Islet-Specific T-Cell Responses and Proinflammatory

Pathophysiology/Complications ORIGINAL ARTICLE

Islet-Specific T-Cell Responses and Proinflammatory Monocytes Define Subtypes of Autoantibody-Negative Ketosis-Prone Diabetes (KPD)

1,2 3 BARBARA M. BROOKS-WORRELL, PHD KEREM OZER, MD that this phenotype comprises two dis- 3 1,2 DINAKAR IYER, PHD RADHIKA NARLA, MD 3 1,2 tinct subtypes distinguished by whether IVONNE CORAZA, JERRY P. PALMER, MD fi 1 3,4 the index DKA episode had a de ned pre- CHRISTIANE S. HAMPE, PHD ASHOK BALASUBRAMANYAM, MD 3,4 cipitant (“provoked” A2b+) or no pre- RAMASWAMI NALINI, MD cipitant (“unprovoked” A2b+) (4). Provoked A2b+ KPD patients have progressive loss of b-cell function after initial OBJECTIVEdKetosis-prone diabetes (KPD) is characterized by diabetic ketoacidosis (DKA) in patients lacking typical features of type 1 diabetes. A validated classification scheme for KPD recovery from the DKA episode, with re- includes two autoantibody-negative (“A2”) phenotypic forms: “A2b2” (lean, early onset, lack- lapse to insulin dependence, no sex pre- ing b-cell functional reserve), and “A2b+” (obese, late onset, with substantial b-cell functional dominance, and an increased frequency reserve after the index episode of DKA). Recent longitudinal analysis of a large KPD cohort of the HLA class II alleles DQB1*0302 revealed that the A2b+ phenotype includes two distinct subtypes distinguished by the index and DRB1*04 associated with suscepti- DKA episode having a defined precipitant (“provoked,” with progressive b-cell function loss over bility to autoimmune type 1 diabetes. In time) or no precipitant (“unprovoked,” with sustained b-cell functional reserve). These three contrast, unprovoked A2b+KPDpa- 2 A KPD subtypes are characterized by absence of humoral islet autoimmune markers, but a role tients have sustained preservation of for cellular islet autoimmunity is unknown. b-cell function after recovery from the RESEARCH DESIGN AND METHODSdIslet-specific T-cell responses and the percent- DKA episode, prolonged insulin indepen- age of proinflammatory (CD14+CD16+) blood monocytes were measured in A2b2 (n =7), dence, male predominance, and an in- provoked A2b+(n = 15), and unprovoked A2b+(n = 13) KPD patients. Genotyping was creased frequency of the protective allele performed for type 1 diabetes–associated HLA class II alleles. DQB1*0602 (4). The unique clinical fea- d 2b 2b2 fi tures and natural histories of these two RESULTS Provoked A +andA KPD patients manifested stronger islet-speci c T-cell subtypes of A2b+ KPD patients suggest responses (P , 0.03) and higher percentages of proinflammatory CD14+CD16+ monocytes (P , 0.01)thanunprovokedA2b+ KPD patients. A significant relationship between type 1 diabetes distinctive underlying pathophysiologic HLA class II protective alleles and negative T cell responses was observed. processes for each. Although an underlying “occult” autoimmune element is sug- CONCLUSIONSdProvoked A2b+KPDandA2b2 KPD are associated with a high fre- gested in the provoked A2b+subtypeby quency of cellular islet autoimmunity and proinflammatory monocyte populations. In contrast, progressive b-cell loss and the presence of 2b unprovoked A + KPD lacks both humoral and cellular islet autoimmunity. type 1 diabetes–associated HLA susceptibility alleles, the unprovoked A2b+subtype could represent a truly nonautoimmune syndrome of KPD. etosis-prone diabetes (KPD), char- absence of b-cell functional reserve (“b+” We have previously shown that the Kacterized by presentation with di- or “b2”) (3) includes two autoantibody- T cells of a significant proportion of abetic ketoacidosis (DKA) in negative A2 phenotypic forms: “A2b2” individuals with an apparent phenotype patients lacking the typical features of (lean, early onset, lacking b-cell functional of type 2 diabetes react strongly to islet autoimmune type 1 diabetes, is a hetero- reserve), and “A2b+” (obese, late onset, antigens, despite lacking b-cell autoanti- geneous syndrome (1,2). A validated clas- with substantial b-cell functional reserve bodies, and that this reactivity is associ- sification scheme for KPD, based on the after the index episode of DKA). Long- ated with low C-peptide levels, indicating presence or absence of b-cell autoanti- term longitudinal follow-up of a large co- underlying cellular immune damage to bodies (“A+” or “A2”) and presence or hort of A2b+ KPD patients has revealed b-cells (5,6). This finding expands the range of diabetic phenotypesdincluding ccccccccccccccccccccccccccccccccccccccccccccccccc those labeled as having “type 2” From the 1Department of Medicine, University of Washington, Seattle, Washington; the 2Department of diabetesdwith a potential pathophysio- 3 Medicine, VA Puget Sound Health Care System, Seattle, Washington; the Translational Metabolism Unit, logic basis in islet autoimmunity. In the Diabetes Research Center, Division of Diabetes, Endocrinology and Metabolism, Baylor College of Medi- fi cine, Houston, Texas; and the 4Endocrine Service, Ben Taub General Hospital, Houston, Texas. current study, we extended these ndings Corresponding authors: Barbara M. Brooks-Worrell, [email protected], and Ashok to the unique, emerging forms of A2 KPD. Balasubramanyam, [email protected]. Specifically, we hypothesized that differ- Received 8 November 2012 and accepted 22 July 2013. ences in cellular immune responses might DOI: 10.2337/dc12-2328 2 © 2013 by the American Diabetes Association. Readers may use this article as long as the work is properly distinguish the three A KPD subtypes cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ (A2b2, unprovoked A2b+, and pro- licenses/by-nc-nd/3.0/ for details. voked A2b+) with regard to a cellular care.diabetesjournals.org DIABETES CARE 1 Diabetes Care Publish Ahead of Print, published online October 15, 2013 Islet autoimmunity distinguishes KPD patients autoimmune etiology. To test this hypoth- radioligand binding assays with recent to the University of Washington, Seattle, esis, we measured islet-specific T-cell re- modifications (8,9). As previously de- WA, for T-cell analysis (10,11). Serum sponses using the validated cellular scribed (1), KPD patients are classified samples were shipped frozen to the Uni- immunoblotting assay and islet autoanti- as A2 if the autoantibody index for all versity of Washington for measurement of body responses to determine the presence three autoantibodies are below the ethnic- islet autoantibodies. of islet autoimmunity in patients carefully specific 98th percentile. Patients were classi- phenotyped for these three KPD subtypes. fied as b+orb2 based on the presence Autoantibody assays We further assessed the percentage of or absence of b-cell functional reserve, Radioligand binding assay. GAD65, proinflammatory (CD14+CD16+) mono- measured by fasting serum C-peptide islet cell antigen (IA-2), and zinc trans- cytes in the peripheral blood of the three concentration and C-peptide response porter isoform 8 (ZnT8) autoantibodies KPD subtypes. to glucagon 1–4 weeks after resolution were determined by radioligand binding In healthy subjects, 90–95% of clas- of ketoacidosis. Cutoffs defining b-cell assay, as previously described (8,9). Re- sical monocytes express high levels of functional reserve were as previously es- combinant [35]S-antigen was produced in the cell surface marker CD14 (CD14+), tablished and longitudinally validated us- an in vitro coupled transcription and without expression of CD16 (CD162). ing receiver-operator characteristic curve translation system with SP6 RNA poly- Inflammation and infection are associ- analysis (1); briefly, b+ status is defined merase and nuclease-treated rabbit retic- ated with the emergence of a distinct by fasting C-peptide .1 ng/mL and peak ulocyte lysate (Promega, Madison, WI). monocyte population characterized C-peptide after glucagon stimulation Sera (2.5 mL) were incubated with by coexpression of CD14 and CD16 .1.5 ng/mL. [35]S-antigen (25,000 of trichloroacetic (CD14+CD16+). CD14+CD16+ mono- Provoked A2b+ KPD patients were acid–precipitable radioactivity). After an cytes secrete high levels of proinflamma- defined by the presence of a clinically de- overnight incubation at 48C, antibody- tory tumor necrosis factor-a and low fined precipitating factor associated with bound [35]S-antigen was separated from levels of anti-inflammatory interleukin- or immediately preceding the index DKA unbound antigen by precipitation with 10, leading to their designation as pro- episode(4).Wehavepreviouslyshown PAS (Invitrogen, Carlsbad, CA). The im- inflammatory monocytes (7). that the phenotype of provoked A2b+ munoprecipitated radioactivity was Our results demonstrate that 1)pro- KPD is characterized by progressive loss counted on a Wallac Microbeta Liquid voked A2b+ KPD, associated with pro- of b-cell function after recovery from the Scintillation Counter (PerkinElmer Life gressive loss of b-cell function, is index DKA episode, no sex predomi- and Analytical Sciences, Inc., Boston, associated with a high frequency of cellular nance, inability to discontinue insulin or MA). Antibody-positive control sera islet autoimmunity and proinflammatory relapsing to insulin requirement after a were included as a standard and used to monocyte populations; 2)unprovoked brief period off insulin, and an increased express immunoglobulin-binding levels A2 KPD is a distinct syndrome of revers- frequency of type 1 diabetes-associated as a relative unit. Samples with a GAD65 ible b-cell dysfunction lacking evidence HLA class II alleles DQB1*0302 and antibody titer .65 units/mL were consid- of humoral or cellular islet autoimmu- DRB1*04 (4). Unprovoked A2b+KPD ered positive for GAD65-antibody. IA2- nity; and 3) a substantial proportion of patients were defined by the absence antibody positivity was established as A2b2 KPD patients, who resemble pa- of a clinically defined precipitating factor exceeding the binding of the 98th percen- tients with type 1 diabetes but lack auto- associated with or immediately preced- tile for healthy controls (30 units/mL). antibodies, have evidence of cellular islet ing the index DKA episode. We have pre- Autoantibody positivity for ZnT8R and autoimmunity. viously shown that the phenotype of ZnT8 W were set at 15 units/mL and 26 unprovoked A2b+ KPD is characterized units/mL, respectively, based on the 98th RESEARCH DESIGN AND by marked improvement and sustained percentile observed in 50 healthy human METHODS preservation of b-cell function after re- control sera. The autoantibody laboratory covery from the index DKA episode, (C.S.H.) participated in the Diabetes Au- Subjects male predominance, insulin indepen- toantibody Standardization Program The protocol was approved by the in- dence, and an increased frequency of (DASP) workshop (12); the GAD65 auto- stitutional review boards for human stud- the protective allele DQB1*0602 (4). antibody assay showed a sensitivity of ies of Baylor College of Medicine, the A2b2 KPDpatientsweredefined ex- 86% and specificity of 93%, and the Harris County Hospital District, and the clusively by A and b status, without re- ICA512/IA-2 autoantibody assay showed University of Washington. All subjects gard to DKA precipitant, because the a sensitivity of 66% and a specificity of enrolled in the study gave informed writ- latter does not affect the natural history 98%. ZnT8 autoantibodies were not in- ten consent. or phenotypic characteristics of this KPD cluded in the workshop. All KPD patients were identified by subgroup. acute presentation with an episode of Blood was drawn for autoantibody, T-cell assay DKA, defined by the presence of all of phenotyping, and T-cell responses during Cellular immunoblotting. Cellular im- the following: anion gap $15, blood pH routine follow-up visits of the subjects to munoblotting was performed on blinded ,7.30, serum bicarbonate #17 mmol/L, the KPD clinic at Ben Taub General freshly isolated peripheral blood mono- serum glucose .200 mg/dL, serum ke- Hospital. All subjects were clinically sta- nuclear cells (PBMCs) to test for the tones $5.2 mmol/L, and urine ketones ble, without evidence of acute illness or presence of islet-reactive T cells, as pre- .13.9 mmol/L. The A2 KPD patients inflammation at the time of the blood viously described (13). Briefly, normal hu- were identified by the absence of draw. Blinded whole-blood samples were man islet cell preparations were subjected GAD65, islet antigen-2, or ZnT8 autoan- transported, using a validated shipping to preparative one-dimensional 10% SDS- tibodies, measured in sera by quantitative protocol, overnight in a blinded fashion PAGE and electroblotted onto nitrocellulose.

2 DIABETES CARE care.diabetesjournals.org Brooks-Worrell and Associates

The nitrocellulose was cut into molecular Statistics (“initial” values in Table 1) (P , 0.01, weight regions (blots) and then solubi- ANOVAs, the Wilcoxon signed rank test, Table 2). As is typical for these KPD sub- lized to form nitrocellulose particles. The and Mann-Whitney tests were used to groups (1), fasting and peak C-peptide nitrocellulose particles containing islet compare groups. Significance was estab- values were essentially unchanged in the proteins were used to stimulate PBMCs, lished at P , 0.05. Multivariate regression A2b2 KPD group 6–12 months after the at a concentration of 3.5 3 105 PBMCs analyses were performed to evaluate positiv- index DKA, whereas they increased sig- per well (5,6,10,11,13,14). Type 1 diabe- ity for monocytes and islet-reactive T-cell nificantly in both A2b+ KPD groups after – tes patients have been shown to respond responsestoBMI,age,sex,anddurationof 6 12 months. HbA1c at the time of the to 4–18 molecular weight proteins and diabetes. index DKA (“initial” HbA1c in Table 2) normal control subjects (without diabe- was similar in the three groups; however, tes) to 0–3 molecular weight regions RESULTSdPatient demographic, clin- after 6–12 months of treatment following (14). Human pancreatic islets were ob- ical, and phenotypic data are reported in an established and standardized protocol tained from the National Institutes of Tables 1 and 2. There were no significant (18) in the same clinic, HbA1c was signif- Health–supported Islet Cell Resource differences among the three groups in age icantly lower in the two A2b+KPD Centers (ICR-ABCC). The specificity of at the time of T-cell testing, sex, or eth- groups compared with the A2b2 KPD the T-cell responses from diabetes pa- nicity (Table 1). The A2b2 KPD patients group (P , 0.01), and this difference tients to islet proteins has been dem- had been diagnosed with diabetes at a was maintained in HbA1c measured close onstrated previously (13). Cellular younger age than the two A2b+KPD to the time of the blood draw for T-cell immunoblotting has been validated to groups (P , 0.01) and had a longer du- testing. have excellent specificity and sensitivity ration of diabetes (P , 0.01, Table 1). The A key characteristic of provoked for distinguishing T-cell responses to islet initial diagnosis of diabetes coincided A2b+ KPD patients is that 3–4 years after proteins between T1D patients and con- with the index DKA (defined as the epi- recovery from the index DKA, their b-cell trol subjects (10,11,15). The specificity sode of DKA that first brought the patient function declines and glycemic control and sensitivity for the cellular immuno- to the attention of the KPD research deteriorates, whereas b-cell function re- blotting assay was similar to those for group) among unprovoked A2b+KPD mains stable and glycemic control is the islet autoantibody assays performed patients, but usually preceded the index maintained in patients with unprovoked in the individual workshops and ex- DKA (by 0–7 years) among provoked A2b+ KPD (4); the number of A2b+ ceeded the specificity and sensitivity of A2b+ KPD patients, and was well be- KPD patients who participated in the cur- other T-cell assays validated at this fore the index DKA among A2b2 KPD rent study and have been monitored long time. patients. Fasting C-peptide and peak enough to compare these outcomes be- C-peptide levels after glucagon stimulation tween provoked and unprovoked A2b+ Phenotyping were significantly lower in the A2b2 KPD KPD patients after 3–4 years was insuffi- PBMCs were stained with antibodies spe- group shortly after the index DKA cient for analysis of this outcome. cific to human CD14 and human CD16 (BD Pharmingen, San Jose, CA). Data acquisition was performed using BD Cell- Table 1dDemographic data of KPD patients quest 5.2.1 software on a BD FACSCali- bur (BD Biosciences). Data analysis was performed using FlowJo 9.3.1 software Unprovoked Provoked 2b 2b 2b2 (Tree Star, Inc., Ashland, OR). The inter- A + KPD A +KPD A KPD n n n assay coefficient of variation for the phe- Characteristics ( =13) ( = 15) ( =7) notyping assays was 10%. Age at diagnosis of diabetes (years) 41 (19–67)* 46 (13–61) 20 (8–44)† Age at time of index DKA (years) 41 (19–67) 47 (18–62) 41 (17–53) HLA genotyping for class II type 1 Duration of diabetes at time diabetes susceptibility alleles of index DKA (years) 0 (0–0) 1 (0–5) 9 (0–26)† HLA genotyping for class II alleles asso- Age at time of T-cell testing (years) 42 (20–69)* 52 (19–62) 45 (27–56) ciated with susceptibility to or protection Duration of diabetes at time from autoimmune type 1 diabetes was of T-cell testing (years) 1.5 (0.2–6) 2.5 (0.9–12) 9 (4–29)† performed at Benaroya Research Institute, Sex (%) fl Seattle, Washington. Brie y, HLA geno- Male 61 60 43 typing for DR1, DR3, DR4, DR13, DR15, Female 39 40 57 DQ2, DQ6, and DQ8 was performed by BMI (kg/m2)31(27–35) 26 (22–43)† 25 (21–29)† fi real-time PCR using TaqMan speci c Race (%) assays, as described previously (16,17). Caucasian 0 0 0 fi Primers and probes speci cforHLA- African American 15 20 29 DRA were included as a template control. Hispanic 85 80 57 Real-time PCR was done using TaqMan Other 0 0 14 GTXpress genotyping master mix (Ap- plied Biosystems, Life Technologies, “Index DKA” is defined as the episode of DKA that first brought the patient to the attention of the KPD research group. Islet autoantibodies were measured 2 to 10 months after the index DKA. Precipitating factors Inc.) on an ABI StepOnePlus Real-Time associated with the index DKA in the Provoked A-b+ KPD subgroup include pancreatitis (n = 5), pneumonia PCR system (Applied Biosystems), as (n = 1), asthma treated with prednisone (n = 1), acute sinusitis (n = 1), perirectal abscess (n = 1), fever of described (17). unknown origin (n = 2), other acute stressful events (n = 4). *Median (range). †P , 0.01. care.diabetesjournals.org DIABETES CARE 3 Islet autoimmunity distinguishes KPD patients

Table 2dClinical data of KPD patients

Unprovoked Provoked A2b+KPD A2b+KPD A2b2KPD Characteristics (n = 13) (n = 15) (n =7) † – – – Initial* HbA1c (%) 14 (8 17) 13 (6 17) 12 (9 15) HbA1c 6–12 months after index DKA (%) 6‡ (5–11) 6‡ (5–12) 9 (7–11) HbA1c at time of T-cell testing (%) 6‡ (5–12) 7‡ (5–14) 9 (7–11) Initial* Fasting C-peptide (ng/mL) 0.7 (0.3–2) 1 (0.2–3) 0.1‡ (0.09–0.5) Initial* Peak C-peptide (ng/mL) 3 (0.9–5) 3 (0.8–7) 0.2‡ (0.09–1) Fasting C-peptide 6–12 months after index DKA (ng/mL) 3‡ (2–6) 2‡ (0.5–4) 0.1 (0.1–1) Peak C-peptide 6–12 months after index DKA (ng/mL) 6 ‡ (3–12) 5 ‡ (2–10) 0.1 (0.1–3) Insulin independent .1 year after index DKA (%) 69 46 0 Diabetes treatment modalities ;1 year after index DKA Insulin only = 2 Insulin only = 5 Insulin only = 7 Oralx only = 9 Oral only = 7 Oral only = 0 Insulin + Oral = 2 Insulin + Oral = 3 Insulin + Oral = 0

* “Initial” refers to time of measurements performed within 4 weeks after the index DKA. †Median (range). ‡P , 0.01. xOral = metformin alone, or metformin + sulfonylurea.

However, a substantially higher fraction significant difference between the pro- protection) were also genotyped. Fre- of unprovoked A2b+KPDpatients voked A2b+ KPD and A2b2 KPD pa- quencies of HLA class II alleles associated (69%) than provoked A2b+KPDpa- tients in frequency of T-cell reactivity to with susceptibility or protection are pre- tients (46%) were insulin-independent, islet proteins. In contrast, the patients sented in Table 3. There was no signifi- with good glycemic control, more than with the distinctive syndrome of unpro- cant relationship between possessing a 1 year after the index DKA episode. voked A2b+ KPD lacked islet autoanti- susceptibility allele and having a positive Islet-specific T-cell reactivities among bodies and islet-reactive T cells, suggesting T-cell test. However, there was a signifi- the three KPD patient groups are shown in that the pathophysiology of this distinc- cant relationship between possessing a Fig. 1. We observed that provoked A2b+ tive phenotype is not associated with islet protective allele and having a negative KPD patients and A2b2 KPD patients had autoimmunity. Differences in the islet- T-cell test, and no patient possessing significantly (P , 0.03) higher T-cell re- specific T-cell responses among the three the strong protective allele DQB1*0602 sponses to islet proteins than unprovoked patient groups in Fig. 1 were not signifi- had a positive T-cell test. A2b+ KPD patients. The T-cell responses cantly affected by BMI, age, sex, or diabe- to the islet proteins in the provoked A2b+ tes duration. CONCLUSIONSdSystemic inflam- KPD and A2b2 KPD patients were similar We next analyzed the percentages of mation has been demonstrated to be in- to what we have observed in other patient proinflammatory monocytes (CD14+ volved in the development of type 2 populations with autoimmune type 1 dia- CD16+) in the peripheral blood of the diabetes. We have used the validated cellular betes (5,6,10,11,13,14,19). There was no three KPD subgroups and observed that immunoblotting assay to study islet-specific the provoked A2b+ KPD patients and the T-cell autoimmunity in autoimmune type 1 A2b2 KPD patients both had a significantly higher frequency of proinflammatory CD14+CD16+ monocytes than the unprovoked A2b+ KPD patients (P , 0.01) (Fig. 2). There were no differences in the frequency of proinflammatory monocytes between the provoked A2b+ KPDpatientsandtheA2b2 KPD patients. Differences in the percentages of proinflammatory monocytes observed in Fig. 2 were not significantly affected by age, sex, or diabetes duration. However, increases in proinflammatory monocytes were negatively correlated with BMI. d HLA genotyping was conducted to Figure 1 Number of islet protein blot sec- determine the frequency of the three A2 tions (cellular immunoblotting) stimulatory to subgroups of KPD patients with HLA T-cells from unprovoked A2b+ KPD patients Figure 2dPercentage of proinflammatory (n =13),provokedA2b+ KPD patients (n = class II alleles (in the DR and DQB loci) (CD14+CD16+) monocytes in the peripheral 15), and A2b2 KPD patients (n =7).The known to be associated with susceptibility blood of unprovoked A2b+ KPD patients (n = horizontal dashed line at the three blot sections to or protection against autoimmune 13), provoked A2b+ KPD patients (n =15), represents the cutoff for T-cell positivity. Sig- type 1 diabetes. Some neutral alleles (as- and A2b2 KPD patients (n = 7). Significance nificance was taken at P , 0.05. sociated with neither susceptibility nor was taken at P , 0.01.

4 DIABETES CARE care.diabetesjournals.org Brooks-Worrell and Associates

Table 3dHLA class II alleles patients, indicate that 1) in the A2 KPD population, having a protective HLA Susceptibility alleles, n/N (%) Protective alleles, n/N (%) class II allele is associated with absence of both humoral (4) and T-cell autoreac- Class II (DR 3/4; DR*0401, tivity to islet antigens and that 2) compared *0404, *0405, *0407; Class II (DR*1501; with the other A2 KPD subgroups, un- KPD subgroup DQB1*0302, *0301, *0201) DQB1*0602) provoked A2b+ KPD, previously shown Unprovoked A2b+ 3/13(23) 3/13(23) to have a lower frequency of susceptibility Provoked A2b+ 4/15 27) 2/15 13) HLA class II alleles (4), also has a higher A2b2 4/7 (57) 1/7 14) frequency of protective alleles. Together with the present results indicating that unprovoked A2b+KPDhasalowfre- quency of T-cell positivity and that and phenotypic type 2 diabetic patients Moreover, provoked A2b+ KPD pa- having a protective allele is strongly asso- (5,6,10,11,13,14,19). These studies under- tients have a high frequency of HLA class ciated with absence of T-cell positivity, score the heterogeneity of type 2 diabetes II alleles linked to susceptibility to auto- these findings strengthen the notion that and demonstrate an overlap between the immune type 1 diabetes (DQB1*0302 the cause of b-cell dysfunction in unpro- causes of b-cell dysfunction in type 1 and and DRB1*04), suggesting that their voked A2b+ KPD is not related to islet type 2 diabetes because they reveal that a pathophysiology includes a component autoimmunity. strikingly high percentage of patients of islet autoimmunity, whereas unpro- In summary, our results lead to the classified as type 2 possess islet-specific voked A2b+ KPD patients lack the in- conclusion that provoked A2b+KPDis autoreactive T cells. Moreover, the pres- creased frequency of susceptibility associated with a high frequency of islet ence of islet-specific T cells in these type alleles (4). Hence, in the current study, autoimmunity and proinflammatory 2 diabetic patients has been linked to a we used cellular immunoblotting and monocyte populations, which could con- more severe b-cell dysfunction (5). flow cytometry to determine whether tribute to the propensity to ketoacidosis The KPD syndromes include pheno- the differences observed in the clinical and the subsequent progressive loss of types that defy classification according characteristics and natural histories of b-cell function in these patients. On the to traditional definitions of type 1 and these three A2 KPD subgroups are re- other hand, unprovoked A2b+KPDisa type 2 diabetes. As we have demon- lated to the presence or absence of islet- distinct syndrome of reversible b-cell dys- strated previously (1,4,8), the “Ab” clas- specificTcellsand/orproinflammatory function and sustained b-cell functional sification of patients shortly after monocytes (CD14+CD16+). Proinflam- reserve comprising patients who lack ev- presentation with the index DKA is matory monocytes (CD14+CD16+) are idence for humoral and cellular islet au- highly accurate in predicting clinical be- increased in a variety of acute infections toimmunity and whose proneness to havior (especially insulin dependence), and severe inflammatory states such as ketosis may be explained by specificde- b-cell functional reserve, and glycemic sepsis. Therefore, all samples from sub- fects in oxidation of ketones and fatty control. In particular, the characteristics jects were obtained when they were clin- acids and in the metabolism of the keto- of the three subgroups of KPD character- ically stable, without evidence of acute genic amino acid leucine (21). Patients ized by absence of islet autoantibodies inflection or inflammation, to avoid the in the A2b2 subgroup are demograph- (the A2 subgroups) are distinctive, potential increases of inflammatory ically different from the other two with different natural histories of b-cell monocyte populations associated with A2 KPD subgroups, being lean, with ear- dysfunction: inflammation or infection. We discov- lier age of onset of diabetes, and completely ered an inverse relationship between insulinopenic. In our earlier analyses, c A2b2 KPDpatientshavenob-cell BMI and the presence of proinflamma- A2b2 KPD patients were presumed to functional reserve shortly after the ep- tory monocytes. Because an association have a “nonautoimmune” etiology of se- isode of DKA and never recover insulin of obesity with increased numbers of vere b-cell dysfunction on the basis of ab- secretory capacity, indicating severe circulating proinflammatory monocytes sent autoantibodies and a frequency of and permanent b-cell dysfunction. has been reported (20), this finding HLA susceptibility alleles that is lower c Provoked A2b+ KPD patients recover suggests that some factors unique to than that of A+b2 KPDpatients(who b-cellfunctionsoonaftertheindex KPD may drive the systemic inflamma- have classic, autoimmune type 1 diabetes) DKA and may maintain b-cell reserve tory response and override the effects (1). The current study shows that ;50% for up to 4 years, but thereafter mani- of obesity per se. Further studies are of these A2b2 KPD patients are in fact fest progressive loss of b-cell function warranted to investigate this aspect of “T-cell positive”. It appears, therefore, and insulin dependence, suggesting an KPD pathophysiology. that A2b2 KPD subsumes multiple etiol- ongoing process of b-cell dysfunction Frequencies of HLA class II alleles ogies of severe b-cell dysfunction, includ- that recovers only transiently upon in- associated with susceptibility to or pro- ing cellular islet autoimmunity (“T-cell tensive glycemic management. tection against autoimmune type 1 di- positive” patients) and monogenic causes c Unprovoked A2b+ KPD patients re- abetes were measured in all of the patients of defective islet development or function, cover b-cell function soon after the in this study to assess genetic contribu- as we have previously reported (22). Thus, index DKA and maintain substantial tions to cell-mediated islet autoimmunity these findings demonstrate that cellular is- and stable b-cell functional reserve and in the A2 KPD patients. The results, in let autoimmunity is present in some but insulin independence for prolonged conjunction with those previously ob- not all A2 forms of KPD and expand the periods of time (usually .4years). tained in a separate, larger cohort of KPD autoimmune basis of b-cell dysfunction care.diabetesjournals.org DIABETES CARE 5 Islet autoimmunity distinguishes KPD patients beyond the boundaries of classic, autoim- clinical outcomes. J Clin Endocrinol Group. Validity and reproducibility of mune type 1 diabetes. Metab 2003;88:5090–5098 measurement of islet autoreactivity 2. Nalini R, Gaur LK, Maldonado MR, et al. by T-cell assays in subjects with early HLA class II alleles specify phenotypes of type 1 diabetes. Diabetes 2009;58:2588– d ketosis-prone diabetes. Diabetes Care 2595 Acknowledgments This study was sup- 2008;31:1195–1200 12. Bingley PJ, Bonifacio E, Mueller PW. Di- ported by R01-DK-083471, the Diabetes 3. Balasubramanyam A, Garza G, Rodriguez abetes Antibody Standardization Pro- Research Center (P30-KD-017047) at the L, et al. Accuracy and predictive value of gram: first assay proficiency evaluation. University of Washington, National Institutes classification schemes for ketosis-prone Diabetes 2003;52:1128–1136 of Health R21-DK-082827, and the Diabetes diabetes. Diabetes Care 2006;29:2575– 13. Brooks-Worrell BM, Starkebaum GA, Research Center (P30-DK-079638) at Baylor 2579 Greenbaum C, Palmer JP. Peripheral College of Medicine. fl 4. Nalini R, Ozer K, Maldonado M, et al. blood mononuclear cells of insulin- No potential con icts of interest relevant to Presence or absence of a known diabetic dependent diabetic patients respond to this article were reported. ketoacidosis precipitant defines distinct multiple islet cell proteins. J Immunol K.O., R.N., and A.B. diagnosed, managed, syndromes of “A-b+” ketosis-prone di- 1996;157:5668–5674 and phenotyped the KPD patients. D.I. per- abetes based on long-term b-cell function, 14. Brooks-Worrell B, Gersuk VH, formed biochemical analyses. I.C. recruited human leukocyte antigen class II alleles, Greenbaum C, Palmer JP. Intermolecular the patients and maintained the KPD database. and sex predilection. Metabolism 2010; antigen spreading occurs during the pre- C.S.H. performed the autoantibody analyses 59:1448–1455 clinical period of human type 1 diabetes. J and analyzed data. B.M.B.-W. performed the 5. Goel A, Chiu H, Felton J, Palmer JP, Immunol 2001;166:5265–5270 T-cell analyses and flow cytometry and pre- fi fl Brooks-Worrell B. T-cell responses to 15. Brooks-Worrell B, Warsen A, Palmer JP. pared gures. R.N. assisted with the ow cy- islet antigens improves detection of au- Improved T cell assay for identification of tometry. B.M.B.-W., J.P.P., and A.B. conceived toimmune diabetes and identifies pa- type 1 diabetes patients. J Immunol the project, analyzed, and interpreted the data, tients with more severe b-cell lesions in Methods 2009;344:79–83 and drafted the manuscript. J.P.P. guided the phenotypic type 2 diabetes. Diabetes 16. Gersuk VH, Nepom GT. A real-time PCR design of the study and analyzed and in- 2007;56:2110–2115 approach for rapid high resolution sub- terpreted the data. All authors reviewed and 6. Brooks-Worrell BM, Reichow JL, Goel A, typing of HLA-DRB1*04. J Immunol edited the manuscript. B.M.B.-W., J.P.P., and Ismail H, Palmer JP. Identi fication of Methods 2006;317:64–70 A.B. are guarantors of this work and, as such, autoantibody-negative autoimmune type 17. Gersuk VH, Nepom GT. A real-time had full access to all the data in the study and 2 diabetic patients. Diabetes Care 2011; polymerase chain reaction assay for take responsibility for the integrity of the data 34:168–173 the rapid identification of the autoim- and the accuracy of the data analysis. 7. Ziegler-Heitbrock L. The CD14+ CD16+ mune disease-associated allele HLA- The authors thank Mary Lou Arvizu, RN, blood monocytes: their role in infection DQB1*0602. Tissue Antigens 2009;73: (Ben Taub General Hospital, Houston, TX) for and inflammation. J Leukoc Biol 2007;81: 335–340 excellent care of the KPD patients, Jessica 584–592 18. Balasubramanyam A, Nalini R, Hampe Reichow (Seattle Institute for Biomedical and 8. Hampe CS, Nalini R, Maldonado MR, CS, Maldonado M. Syndromes of ketosis- Clinical Research, Seattle, WA) for help with et al. Association of amino-terminal- prone diabetes mellitus. Endocr Rev the manuscript, Helena Reijonen, PhD, Karen specific antiglutamate decarboxylase 2008;29:292–302 Cerosaletti, PhD, and Vivian Gersuk, PhD, (GAD65) autoantibodies with beta-cell 19. Brooks-Worrell BM, Juneja R, Minokadeh (Benaroya Research Institute, Seattle, WA) for functional reserve and a milder clinical A, Greenbaum CJ, Palmer JP. Cellular performing HLA typing and analysis, and phenotype in patients with GAD65 an- immune responses to human islet pro- Dr. Edward J Boyko (VA Puget Sound Health tibodies and ketosis-prone diabetes teins in antibody-positive type 2 diabetic Care System and the University of Washington) mellitus. J Clin Endocrinol Metab 2007; patients. Diabetes 1999;48:983–988 for assistance with statistical analyses. 92:462–467 20. Ghanim H, Aljada A, Hofmeyer D, Syed T, This study was presented as a poster pre- fi 9. Vaziri-Sani F, Oak S, Radtke J, et al. ZnT8 Mohanty P, Dandona P. Circulating sentation at the 71st Scienti cSessionsofthe autoantibody titers in type 1 diabetes pa- mononuclear cells in the obese are in a American Diabetes Association, San Diego, fl – tients decline rapidly after clinical onset. proin ammatory state. Circulation 2004; California, 24 28 June 2011. Autoimmunity 2010;43:598–606 110:1564–1571 10. Seyfert-Margolis V, Gisler TD, Asare AL, 21. Patel SG, Hsu JW, Jahoor F, et al. Patho- et al. Analysis of T-cell assays to measure genesis of A⁻b⁺ ketosis-prone diabetes. References autoimmune responses in subjects with Diabetes 2013;62:912–922 1. Maldonado M, Hampe CS, Gaur LK, et al. type 1 diabetes: results of a blinded con- 22. Haaland WC, Scaduto DI, Maldonado Ketosis-prone diabetes: dissection of a trolled study. Diabetes 2006;55:2588– MR, et al. A-b-subtype of ketosis-prone heterogeneous syndrome using an im- 2594 diabetes is not predominantly a mono- munogenetic and beta-cell functional 11. Herold KC, Brooks-Worrell B, Palmer J, genic diabetic syndrome. Diabetes Care classification, prospective analysis, and et al.; Type 1 Diabetes TrialNet Research 2009;32:873–877

6 DIABETES CARE care.diabetesjournals.org