β-Cell Secretory Capacity Predicts Metabolic Outcomes Over 6 Years After Human Islet Transplantation Free

Allogeneic islet transplantation improves glycemic control, restores impaired awareness of hypoglycemia, and prevents severe hypoglycemia in individuals with long-standing C-peptide–negative type 1 diabetes (T1D) (1). After transplantation by intraportal infusion, pancreatic islets embed within the hepatic sinusoids and undergo revascularization by the hepatic arterial system over several weeks in a process known as engraftment (1). Clinical outcomes are highly dependent on the engrafted functional β-cell mass (2,3), with early β-cell function predicting long-term insulin independence and graft survival (4–6). Islet graft β-cell function, assessed by either fasting composite measures, such as the BETA-2 score, mixed-meal tolerance test (MMTT)–stimulated C-peptide, or β-cell secretory capacity measured as the acute insulin response potentiated (AIR_pot) by a hyperglycemic clamp, are predictive of posttransplantation glycemic control (7–13). However, attrition in graft β-cell function is observed over time (4–6,14). How declining islet β-cell function may be related to changes in engrafted β-cell mass, measured by the gold-standard β-cell secretory capacity in response to glucose-potentiated arginine (GPA) testing, has not been assessed beyond 1 year after allogeneic islet transplantation (1,2,8,15).

Transplantation under the Clinical Islet Transplantation 07 (c7) protocol (16), which includes peritransplantation T-cell–depleting antibody, tumor necrosis factor-α inhibition, heparinization, and intensive insulin therapy, improved engrafted β-cell mass in comparison with earlier protocols despite transplantation of fewer islets isolated more often from a single-donor pancreas (1–3). We sought to evaluate changes in β-cell secretory capacity in individuals with T1D undergoing transplantation under the c7 protocol over long-term follow-up and examine associations with insulin independence, physiologic parameters of islet cell hormone metabolism, and routine clinical measures of β-cell function (BETA-2 score and MMTT C-peptide) and glycemic control by continuous glucose monitoring (CGM).

Eleven individuals underwent allogeneic islet transplantation under the c7 protocol (16) at the University of Pennsylvania between 2008 and 2012 and were included in this analysis. Participants were recruited as per c7 trial criteria with study procedures, islet infusion characteristics, and early outcomes for the islet-alone phase 3 trial (16) (ClinicalTrials.gov identifier NCT00434811) and University of Pennsylvania subcohort previously described (2). All individuals had C-peptide–negative T1D (≥5 years duration) complicated by impaired awareness of hypoglycemia and marked glycemic lability with at least one severe hypoglycemia event in the year before enrollment. Individuals received an infusion of purified human pancreatic islets isolated from a single ABO-compatible deceased donor pancreas (≥5,000 islet equivalents [IEQ]/kg recipient body weight) by percutaneous transhepatic portal vein catheterization with the goal of insulin independence (2,16). Immunosuppression was induced with T-cell–depleting rabbit ATG and tumor necrosis factor-α inhibition with etanercept. Intensive insulin therapy was continued as required for at least 2 months after the first and 1 month after the second infusion and tapered as tolerated for the day-75 assessment (2). Individuals who remained insulin dependent at day 75 were eligible to receive another islet infusion of ≥4,000 IEQ/kg within 240 days of initial transplantation under the interleukin-2 receptor antagonist basiliximab plus etanercept. Insulin independence was defined as >7 days without insulin in the context of good glycemic control (HbA_1c <7% [53 mmol/mol]; fasting glucose <126 mg/dL [7.0 mmol/L]; MMTT 90-min glucose <180 mg/dL [10 mmol/L]) and MMTT-stimulated C-peptide ≥0.5 ng/mL (16). Maintenance immunosuppression included low-dose tacrolimus (target trough 3–6 μg/L), a calcineurin inhibitor, and sirolimus (target trough 10–15 μg/L for 3 months, then 8–12 μg/L), a mammalian target of rapamycin inhibitor, with sirolimus converted to mycophenolic acid if not tolerated (2, 16).

Paired MMTT and GPA procedures were performed on successive days at day 75 posttransplantation and annually thereafter. After the 2-year c7 protocol, 10 individuals were enrolled in the c8 follow-up protocol with annual MMTT/GPA testing up to 8 years (4) (ClinicalTrials.gov identifier NCT01369082). In addition, seven individuals participated in a separate University of Pennsylvania study at a median of 7 (interquartile range 5–7) years posttransplantation (17) with clinical data included if not captured under the c8 study. An age-, sex-, and BMI-matched control group of 11 individuals without diabetes also underwent GPA procedures (2). A separate control group (n = 11) of individuals without diabetes unmatched for age but with nearly identical β-cell secretory capacity was used for glucagon comparisons because of a change in assay from the original control group. Blinded CGM data were collected at day 75 and annually as part of the c7 study (72 h; MiniMed; Medtronic, Northridge, CA) (18) and later University of Pennsylvania follow-up (7 days; iPro2 Professional CGM; Medtronic) (17).

Approval of all study protocols was obtained by the institutional review board of the University of Pennsylvania. All participants provided written informed consent for participation and follow-up.

At 0800 after an overnight fast, an antecubital or forearm vein catheter was placed for blood sampling, with fasting samples for C-peptide and glucose obtained before consumption of a standardized liquid meal (6 mL/kg up to 360 mL; Boost High Protein) over 5 min (4,16). Additional blood samples for C-peptide and glucose were obtained at 60 and 90 min postingestion. Samples were centrifuged at 4°C, separated, frozen at −80°C, and shipped together with samples for HbA_1c and creatinine to the University of Washington (Seattle, WA) for central laboratory analysis (4,16).

GPA procedures were performed as previously described (2). At 0700 after an overnight fast, i.v. catheters were placed, one in an antecubital vein for infusion and the other in the contralateral hand or forearm for blood sampling, with heating pads placed to promote arterialization of venous blood. Baseline blood samples were taken at t = −5 and −1 min before injection of 10% arginine (5 g) over a 1-min period starting a t = 0 min. Additional blood samples were taken at t = 2, 3, 4, and 5 min. At t = 10 min, a hyperglycemic clamp was initiated with a variable rate of 20% glucose infused to achieve target plasma glucose of 230 mg/dL (19). Plasma glucose was measured every 5 min by the glucose oxidase method using a bedside automated glucose analyzer (YSI 2300; Yellow Springs Instruments, Yellow Springs, OH) to direct the glucose infusion rate. After 45 min of glucose infusion (at t = 55 min), a second arginine pulse was injected with identical sampling (2).

GPA samples were collected on ice into tubes containing EDTA and protease inhibitor cocktail. Samples were centrifuged at 4°C, separated, and frozen at −80°C for subsequent analysis (2). Plasma glucose was confirmed in duplicate by the glucose oxidase method using an automated glucose analyzer (YSI 2300). Plasma insulin, C-peptide, proinsulin, and glucagon were measured in duplicate by double-antibody radioimmunoassays (Millipore, Billerica, MA). Active amylin (islet amyloid polypeptide) was measured by ELISA (Millipore).

BETA-2 scores were calculated from fasting glucose, C-peptide, HbA_1c, and insulin dose using an online calculator (20). Acute insulin (AIR_arg), C-peptide, proinsulin, and glucagon responses to arginine under fasting and glucose-potentiated (glucose-inhibited for glucagon) conditions were calculated as means of +2-, 3-, 4-, and 5-min values after injection of arginine minus the means of baseline values (19,21,22). The response to arginine during the 230-mg/dL clamp enabled the determination of glucose potentiation of arginine-induced insulin secretion (AIR_pot) or β-cell secretory capacity, which has been shown to correlate highly with the maximal β-cell secretory response at 340 mg/dL in islet transplant recipients (23). Functional β-cell mass was defined as the percent β-cell secretory capacity relative to the mean of responses of controls without diabetes representing normal (8). Insulin sensitivity was assessed by M/I, determined by dividing the mean glucose infusion rate during the 230-mg/dL glucose clamp (M) by the mean prestimulus insulin level (I) between 40 and 45 min of glucose infusion. Disposition index, a measure of insulin secretion accounting for insulin sensitivity, was calculated by AIR_arg × M/I (22). Proinsulin secretory ratios (PISRs) were calculated as the molar concentration of the acute proinsulin response divided by the acute insulin response × 100 (22). Insulin clearance was determined by the molar ratio of acute C-peptide to acute insulin responses (24). Insulin-to-amylin ratios were calculated as the molar concentration of insulin divided by active amylin (25).

Data are reported as mean ± SD or median (interquartile range) unless otherwise noted. Analyses were performed using Statistica (version 13.5.0; StatSoft, Inc., Tulsa, OK) and R (version 4.0.2 2020; R Core Team). Changes in longitudinal outcomes over follow-up were assessed by trend test, with quasiexperimental comparison with the control group where applicable. For one participant with follow-up data available for both 6 and 7 years, 6-year data were included. Relationships between GPA and fasting/MMTT measures of β-cell function were assessed by partial correlation coefficient (PCC) analysis accounting for repeated measures (26). Receiver operating characteristic (ROC) curves were used to assess the ability of functional β-cell mass to predict insulin independence and determine MMTT and CGM threshold predictors of β-cell mass. Optimal ROC curve cut points of maximal combined sensitivity and specificity were determined by the Youden index. Graphs were prepared using OriginPro (version 2024; OriginLab Corporation, Northampton, MA). Results were considered significant at P < 0.05 (two tailed).

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

Eleven individuals (four men, seven women) age 45 ± 10 years with T1D duration 28 ± 13 years received 697,762 ± 161,859 IEQ (10,090 ± 2,206 IEQ/kg) over one (n = 7) or two (n = 4) infusions, with a time to second transplantation of 5 ± 2 months (Supplementary Table 1). Eleven, 10, nine, and six individuals were observed to 2, 3, 5, and 6–7 years, respectively. Participant follow-up for clinical data was 6 (5–7) and for metabolic outcomes 5 (3–6) years. Sixty-three paired MMTT/GPA procedures were completed, representing approximately seven repeated measures per participant, with paired CGM data at 33 time points.

All individuals became insulin independent by day 75 after last infusion. Insulin independence was maintained in all participants at 1 year, with 10 of 11, eight of 10, seven of nine, and three of five maintaining insulin independence at 2, 3, 5, and 6–7 years, respectively (Supplementary Fig. 1). One insulin-independent individual withdrew after completion of the c7 study. Of the five individuals who returned to insulin therapy, four required low-dose basal insulin, and one experienced graft failure because of recurrent autoimmune diabetes (27). HbA_1c remained ≤6.5% (48 mmol/mol) (12,28) in all participants over 6–7 years posttransplantation (Fig. 1A). For those who returned to insulin, the dose remained <50% of pretransplantation requirements (Fig. 1B), meeting Igls criteria for good islet graft function (29). Nevertheless, HbA_1c and insulin requirements increased over time (β 0.06%; P = 0.001 and β 0.007 units/kg per day; P < 0.01, respectively, per year). Participant weight and BMI remained stable (Fig. 1C). Islet graft β-cell function assessed by BETA-2 score decreased with time (P < 0.001), with the mean below the threshold associated with insulin independence at 6–7 years (Fig. 1D) (10). Fasting C-peptide and ratio of C-peptide to glucose were similarly affected (Fig. 1E and F) (13), with no change in fasting glucose. MMTT-stimulated C-peptide and ratio of C-peptide to glucose remained stable over 3 years before declining with time (P ≤ 0.05) (Fig. 1G and H). Post hoc analysis with repeated measures ANOVA modeling the within-participant correlation by a compound symmetry variance-covariance structure assessed sequential time points using Tukey adjusted pairwise comparison and did not demonstrate a statistically significant inflection point for any measure of β-cell function.

No severe hypoglycemia events occurred over follow-up. Renal function declined modestly, with a gradual increase in creatinine (β 0.02 mg/dL per year; P < 0.01) and decrease in estimated glomerular filtration rate (β −2.17 mL/min/1.73 m² per year; P < 0.001), with estimated glomerular filtration rate generally maintained >60 mL/min/1.73 m² (28) (Supplementary Fig. 2). No change was observed in urinary albumin excretion; however, one individual developed macroalbuminuria, which resolved with a change from sirolimus to mycophenolic acid (30). Two participants converted from sirolimus to mycophenolic acid for other adverse effects and one for pregnancy (31). One individual developed breast cancer at 4.5 years, requiring temporary insulin during adjuvant chemotherapy. Immunosuppression levels were maintained per protocol, which included a reduction in sirolimus trough concentrations (Supplementary Fig. 2).

AIR_arg under fasting conditions was comparable to that of controls without diabetes (Fig. 2A) and stable to 3 years before decreasing (P < 0.001) as insulin sensitivity increased (P < 0.001) (Fig. 2B), such that there was no change in the disposition index (P = 0.42) (Fig. 2C). Accordingly, AIR_arg and insulin sensitivity were related by a hyperbolic function (r² = 0.48; P < 0.001) (Fig. 2D), as has been previously described in humans with native pancreatic islet function (32). Similar relationships existed between β-cell secretory capacity and insulin sensitivity. AIR_pot (Fig. 2E) was lower than normal but remained stable to 4 years before declining (β −3.89 μU/mL per year; P < 0.05). β-Cell secretory capacity assessed by AIR_pot correlated highly with ACR_pot over time (PCC 0.56; P < 0.001) (Table 1). The acute proinsulin response to GPA was also correlated with β-cell secretory capacity (PCC 0.29; P < 0.05) (Table 1). Nevertheless, when assessed as the proinsulin secretory ratio, both PISR_arg and PISR_pot increased over time (P < 0.01) (Fig. 2F), with a negative correlation of PISR_pot with β-cell secretory capacity by trend (PCC −0.26; P = 0.06) (Table 1). Fasting glucagon increased posttransplantation (P < 0.001) (Fig. 2G), with the increase after 4 years mirroring and negatively correlated with β-cell secretory capacity decline (PCC −0.32; P = 0.01). Similar to the MMTT visits above, fasting glucose assessed before GPA testing remained unchanged over follow-up (Supplementary Fig. 3). There were no changes observed in acute glucagon responses to arginine or glucose-inhibited arginine (Supplementary Fig. 3). Insulin clearance under glucose-potentiated conditions tended to increase as β-cell secretory capacity declined (PCC −0.29; P < 0.05) (Fig. 2H and Table 1).

All measures of β-cell function were positively correlated with β-cell secretory capacity (Table 2). Stimulated MMTT measures were strongly related to AIR_pot (PCC ≥0.50; P < 0.001), with overlapping CIs. The BETA-2 score performed comparably well (PCC 0.49; P < 0.001). Relationships with C-peptide were improved by assessment of the ratio of C-peptide to glucose.

One recipient achieved a functional β-cell mass comparable to the mean among controls without diabetes (functional β-cell mass of 100%) (Fig. 3A) after a single infusion of 12,073 IEQ/kg and maintained insulin independence over 6 years of follow-up; however, the detection of a de novo donor-specific antibody at 4 years was associated with a subsequent 45% reduction in functional β-cell mass (Supplementary Fig. 1). Insulin-independent visits were associated with a mean functional β-cell mass of >40% (Fig. 3B). ROC curve analysis supported functional β-cell mass as a strong predictor of insulin independence (area under the curve [AUC] 0.84; 95% CI 0.69–0.99; P = 0.001), with a functional β-cell mass of 40% having 100% specificity and 52% sensitivity. Across the cohort, marked declines or persisting low β-cell mass was associated with return to exogenous insulin (Fig. 3A and C). Low functional β-cell mass was associated with occasional need for temporary insulin use or insulin dependence. Four individuals with functional β-cell mass 16–40% required periods of temporary insulin during episodes of β-cell stress, such as illness, medication use (glucocorticoid, antidepressant, or progesterone-only contraceptive), or pregnancy (31) (Supplementary Fig. 1).

Return to insulin after decline in functional β-cell mass was associated with immune mechanisms in four participants (Supplementary Fig. 1). Markers of alloimmune recognition were evidenced through development of de novo donor-specific antibody in two individuals, with a concurrent increase in islet cell autoantibody (ICA) titer in one participant and preceded by GAD-65 autoantibody seroconversion to low-level titers in the other participant. One participant who had seroconverted for ICA subsequently experienced an increase in ICA and GAD-65 autoantibody titers and later had a decline in β-cell mass. Another recipient experienced autoimmune graft loss after the development of de novo autoantibodies against GAD-65 to marked titers before recurrence of hyperglycemia and loss of donor-derived islet β-cell–specific exosomes (27).

Nonimmunologic graft exhaustion was suggested by increases in PISR_pot with declining β-cell secretory capacity in two participants (33,34). In addition, disproportionate increases in active amylin relative to insulin, as indicated by reduction in insulin-to-amylin ratio over follow-up, was observed as individuals experienced a decline in β-cell secretory capacity (Supplementary Fig. 4) (25).

Measures of β-cell function were strong predictors of functional β-cell mass of 40% (Table 3). Ratios of MMTT C-peptide to glucose demonstrated improved discriminative ability to predict functional β-cell mass beyond C-peptide alone. Sixty-min ratio of C-peptide to glucose demonstrated an ROC AUC of 0.89 (95% CI 0.81–0.97; P < 0.001), with a threshold of 0.13 nmol/mmol predicting a functional β-cell mass >40% with 83% sensitivity and 88% specificity. As a fasting measure, BETA-2 score demonstrated excellent ability to predict functional β-cell mass of 40% (AUC 0.85; 95% CI 0.76–0.94; P < 0.001), with a threshold of 19.

Paired GPA/CGM data demonstrated significant negative hyperbolic relationships between functional β-cell mass and percentage of time in hypoglycemia <60 (r² = 0.11; P = 0.01) (Fig. 4A) and <70 mg/dL (r² = 0.05, P = 0.02) and in hyperglycemia >180 mg/dL (r² = 0.28; P < 0.001) (Fig. 4B), whereas functional β-cell mass was related to time in range 70–180 mg/dL by exponential function (r² = 0.37; P < 0.001) (Fig. 4C). Functional β-cell mass was a strong predictor of most optimal glycemic targets (ROC AUC ≥0.84; P ≤ 0.05) except percentage of time spent <54 (≤1% time in all but one individual) or <70 mg/dL (AUC <0.60) (Table 4). Excellent glycemic outcomes, including ≤1% time below 60 mg/dL, ≤2% time above 180 mg/dL, and ≥90% time in range (70–180 mg/dL) were predicted by a functional β-cell mass of >20%. However, targets of glucose variability in nondiabetic ranges (SD and coefficient of variation (CV) <20 mg/dL and <20%, respectively) and time in range ≥95% were best predicted by functional β-cell mass >30% (Table 4 and Supplementary Fig. 5).

Islet transplantation in T1D leads to sustained metabolic benefits over long-term follow-up. In this cohort, transplantation under the c7 protocol resulted in insulin independence for all recipients, with most remaining insulin independent after 5 years. Acute insulin responses to arginine under fasting conditions were normal; however, the reserve capacity for insulin secretion was less than that in individuals without diabetes for all except one recipient. Islet graft β-cell secretory capacity remained stable for >3 years before a gradual decline, which in part may be explained by a functional adaptation of transplanted β-cells to enhanced insulin sensitivity to maintain glucose tolerance, such that in the absence of immunologic recognition, islet grafts had sufficient reserve capacity to resist metabolic exhaustion despite chronic immunosuppressive drug therapy.

Average functional β-cell mass of this cohort was 40–50% normal after engraftment, surpassing the 25% engrafted functional β-cell mass achieved with the preceding Edmonton protocol, where the marginal reserve capacity had already declined over the first year posttransplantation (2). Functional β-cell mass >40% normal was predictive for maintaining insulin independence in our cohort over follow-up. However, good glycemic outcomes were attained with a functional β-cell mass >20%, although this was associated with a requirement for exogenous insulin under periods of islet stress or insulin dependence. Functional β-cell mass of ∼25% normal has been reported in individuals with presymptomatic (35) and recent-onset T1D (36), suggesting this represents a threshold of marginal islet β-cell mass to maintain glycemic control before deterioration to clinical diabetes. Keymeulen et al. (37) also reported a functional β-cell mass of ∼25% normal in a T1D cohort with intrahepatic islet grafts and insulin independence at 1 year. Although glycemic benefit was reported through improved glucose variability and no time spent <70 mg/dL (37), this group subsequently reported higher functional β-cell mass targets of >37% as necessary for reducing the CV for fasting glucose <25% at 6 months posttransplantation (8). Therefore, our findings add additional support that higher functional β-cell mass targets (>40%) are necessary for achieving and maintaining insulin independence and sustained islet graft function that can resist metabolic exhaustion over time.

Long-term outcome data have reported strong associations with insulin independence and graft survival (5). However, gradual graft attrition and loss of insulin independence have been observed, and supplemental islet infusions have been administered to sustain graft function through recovery of a reserve capacity for insulin secretion. One center reported three or more islet infusions in 42% of individuals with sustained graft function >10 years after first transplantation (5). Repeated infusions were not permitted in CIT07, and our data showed gradual attrition in functional β-cell mass associated with return to exogenous insulin in two individuals with no evidence of alloimmune sensitization or autoimmune recognition (Supplementary Fig. 1). Evidence for metabolic exhaustion from a depleted secretory reserve is supported by a disproportionate release of proinsulin, evidencing recruitment of immature β-cell secretory granules as AIR_pot declined. These results are consistent with those reported from autologous islet transplant recipients in whom incomplete β-cell prohormone processing at 3 months posttransplantation was associated with insulin dependence at 1 year (38), indicating islet grafts with a marginal engrafted functional β-cell mass. Moreover, the disproportionate release of active amylin relative to insulin observed in individuals as secretory reserve declined to a marginal mass provides further evidence of β-cell stress (25,39) and may itself lead to further β-cell mass reduction because amylin can deposit as fibrils within the islet with β-cell–toxic effects (1,40). In contrast, markers of autoimmune recurrence and alloimmune rejection were observed in four participants beyond 12 months posttransplantation and were associated with a subsequent decline in β-cell secretory capacity (Supplementary Fig. 1). Although T-cell responses were not assessed in the current study, graft dysfunction driven by alterations in the phenotype of islet-specific T-cell response on immune reconstitution after T cell–depleting induction, as previously reported for alemtuzumab (41), cannot be excluded.

Similar to native pancreas, insulin secretion declined with increasing insulin sensitivity, such that the disposition index remained stable and suggested appropriate compensation by intrahepatic islets for insulin secretion in response to varying demand (32). Changes in insulin sensitivity were not associated with changes in sirolimus trough concentrations in this study (data not shown). Reduced insulin clearance after intrahepatic islet transplantation has previously been reported (2), although not uniformly, perhaps because of variations in engrafted islet mass and insulin response (1). In our cohort, insulin clearance after arginine injection under glucose potentiation at day 75 was significantly lower than that in controls and negatively correlated with β-cell secretory capacity, suggesting reduced hepatic extraction of insulin through saturation of receptors by the augmented response from intrahepatic islet grafts (1). Both the acute and glucose-inhibited arginine-induced glucagon responses were comparable to those in controls over follow-up, suggesting normalization of α-cell suppression by hyperglycemia (1). However, fasting glucagon increased over time and was inversely correlated with β-cell secretory capacity. In the absence of changes in fasting glucose over follow-up, the increasing fasting glucagon might be explained by either a reduction of intraislet (paracrine) insulin suppression of transplanted α-cell glucagon production (38) or a reduction in the endocrine regulation of native α-cells by the islet graft.

Clinical measures of β-cell function strongly correlated with β-cell secretory capacity. Ratios of MMTT-stimulated C-peptide to glucose strongly predicted a functional β-cell mass of 40% normal. Moreover, the fasting BETA-2 score demonstrated excellent discriminatory ability, with a cutoff >19 predicting functional β-cell mass of 40% normal with 80% sensitivity and 76% specificity. Our thresholds support aiming for higher targets than those previously proposed (5,13). Marfil-Garza et al. (5) analyzed data from the Edmonton cohort, reporting a BETA-2 score >15 in the first year posttransplantation as predictive of sustained graft survival. However, in this group with sustained long-term islet graft function (median 7.4 years), only 39% maintained insulin independence, with a median insulin independence duration of 3 years (5). Our data indicate that a higher BETA-2 score is necessary to maintain longer-term insulin independence. Additionally, a recent analysis by the Collaborative Islet Transplant Registry including almost 700 islet recipients reported a cutoff of 0.04 nmol/mmol for the ratio of fasting C-peptide to glucose predicted insulin independence (13), with our study reporting 0.07 nmol/mmol. Remarkably, however, the cutoff predictive of insulin independence for the ratio of MMTT-stimulated C-peptide to glucose in the Collaborative Islet Transplant Registry analysis was 0.13 nmol/mmol (13), the same as that reported here for the 60-min ratio of C-peptide to glucose, which in our analysis showed the best discriminatory ability to predict target functional β-cell mass (83% sensitivity, 88% specificity). Therefore, our data further validate this registry analysis against gold-standard assessment of β-cell secretory capacity.

Consistent with published data, glycemic benefits to avoid hypoglycemia were attained at lower functional mass, with higher β-cell mass further reducing glucose variability to facilitate insulin independence (7,9,12,13). Indeed, in our cohort, significant hypoglycemia exposure was avoided in nearly all participants. Nonetheless, percentage of time spent <60 mg/dL, previously shown to be a more effective predictor of the autonomic symptom response to hypoglycemia (42), was strongly predicted by functional β-cell mass, with good glycemic outcomes of ≤1% time spent <60 mg/dL, ≥90% time in range, and ≤2% time above range predicted by a functional β-cell mass >20%. Reducing glucose variability to SD <20 mg/dL and CV <20% and further increasing time in range ≥95% require a functional β-cell mass >30%.

Limitations of this single-center study include the small cohort with loss to follow-up or withdrawal (including one with graft failure), which may have introduced bias through inclusion of those with maintained islet graft function, and the small numbers do not permit complex analyses. Nevertheless, we present unique paired metabolic testing data over a long duration of follow-up using gold-standard procedures to examine changes in functional β-cell mass after transplantation not previously reported in the alloislet setting. PCC analyses enabled the assessment of relationships with repeated measures over time, and we present association with measures rather than imply causation for exploration of mechanisms that require further assessment in a larger cohort. As clinical β-cell replacement expands to the delivery of stem cell–derived islets, a clearer understanding of the targets for attaining sufficient β-cell mass is required. This study validates a target engrafted functional β-cell mass >40% for deceased donor and also likely stem cell–derived islet approaches and informs prediction of the attained β-cell mass using readily available clinical measures. Moreover, the intrahepatic site, which relies on portal vein catheterization for islet infusion, is affected by an estimated ∼25% islet loss before engraftment (1); therefore, isolation and transplantation of a greater islet β-cell mass will be required to reach an engrafted target of >40%. This may be achieved more readily with stem cell–derived islet approaches where donor availability/isolation variables are not a factor. Further validation of β-cell mass targets will be required for nonhepatic transplantation sites because of varying effects on insulin clearance with systemic rather than portal insulin delivery.

In summary, β-cell secretory capacity is highly predictive of metabolic outcomes after islet transplantation. Intrahepatic β-cell replacement approaches should target a functional β-cell mass >40% normal to provide sufficient islet reserve for sustained insulin independence. Nonetheless, marked glycemic benefits are still attained with functional β-cell mass >20% normal. Ratio of MMTT C-peptide to glucose and BETA-2 score can inform changes in β-cell secretory capacity in the clinical setting.

Acknowledgments. The authors thank the islet recipients for their participation, the nursing staff of the Penn Clinical & Translational Research Center for their patient care and technical assistance, Dr. Santica Marcovina of the University of Washington Northwest Lipid Metabolism and Diabetes Research Laboratories for performance of the central biochemical assays, Dr. Heather Collins of the Penn Diabetes Research Center for performance of the radioimmunoassays and ELISAs, and Huong-Lan Nguyen of the Penn Institute for Diabetes, Obesity & Metabolism for laboratory assistance.

Funding. This work was performed as a project of the Clinical Islet Transplantation Consortium, a collaborative clinical research program headquartered at the National Institute of Diabetes and Digestive and Kidney Diseases and the National Institute of Allergy and Infectious Diseases, and was supported by Public Health Services Research Grants U01 DK070430 (A.N.), R01 DK091331 (M.R.R.), UL1 TR000003 (Penn Clinical & Translational Research Center), UL1 TR001878 (University of Pennsylvania Center for Human Phenomic Science), and P30 DK19525 (Penn Diabetes Research Center Radioimmunoassay & Biomarkers Core); by the W.W. Smith Charitable Trust; by the Schiffrin Award in Autoimmune Research (M.R.R.); and by the Charles B. Humpton Jr. Endowed Fellowship in Diabetes Research (A.J.F.).

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

Author Contributions. A.J.F. was responsible for database assembly, data analysis, and preparation of the first draft of the manuscript. A.M.M. supported data analysis and graphical presentation of data. R.J.G. supported statistical analysis. E.M., C.D.-B., and A.J.P. participated in the conduct of the study and data collection and researched data. C.L. contributed to the study design and conduct and researched data. A.N. contributed to the study design and conduct, researched data, and revised the manuscript critically for important intellectual content. M.R.R. designed and conducted the study, researched data, and led manuscript revisions. All authors reviewed and edited the manuscript. M.R.R. 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. Portions of this work were presented as an abstract at the American Diabetes Association 84th Scientific Sessions, Orlando, FL, 22 June 2024.