figure b

Introduction

Islet transplantation, a minimally invasive transplant treatment modality [1], restores near-normoglycaemia and awareness of hypoglycaemia, and provides protection from severe hypoglycaemic events (SHEs) in recipients with type 1 diabetes and impaired awareness of hypoglycaemia (IAH) [2,3,4,5,6,7,8]. The benefits of islet transplant alone (ITA) must be weighed against the risks associated with immunosuppression. In recipients of islet-after-kidney (IAK) and simultaneous-islet-kidney (SIK) transplants, who are immunocompromised to prevent kidney allograft refection [9,10,11,12], the added risk of an islet transplant is largely limited to rare procedural complications when following best clinical practices [13].

For islet transplantation outcomes to continue to improve, an analysis of the factors associated with favourable long-term outcomes is required. The Collaborative Islet Transplant Registry (CITR) is an international registry positioned to conduct such an analysis [14]. It was established in 2001 and has since been funded by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) in the USA to compile, analyse and communicate data from all transplant centres in the USA and Canada and from participating European and Australian centres, where islet transplantation is now approved by several national health authorities. The CITR database as of 6 January 2017 included recipient, donor, islet and study protocol information from 877 ITA and 207 IAK/SIK transplants carried out between 1999 and 2015, with at least 1 year of follow-up, constituting the single largest collection of human islet transplant data in the world [15]. Of the 877 ITA recipients, the 398 ITA recipients with type 1 diabetes who were C-peptide negative (<0.1 nmol/l) and who experienced SHEs pre-transplant were included in this analysis. Most (74%) of the included recipients received at least two islet infusions, with 24% receiving three or more (electronic supplementary material [ESM] Table 1). The mean cumulative number of islets infused per recipient was 863,000 (ESM Table 2).

We sought to identify those recipient, donor, islet and medical management characteristics that were associated with the highest rates of successful study outcomes in ITA recipients. The four factors confirmed as strong predictors via multivariate models were used to define a subgroup having all the favourable factors that were common to all the study outcomes (4CFF) and those with fewer than all the favourable factors (<4CFF). We then computed and compared the prevalence rates of the outcomes between these two subgroups to confirm the effect of the identified factors. These estimates can reasonably be used to project unbiased rates of clinical success of islet transplantation in patients managed with current best clinical practices.

Methods

Study design and participants

CITR is a prospective, observational, voluntary registry of islet allograft recipients, capturing data on about 80% of all allogeneic islet transplants at participating institutions since 1999. CITR receives US pancreas donor data from the US United Network for Organ Sharing (UNOS) [16], and relevant data from the Clinical Islet Transplant Consortium (CIT) funded by the National Institutes of Health [17]. Human islets checklists for each islet preparation included in this analysis are provided as ESM.

The current analysis included 398 individuals with type 1 diabetes who received an allogeneic ITA between 1999 and 2015, with pre-transplant fasting C-peptide <0.1 nmol/l, at least one SHE experienced in the year prior to transplant (defined as an episode with symptoms consistent with hypoglycaemia, in which the individual required the assistance of another person and which was associated with a blood glucose level of <2.77 mmol/l or prompt recovery after carbohydrate [oral] or glucagon or glucose [i.v.] [18]), and providing written informed consent to register in CITR. Most recipients (99%) showed evidence of IAH pre-transplant (Gold or Clarke questionnaire). Reasons for excluding participants from the current analysis are presented in ESM Table 3. The registry protocol was reviewed and approved by each participating centre’s institutional review board. This cohort had predictor data at baseline and clinical outcomes reported at 1–5 years following the last islet infusion. The CITR represents the most current and most comprehensive global resource for human islet transplantation data, with the longest follow-up consistently monitored throughout the life of the registry, including ascertainment of safety data.

Outcomes

The following outcomes were analysed: (1) HbA1c <53 mmol/mol (7.0%) and absence of SHEs; (2) HbA1c <53 mmol/mol (7.0%); (3) absence of SHEs; (4) fasting C-peptide ≥0.1 nmol/l; (5) fasting glucose of 3.3–7.8 mmol/l; and (6) insulin independence, defined as >14 days with no exogenous insulin use (intervals of <5 days at minimal levels were not considered as return to insulin use). The main safety measures were the incidence of serious adverse events (SAEs) deemed by the local investigator, and adjudicated by the data committee, as possibly or definitely related to the islet product/infusion procedure or to the immunosuppression regimen per recipient from first infusion through to 5 years post last infusion (the period of safety risk), and the outcome of all immunosuppression-related SAEs in the recipient during their period of follow-up (ESM Tables 4–10). SAEs are defined in the ESM Methods. Additionally, we analysed total (all-cause) mortality, regardless of relatedness to the islet transplant procedure or immunosuppression.

Follow-up and data collection

CITR efficacy and safety data are collected at protocol-designated time points through 10 years from last transplant, regardless of current or previous total islet graft failure (C-peptide <0.1 nmol/l or undetectable for three consecutive visits without recovery, or re-transplant). In the event of complete graft failure and no data collected at the visit, the following data were imputed: insulin use was set as ‘using exogenous insulin’; and C-peptide was set at <0.1 nmol/l. HbA1c, fasting glucose and SHEs were set to missing and analysed as missing at random (outcome not related to absence of data). ESM Table 11 shows the number of participants in the defined cohort who were expected at each annual follow-up visit, and the percentage on whom each study endpoint was available. Immunosuppression was coded as each agent class (T cell-depleting [TCD] antibodies, calcineurin inhibitors, etc.) as ever having been administered from first transplant or not. Often, two or more agents were given in various combinations [15]. About 74% of the present cohort received more than one islet transplant. Analysing outcomes following the last islet infusion is the only approach that accounts for all treatment given. The Data Coordinating Center has verified participant voluntary informed consent and adverse events reporting.

During its 20 years of operation, CITR has utilised best data collection practices including regular conference calls and site visits to monitor data reporting, regular meetings of the oversight committees and the Executive Committee, regular reviews by the NIDDK, and a rigorous reporting system for relevant adverse events and other safety outcomes. Data on SHEs was ascertained at every scheduled follow-up visit: whether there were any events, how many and whether they involved the help of another person, a seizure or loss of consciousness, ambulance or hospitalisation. The occurrence or absence of SHEs was an auditable item during site reviews.

Statistical considerations

All available donor, islet, recipient and immunosuppression variables were analysed univariately for association with each of the study outcomes. The 108 predictors analysed are presented in ESM Table 1. In the case of recipients with multiple infusions or donors, information was collapsed by logical combination (e.g. an infusion product derived from a female donor and a male donor is termed ‘Mixed’), averaging (e.g. viability, stimulation index, etc.) or summation (e.g. total beta cells, total islet equivalents [IEQs] infused, etc.). For each outcome, predictors associated at p<0.10 and exhibiting at least 10% higher rates of success were then included in a multivariate model and significant predictors from the multivariate models were further identified. All modelling was done by outcome via logistic general estimating equations with repeated measures per participant (multiple annual visits). The observed prevalence of outcome among recipients with the significant predictors was then compared directly with that among recipients without at least one significant predictor, by outcome (ESM Fig. 1). Factors that did not increase the observed prevalence of outcome across all outcomes were eliminated from the final model. A small number of predictors were found to be associated with the highest prevalence of successful study outcomes across all the outcomes investigated. These are termed the 4CFF; that is, common to all the six outcomes.

The main safety measure was the outcome from all immunosuppression-related SAEs reported for each recipient from first infusion, rated as follows: 0, no adverse event reported; 1, recovered; 2, recovered with sequelae/not recovered; 3, disability; and 4, death. The likelihood of this outcome measure was compared between the 4CFF subgroup vs the <4CFF, and across immunosuppression subgroups, using multinomial logistic regression with ORs.

As an observational study, there were no control groups and no pre-determined hypothesis of efficacy. All p values are considered exploratory. We interpret p values of <0.001 as indicative of an important effect, p values of 0.001–0.099 as highly suggestive of an effect, and p>0.01 as not significant. Continuous data are reported as mean±SD.

Results

Baseline

ESM Table 2 shows key baseline characteristics of the study cohort. By the cohort definition, all were ITA recipients with SHEs pre-transplant (99% had partial or total unawareness of hypoglycaemia) and C-peptide <0.1 nmol/l. Mean±SD age was 46±10 years, with 30±11 years of diabetes, 0.53±0.18 U/kg daily insulin usage, 9.4±4.8 mmol/l fasting glucose and 61.5±13.9 mmol/mol HbA1c (7.8±3.4%). Fifty-four per cent were treated with induction regimens that included TCD antibodies and/or TNF-α inhibition and 78% were maintained on regimens of calcineurin and mechanistic target of rapamycin (mTOR) inhibition, although these rates varied substantially across the eras (Table 1). For islet preparations, mean±SD cold ischaemia time was 7.7±4.7 h and culture time was 20.5±16.7 h (which included 19% not cultured). Mean±SD total IEQs infused over one to six infusions was 863,000±395,000.

Table 1 Percentage of recipients with 4CFF and with each individual favourable factor, by era
Full size table

The common favourable factors

The univariate associations of baseline variables with the six main study outcomes are presented in ESM Table 1. Four factors were strongly associated with higher prevalence (≥10%) across all the desired outcomes of islet transplantation: recipient age ≥35 years; a total of ≥325,000 IEQs infused over one to several infusions; induction with TCD antibodies and/or TNF-α inhibition; and maintenance with calcineurin inhibitor (CNI) and mTOR inhibitor for at least some portion of follow-up. These four factors defined the subgroups having four (4CFF, N=126) and those having fewer than all four common favourable factors (<4CFF, N=272).

Main study outcomes

The observed prevalence rates of the study outcomes for the 4CFF vs <4CFF subgroups of ITA recipients with SHEs and negative C-peptide are shown in Fig. 1. The differences in prevalence over 5 years were significant (p≤0.01) for each one of the outcomes, with at least a 10 percentage point superiority in the 4CFF subgroup at 5 years post last transplant. The error margin of these estimates was 3–4%.

Fig. 1
figure 1

Observed prevalence rates of the study outcomes in the 4CFF subgroup (favourable factors, n=126) vs the <4CFF subgroup (others, n=272): HbA1c <53 mmol/mol (7.0%) and absence of SHEs (a); HbA1c <53 mmol/mol (7.0%) (b); absence of SHEs (c); C-peptide ≥0.1 nmol/l (d); fasting glucose 3.3–7.8 mmol/l (e); and insulin independence (f) in ITA recipients with SHEs and pre-transplant negative C-peptide at baseline and years 1–5 post last infusion

Full size image

HbA1c <53 mmol/mol (7.0%) and absence of SHEs on available data (Fig. 1a) started at 83% at 1 year post last infusion, declining to 73% at 5 years for the 4CFF subgroup, compared with 68% at 1 year declining to 47% at 5 years for the <4CFF subgroup (p<0.0001). HbA1c <53 mmol/mol (7.0%) rates (Fig. 1b) were also significantly and importantly higher for the 4CFF subgroup: 86% at 1 year declining to 76% at 5 years, compared with 71% at 1 year and 54% at 5 years for the <4CFF subgroup (p<0.0001). Absence of SHEs (Fig. 1c) showed 93–95% positive outcome persisting throughout the 5 year follow-up for the 4CFF subgroup, compared with 93% declining to 89% in <4CFF recipients (p=0.01). The prevalence of C-peptide ≥0.1 nmol/l (Fig. 1d) in the 4CFF subgroup started at 92% at 1 year and declined to 68% at 5 years post transplant, while in <4CFF recipients it started at 77% declining to 47% at 5 years (p=0.0002). Fasting glucose of 3.3–7.8 mmol/l (Fig. 1e) was maintained at a prevalence of 87% over 5 years following the last infusion for 4CFF recipients, while the rate declined from 73% to 65% over the 5 years for the <4CFF recipients (p<0.0001). In terms of fasting glucose as a continuous measure, the 4CFF subgroup showed improved fasting glucose from 8.6±4.2 mmol/l at baseline to 6.2–6.3±1.8–1.6 over years 1–5, compared with 9.8±5.1 at baseline to 7.0–7.4±3.0–3.8 over years 1–5 in the <4CFF subgroup (p=0.0001; ESM Fig. 2). In addition to significantly lower levels of fasting glucose in the 4CFF group, the range of fasting glucose was remarkably narrower. Insulin independence rates (Fig. 1f) following last infusion were much higher in the 4CFF subgroup than in the <4CFF subgroup: 75% at 1 year declining to 53% at 5 years vs 56% at 1 year declining to 27% at 5 years, respectively (p<0.0001). In both cohorts (4CFF and <4CFF), the proportion of recipients with HbA1c <53 mmol/mol (7.0%) at 5 years post last infusion was higher in insulin-independent vs insulin-dependent islet transplant recipients (ESM Table 12).

In 2012, the CITR reported improved outcomes with potent induction immunotherapy including TCD antibodies and TNF-α inhibition [19]. The present analysis (with a larger dataset) yielded the factor of ‘induction with TCD antibodies and/or TNF-α inhibition’ as one of the 4CFF. We subdivided the 4CFF subgroup by induction regimen (4CFF, and those with the other three favourable factors but neither TCD antibodies nor TNF-α inhibition) to compare against those not given either TCD antibodies or TNF-α inhibitors, as shown in Fig. 2. For the insulin independence outcome, with substantially more observations in the present analysis, we found that the TCD antibody alone, the TNF-α inhibitor alone, and the TCD antibody plus TNF-α inhibitor groups were very similar to each other in their prevalence rates, and all the rates were higher than in the induction group receiving neither TCD nor TNF-α inhibition (p=0.03).

Fig. 2
figure 2

Prevalence of insulin independence 1–5 years post last infusion, comparing individual induction immunosuppression regimens among recipients aged ≥35 years, with ≥325,000 IEQs infused and maintenance on mTOR inhitor + CNI (‘all other common factors favourable group’, n=238) (a) vs recipients aged <35 years, with <325,000 IEQs infused or maintenance on other than mTOR inhibitor + CNI (‘not all other common factors favourable group’, n=160) (b); p<0.01 for each comparison except TCD+TNFa (p=0.07). The precipitous drop over 5 years in the ‘neither TCD/TNFa’ group was significant at p<0.0001. There was no significant difference between TCD+TNFa, TNFa alone, and TCD alone after controlling for other favourable factors (p=0.5); TCD, TCD antibody; TNFa, TNF-α inhibitor

Full size image

Additionally, in 2012 the CITR reported improved outcomes in recipients of ITA over 4 year eras from 1999 through 2010 [14]. The CITR 8th and 9th Annual Reports both note a subsequent decline in successful outcomes in the most recent era [20, 21]. A potential explanation lies in the significantly lower proportion of recipients in the recent vs the 2007–2010 era who received induction with TCD antibody and/or TNF-α inhibition and maintenance with mTOR inhibitor and CNI and who thus met all 4CFF: 25% in 1999–2002 increasing to 50% in 2007–2010, then declined markedly to 23% in 2011–2015 (Table 1).

ESM Fig. 3 shows that, within each era, the main effect of the 4CFF vs <4CFF for each of the six outcomes remains highly significant (p<0.0001 to p<0.02), in the same direction and similar magnitude, while era explains very little of the variation in the outcomes (p=0.1 to p=0.8).

The impact of significantly reduced usage of mTOR inhibitors plus CNIs for immunosuppression maintenance in the most recent era is shown in ESM Table 13, which compares the prevalence rates of outcomes for the 4CFF subgroup vs the subgroup with the other three CFFs but not mTOR inhibitor plus CNI maintenance. For all outcomes (except absence of SHEs, for which high success rates were achieved in both groups), the 4CFF subgroup (which includes maintenance with mTOR inhibitors plus CNI) achieved superior outcomes throughout 1–5 years compared with the CFF subgroup (which excluded this regimen) (ESM Table 13).

Excluding the age requirement from the 4CFF subgroup (Table 2) reduced the rates of the successful outcomes by about five percentage points throughout the 5 years of follow-up, with virtually no impact on the <4CFF subgroup (except for absence of SHEs, which remained similar).

Table 2 Prevalence of the study outcomes for the 4CFF subgroup vs a subgroup with all common favourable factors except age
Full size table

Safety

With no differences between groups, the incidence of product/infusion procedure-related SAEs in the entire cohort was 0.4 events per person, including 0.008 product/infusion procedure-related SAEs per person that resulted in sequelae (data not shown). No permanent disabilities or deaths deemed to be related to an infusion procedure were reported. The 4CFF subgroup experienced higher rates of immunosuppression-related SAEs from first infusion to 5 years post last infusion (period of exposure) when compared with the <4CFF subgroup (mean±SD 1.02±1.39 vs 0.58±1.04 immunosuppression-related SAEs per person, p=0.0005, Fig. 3). However, the prevalence of immunosuppression-related SAEs that did not completely recover and that were associated with sequelae, disability or death (Table 3), was comparable in the 4CFF and <4CFF subgroups (mean±SD 0.056±0.26 per person vs 0.074±0.34, p=0.6). When included in a multivariate model that included grou**s by immunosuppression agents used, adjusted for the favourable factors and for era (which impacted the outcomes as per above), only the use of mTOR inhibitors was associated with both higher rates and worse outcomes of immunosuppression-related SAEs (p=0.01, Table 4; ESM Fig. 4). At 1 year post last infusion, 59% of recipients in the 4CFF group were on maintenance with mTOR inhibitors, drop** to 30% at 5 years. In the <4CFF group, the percentage of recipients on mTOR inhibitors was 55% at 1 year and 40% at 5 years (ESM Fig. 5).

Fig. 3
figure 3

Number of SAEs deemed related or possibly related to immunosuppression per recipient (a) and deemed related or possibly related to infusion procedure (b), for the 4CFF subgroup (favourable factors, n=126) vs the <4CFF subgroup (others, n=272). The mean is indicated by a blue diamond and the median by a blue line. A black box shows the IQR with whiskers for 1.5 times the IQR. Outliers are plotted as black circles

Full size image
Table 3 Outcomes per recipient of immunosuppression-related SAEs and procedure-related SAEs for the 4CFF vs the <4CFF subgroups
Full size table
Table 4 Outcomes of immunosuppression-related SAEs per recipient
Full size table

Alloantibodies (measured as panel reactive alloantibodies to HLA class 1 or 2) were present at baseline in 17% for class 1 and 15% for class 2 of all recipients; those proportions increased at the time of islet graft failure (32% for class 1 and 27% for class 2). After graft failure and discontinuation of immunosuppression, allosensitisation further increased to 58% for class 1 and 50% for class 2 (ESM Table 14). Alloantibody results were similar between 4CFF and <4CFF subgroups, whereas islet graft failure was more common in the <4CFF (19%) than the 4CFF (9%) subgroup (data not shown).

No deaths from any cause were reported in the 4CFF subgroup. Among the <4CFF subgroup, five (2%) all-cause deaths were reported. Two (1%) were deemed to be related or possibly related to immunosuppression. The two deaths deemed related were due to viral meningitis and renal failure, respectively (ESM Table 4). No permanent disability due to an immunosuppression-related SAE occurred in recipients in the 4CFF subgroup, compared with 1% in the <4CFF subgroup. Within the 4CFF subgroup, malignancy related or possibly related to immunosuppression occurred in three recipients (2%): one with squamous cell carcinoma; and two with papillary carcinoma (ESM Table 6). Within the <4CFF subgroup malignancy related or possibly related to immunosuppression was reported in ten recipients (3.7%): two squamous cell carcinoma; three basal cell carcinoma; two breast cancer; and three ovarian cysts (ESM Table 6). Infection deemed related or possibly related to immunosuppression (ESM Table 7) occurred in a slightly higher, not statistically different (ESM Table 15), percentage of recipients in the 4CFF subgroup (9.5% vs 7.4% in the 4CFF vs <4CFF subgroup, respectively).

Discussion

This CITR report on 398 human islet allograft recipients with type 1 diabetes, SHEs and C-peptide <0.1 nmol/l pre-transplant presents the most comprehensive analysis of islet transplant outcomes to date and provides new insights not available through single-centre or even consortia reports [2,3,4,5,6,7,8]. First, perhaps unexpectedly, only four factors were strong predictors of all six studied favourable and clinically relevant study outcomes in ITA recipients: recipient age ≥35 years, total IEQs ≥325,000 infused over one to several infusions, induction with TCD antibodies and/or TNF-α inhibition, and maintenance with CNIs and mTOR inhibitors for at least some duration of follow-up. Second, at 5 years after the last islet infusion, of the recipients with type 1 diabetes complicated by SHEs meeting these 4CFFs, 95% experienced persistent resolution of SHEs. Third, 73% had HbA1c <53 mmol/mol (7.0%) and absence of SHEs, 68% showed C-peptide ≥0.1 nmol/l and 87% had fasting glucose 3.3–7.8 mmol/l. These observed prevalence rates were significantly higher than in the remaining recipients in the <4CFF subgroup. Fourth, insulin independence rates following last infusion were much higher in the 4CFF subgroup than in the <4CFF subgroup: 75% at 1 year declining to 53% at 5 years vs 56% at 1 year declining to 27% at 5 years (p<0.0001). Fifth, the incidence of product/infusion procedure-related SAEs has decreased over time and is extremely low with current refined and controlled techniques [13]. Finally, the incidence of immunosuppression-related SAEs per person was higher in 4CFF (1.02) than in <4CFF recipients (0.58; p=0.0005), and associated with higher utilisation of mTOR inhibitors in 4CFF recipients. However, the incidence of immunosuppression-related SAEs that resulted in sequelae, permanent disability or death was similarly low in both the 4CFF (0.06 per person) and <4CFF (0.07 per person) groups. There was no death or permanent disability in the 4CFF group followed for 5 years after islet transplantation.

The variation over CITR’s lifetime in utilising all the (now-known) factors and practices that optimise outcomes of ITA did not in any way mitigate the consistent beneficial clinical effect of the 4CFF. These 4CFF are few, and except for recipient age, controllable and achievable with current islet manufacturing and medical management practices. If the other three factors are observed, the outcome levels are generally only about 5% lower than in the subgroup restricting age to ≥35 years, except for absence of SHEs which is similarly high and persistent at the 90–95% level through 5 years. Age alone should therefore not be considered a disqualifying criterion of islet transplantation for individuals with type 1 diabetes who experience SHEs.

The cumulative data in this cohort indicate that either T cell depletion or TNF-α inhibition should be considered as part of adequate induction immunosuppression. Combining them may not increase their effectiveness, although differences may exist between the timing and use of individual TCD antibodies and the need for TNF-α inhibition to mitigate potential detrimental effects of released cytokines. Factors that are definitely not predictive are not as reliably detectable as those showing large, consistent benefits in this type of exploratory analysis.

The impressive protection from SHEs at 5 years in 95% of 4CFF recipients and in over 80% of all recipients with SHEs is likely related to the demonstrated efficacy of islet transplants in abolishing time spent in hypoglycaemia, re-establishing partial glucagon and adrenaline (epinephrine) counter regulation, restoring autonomic symptom perception and normalising endogenous glucose production in response to insulin-induced hypoglycaemia in individuals with long-standing type 1 diabetes and IAH [22]. Minimal islet graft function is sufficient to abrogate SHEs [23]. Protection from SHEs is maintained for 5 years after the final islet infusion even after complete loss of islet function, which occurred in 16% of 4CFF and 32% of <4CFF recipients within 5 years post last infusion.

Alloantibodies significantly increased after islet graft failure and discontinuation of immunosuppressive drugs, as previously shown [24,25,26,27,28,29]. These alloantibodies after islet graft failure can persist for many years [28]. Persistence of allosensitisation in this recipient population is of clinical importance as it may result in longer transplant waiting-list times for identification of a suitable donor in case of requiring a subsequent transplant [28].

These promising long-term outcome results of islet transplantation are relevant to individuals with type 1 diabetes in whom SHEs are refractory to medical intervention involving structured education, novel insulin analogues, real-time continuous glucose monitoring and sensor-augmented insulin pumps with low glucose suspend technology [30,31,32,33,34,35]. While these interventions reduce the number of hypoglycaemic events and should always be evaluated [36], approximately 50% of individuals with type 1 diabetes and recurrent SHEs so treated by a specialist hypoglycaemia service fail to experience resolution of SHEs [37]. In these non-responders, the morbidity of persistent SHEs is substantial [36, 37]. When SHEs persist despite educational programmes and current diabetes technologies, randomisation in clinical trials of affected individuals to those failed interventions would present a concern. This conundrum explains the lack of a medical treatment comparator arm in all clinical trials except one [8] of islet transplantation in patients with type 1 diabetes and persistent SHEs.

As an observational study, this analysis is limited by the declining completeness of data with increasing length of follow-up, likely due to increasing incidence of graft loss and fewer contacts with the transplant clinic as follow-up time increases. At 5 years post transplant, data availability varied from 56% to 81%. Protection from SHEs was a main outcome measure; data completeness for SHEs was 78% at 5 years after the last islet infusion in 4CFF recipients. The incompleteness of outcomes data could have affected the main findings and conclusions of this study. Additional limitations include the potential inaccuracy of retrospectively collected SHE data, absent documentation of euglycaemia as part of the definition of insulin independence, and the impossibility of randomised comparisons such as immunosuppression regimens. The predictors of outcome identified in ITA recipients may not be generalisable to other recipient categories. Over the eras encompassed by the CITR, the choice of immunosuppression protocol was largely guided by incremental advances in small single-cohort islet transplantation clinical trials, with the accumulating evidence evaluated on a relatively qualitative basis.

Taken together, this analysis by the CITR of 5 year outcomes in a large cohort of pre-transplant C-peptide-negative islet transplant recipients with type 1 diabetes and SHEs provides important new insights, guides transplant centres in develo** and refining islet/beta cell transplant protocols, assists diabetes care teams in considering islet transplantation as an intervention, and thereby facilitates the informed utilisation of a scarce and costly transplant treatment modality for the benefit of individuals with type 1 diabetes and recurrent SHEs.