Diabetes mellitus affected an estimated 537 million people worldwide in 2021 and, according to the International Diabetes Federation, this number could rise to 783 million by 20451. It imposes substantial premature morbidity and mortality. The disease is primarily divided into type 1 (T1D) and type 2 (T2D). T1D is an autoimmune disease in which autoreactive lymphocytes destroy pancreatic β-cells, eliminating endogenous insulin secretion2.

Individuals with T1D therefore depend on lifelong exogenous insulin. By contrast, T2D represents ~90 % of all diabetes cases and results from a combination of peripheral insulin resistance and insufficient compensatory insulin secretion by β-cells. Ongoing research continues to clarify the molecular mechanisms that drive T2D onset and progression3.

Monogenic forms of diabetes also exist, most notably neonatal diabetes mellitus (NDM) and maturity-onset diabetes of the young (MODY)4. NDM presents within the first six months of life and affects approximately 1 in 300 000–400 000 live births5, 6. MODY, an autosomal-dominant β-cell disorder that typically manifests in adolescence or early adulthood, accounts for <5 % of all diabetes cases7, 8. Molecular testing has so far delineated 14 MODY subtypes9.

Because stem cells can modulate immune responses and differentiate into insulin-producing β-like cells, they are being explored as innovative therapies for diabetes10, 11, 12. Stem cells are classically categorized as totipotent, pluripotent, multipotent, oligopotent, or unipotent, depending on their developmental potential13. In a pioneering trial, Voltarelli et al. (2007) infused autologous hematopoietic stem cells (HSCs) into patients with recent-onset T1D and reported partial remission14. Subsequently, Bhansali et al. (2009) showed that bone-marrow-derived stem cells can safely improve β-cell function in T2D15. Numerous trials have since evaluated various stem-cell sources; however, the optimal cell type, dose, and delivery route remain to be defined, and severe infections have occasionally been observed16. Advances in stem-cell derivation and bioprocessing aim to provide an unlimited, donor-independent supply of transplantable β-cells for regenerative medicine17 (Figure 1).

Figure 1

Embryonic stem cells (ESCs) are derived from the inner cell mass of the blastocyst, whereas induced pluripotent stem cells (iPSCs) are reprogrammed from somatic cells by ectopic expression of factors such as OCT3/4, SOX2, KLF4, and c-MYC. Both ESCs and iPSCs are pluripotent, meaning they can give rise to virtually any cell type, including cardiomyocytes, neurons, and pancreatic β-cells18, 19 (Figure 2).

Figure 2

Precision genome-editing, particularly CRISPR–Cas9, now allows targeted modification of transcription factors and signaling pathways in human pluripotent stem cells (hPSCs). Knockout of ARX, NKX6-1, or NEUROD1 clarifies endocrine lineage specification, whereas knock-in reporters such as PDX1-GFP or INS-mCherry permit real-time monitoring of β-cell maturation. Correction of pathogenic variants (e.g., HNF1A, INS) in patient-specific iPSCs paves the way for personalised cell-replacement therapies20. Single-cell RNA-seq and ATAC-seq provide high-resolution atlases of in-vitro differentiation, revealing rare progenitors, off-target populations, and regulators of β-cell maturation; only a fraction of derived cells fully resembles fetal or adult islets21. Engineering 3D microenvironments—vascularised islet organoids, hydrogel scaffolds, organ-on-a-chip systems—improves glucose responsiveness, endocrine maturity, and cell survival by recapitulating paracrine cues, extracellular-matrix signals, and biomechanical stiffness22. Together, these innovations are propelling stem-cell science from proof-of-concept studies toward robust, scalable, and clinically translatable regenerative therapies for diabetes.

Diabetes mellitus (DM) comprises a spectrum of metabolic disorders characterized by persistent hyperglycemia, yet the etiology and underlying mechanisms differ among subtypes. Appreciating these distinctions is critical to targeted prevention, accurate diagnosis, and effective therapy. The principal forms are type 1 diabetes (T1D), type 2 diabetes (T2D), gestational diabetes mellitus (GDM), and monogenic variants such as neonatal diabetes and maturity-onset diabetes of the young (MODY).

Pancreatic β-cells within the islets of Langerhans orchestrate insulin secretion and are therefore central to glucose homeostasis. In diabetes, hyperlipidaemia, hyper-glycaemia and chronic inflammation converge to provoke endoplasmic-reticulum (ER) stress, oxidative stress and mitochondrial dysfunction, ultimately driving β-cell death and de-differentiation. Although ER and mitochondrial stress individually impair β-cell viability, recent work highlights their synergistic amplification of reactive oxygen-species (ROS) generation42, 43. β-cells inherently produce high ROS yet possess only modest antioxidant defences, rendering them exceptionally vulnerable to oxidative injury and functional collapse44, 45.

In type 1 diabetes (T1D) the immune system eliminates ≈90 % of β-cells, causing an early fall in insulin output that precedes overt hyper-glycaemia. Intriguingly, residual β-cells often persist in people with long-standing T1D46, implying that low-level endogenous insulin secretion can continue47, 48. In type 2 diabetes (T2D) approximately half of the original β-cell mass remains at diagnosis49, 50. Butler et al. analysed 124 human pancreata and found β-cell apoptosis was elevated ten-fold in lean T2D and three-fold in obese T2D, independent of auto-immunity51. These observations underscore the importance of preserving or restoring β-cell mass to maintain euglycaemia, making β-cell replacement a rational strategy for both T1D and T2D52. Whole-pancreas or islet transplantation can normalize glycaemia but is limited by donor scarcity, surgical risk and lifelong immuno-suppression. Consequently, stem-cell-derived β-cells have emerged as an attractive, potentially unlimited alternative source53, 54.

PSC populations—embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs)—combine unlimited self-renewal with the capacity to generate derivatives of all three germ layers. ESCs originate from the inner cell mass of the blastocyst, whereas iPSCs arise when somatic cells are reprogrammed by OCT4, SOX2, KLF4 and c-MYC expression (the Yamanaka factors)55. Pluripotency is sustained by interconnected transcription-factor networks, epigenetic regulators and signalling cascades including WNT, TGF-β/Activin/Nodal and FGF. This developmental plasticity underpins applications in regenerative medicine, disease modelling and high-throughput drug discovery. Nonetheless, genetic instability, tumourigenicity and line-to-line variability mandate rigorous quality control.

The 2006 discovery of iPSCs by Takahashi & Yamanaka revolutionised the field by providing an ethically acceptable ESC surrogate56, 57, 58, 59, 60, 61. Advances in vector design—from integrating retroviruses to non-integrating episomes, Sendai virus and mRNA—have enhanced reprogramming safety and efficiency62, 63, 64. Today, patient-specific iPSC lines enable precise disease modelling and personalised cell-based therapies, underscoring PSCs’ transformative potential65.

To generate functional cell types, researchers recapitulate embryogenesis in vitro using serum-free media, defined growth factors and stepwise signalling cues. For example, VEGF plus other angiogenic factors yield endothelial cells66; multistage Activin-A/retinoic-acid/Notch modulation directs PSCs to pancreatic β-like cells67; BMP4, FGF and HGF drive hepatic specification toward hepatocyte-like cells68; and 3-D culture of porcine iPSCs with retina-specific factors forms laminated retinal organoids69. Ongoing protocol optimisation is boosting purity, scalability and reproducibility, accelerating deployment of PSC-derived cells in drug screening, disease modelling and regenerative therapy.

Research involving pluripotent stem cells, including embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs), holds tremendous promise for regenerative medicine and disease modeling; however, it is also accompanied by significant ethical and clinical challenges that demand careful analysis and innovative solutions70. The main ethical controversy surrounding ESCs arises from the need to destroy embryos during derivation, prompting intense debate over the moral status of the embryo and leading to diverse regional regulations that, in turn, shape funding, research priorities, and clinical translation71. Although iPSCs circumvent embryo destruction and have revolutionized the field, they introduce fresh ethical and safety concerns that likewise require rigorous oversight72.

Early‐generation iPSC protocols relied on oncogenic transcription factors, raising fears of tumorigenesis and underscoring the necessity for safer reprogramming methods that preserve genomic integrity73. Integration-free techniques now mitigate many of these risks, yet confirming both genomic and epigenetic stability in patient-derived iPSC lines remains a critical bottleneck before clinical deployment74. Because undifferentiated PSCs can form teratomas, transplantation therapies must meet stringent safety thresholds and demonstrate robust efficacy75. Even autologous iPSC-based therapies face variability in line-to-line quality, complicating standardization and quality control76, 77.

Successful differentiation of PSCs into insulin-producing β-cells hinges on precise control of the culture microenvironment and a deep understanding of developmental signaling cues. Human ESCs, derived from the inner cell mass of blastocysts, possess unique epigenetic landscapes that preserve pluripotency; establishing stable hESC lines that faithfully recapitulate primary ESC characteristics is therefore essential for an unlimited cell supply78, 79.

Step-wise protocols have converted hESCs into insulin-secreting cells, initially achieving ~12 % efficiency but with limited glucose responsiveness. Subsequent optimization—such as fine-tuning in-vitro glucose levels and modulating growth factors like Transforming Growth Factor (TGF)—has boosted yields to ~25 %80, 81. These findings underscore the need for early-stage interventions that protect or regenerate endogenous β-cell mass, or for replacement strategies using engineered stem-cell-derived islet-like clusters82.

Standard differentiation proceeds through definitive endoderm (SOX17⁺), pancreatic progenitor (PDX1⁺), endocrine progenitor (NGN3⁺), and finally mature β-cell stages. At each step, specific factors are applied, and stage-specific markers confirm proper lineage progression83.

Transplantation studies indicate that the in-vitro microenvironment strongly affects pancreatic progenitor expansion and maturation. Using immature pancreatic progenitors rather than fully differentiated cells consistently enhances in-vivo β-cell development. For example, pancreatic precursors derived from hESCs become glucose-responsive, insulin-secreting β-cells after implantation under the kidney capsule or into adipose tissue of streptozotocin (STZ)-induced diabetic mice84. Likewise, iPSC-derived grafts placed in both T1D and T2D mouse models acquire glucose-regulated insulin secretion and reduce hyperglycaemia85. Transplantation of non-human-primate iPSCs achieves comparable glycaemic improvement in murine diabetes models86. In the NOD mouse, iPSC-derived insulin-producing cells transplanted into the kidney respond appropriately to rising glucose concentrations87. Collectively, these findings show that extensive pre-transplant in-vitro patterning is essential, whereas site-specific in-vivo cues complete endocrine maturation.

Efficient expansion of β-cell mass from endogenous sources requires simultaneously limiting β-cell apoptosis and stimulating new-cell formation. Finegood et al. (1995) quantitatively analysed β-cell turnover by BrdU/thymidine labelling in rat pancreas118. They calculated a daily turnover of ~2 % of β-cells in adult rodents. Extended BrdU exposure in adult mice showed that roughly 1 in 1,400 β-cells divides each day. Assuming no input from neogenesis, trans-differentiation or other sources, the daily growth rate equals 0.070 %119, 120. Even with zero β-cell death, replacing one-half of lost β-cell mass would therefore require ≈1,429 days—far longer than the average mouse life span. Human calculations are limited, as BrdU cannot be used ethically; Ki67 staining nevertheless indicates an even slower turnover that can rise several-fold during pregnancy121, 122.

An alternative strategy for diabetes therapy is to enhance endogenous β-cell renewal. Evidence shows that β-cell mass is plastic and adapts to changing secretory demand. Two main mechanisms are proposed: (i) replication of existing β-cells and (ii) differentiation of progenitors, possibly within the ductal epithelium. Replication has been documented in mice, rats and humans, and lineage-tracing in postnatal mice demonstrates that most new β-cells derive from pre-existing ones. The close anatomical relationship between β-cells and pancreatic ducts suggests a potential ductal progenitor source, but cross-sectional studies cannot yet pinpoint the exact origin of mature β-cells118, 119, 120 (Figure 3). Thus, insulin-positive cells observed near ducts in adult tissue may simply reflect residual patterns of fetal pancreas development rather than active duct-derived neogenesis119, 121, 122.

Figure 3

The pathogenesis of different diabetes subtypes is not fully understood. To address this gap, researchers generate patient-specific pluripotent stem cells (PSCs) from diabetic individuals as versatile in-vitro disease models. These cells can be differentiated into pancreatic lineages for mechanistic studies or transplantation, thereby providing new insights and enabling improved therapeutic strategies123.

Pluripotent stem cells (PSCs) possess the remarkable ability to differentiate into almost any cell type, making them indispensable tools for diabetes research65. Accordingly, diabetic mouse models are essential for dissecting disease mechanisms and testing emerging therapies. Through controlled in-vitro differentiation, investigators can generate insulin-producing pancreatic β-cells, endothelial cells, and immunomodulatory mesenchymal stem cells (MSCs) from PSCs131. When these PSC-derived cells are transplanted into diabetic mice, their effects on tissue regeneration, glycaemic control, and immune modulation can be quantified with precision132.

Transplantation of PSC-derived pancreatic β-cells offers the possibility of restoring endogenous insulin production and alleviating the consequences of insulin deficiency. Outcomes are monitored by evaluating cell survival, engraftment within host pancreatic tissue, and the re-establishment of normoglycaemia133. Such analyses are critical, as loss of functional β-cells is central to diabetes pathophysiology134.

PSC-derived endothelial cells additionally target the vascular complications of diabetes, supporting vascular repair and restoring blood flow to damaged tissues and organs135. Likewise, PSC-derived MSCs exert potent anti-inflammatory and immunomodulatory effects that counter the inflammatory milieu associated with diabetes. Mouse models allow researchers to quantify their capacity to dampen immune dysregulation, protect pancreatic cells, and accelerate tissue repair136.

Collectively, PSC-based interventions in diabetic mouse models refine our understanding of the disease and accelerate the development of next-generation therapies for patients.

With its ability to precisely fix disease-causing mutations and improve cell function, CRISPR-Cas9 technology has become a cornerstone in the genetic engineering of PSCs. To restore normal insulin expression in iPSC-derived β-cells, CRISPR has been used to correct mutations in genes such as HNF1A and GCK, thereby restoring normal insulin expression in monogenic diabetes forms like MODY (Maturity-Onset Diabetes of the Young)137. Additionally, CRISPR is being investigated to generate universal, immune-evasive β-cell grafts for allogeneic transplantation by deleting immune-recognition molecules such as HLA.

PSC-derived three-dimensional (3D) pancreatic organoids have revolutionized both in vitro modeling and in vivo transplantation. These organoids mimic native islet architecture and function, including vasculature formation, cell-cell communication, and glucose-stimulated insulin release. Stepwise development of PSCs into mature β-cells within organoids is now possible thanks to protocols guiding differentiation through definitive endoderm, pancreatic progenitors, and endocrine precursors. To further enhance vascular integration and insulin-release kinetics, these organoids can be co-cultured with endothelial cells or embedded in extracellular-matrix hydrogels138.

The field is now entering the clinic, with multiple stem-cell-derived products in human trials. ESC-derived pancreatic islet cells are used in Vertex Pharmaceuticals’ VX-880 program, which has successfully restored insulin production in type 1 diabetes patients with undetectable C-peptide levels139. Similarly, ViaCyte’s PEC-Direct and PEC-Encap systems administer PSC-derived β-cell progenitors in encapsulated devices, aiming to achieve long-term insulin independence in early-phase trials and preclinical models140. However, foreign-body responses impaired graft vascularization and function in PEC-Encap (VC-01). In Phase 1/2 trials, PEC-Direct—an open device that permits vascularization but requires systemic immunosuppression—generated partial C-peptide and insulin secretion yet did not fully restore glycaemic control.

Patient-specific iPSC-derived β-cells are now being integrated into high-throughput drug-screening platforms to identify compounds that enhance insulin secretion or promote β-cell survival, creating a rapid feedback loop to clinical observation. The convergence of PSCs, CRISPR editing, and organoid technology is enabling high-throughput modeling of both monogenic and complex polygenic diabetes. Researchers now generate patient-specific, CRISPR-edited islet organoids to dissect gene function, predict therapeutic response and validate candidate drugs. These models recapitulate immune infiltration, ER stress, and β-cell dysfunction—hallmarks of both type 1 and type 2 diabetes.

Strict adherence to Good Manufacturing Practice (GMP) is the cornerstone for translating pluripotent stem cell (PSC) and induced pluripotent stem cell (iPSC) products from the laboratory to the clinic. By controlling cell sourcing, scale-up, contamination risk, and batch-to-batch consistency, GMP safeguards product safety, reproducibility, and regulatory acceptance.

Recent literature therefore centres on the transition from bench-scale PSC cultures to pilot- and full-scale manufacturing. Huang et al. (2020) present a comprehensive GMP-compatible suspension-bioreactor workflow, detailing media optimisation, shear-stress management, and process automation required for clinical-grade expansion141. Similarly, Martins and Ribeiro (2025) describe the creation of GMP Master and Working Cell Banks, highlighting donor eligibility, traceability, and high-throughput quality control under EU and U.S. regulations142.

Thon and Karlsson (2017) show that feeder-free, xeno-free, closed-system bioreactors markedly improve reproducibility of platelet-producing PSC derivatives and facilitate regulatory approval143. Nath et al. (2020) further refine stirred-tank designs with in-line sensors, automated batch records, and continuous environmental monitoring, fully embedding GMP in iPSC expansion and differentiation144.

To minimise lot variability, Wong et al. (2017) introduced the CryoPause® workflow, which cryopreserves fully characterised PSCs in a ready-to-use format, enabling immediate and parallel differentiation across GMP sites145.

Demonstrating clinical relevance, Surendran et al. (2025) report a scalable, allogeneic retinal-pigment-epithelium (RPE) manufacturing process that meets FDA Investigational New Drug criteria via GMP-aligned cryopreservation, sterility, and endotoxin removal146. Likewise, Couture et al. (2014) provide detailed standard operating procedures (SOPs) for spinner-flask suspension culture that reliably generate undifferentiated PSCs at pilot scale147.

Pluripotent stem cells (PSCs) now allow researchers to reproduce diabetic pathology in vitro and to design cell-replacement strategies that were unthinkable a decade ago. Recent work has shown that patient-specific PSC lines can be differentiated into pancreatic lineages, providing platforms both for mechanistic studies and for the development of autologous therapies148, 149. Before these advances can reach the clinic, however, several obstacles must be overcome. First, robust, tumor-safe differentiation protocols are required to minimize the risk of teratoma formation149. Human embryonic stem cells (hESCs) additionally raise ethical concerns and problems of immune incompatibility, which currently restrict their clinical use. Somatic-cell nuclear transfer (SCNT) has recently been used to create hESC lines that are human-leukocyte-antigen (HLA) matched to individual patients, potentially reducing rejection150.

Although induced PSCs (iPSCs) represent a landmark step toward autologous β-cell replacement, important challenges remain. Comprehensive in-vitro assays and long-term transplantation studies are still needed to confirm the function, safety and durability of iPSC-derived β-cells. Genomic integrity is a key issue because some reprogramming methods employ integrating viral vectors that can introduce oncogenic or disruptive mutations150, 151. Even non-viral techniques can generate copy-number variations or point mutations that confound disease modelling; therefore, integration-free reprogramming and rigorous genomic screening should be standard practice.

Another hurdle is incomplete maturation. Cells produced in most differentiation protocols still express early developmental markers such as DPPA4, LIN28A and LIN28B, indicating a stage equivalent to < 6.5-week human embryos and explaining their poor glucose responsiveness152, 153. Refinement of stage-specific cues and 3-D culture systems is required to obtain fully mature, glucose-sensitive β-cells. Patient-to-patient and clone-to-clone variability further complicate individualized therapies; future work must elucidate and minimise these clonal differences.

Emerging evidence also overturns the assumption that autologous iPSC derivatives are intrinsically immune-privileged. Deuse et al. (2019) demonstrated that de-novo mitochondrial DNA (mtDNA) mutations can provoke T-cell-mediated rejection in syngeneic hosts154; similar observations were made by Sercel et al. (2021) and Bogomazova et al. (2024)155, 156. Hence, routine immunogenomic screening—especially of mtDNA—should precede any iPSC-based transplantation, even in autologous settings.

Whether iPSCs can fully replace hESCs remains unresolved, and the limited comparative data published so far underline the value of continuing hESC research alongside iPSC efforts127.

PSC research continues to attract intense interest because these cells can generate isogenic models of diabetes and theoretically provide unlimited supplies of patient-matched β-cells. To translate this promise into therapy, investigators must eliminate undifferentiated contaminants, perfect maturation protocols, and address ethical and immunological barriers. The creation of SCNT-derived, HLA-matched hESCs and the advent of footprint-free iPSC technologies offer encouraging solutions. Combined with clinical-grade differentiation, 3-D islet-organoid platforms and precise CRISPR editing, PSC technology is progressing from glucose-control adjuncts toward durable, potentially curative interventions. Multiple allogeneic islet products are already in human trials, and genetically engineered, patient-specific β-cells are close behind. With continued multidisciplinary effort, PSC-based precision therapies are expected to transform diabetes management in the near future.

3D: Three-Dimensional, ARX: Aristaless Related Homeobox, ATAC-seq: Assay for Transposase-Accessible Chromatin with sequencing, BMP4: Bone Morphogenetic Protein 4, BrdU: Bromodeoxyuridine, Cas9: CRISPR-associated protein 9, CD8⁺: Cluster of Differentiation 8 Positive, CRISPR: Clustered Regularly Interspaced Short Palindromic Repeats, CXCR4: C-X-C Motif Chemokine Receptor 4, DE: Definitive Endoderm, DM: Diabetes Mellitus, DPPA4: Developmental Pluripotency Associated 4, EGF: Epidermal Growth Factor, ER: Endoplasmic Reticulum, ESCs: Embryonic Stem Cells, FDA: Food and Drug Administration, FGF: Fibroblast Growth Factor, FOXA2: Forkhead Box A2, GAD65: Glutamic Acid Decarboxylase 65, GCK: Glucokinase, GDM: Gestational Diabetes Mellitus, GDF-8: Growth and Differentiation Factor-8, GFP: Green Fluorescent Protein, GMP: Good Manufacturing Practice, GSC: Goosecoid Homeobox, GSK3: Glycogen Synthase Kinase 3, HEDGEHOG: a signaling pathway, HGF: Hepatocyte Growth Factor, HLA: Human Leukocyte Antigen, HNF1A: Hepatocyte Nuclear Factor 1 Alpha, hPSCs: Human Pluripotent Stem Cells, HSCs: Hematopoietic Stem Cells, IA-2: Insulinoma-Associated protein 2, IDE1/IDE2: Inducers of Definitive Endoderm 1 and 2, IGF-1: Insulin-like Growth Factor 1, IL-6: Interleukin-6, INS: Insulin, iPSCs: Induced Pluripotent Stem Cells, ISL1: ISL LIM Homeobox 1, KLF4: Krüppel-like factor 4, MAFA: MAF BZIP Transcription Factor A, MODY: Maturity-Onset Diabetes of the Young, MSCs: Mesenchymal Stem Cells, mtDNA: Mitochondrial DNA, MYC: Myelocytomatosis oncogene, NDM: Neonatal Diabetes Mellitus, NEUROD1: Neuronal Differentiation 1, NGN3: Neurogenin 3, NKX6-1: NK6 Homeobox 1, NOD: Non-Obese Diabetic, OCT3/4: Octamer-Binding Transcription Factor 3/4, PAX4: Paired Box 4, PDX1: Pancreatic And Duodenal Homeobox 1, PGD: Pre-implantation Genetic Diagnosis, PI3K: Phosphoinositide 3-Kinase, PSCs: Pluripotent Stem Cells, RA: Retinoic Acid, RNA-seq: RNA sequencing, ROS: Reactive Oxygen Species, RPE: Retinal Pigment Epithelium, SCNT: Somatic Cell Nuclear Transfer, SOPs: Standard Operating Procedures, SOX2: SRY-Box Transcription Factor 2, SOX17: SRY-Box Transcription Factor 17, STZ: Streptozotocin, T1D: Type 1 Diabetes, T2D: Type 2 Diabetes, TGF-β: Transforming Growth Factor Beta, TNF-α: Tumor Necrosis Factor Alpha, VEGF: Vascular Endothelial Growth Factor, VMAT2: Vesicular Monoamine Transporter 2, Wnt: Wingless-related integration site, ZnT8: Zinc Transporter 8

None.

All the authors have contributed substantially to the work. All authors read and approved the final manuscript.

None.

Not applicable.

Not applicable.

Not applicable.

The authors declare that they have used generative AI and/or AI-assisted technologies in the writing process before submission, but only to improve the language and readability of their paper.

The authors declare that they have no competing interests.