HLA-A*24 Carrier Status and Autoantibody Surges Posttransplantation Associate With Poor Functional Outcome in Recipients of an Islet Allograft

The success rate of an intraportal islet graft for achieving insulin independence critically hinges on the amount of donor tissue implanted (1). However, a positive outcome can be counteracted by other factors, including preexisting autoreactive T cells, HLA alloantibodies, or high lymphocyte counts (2–4). In recipients of a sufficiently large intraportal islet allograft (5,6), we investigated whether the presence of HLA-A*24, HLA-DQ2/DQ8, or rising islet autoantibody levels, previously shown to independently predict rapid β-cell loss in prediabetes (7,8), associates with poor functional graft outcome.

Forty-one patients with C-peptide–negative type 1 diabetes (median age 45 years) with a history of recurrent hypoglycemia and negative for preexisting HLA class I and class II antibodies (4) received ≥2.3 million β-cells (∼4,600 islet equivalents)/kg body wt in one to two intraportal islet implantations (5,6) (clinical trial reg. nos. NCT00798785 and NCT00623610) under antithymocyte globulin-mycophenolate mofetil-tacrolimus therapy. After standardized islet isolation and culture, the cellular composition of the grafts was analyzed before infusion.

Plasma C-peptide and proinsulin were measured with a time-resolved fluorescence immunoassay (AutoDELFIA; PerkinElmer, Turku, Finland) (9), HbA_1c levels with high-performance liquid chromatography (Tosoh Bioscience, Tessenderlo, Belgium), and lymphocyte subclasses with an EPICS XL flow cytometer (Beckman Coulter, Miami, FL). Insulin independence was defined as insulin dose ≤8 units ⋅ kg⁻¹ ⋅ day⁻¹ and C-peptide negativity as two consecutive random C-peptide values ≤0.05 ng/mL. Autoantibodies against GAD (GADA), against IA-2 (IA-2A), and against ZnT8 (ZnT8A) were determined by liquid phase radiobinding assays (10) before the first implantation, 6 weeks and 6 months thereafter, and 6 weeks after the second transplantation, using the 99th percentile of controls as cutoff (10). Significant autoantibody surges were defined according to Decochez et al. (11). Assays were validated by participation in successive Diabetes Antibody Standardization Programs (10). HLA-DQ2/DQ8 and -A*24 were determined by allele-specific oligotyping (8). The hyperglycemic clamp procedure was performed as previously described (6). C-peptide measurements at 120, 135, and 150 min of hyperglycemia (180 mg/dL) were used to calculate the C-peptide area under the curve (AUC).

The data were analyzed with Mann-Whitney U, χ², or Fisher exact tests. Kaplan-Meier analysis with log-rank test and Cox proportional hazards model, performed by forward stepwise method, were used to assess the development of insulin independence or C-peptide negativity. All tests were performed two sided with SPSS 20 for Windows (IBM Corporation, Chicago, IL). P < 0.05 was considered significant.

Two patients had positive tests for all three autoantibodies, 6 for two (GADA and IA-2A), 18 for one (GADA n = 11, IA-2A n = 4, ZnT8A n = 3), and 15 for none. Ten patients experienced a significant increase in autoantibody level, including seven in whom an additional autoantibody developed within 6 months (GADA in five and IA-2A in two) and three with increases of preexisting GADA. Thirteen patients carried HLA-A*24 and seven HLA-DQ2/DQ8.

Pretransplant autoantibody levels did not correlate with AUC C-peptide or insulin dose (data not shown). However, patients with an autoantibody surge ≤6 months after the first implantation (n = 10) tended to have lower C-peptide levels and higher insulin doses despite a trend toward higher HbA_1c during 18 months posttransplantation than patients without such a rise (n = 31) (Fig. 1A–C). Likewise, patients positive for HLA-A*24 (n = 13) experienced lower C-peptide levels and higher insulin doses and HbA_1c during follow-up than those negative for HLA-A*24 (n = 28) (Fig. 1D–F). Both groups did not differ in the frequency of receiving a graft containing one or more HLA-A*24–positive donor or in the number of HLA-A*24–positive donors per graft (data not shown). In patients who were positive for HLA-A*24 or who experienced an early autoantibody surge (n = 19) compared with patients lacking both risk factors, differences in C-peptide level (P ≤ 0.003), insulin dose (P ≤ 0.001), and HbA_1c (P ≤ 0.018) were strengthened (Fig. 1G–I). Outcome was unaltered by the presence or absence of HLA-DQ2, -DQ8, or -DQ2/DQ8 (data not shown).

Patients without an autoantibody surge more often tended to become insulin independent and to remain C-peptide positive 18 months posttransplantation than those with such a surge (Supplementary Fig. 1A and B); the same applied for patients negative for HLA-A*24 versus those positive for HLA-A*24 (Supplementary Fig. 1C and D). All recipients lacking both risk factors (n = 22) remained C-peptide positive after 18 months (P < 0.001 vs. 47% if one or more risk factors), and 68% of them achieved insulin independence (P = 0.002 vs. 16% if one or more risk factors) (Supplementary Fig. 1E and F).

In the poor response group (n = 19) only four patients presented both risk factors (Supplementary Fig. 2); hence, the presence of either one explains a larger fraction of recipients with poor outcome. Progression rates to insulin independence or C-peptide negativity were not different according to the number of risk factors present (data not shown).

The subgroup with HLA-A*24 or autoantibody surge (n = 19) did not differ from the rest in patient or graft characteristics except for a lower age and, logically, a higher prevalence of HLA-A*24 and autoantibody surges (Table 1). To identify predictors of insulin independence (not performed for the development of C-peptide negativity due to low event numbers) all variables from Table 1 were tested in univariate Cox regression. Those with P < 0.1 (tacrolimus level over first 6 months, baseline HbA_1c) were entered in a multivariate model together with HLA-A*24/autoantibody surge. Only the latter remained significant (P = 0.007; 95% CI for hazard ratio 0.053–0.635).

Recipient HLA-A*24 carrier status and autoantibody surges, mostly GADA, within 6 months after the first islet implantation associate with poor functional outcome. The joint absence of both risk factors identified individuals with higher and more persistent C-peptide release, lower insulin dose and HbA_1c, and higher probability of insulin independence during 18 months than in participants presenting at least one marker. This different outcome could not be explained by differences in recipient age, immune status, graft composition, or immune suppression.

The relatively large size and homogeneity of the cohort is a strength of this study. Nevertheless, our observations could have benefited from larger patient numbers in the subgroups that turned out relevant. All patients received sufficient β-cells to generate plasma C-peptide levels at posttransplantation month 2 in C-peptide–negative recipients (2,5) under similar immunosuppression.

The current results indicate that similar to observations in pre–type 1 diabetes, the presence of HLA-A*24 or autoantibody surges predicts β-cell loss during an inflammatory and/or immune process in islet allograft recipients (7,8). The association of early autoantibody rises with poor graft outcome confirms previous observations and has been explained as the consequence of autoantigen release after β-cell death (12). The negative influence of HLA-A*24 is, to the best of our knowledge, novel and consistent with its association with accelerated progression to clinical onset in prediabetes (8). Efficient HLA-A*24–restricted presentation of autoantigen fragments to cytotoxic T cells (13) may play a role in the accelerated functional loss posttransplantation, but HLA-A*24 expression on non–β-cells is likely to be involved because it is the recipients’ HLA-A*24 that affects outcome. The combination of HLA-A*24 and autoantibody surges is consistent with reports that HLA-A*24 associates with attenuated humoral responses to islet autoantigens in prediabetes and diabetes (14). The presence of HLA-DQ2/DQ8, another independent predictor of diabetes in risk groups (8), was not associated with poor graft outcome.

In conclusion, in recipients of an intraportal islet allograft, the presence of HLA-A*24 or an early autoantibody surge associates with poor metabolic outcome. Autoantibody monitoring creates a time window during which immunosuppression could be changed or another therapeutic intervention planned. When comparing efficacy of transplantation protocols or prognostic value of novel biomarkers, results should be adjusted for HLA-A*24 carrier status.

Acknowledgments. The authors thank Monique Robyn, Brigitta Swennen, Katrien Rouffe, and Nadine Pardon (University Hospitals Leuven); Koen Verbeeck, Sofie Vandenhoeck, Soniya Thomas, and Jenny Van Den Brande (University Hospitals Brussels); and Rie Braspenning (Antwerp University Hospital) for completing all clamp tests. The authors also thank the Eurotransplant International Foundation and its transplant surgeons and coordinators for organ procurement. Finally, the authors thank the staff of the Beta-Cell Bank, the Belgian Diabetes Registry, the Diabetes Research Center of the Vrije Universiteit Brussel, and the Clinical Biology Department of the University Hospital Brussels.

Funding. This study was supported by the JDRF (grant 4/2005/1327) and the W. Gepts Fund of University Hospitals Brussels. P.G. is funded by the Clinical Research Foundation of the University Hospitals Leuven, Katholieke Universiteit Leuven. B.K. is senior clinical investigator of the Research Foundation Flanders (Fonds Wetenschappelijk Onderzoek-Vlaanderen).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.D. and F.K.G. contributed to the study design, literature search, data collection, data analysis and interpretation, and writing of the manuscript. E.M.B. contributed to the study design, literature search, and data collection, analysis, and interpretation. B.J.V.d.A., P.G., R.H., D.L., U.V.d.V., and Z.L. researched data and reviewed the manuscript. B.O.R. contributed to the data research, discussion, and review and editing of the manuscript. D.G.P. contributed to the discussion and editing of the manuscript. B.K. contributed to the study design, literature search, data collection, data analysis and interpretation, discussion, and editing of the manuscript. F.K.G. and B.K. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.