Islet Transplantation Versus Standard of Care for Type 1 Diabetes Complicated by Severe Hypoglycemia From the Collaborative Islet Transplant Registry and the T1D Exchange Registry Free

Deceased donor islet transplantation was recently approved by the U.S. Food and Drug Administration as the first cellular therapy (Lantidra; CellTrans, Inc.) for adults with type 1 diabetes who are unable to approach target HbA_1c because of current repeated severe hypoglycemia events (SHEs) despite intensive diabetes management and education (1). Long-term follow-up of the Clinical Islet Transplantation Consortium multicenter phase 3 trial of islet-alone transplantation involving 48 individuals from this population demonstrated islet graft survival in 84% of recipients, with HbA_1c maintained at <7.0% in 77% and ≤6.5% in 74%, absence of SHEs in >90%, and ∼50% remaining insulin independent at a median follow-up of 6 years (2). Most other clinical trials of islet transplantation have involved relatively small single-center cohorts, with larger nontrial cohorts available outside the U.S. where regulatory approval and national reimbursement were first achieved (3). The few comparative cohorts analyzed have all crossed over to islet transplantation intervention because of ethical concerns for ongoing SHEs experienced in a population not undergoing transplantation (4,5), including in the only randomized clinical trial, which was 6 months in duration (6). Therefore, direct comparison of outcomes for islet transplant recipients with those of a medically managed group has been largely lacking in the literature.

Despite its consistently reported metabolic benefits, islet transplantation currently requires chronic immunosuppression to prevent alloimmune rejection and autoimmune recurrence; therefore, understanding longer-term benefits and risks associated with islet transplantation compared with ongoing standard of care is critically important. Both diabetes and the immunosuppression required to support a functioning islet graft may affect kidney function, and whether the anticipated improvement in glycemic control expected with islet transplantation may offset any decline in kidney function associated with the immunosuppression regimen is unknown.

To better understand both the long-term benefit for glycemic control and risk of immunosuppression to kidney function associated with islet transplantation when compared with ongoing standard of care, we conducted a case-control comparison of islet transplantation with standard of care in the U.S. by selecting patients from the Collaborative Islet Transplant Registry (CITR) and the T1D Exchange (T1DX) Clinic Network and Registry with at least one SHE in the year before transplantation or enrollment, respectively, and compared outcome measures of glycemic control, SHEs, insulin use and delivery, and kidney function over 5 years of follow-up.

CITR represents the majority of islet transplantation activity in the U.S., Canada, Australia, and JDRF-sponsored programs in Europe and presently includes >1,000 islet transplant–alone recipients (7). The T1DX Clinic Network and Registry represents a diverse array of pediatric and adult specialized diabetes practices in the U.S. and includes >3,000 individuals who completed 5 years of follow-up (8). The present analysis focused on islet transplant–alone case subjects from CITR conducted in the U.S. for comparison with control patients with type 1 diabetes from the T1DX Clinic Network and Registry with no history of previous transplantation. Registrants in the T1DX Clinic Network received their diabetes care from specialized diabetes practices with access to the latest insulin delivery and continuous glucose monitoring (CGM) technologies; standard of care reflected their observed treatment without any specified intervention. All U.S. cities with a CITR clinical center also contained at least one adult T1DX Clinic Network site (Supplementary Fig. 1).

Case subjects were selected from CITR based on the following criteria: recipient with type 1 diabetes of islet-alone transplantation, no history of prior pancreas or kidney transplantation, and presence of at least one SHE with seizure or loss of consciousness in the year before transplantation. A period up to 5 years of follow-up from last infusion was examined as available for each of the 71 selected CITR participants. Control subjects were selected from the T1DX Clinic Network and Registry based on the following criteria: type 1 diabetes; no history of prior islet, pancreas, or kidney transplantation; 5-year follow-up data available; and presence of at least one SHE with seizure or loss of consciousness in the year before initial enrollment in the registry. Although an SHE can be defined as any episode of severely low blood glucose characterized by altered mental and/or physical status requiring assistance for treatment of hypoglycemia (9), T1DX limited capture of SHEs to those that resulted in seizure or loss of consciousness. Therefore, to harmonize data representation between the registries, SHEs for both CITR and T1DX included only those with seizure or loss of consciousness. SHEs were reported to CITR by the transplantation centers that collected data prospectively during the conduct of phase 1–3 trials of islet transplantation and were reported to T1DX by registrant-completed questionnaire. Baseline for the purposes of the present analysis was taken from an individual’s enrollment in either CITR or the T1DX Registry.

The primary outcome was HbA_1c <7.0% and absence of an SHE assessed annually; secondary outcomes included HbA_1c ≤6.5% and absence of an SHE, HbA_1c <7.0%, HbA_1c ≤6.5%, change from baseline in HbA_1c, incidence of SHEs, insulin requirements and delivery modality, and kidney function assessed as both the estimated glomerular filtration rate (eGFR) and change from baseline in eGFR. The CITR statistician (C.M.B.) produced the original prespecified statistical analysis plan (Supplementary Fig. 2) in collaboration with the T1DX statistician (N.C.F.) and with input from all authors. Note that main cohort results are presented in this report and were performed as possible based on actual data availability from both registries. This investigation was undertaken jointly between CITR investigators (M.R.R., R.A., D.A.B., M.D.B., B.J.H., F.K., A.B., F.B.B., E.H.P.), T1DX investigators (M.R.R., K.M.M.), and the National Institute of Diabetes and Digestive and Kidney Disease (T.L.E.).

Continuous variables are summarized by mean ± SD, and categorical variables are summarized by percentage. To identify significant differences at baseline between the cohorts, tests of means (two-sample t test) or proportions (Pearson χ² test), respectively, were performed. Comparisons of continuous outcomes between cohorts were conducted using mixed models for repeated measures, and categorical outcomes were modeled via logistic generalized estimating equations with repeated measures. Baseline clinical and demographic variables of interest were considered for inclusion as covariates. Covariates with a statistically significant difference in outcome or at baseline between cohorts were retained in final models, in addition to study and visit. Time-to-event outcomes, including mortality and SHEs, were analyzed using the Kaplan-Meier method. The Wilcoxon test was used to determine significant differences in time to event between the CITR islet transplantation case cohort and the T1DX control cohort. Analyses were performed using SAS 9.4. Significance was considered at P < 0.01 (two tailed) to account for multiple comparisons.

The data sets generated during and/or analyzed in the current study are available from the corresponding author on reasonable request.

Case subjects from CITR (n = 71) had baseline assessments completed between 2000 and 2014, whereas control subjects from T1DX (n = 213) underwent baseline assessment between 2010 and 2012 (CONSORT flow diagram is presented in Supplementary Fig. 3). The cohorts from CITR and T1DX were similar in age and diabetes duration, with the CITR cohort having a higher proportion of women (69% vs. 51%; P = 0.009), lower BMI (23 ± 2 vs. 27 ± 5 kg/m²; P < 0.0001), lower daily insulin requirements (0.52 ± 0.14 vs. 0.64 ± 0.31 units ⋅ kg⁻¹ ⋅ day⁻¹; P < 0.0001), and lower HbA_1c (7.3 ± 1.0% vs. 7.7 ± 1.4%; P = 0.003) and experiencing a slightly higher frequency of SHEs in the year before the baseline assessment (Table 1). Baseline C-peptide was similarly undetectable or very low in both cohorts, and there was no baseline difference in kidney function based on the Chronic Kidney Disease Epidemiology Collaboration equation (10) between the cohorts, although a higher percentage of T1DX participants had lower categorized eGFR values at baseline (Table 1). Baseline variables were considered in initial full models, and statistically significant baseline covariates were retained in final models. Two CITR participants with missing baseline HbA_1c data were dropped from modeling of HbA_1c outcomes. CGM use was not reported for CITR participants; however, of the T1DX participants, 23% reported CGM use at baseline.

The primary outcome of HbA_1c <7.0% and absence of an SHE was met by 71–80% in CITR versus 21–33% in T1DX over 5 years (P < 0.001) (Fig. 1A), and the outcome of HbA_1c ≤6.5% and absence of an SHE was met by 60–75% in CITR versus 10–20% in T1DX (P < 0.001) (Fig. 1B). The proportion with HbA_1c <7.0% was 73–88% in CITR versus 32–36% in T1DX (P < 0.001) (Fig. 1C), the proportion with HbA_1c ≤6.5% was 63–80% in CITR versus 16–22% in T1DX (P < 0.001) (Fig. 1D), and the change from baseline in HbA_1c was −0.77 to −1.02 percentage points in CITR versus −0.06 to +0.03 percentage points in T1DX (P < 0.001) (Fig. 1E). The CITR cohort experienced 17 SHEs (in six of 71 participants) over 225 person-years of follow-up, fewer than the T1DX cohort experienced at 97 SHEs (in 46 of 213 participants) over 666 person-years of follow-up (7.6 vs. 14.6 SHEs per 100 person-years; hazard ratio 0.402; 95% CI 0.192–0.841; P = 0.016) (Fig. 1F), but did not meet prespecified criteria for statistical significance. A listing of SHEs is included as Supplementary Fig. 4. Four of the six CITR participants who experienced at least one SHE during follow-up had undetectable fasting C-peptide consistent with islet graft failure, and all were insulin dependent.

Insulin requirements decreased to 0.07–0.15 units ⋅ kg⁻¹ ⋅ day⁻¹ in CITR versus remaining at 0.56–0.67 units ⋅ kg⁻¹ ⋅ day⁻¹ in T1DX over 5 years (P < 0.001) (Fig. 2A), with 52–75% insulin independent in CITR (Fig. 2B). The reduction in insulin requirements in CITR and, for a majority, the achievement of insulin independence were accompanied by the restoration of clinically significant endogenous insulin secretion, as indicated by mean levels of C-peptide ranging from 0.24 to 0.41 nmol/L (Fig. 2C). Insulin delivery by continuous subcutaneous insulin infusion or pump decreased to 4–7% in CITR versus increasing to 53–66% in T1DX (P < 0.001) (Fig. 2D). A total of 18–41% of the T1DX participants reported CGM use over the 5-year follow-up.

Kidney function, measured by eGFR, decreased to 71–84 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² in CITR versus 81–87 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² in T1DX over 5 years (P < 0.001) (Fig. 3A), and the change from baseline in eGFR was −8.8 to −20 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² in CITR versus −1.3 to −6.5 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² in T1DX (P < 0.001) (Fig. 3B). A waterfall plot by year is also provided for enhanced visualization and comparison of change from baseline in eGFR between studies (Supplementary Fig. 5A) and for CITR participants by insulin independence status (Supplementary Fig. 5B). Using the model predicting change in eGFR status, including CITR participants with insulin independence status compared with all insulin dependence and retaining covariates with an association with the outcome, a significant relationship existed between insulin independence status and greater decrease in eGFR (P = 0.002). This relationship was further examined via summaries of HbA_1c (Supplementary Fig. 5C) and eGFR (Supplementary Fig. 5D) values over time for CITR participants, by insulin independence status. The greatest mean decrease in eGFR (−8.8 mL ⋅ min⁻¹ ⋅ 1.73 m⁻²) occurred in CITR participants from baseline to year 1 and seemed to be driven by the insulin-independent group and consistent with HbA_1c value change, reflecting a greater reduction in average glycemia. Mean annual change over subsequent years for CITR participants was −2.7 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² and for T1DX participants was −0.6 mL ⋅ min⁻¹ ⋅ 1.73 m⁻².

There was only one reported death among the 71 CITR participants and six deaths among the 213 T1DX participants; given these small numbers, formal analysis was not pursued. A total of 55 possibly related or related serious adverse events (SAEs) were reported from the CITR participants, with 29 participants reporting at least one related SAE. Of these 55 SAEs, 13 were determined to be possibly related or related to the islet infusion procedure and 47 to the immunosuppression regimen. Only three events were noted as not recovered/not resolved. The most commonly reported SAEs were abnormal neutrophil count (n = 8) and infection (n = 7). Five cancer events were reported among four participants: two instances of squamous cell carcinoma in one individual, an unspecified malignancy in another, papillary thyroid cancer in another, and metastatic lung adenocarcinoma in an individual who died and had a history of long-term tobacco use. Panel-reactive alloantibodies were present in six of the 71 CITR participants at baseline and became positive in 14 of the 65 who were initially negative during follow-up, seven of whom experienced islet graft failure.

Only 29 of the 71 CITR participants had baseline retinopathy status information, reported as present in 13 and absent in 16. Of the 16 CITR participants who reported no retinopathy at baseline, none developed retinopathy over the 5-year follow-up period. Of the 13 CITR participants who reported retinopathy at baseline (n = 9 nonproliferative and n = 4 proliferative), eight reported eye surgery at baseline, and one participant reported postbaseline eye surgery during the 5-year follow-up period. Six participants were reported as no longer having retinopathy at their last nonmissing status during the 5-year follow-up period; remaining participants did not indicate a change in retinopathy stage.

The current study demonstrates the long-term durability of islet transplantation performed by CITR centers in the U.S. to achieve near-normal glycemic control in the absence of an SHE, which occurred more often than observed with standard of care delivered by specialized diabetes practices of the T1DX Clinic Network in the U.S. The improved glycemic control in CITR was associated with clinically significant reductions in insulin requirements, such that a majority achieved and maintained insulin independence, with ongoing evidence by C-peptide levels of islet β-cell graft function. However, kidney function declined at a greater rate in the CITR cohort when compared with the T1DX cohort, an effect likely explained by the ongoing requirement for calcineurin inhibitor–based immunosuppression to protect the islet graft from alloimmune rejection and autoimmune recurrence. Together, these results help to frame the anticipated long-term metabolic benefits of restoring β-cell function through islet transplantation against the current immunosuppression risks to kidney function.

The improved glycemic control observed with islet transplantation relative to standard of care translates to a more than twofold increase in the proportion maintaining HbA_1c <7.0% and an almost fourfold increase in the proportion maintaining near-normal glycemic control with HbA_1c ≤6.5% over the 5-year period of observation. This is remarkable, because the increased risk of cardiovascular and all-cause mortality in type 1 diabetes continues to decline as HbA_1c decreases below 7.0% (11), with the lowest mortality rates observed with HbA_1c ≤6.5% (12). Importantly, the higher likelihood of achieving on-target (HbA_1c <7.0%) and near-normal (HbA_1c ≤6.5%) glycemic control with islet transplantation than with standard of care was sustained over 5 years, along with a reduced risk of experiencing SHEs. These results support international consensus recommendations for the goals of β-cell replacement therapy to achieve HbA_1c <7.0% (and optimally HbA_1c ≤6.5%) in the absence of SHEs (13), which is in contrast with current American Diabetes Association standard-of-care recommendations to set less-stringent goals (e.g., HbA_1c up to 8%) for individuals experiencing severe or frequent hypoglycemia (14). In practice, setting a higher HbA_1c goal for individuals experiencing SHEs can be challenging for those fearful of the complications of sustained hyperglycemia. At baseline, HbA_1c was lower in the CITR than in the T1DX cohort and likely explained by greater hypoglycemia exposure in the CITR relative to the T1DX cohort, which is supported by the higher incidence of SHEs reported at baseline in CITR.

A reduction in insulin requirements is expected with islet transplantation, and the present report confirms that most recipients can expect to experience a period of prolonged insulin independence. Indeed, the restored levels of C-peptide observed in this CITR cohort are consistent with those required to meet the glycemic goals for β-cell replacement as well as insulin independence (15). For those individuals in the CITR cohort remaining on or returning to insulin, insulin requirements remained substantially lower than those in the T1DX cohort. Such low insulin requirements can be delivered more simply, clinically often achieved with a single daily injection of long-acting insulin. Although the number of daily injections was unfortunately not collected during follow-up in CITR, very few islet recipients used a pump for insulin delivery, whereas an increasing majority of registrants in the T1DX cohort used an insulin pump. Therefore, the simplification of insulin delivery may itself represent an important outcome and potential benefit of current and future β-cell replacement therapies (16). This reduction in insulin requirements may explain much of the reduction in SHEs; however, the restoration of a glucagon response to insulin-induced hypoglycemia afforded by islet transplantation (17) and the resultant avoidance of even subclinical hypoglycemia can reverse hypoglycemia-associated autonomic failure in islet transplant recipients previously experiencing SHEs (18).

The current requirement for chronic immunosuppression to prevent alloimmune rejection and autoimmune recurrence affecting transplanted islets in type 1 diabetes carries important long-term risks that limit broader application of β-cell replacement to individuals other than adults experiencing problematic hypoglycemia or individuals who may already require immunosuppression to support another organ transplant. In our analysis, kidney function as measured by eGFR declined more over the 5-year follow-up in the CITR than in the T1DX cohort. The greatest annual decline in eGFR (−8.8 mL ⋅ min⁻¹ ⋅ 1.73 m⁻²) was observed during the first year posttransplantation, which may be attributable to both the initiation of immunosuppression, because calcineurin inhibitor–based immunosuppression can induce glomerular afferent vasoconstriction (19), and the normalization of glycemia can reduce glomerular hyperfiltration (20). Evidence for an effect of reduced glomerular hyperfiltration in the CITR cohort examined here comes from the greater reduction in eGFR observed in the first year posttransplantation among recipients who achieved insulin independence who also had greater reduction in HbA_1c than those who remained insulin dependent. Nevertheless, eGFR continued to decline at a rate of −2.7 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² per year of remaining follow-up, compared with a decline of −0.6 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² per year of follow-up for the T1DX cohort. This eGFR decline in the CITR cohort was greater than that seen in the Clinical Islet Transplantation Consortium islet-alone phase 3 trial, which was −6.9 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² in the first year and −1.3 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² thereafter, more consistent with the current T1DX cohort. In contrast, an earlier study of both islet transplantation and islet transplantation–eligible cohorts of individuals with type 1 diabetes observed a lower rate of decline in measured GFR of −1.3 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² in the islet transplant recipients relative to a rate of −3.0 mL ⋅ min⁻¹ ⋅ 1.73 m⁻² in those receiving medical therapy (4). Differences in reported rates of eGFR decline after islet transplantation or standard of care may be related to differences in baseline eGFR or, for islet transplantation cohorts, the combination and levels of immunosuppression drugs used. Together, these data highlight the risk of progression of underlying diabetic kidney disease in individuals with long-duration type 1 diabetes, which, if mitigated by the improved glycemic control seen with islet transplantation, may still be offset by the risk of nephrotoxicity associated with calcineurin inhibitor–based immunosuppression. These findings support ongoing efforts to identify clinically safe and efficacious calcineurin inhibitor–free immunosuppression that may be possible through costimulation blockage (ClinicalTrials.gov identifier NCT06305286) (21). Such advances in immunosuppression approaches will be necessary to broaden eligibility for islet transplantation to those with established kidney disease.

Additional risks of immunosuppression include infection and certain cancers, such as nonmelanoma skin cancers, which were not captured in the T1DX Registry and so could not be compared between cohorts. In the CITR cohort, the incidence of immunosuppression-related infections and cancers was low and consistent with prior reports for islet transplantation (2,3,7). Panel-reactive alloantibodies developed in ∼20% of those without alloantibodies present before transplantation, half of whom experienced islet graft failure. Prevalence rates for panel-reactive alloantibody sensitization >50% have been reported after the withdrawal of immunosuppression for islet graft failure in recipients of islet-alone transplantation (7), which theoretically may result in longer waiting time for a future organ transplantation. Limited retinopathy data available in CITR indicate remarkable stability and possible improvement of retinal disease, consistent with a prior report (4).

Strengths of the current study include the relatively large number of individuals with type 1 diabetes included who were experiencing SHEs resulting in seizure or loss of consciousness at baseline, the temporal and geographic overlap of the two registries compared, and the prospectively collected 5-year follow-up data obtained in both CITR and T1DX. There are nonetheless a few important limitations. First, our analyses were constrained by those measurements and outcomes collected similarly in the two registries. This includes the outcome of SHEs being restricted to those resulting in seizure or loss of consciousness. Second, despite adjustment for differences in baseline covariates between the study cohorts, retrospective case-control studies are susceptible to unidentified confounders, potentially explaining the differences in outcomes between groups. We did not have measures of socioeconomic status or access to care, but all participants received specialized transplantation or diabetes practice care in overlapping geographic regions. Both registries were affected by missing data, although the proportion of missing data for the primary and secondary outcomes at various annual time points was similar between the two registries. Third, technology use in the management of type 1 diabetes has further increased since these T1DX Clinic Network data were collected; however, more recent cross-sectional data from the ongoing virtual T1DX Registry indicates that impaired awareness of hypoglycemia remains prevalent in ∼30% of adults with type 1 diabetes despite use of CGM and/or automated insulin delivery, and ∼20% have experienced at least one SHE in the prior year, including 17% of those using automated insulin delivery (22). Although this survey study did not report on HbA_1c for the subgroup experiencing SHEs, only 58% overall reported achieving HbA_1c <7.0% and just 35% reporting achieving HbA_1c <6.5% (22). Therefore, even with the best currently available standard of care, hypoglycemia remains a significant clinical issue, and many individuals with type 1 diabetes remain unable to meet current recommended targets for glycemic control, problems that may be addressed through islet transplantation where available.

In conclusion, individuals with long-standing type 1 diabetes complicated by SHEs experience long-term benefit for glycemic control and reduction in severe hypoglycemia as well as risk of immunosuppression to kidney function when compared with ongoing standard of care. When considering islet transplantation for such individuals with problematic hypoglycemia, the anticipated benefits of achieving recommended targets for HbA_1c in the absence of SHEs together with more simplified or no insulin requirements must be balanced against the risk of a potentially greater decline in kidney function as well as that for less common immunosuppression-related infections and malignancies over the long term (3). Future work is required to extend the eligibility for islet transplantation in the U.S. beyond adults with severely problematic hypoglycemia, in particular, to those individuals already requiring immunosuppression in support of another organ transplant, while continuing efforts to reduce or eliminate the requirement for chronic immunosuppression through the achievement of operational tolerance (23) or immune evasiveness of transplanted islets.

Acknowledgments. The authors thank Dr. Roy W. Beck from the Jaeb Center for Health Research for input on the design and preliminary analyses of this work and the individuals with type 1 diabetes who participated in CITR and the T1DX Registry. A list of the participating CITR clinical centers can be found in Supplementary Fig. 6.

M.R.R. is an editor of Diabetes Care but was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Funding. CITR is supported by Public Health Services research grant U01-DK133709 from the National Institutes of Health, and the T1DX Clinic Network was supported by a grant from the Leona M. and Harry B. Helmsley Charitable Trust.

Duality of Interest. M.R.R. serves on a scientific advisory board for Vertex Pharmaceuticals and a data and safety monitoring board for Sernova Corp. and has received research support from Dompé farmaceutici S.p.A. and Tandem Diabetes Care. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.R.R. wrote the first draft of the manuscript. M.R.R., C.M.B., and N.C.F. conceived of and designed the study. All authors collected and/or analyzed data; contributed to the interpretation of the results; and reviewed, edited, and approved the final version of the manuscript. E.H.P. 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.

Prior Presentation. Preliminary results from this work were presented orally in abstract form at the 79th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 7–11 June 2019, and at the 17th World Congress of the International Pancreas & Islet Transplant Association, Lyon, France, 2–5 July 2019.

Handling Editors. The journal editor responsible for overseeing the review of the manuscript was Mark A. Atkinson.