in scatterplot graphs represent medians and interquartile range values. *P , 0.05; ***P , 0.001. and 90-min time points. Table 2 displays findings, targeted mass spectrometry without diabetes. In adults without di- median baseline fasting and stimulated analysis was performed on 10 sam- abetes, proinsulin levels increased proinsulin values grouped by C-peptide ples from individuals with undetect- approximately threefold with mixed- positivity. As anticipated, nearly all sub- able stimulated C-peptide measured by meal stimulation (Table 2, with indi- jects who were C-peptide positive at the TOSOH assay (,0.017 nmol/L) but vidual increments during stimulation baseline had detectable fasting or stim- detectable proinsulin measured by the plottedinFig.1C). In contrast, among ulated proinsulin, including 100% in the Millipore radioimmunoassay (Supple- groups with long-standing type 1 dia- highest C-peptide group and 98.3% mentary Tables 5 and 6 and Supplemen- betes, proinsulin secretion increased in the group with C-peptide $0.017 tary Figs. 2 and 3). The presence of with stimulation only in subjects with but ,0.2 nmol/L. Unexpectedly, proin- serum proinsulin in the absence of de- the highest C-peptide values. In this sulin was also detectable in 89.9% (89 tectable serum C-peptide was confirmed group, proinsulin secretion increased of 99) of subjects with serum C-peptide in all 10 samples. Taken together, these ;1.5-fold with meal stimulation (P , levels ,0.017 nmol/L. Of note, median data point to the continued presence 0.001 compared with fasting values) fasting proinsulin values among each of of proinsulin production and secretion, (Table 2 and Fig. 1C). No significant the groups with type 1 diabetes (13.5, even in the presence of minimal stimu- increase in proinsulin was present 10.9, and 15.2 pmol/L) were well above lated C-peptide secretion. with mixed-meal stimulation among the recommended level of detection for To define relationships between pro- the groups with C-peptide ,0.017 or the assay (3.1 pmol/L) and fell in the insulin and C-peptide secretion, we ex- C-peptide $0.017but ,0.2 nmol/L (Table midpoint of the standard curve for the amined changes in proinsulin levels 2 and Fig. 1C). Along these lines, corre- assay. Fasting proinsulin concentrations in response to MMTT stimulation by lation analysis by C-peptide category were not significantly different between comparing fasting and stimulated values revealed that stimulated C-peptide any of the groups, and there was no of proinsulin among the different and proinsulin values were correlated reduction in presence or level of fasting C-peptide categories and compared only in the group with high residual proinsulin with increasing diabetes du- these values with those obtained in a C-peptide secretion (correlation coef- ration (Fig. 1B). For validation of these small cohort of young control subjects ficient of 0.54). 262 Persistent Serum Proinsulin in Type 1 Diabetes Diabetes Care Volume 42, February 2019

Overthecourseof4yearsoffollow- proinsulin in 16% of samples from analysis indicates that in long-standing up, stimulated serum proinsulin levels C-peptide–negative subjects with long- type 1 diabetes, the relationship be- tended to stay relatively stable. Figure standing type 1 diabetes. However, this tween meal stimulation and proinsulin 1D displays individual stimulated pro- analysis was performed on randomly secretion is influenced by the level of insulin values for the groups with collected samples without regard for ambient and residual b-cell function. C-peptide $0.017 but ,0.2 nmol/L and meal stimulation, and all previous studies Only subjects with highly functional C-peptide $0.02 nmol/L at each study were cross-sectional in nature (24). Here, b-cells (stimulated C-peptide levels visit. Among all participants, the intra- we analyzed longitudinal fasting and $0.2 nmol/L) exhibited increased pro- class correlation coefficient for re- stimulated serum samples from subjects insulin levels with meal stimulation. No peated proinsulin values was 0.795, with established type 1 diabetes, using significant increase in proinsulin values confirming a strong clustering of re- a proinsulin radioimmunoassay with was observed with stimulation among peated measurements over the dura- negligible cross-reactivity to insulin or the groups with undetectable (,0.017 tion of the study. C-peptide. In a cohort with a wide distri- nmol) or low stimulated C-peptide val- We have previously shown that circu- bution of age at diagnosis and duration of ues. Along these lines, stimulated pro- lating PI:C ratios are elevated in pediatric disease, we found that almost all individ- insulin and C-peptide values were subjects with new-onset type 1 diabetes uals tested had detectable serum pro- correlated only in the group with high compared with matched healthy control insulin under fasting or meal-stimulated residual stimulated C-peptide secretion. subjects (15). Moreover, elevations in the conditions, including 89.9% of subjects The etiology of this observation is un- circulating PI:C ratio were associated with undetectable serum C-peptide clear but will require further testing in with clinical progression to type 1 di- (,0.017 nmol/L). Median values for human samples to explore underlying abetes in autoantibody-positive individ- the cohorts were well within the limits mechanisms. uals, suggesting utility of the PI:C ratio of detection for the proinsulin radioim- Many individuals in this cohort were as an indicator of b-cell stress (14). To munoassay used and fell in the midpoint diagnosed with type 1 diabetes as adults. define whether circulating PI:C ratios ofthe standard curve for the assay.Infact, Still, 130 of 133 (97.7%) of subjects di- differed among the groups with long- fasting proinsulin values were similar agnosed as children had detectable pro- standing type 1 diabetes and control among all groups, irrespective of stim- insulin, suggesting that applicability of subjects, we calculated fasting and stim- ulated C-peptide status. Taken together, these findings is not limited to individuals ulated PI:C ratios for subjects with de- these data indicate that the vast ma- with a later age of diagnosis. Because tectable C-peptide and control subjects jority of subjects with long-standing only a small number of individuals without diabetes. Similar to findings type 1 diabetes retain the ability to exhibited undetectable proinsulin and observed at diabetes onset, fasting PI: initiate preproinsulin production and C-peptide values, we were unable to C ratios were increased in both groups secrete proinsulin. adequately examine differences in clin- with type 1 diabetes and detectable These data provide an important clin- ical characteristics of this group, but C-peptide compared with control sub- ical measure that substantiates recently future analyses are warranted to explore jects without diabetes (P , 0.001 for published findings quantifying elevations factors related to absolute proinsulin and each group compared with control sub- in proinsulin at the level of the islet. This C-peptide deficiency. Additionally, some jects) (Table 2). Consistent with this includes increases in islet PI:insulin area subjects in the T1D Exchange cohort measure as a reflection of b-cell dys- in euglycemic individuals with positive were diagnosed with type 1 diabetes function, PI:C ratios were the highest islet autoantibodies as well as in subjects using clinical parameters, and we can- in the group with the lowest residual with recent-onset type 1 diabetes (17). not exclude the possibility that some stimulated C-peptide secretion (P , In pancreatic sections from donors with individuals may have been misdiag- 0.001 compared with group with C- long-standing type 1 diabetes, persis- nosed. We calculated PI:C ratios as a peptide values $0.02 nmol/L) (Table 2). tence of islet insulin mRNA and proin- potential proxy for b-cell stress, which sulin protein, despite reduced islet is the standard method of analysis in CONCLUSIONS insulin and C-peptide content, was the field (14). While our data identified Recently, several independent groups also recently reported (18). Our study an increase in PI:C ratios in subjects with have reported that a substantial per- is the first to examine longitudinal rela- long-standing type 1 diabetes com- centage of individuals retain the ability tionships of circulating C-peptide and pared with a small cohort of healthy to secrete C-peptide many years after proinsulin values in established type 1 control subjects without diabetes, we the diagnosis of type 1 diabetes (3–8). diabetes. Our data show that although acknowledge that differences between However, less is known about the C-peptide levels decreased gradually, thegroupsweredrivenbylowerlevels relationship between proinsulin and proinsulin levels remained stable over of C-peptide among subjects with di- C-peptide secretion in long-standing 4 years of follow-up. abetes. type 1 diabetes. Older analyses of sub- In light of these findings, we suggest Finally, our proinsulin and C-peptide jects with type 1 diabetes detected that detectable proinsulin in subjects assays exhibit different sensitivities. Al- circulating proinsulin in subjects with with low or undetectable C-peptide lev- though more sensitive C-peptide assays undetectable fasting C-peptide (21–23). els provides additional information are available, the TOSOH assay used In addition, a recent report from a subset regarding b-cell hormone production here has been most widely used in of adult subjects from the T1D Exchange over that afforded by measurement large clinical networks including the registry described detectable circulating of C-peptide alone. Interestingly, our T1D Exchange, TrialNet, nPOD (Network care.diabetesjournals.org Sims and Associates 263

for Pancreatic Organ Donors With Dia- is uncertain. However, the observation the manuscript. R.G.M. interpreted data and betes), and the Immune Tolerance of persistent proinsulin production in edited the manuscript. J.Na. interpreted data Network (3,19,25,26). Additionally, stan- late-stage disease raises the tantaliz- andeditedthemanuscript.T.L.M.performed experiments and edited the manuscript. S.M. dardization of C-peptide assays is not ing proposition that therapies aimed performed experiments, interpreted data, and performed at lower levels of C-peptide, at b-cell health could have utility in edited the manuscript. W.-J.Q. designed experi- such as those observed in many of our improving insulin secretion in type 1 ments, interpreted data, and edited the manu- subjects. The National Institutes of Health diabetes. script. L.Y. performed experiments, interpreted data, and editedthe manuscript.A.C.S. performed has recently highlighted issues with re- experiments, interpreted data, and edited the search rigor and reproducibility, related manuscript. M.Y.-S. assisted with sample ac- in large part to specificity of assay re- quisition, interpreted data, and edited the Acknowledgments. For this project serum manuscript. C.M.S. assisted with sample ac- agents. To address this, we performed samples were used and subject data were de- quisition, interpreted data, and edited the multiple assay validation experiments, rived from subjects participating in the Residual manuscript. R.V.C. designed and performed ex- C-peptide in Type 1 Diabetes Study conducted including analysis of sensitivity, intra- periments, interpreted data, and edited the man- under the auspices of the T1D Exchange. The and interassay variability, proinsulin uscript. P.A. performed analyses, evaluated data, authors acknowledge the T1D Exchange Biobank recovery, reproducibility with mass and edited the manuscript. C.J.G. directed the and participating investigators and staff in the Residual C-peptide in Type 1 Diabetes Study, spectrometry testing, and testing for T1D Exchange Clinic Network (listed in Sup- planned analysis and evaluated data from the specificity, with testing for human in- plementary Data) for subject recruitment, clinical study, and edited the manuscript. C.E.-M. evaluation, and follow-up. The authors also ac- sulin, human C-peptide, proinsulin in planned analyses, interpreted data, and wrote knowledge Anthony Acton (Indiana University) the context of insulin autoantibodies, the manuscript. E.K.S., H.T.B., J.Ny., L.H., A.K.D., and Wojciech Grzesik (Eastern Virginia Medical C.S., L.A.D., J.B., M.A.M., R.G.M., J.Na., T.L.M., and proinsulin in individuals after pan- School) for technical assistance with proinsulin S.M., W.-J.Q., L.Y., A.C.S., M.Y.-S., C.M.S., R.V.C., createctomy requiring exogenous in- measurement and laser capture microdissec- P.A., and C.J.G. approved the final version of the sulin analogs. These results were all tion (performed in association with Eastern manuscript.E.K.S.andC.E.-M.aretheguarantorsof Virginia Medical University). reassuring. However, we cannot guaran- this work and, as such, had full access to all Funding. This work was supported by National tee with absolute certainty that our pro- the data in the study and take responsibility Institute of Diabetes and Digestive and Kidney for the integrity of the data and the accuracy insulin assay is binding only intact Diseases grant K08-DK0103983 to E.K.S.; a Pe- of the data analysis. proinsulin or proinsulin split products diatric Endocrine Society Clinical Scholar Award Prior Presentation. Parts of this study were to E.K.S.; National Institutes of Health grants in sera from our subjects with type 1 presented in abstract form at the 76th Scientific DP3 DK110844 (to W.-J.Q.), R01-DK-093954 (to diabetes. Analysis of additional cohorts Sessions of the American Diabetes Associa- C.E.-M.), R01-DK-48280 (to P.A.), and UC4- using different proinsulin assays, includ- tion, New Orleans, LA, 10–14 June 2016; the DK-104166 (to C.E.-M. and R.G.M.); U.S. De- 2016 HIRN Annual Investigator Meeting, Be- ing those measuring both intact and partment of Veterans Affairs Merit Award thesda, MD, 25–26 May 2016; the 2017 JDRF total proinsulin, should be performed. I01BX001733 (to C.E.-M.); JDRF Strategic Re- nPOD Annual Scientific Meeting, Fort Lauderdale, search Agreement 2-SRA-2017-498-M-B to E.K.S.; Notwithstanding these limitations, FL, 19–22 February 2017; and the Endocrine JDRF Pioneer Award and Strategic Research our findings demonstrate that persistent Society’s 99th Annual Meeting & Expo, Orlando, Agreement (to C.E.-M., L.A.D., and J.B.); JDRF circulating proinsulin can be detected in FL, 1–4 April 2017. grant 47-2014-299-Q-R (to J.B. and C.E.-M.); and almost all subjects with long-standing JDRF grant 47-2012-744 (to L.A.D.). Mass spec- type 1 diabetes, including 89.9% of trometry-based proinsulin work was performed those with low or absent C-peptide. in the Environmental Molecular Sciences Labo- References fi The ability to increase proinsulin se- ratory, a national scientific user facility spon- 1. American Diabetes Association. 2. Classi ca- sored by the Department of Energy and located tion and diagnosis of diabetes: Standards of cretion under conditions of meal stimu- at Pacific Northwest National Laboratory, which Medical Care in Diabetesd2018.DiabetesCare lation occurred only in those patients is operated by Battelle Memorial Institute for the 2018;41(Suppl. 1):S13–S27 with significant C-peptide levels. To- Department of Energy under Contract DE-AC05- 2. Campbell-Thompson M, Fu A, Kaddis JS, et al. gether, these observations suggest a po- 76RL0 1830. Core services for this work were Insulitis and b-cellmassinthenaturalhistory – tential hierarchy of b-cell dysfunction, provided by the Diabetes Research Center grant of type 1 diabetes. Diabetes 2016;65:719 b P30-DK-097512 (to Indiana University School of 731 which begins with a healthy -cell that Medicine). T1D Exchange is a program of Unitio, 3. Davis AK, DuBose SN, Haller MJ, et al.; T1D secretes mostly C-peptide. Early defects Inc., and supported by the Leona M. and Harry B. Exchange Clinic Network. Prevalence of detect- are characterized by increased proin- Helmsley Charitable Trust. able C-peptide according to age at diagnosis and sulin secretion with relatively intact Duality of Interest. No potential conflicts of duration of type 1 diabetes. Diabetes Care 2015; – C-peptide secretion. Increased progres- interest relevant to this article were reported. 38:476 481 4. Oram RA, Jones AG, Besser RE, et al. The b Author Contributions. E.K.S. planned analyses, sion of -cell dysfunction is character- evaluated and interpreted data, and wrote the majority of patients with long-duration type 1 ized by lower C-peptide secretion but manuscript. H.T.B. performed statistical analysis, diabetes are insulin microsecretors and have retained ability to increase proinsulin in evaluated and interpreted data, and edited the functioning beta cells. Diabetologia 2014;57: response to stimulation. Ultimately, as manuscript. J.Ny. performed analyses, evaluated 187–191 C-peptide levels fall further, proinsulin data, and edited the manuscript. L.H. performed 5. Oram RA, McDonald TJ, Shields BM, et al.; experiments, interpreted data, and wrote and UNITED Team. Most people with long-duration secretioninresponsetomealstimula- edited the manuscript. A.K.D. designed and type 1 diabetes in a large population-based study tion is blunted. Whether this hierarchy implemented the Residual C-peptide in Type 1 are insulin microsecretors. Diabetes Care 2015; represents distinct stages of disease Diabetes Study. C.S. designed supplemental 38:323–328 that are common among all individuals studies to validate proinsulin measurements, 6. Wang L, Lovejoy NF, Faustman DL. Persis- interpreted data, and edited the manuscript. tence of prolonged C-peptide production in or whether aspects of this framework L.A.D. interpreted data and edited the manu- type 1 diabetes as measured with an ultrasen- can be applied to dissect pathophysio- script. J.B. interpreted data and edited the sitive C-peptide assay. Diabetes Care 2012;35: logical heterogeneity in type 1 diabetes manuscript. M.A.M. collected data and edited 465–470 264 Persistent Serum Proinsulin in Type 1 Diabetes Diabetes Care Volume 42, February 2019

7. Keenan HA, Sun JK, Levine J, et al. Residual isolation using a combination of purified colla- measurement of proinsulin in human serum. insulin production and pancreatic b-cell turnover genase and neutral protease. J Vis Exp 2012;67: Diabetes 1992;41:1084–1090 after 50 years of diabetes: Joslin Medalist Study. e4317 21. Heding LG, Ludvigsson J. Human proinsulin Diabetes 2010;59:2846–2853 14. Sims EK, Chaudhry Z, Watkins R, et al. in insulin-treated juvenile diabetics. Acta Pae- 8. McGee P, Steffes M, Nowicki M, et al.; DCCT/ Elevations in the fasting serum proinsulin–to–C- diatr Scand Suppl 1977;270:48–52 EDIC Research Group. Insulin secretion mea- peptide ratio precede the onset of type 1 di- 22. Sobey WJ, Beer SF, Carrington CA, et al. sured by stimulated C-peptide in long-established abetes. Diabetes Care 2016;39:1519–1526 Sensitive and specific two-site immunoradio- type 1 diabetes in the Diabetes Control 15. Watkins RA, Evans-Molina C, Terrell JK, et al. metric assays for human insulin, proinsulin, and Complications Trial (DCCT)/Epidemiology of Proinsulin and heat shock protein 90 as bio- 65-66 split and 32-33 split proinsulins. Biochem Diabetes Interventions and Complications (EDIC) markers of beta-cell stress in the early period J 1989;260:535–541 cohort: a pilot study. Diabet Med 2014;31:1264– after onset of type 1 diabetes. Transl Res 2016; 23. Heding LG. Insulin, C-peptide, and proinsulin 1268 168:96–106.e1 in nondiabetics and insulin-treated diabetics. Char- 9. Greenbaum CJ, Anderson AM, Dolan LM, 16. Truyen I, De Pauw P, Jørgensen PN, et al.; acterization of the proinsulin in insulin-treated et al.; SEARCH Study Group. Preservation of Belgian Diabetes Registry. Proinsulin levels and diabetics. Diabetes 1978;27(Suppl. 1):178–183 b-cell function in autoantibody-positive youth the proinsulin:c-peptide ratio complement au- 24. Steenkamp DW, Cacicedo JM, Sahin-Efe A, with diabetes. Diabetes Care 2009;32:1839– toantibody measurement for predicting type 1 Sullivan C, Sternthal E. Preserved proinsulin 1844 diabetes. Diabetologia 2005;48:2322–2329 secretion in long-standing type 1 diabetes. En- 10. Liu M, Wright J, Guo H, Xiong Y, Arvan P. 17. Rodriguez-Calvo T, Zapardiel-Gonzalo J, docr Pract 2017;23:1387–1393 Proinsulin entry and transit through the endo- Amirian N, et al. Increase in pancreatic proinsulin 25. Greenbaum CJ, Beam CA, Boulware D, et al.; plasmic reticulum in pancreatic beta cells. Vitam and preservation of b-cell mass in autoantibody- Type 1 Diabetes TrialNet Study Group. Fall in Horm 2014;95:35–62 positive donors prior to type 1 diabetes onset. C-peptide during first 2 years from diagnosis: 11. Cnop M, Welsh N, Jonas JC, Jorns¨ A, Lenzen S, Diabetes 2017;66:1334–1345 evidence of at least two distinct phases from Eizirik DL. Mechanisms of pancreatic beta- 18. Wasserfall C, Nick HS, Campbell-Thompson composite Type 1 Diabetes TrialNet data. Di- cell death in type 1 and : many M, et al. Persistence of pancreatic insulin mRNA abetes 2012;61:2066–2073 differences, few similarities. Diabetes 2005;54 expression and proinsulin protein in type 1 diabe- 26. Herold KC, Gitelman SE, Ehlers MR, et al.; (Suppl. 2):S97–S107 tes pancreata. Cell Metab 2017;26:568–575.e3 AbATE Study Team. Teplizumab (anti-CD3 mAb) 12. Marhfour I, Lopez XM, Lefkaditis D, et al. 19. Beck RW, Tamborlane WV, Bergenstal RM, treatment preserves C-peptide responses in Expression of endoplasmic reticulum stress Miller KM, DuBose SN, Hall CA; T1D Exchange patients with new-onset type 1 diabetes in a markers in the islets of patients with type 1 Clinic Network. The T1D Exchange clinic registry. randomized controlled trial: metabolic and diabetes. Diabetologia 2012;55:2417–2420 J Clin Endocrinol Metab 2012;97:4383–4389 immunologic features at baseline identify a 13. Stull ND, Breite A, McCarthy R, Tersey SA, 20. Bowsher RR, Wolny JD, Frank BH. A rapid subgroup of responders. Diabetes 2013;62: Mirmira RG. Mouse islet of Langerhans and sensitive radioimmunoassay for the 3766–3774