Encapsulated stem cell-derived β cells exert glucose control in patients with type 1 diabetes

Type 1 diabetes involves autoimmune destruction of pancreatic β cells, eliminating endogenous, cell-regulated insulin secretion. While exogenous insulin mitigates hyperglycemia, it cannot fully replicate physiologic control and carries hypoglycemia risk. Allogeneic human islet transplantation can normalize glucose control in many recipients, but wider use is limited by donor scarcity and the need for lifelong immunosuppression. Human pluripotent stem cell-derived pancreatic endoderm or β cell preparations combined with macroencapsulation devices offer a renewable cell source and a retrievable implant platform. Prior first-in-human trials of an open, perforated subcutaneous device (PEC-Direct) containing hESC-derived pancreatic endoderm (PEC-01) showed formation of insulin-positive cells and meal-stimulated C-peptide but with levels below the ~0.1 nmol l−1 threshold generally considered metabolically meaningful, thus without clear glycemic benefit. This study tests whether increasing the implanted cell dose and using a membrane perforation configuration associated with improved cell survival can achieve clinically relevant β cell function and improved glucose control.

The authors review evidence that islet transplantation can restore glycemic control and reduce hypoglycemia in T1D but is constrained by donor availability and immunosuppression risks. They describe advances in directing hPSCs to pancreatic endoderm and β cell lineages and successful correction of diabetes in immunodeficient rodents with such cells, including in macroencapsulation devices. Open, perforated devices allow vascular ingrowth at the cost of requiring immunosuppression. Previous PEC-Direct clinical studies detected insulin expression and glucose-responsive C-peptide release from implants but with C-peptide levels below 0.1 nmol l−1 and no measurable improvement in glycemic outcomes, motivating device configuration and dosing changes in subsequent cohorts.

Design: First-in-human, phase 1/2, open-label, multicenter study (NCT03163511) evaluating safety, tolerability, and efficacy of VC-02 (PEC-01 cells in a perforated macroencapsulation device, PEC-Direct) in adults with T1D and hypoglycemia unawareness. This interim report covers one cadre of 10 patients enrolled between August 2020 and October 2021 at five North American and European centers. Participants: Inclusion included adults with ≥5 years T1D, hypoglycemia unawareness (Clarke score ≥4) or glycemic lability, stable treatment, CGM use, HbA1c ≤10%, and stimulated C-peptide <0.07 nmol l−1 at screening. Key exclusions included prior islet/pancreas transplant, frequent severe hypoglycemia, significant comorbidities, and detectable stimulated C-peptide ≥0.07 nmol l−1. Intervention: Subcutaneous implantation of PEC-Direct devices into abdominal wall/flanks under general anesthesia. Induction immunosuppression with anti-thymocyte globulin (ATG), then maintenance tacrolimus and mycophenolate mofetil adjusted per labs. Each patient received 8 or 10 large dose-finding devices (each ~75×10^6 PEC-01 cells) and 2–3 smaller sentinel devices (each ~7×10^6 PEC-01 cells) for explant analysis. Devices used an optimized membrane perforation density/pattern (patent US16/347,790) intended to enhance engraftment. Assessments: Mixed meal tolerance tests (MMTT) were performed at baseline and every 3 months. Plasma C-peptide and glucose were sampled every 30 min; the minute 90 value (with concurrent glycemia ≥250 mg dl−1 per protocol) defined primary/secondary endpoints. Continuous glucose monitoring (Dexcom G6) provided time-in-range (TIR 71–180 mg dl−1), time-above-range (TAR >180 mg dl−1), time-below-range (TBR ≤70 mg dl−1), and glucose management indicator (GMI) over 3-month intervals. Insulin dosing (IU/kg/day) was averaged over prespecified periods. Adverse events were recorded, including treatment-emergent (TEAE) and serious (TESAE). Endpoints: Primary efficacy endpoint was an increase in MMTT-stimulated C-peptide above the LOD (0.03 nmol l−1) at 6 months. Secondary endpoints included proportion achieving C-peptide >0.07 nmol l−1, changes in insulin dose (including ≥50% reduction and insulin independence), CGM metrics (TIR, TAR, TBR), hypoglycemic event frequency, and HbA1c/GMI changes up to 12 months. Laboratory methods: Glucose (hexokinase), C-peptide (chemiluminescent), and HbA1c (ion exchange HPLC) measured by a central lab. Proinsulin (SIMOA) was assessed at month 12 for the top responder (case 1). CGM auto-mode was standardized before MMTT. Histology: Retrieved sentinel devices (cases 1 and 4) underwent paraffin embedding and staining. Donor (male) vs recipient (female) cells were distinguished in case 1 using RNAscope for KDM5D. Immunostaining included insulin, glucagon, somatostatin, CHRA, CKs, CD34, and immune markers (CD4, CD8, CD20, CD68). Morphometric quantification used the Cavalieri method to estimate volumes of total cellular mass, endocrine markers (CHRA), ductal epithelium (CK), insulin-positive (β), and glucagon-positive (α) cell masses, enabling estimation of donor vs recipient cell contributions and recovery relative to initial loaded cell mass.

Efficacy: At month 3, 5/10 recipients had detectable MMTT-stimulated C-peptide (0.05–0.07 nmol l−1). At month 6, 4/10 met the primary endpoint (detectable above LOD). Three of these four (cases 1–3) reached and sustained C-peptide >0.07 nmol l−1 through month 12, with values 0.10–0.23 nmol l−1. Case 1 reached 0.23 nmol l−1 by months 9–12 and showed a threefold rise from month 3 to 9. Glycemic control: The three patients who achieved ≥0.1 nmol l−1 C-peptide showed improved CGM metrics at months 6 and 9, two sustained to month 12, including increased TIR and decreased TAR toward consensus targets (TIR >70%, TAR <25%). Case 1 improved TIR from 55% pre-implant to 85% at month 12. GMI decreased in parallel with CGM improvements. These improvements occurred with reduced exogenous insulin dosing: cases 1 and 2 used at least 15% less insulin than pre-implant or the first 3 months while achieving better control; case 3 had a 29% insulin reduction at month 9 with improved control but not sustained at month 12, confounded by initiation of liraglutide. Safety: TEAEs were common (notably procedural pain) but none led to study withdrawal. Two TESAEs occurred: one related to surgery and one to immunosuppression. One nonresponder (case 8) developed immunosuppression-related nephrotoxicity and bone marrow toxicity. Histology and cell composition: In case 1 at month 6, donor cells constituted ~32% of inner chamber cells. The insulin-positive (β) cell mass represented ~3% of total inner-chamber cells and ~4% of the initial loaded cell mass; α cell mass was fivefold larger (~16% of cells). Capillary structures (CD34-positive) were present near endocrine clusters. Immune cell accumulations were observed around device mesh externally; immune cells were virtually absent inside chambers. No teratomas were detected. In case 4, donor cells were outnumbered by recipient fibroblasts by month 3 and more so by month 9; β cell mass was <1% of chamber cells at month 9, consistent with lower C-peptide (<0.1 nmol l−1). Dose/device: Patients received 8–10 large devices (~75×10^6 cells each) plus sentinel units, representing a 2–3× higher total cell dose than earlier cohorts and using an optimized perforation pattern associated with improved engraftment in prior work. Overall: Achieving and sustaining C-peptide ≥0.10 nmol l−1 correlated with improved glycemic control and lowered insulin dose, indicating clinically relevant implant function in 3/10 patients by month 6 onward.

The study demonstrates that a subcutaneous, open macroencapsulation device carrying hESC-derived pancreatic endoderm can differentiate to a β cell mass that produces metabolically meaningful C-peptide and exerts measurable glucose control in a subset of T1D patients under immunosuppression. Increasing the implanted cell dose and using a pore configuration previously linked to better survival likely contributed to the higher β cell function observed relative to earlier cohorts. Sustained C-peptide ≥0.10 nmol l−1 was associated with improved CGM-derived endpoints and reduced insulin needs, while nonresponders did not exhibit this pattern, supporting a causal contribution of the implant rather than trial participation effects. Nevertheless, the β cell mass formed was small (≈4% of initial cell mass in the best case, with β cells ~3% of chamber cells and α cells ~16%), insufficient to normalize glycemia during prolonged stimulation, as evidenced by rising proinsulin and declining C-peptide under sustained hyperglycemia. Compared with intrahepatic islet transplantation, which can achieve much higher C-peptide and insulin independence in many patients, subcutaneous extrahepatic approaches remain less effective, consistent with historical outcomes. Differences between rodent and human outcomes highlight challenges in scaling engraftment, vascularization, and survival; donor cell loss and recipient cell infiltration were greater clinically than in animal models. The data suggest opportunities to enhance efficacy via improved device design, cell survival and differentiation strategies, and potentially pro-angiogenic adjuncts, while maintaining retrievability and safety oversight.

This interim analysis shows that device-delivered, stem cell-derived β cells can achieve sustained C-peptide ≥0.10 nmol l−1 in a subset of T1D patients and improve glucose control with reduced insulin dosing by 6–12 months post-implant. Histology links in vivo function to a modest β cell mass, indicating current limitations in cell survival and differentiation. The findings support further development of hPSC-derived pancreatic endoderm in retrievable, perforated devices, with optimization of cell dose, membrane pore configuration, vascularization, and strategies to enhance donor cell survival and β cell differentiation. Future studies should refine implant strategies, evaluate immune-evasive or genetically modified cells, and aim to approach the efficacy of intrahepatic islet transplants while ensuring safety in an accessible site.

Small sample size (n=10) and open-label design without a randomized control group limit generalizability. Results are interim (12 months) and reflect a single device configuration/cadre. Immunosuppression is required, introduces adverse effects, and may confound HbA1c; one responder initiated liraglutide, confounding interpretation at 12 months. C-peptide was assessed under hyperglycemic MMTT conditions (≥250 mg dl−1), which, while standardized, may not reflect everyday physiology. Histologic analyses were limited to sentinel devices from two patients and may not fully represent all implants. Donor cell loss, recipient cell infiltration, and relatively low β-to-α cell ratios constrain efficacy. Long-term durability, safety, and broader reproducibility remain to be established.