Stem cell therapy for diabetes mellitus and its complications — the current evidence, by diabetes type and by indication
## ⚠️ Before anything else: insulin
Nothing on this page is a reason to reduce, delay or stop insulin, or any other prescribed diabetes medication.
Some of the trials described below recorded reductions in insulin requirement. Every one of those reductions
happened **inside a clinical trial, under specialist supervision, with frequent monitoring, in a defined
patient population**. They are observations about what happened to those participants — not instructions,
and not something to attempt independently.
In type 1 diabetes this matters absolutely. Stopping or substantially reducing insulin in type 1 diabetes
can cause diabetic ketoacidosis, which develops within hours to days and can be fatal.
Any change to insulin belongs to the specialist team who prescribed it.
Two Different Diseases
Diabetes mellitus affects approximately 537 million adults worldwide, projected to reach 783 million by 2045 [1]. It is not one condition.
Type 1 diabetes is autoimmune: the immune system destroys the insulin-producing beta cells of the pancreas. Insulin production ceases, and lifelong insulin replacement is required to survive. The therapeutic problem is replacing or protecting beta cells while stopping the autoimmune attack.
Type 2 diabetes, around 90% of cases [2], begins with insulin resistance — tissues respond poorly to insulin — followed by progressive beta-cell exhaustion as the pancreas compensates and eventually fails. Beta cells are present but overworked and dysfunctional. The therapeutic problem is restoring insulin sensitivity and preserving remaining beta-cell function.
⚠️ This distinction governs everything below. A study showing benefit in type 2 diabetes says nothing about type 1, and vice versa. Much of the confusion in this field comes from moving results between them.
Type 1 Diabetes
Stem cell-derived islets: the most advanced programme, and its cost
The furthest-developed approach generates insulin-producing islet cells from pluripotent stem cells and transplants them. Differentiation protocols have advanced substantially, including the acquisition of dynamic glucose-responsive function in stem cell-derived beta cells [5]and the identification of regulators of functional maturation [26].
The clinical programme has now been reported in the New England Journal of Medicine [4]. In a phase 1–2 study of zimislecel, an allogeneic stem cell-derived islet therapy infused into the portal vein:
- 14 participants completed at least 12 months of follow-up
- C-peptide was undetectable in all 14 at baseline; after infusion, all showed engraftment and islet function
- All participants received immunosuppressive therapy
- Neutropenia was the most common serious adverse event, in 3 participants
- ⚠️ Two of the 14 participants died — one from cryptococcal meningitis, one from severe dementia
- The authors note that all analyses were interim and not prespecified
This is a genuine scientific achievement. Restoring measurable endogenous insulin production in people who had none is something no previous therapy has done. It is also a transplant procedure requiring lifelong immunosuppression, and cryptococcal meningitis is an opportunistic infection — the characteristic hazard of suppressing the immune system. In a 14-person study, two deaths is not a footnote.
Reviews of the wider type 1 cell-therapy trial landscape are available [3].
What MSCs have and have not achieved in type 1
Mesenchymal stem cells have been investigated in type 1 diabetes for their immunomodulatory properties — suppression of T-cell proliferation and secretion of anti-inflammatory mediators [7], with the rationale set out in earlier reviews [8][23].
The most informative synthesis is a participant-level meta-analysis of 22 clinical trials and 524 patients [14], and its central finding is that cell type matters more than anything else:
- CD34+ haematopoietic stem cells: 58.9% of type 1 patients became insulin independent, for a mean of 16 months — the best outcome of any modality assessed
- Umbilical cord blood: uniformly negative results — the poorest outcome of any modality assessed
- Umbilical cord MSCs outperformed bone-marrow MSCs in type 1
- "Diabetic ketoacidosis impedes therapeutic efficacy"
⚠️ Two things follow that are easily lost. First, the strongest result belongs to haematopoietic stem cells, delivered with conditioning regimens carrying their own serious risks — not to the mesenchymal cells offered in wellness settings. Second, cord blood — a product frequently marketed for this indication — did not work.
A more recent meta-analysis of 8 controlled trials found that in type 1 patients, daily insulin requirement decreased by a mean of 0.24 units/kg/day (95% CI 0.18–0.29) [12]. That is a real, measured, trial-supervised reduction. It is not a licence to reduce insulin. See the notice at the top of this page.
Type 2 Diabetes
The best controlled evidence
A single-centre, double-blinded, randomised, placebo-controlled phase II trial enrolled 91 Chinese adults with type 2 diabetes, randomised to intravenous umbilical-cord MSCs (n = 45) or placebo (n = 46), given three times at 4-week intervals, with 48 weeks of follow-up [11].
At 48 weeks, 20% of the MSC group reached the primary composite endpoint versus 4.55% of the placebo group.
The authors' conclusion is appropriately measured: UC-MSC transplantation "could be a potential therapeutic approach" for this population. One centre, 91 patients, one ethnic population, 48 weeks.
The pooled picture
Across 8 controlled trials and 334 patients (172 treated, 162 control), umbilical-cord MSC therapy lowered HbA1c by a mean difference of 1.06 percentage points (95% CI 0.85 to 1.27), with no heterogeneity between studies (I [2]= 0%) [12]. For context, that magnitude is comparable to established oral glucose-lowering agents. The authors call for multicentre trials to confirm long-term efficacy and safety.
A separate meta-analysis of 13 studies found a significant HbA1c improvement — but measured as a change from baseline within the treated group, not against a control [13]. Its comparisons of fasting plasma glucose (P = 0.339) and fasting C-peptide (P = 0.231) were not significant. A before-and-after change within one group cannot separate the treatment from natural course, co-intervention or regression to the mean, and should not be read as an effect.
The earliest signal, and what it actually was
An often-cited 2009 study enrolled 10 patients with type 2 diabetes of more than five years' duration, who had failed triple oral therapy, were on insulin at ≥0.7 U/kg/day, and were glutamic acid decarboxylase antibody negative — that is, screened specifically to exclude autoimmune diabetes [10]. They received autologous bone marrow-derived mononuclear cells.
Seven of ten were responders, with insulin requirement reduced by 75%. Three were able to discontinue insulin completely — though in one this was short-lived. Mean HbA1c fell by 1 percentage point. Fasting and stimulated C-peptide improved significantly (P = 0.03). Responders lost a mean 5.5 kg; non-responders gained 2.2 kg. No serious adverse effects were reported.
⚠️ Read the population carefully. These were people with type 2 diabetes and demonstrable residual beta-cell function, whose insulin was adjusted downward by their clinicians as their own insulin secretion improved, under trial monitoring. That is a different situation from type 1 diabetes in every respect that matters.
Mechanisms — and what kind of evidence they are
Proposed mechanisms for type 2 include immunomodulation of chronic low-grade inflammation, support of remaining beta-cell function, and improvement in peripheral insulin sensitivity [6].
The insulin-sensitivity work is frequently cited without its setting. Umbilical-cord MSC exosomes reversed peripheral insulin resistance and reduced beta-cell apoptosis in a rat model of type 2 diabetes induced by high-fat diet and streptozotocin [9]. Related work on exosome protection of beta cells is in vitro [24], and reviews of exosomes in diabetes and its complications remain largely preclinical [25].
Note also that MSCs harvested from people with metabolic syndrome or type 2 diabetes show impaired function from oxidative stress [22]— which is an argument about autologous cell quality, not about efficacy.
Complications: Where the Evidence Is Strongest
The page's most defensible territory is not glycaemic control. It is complications — and here the evidence deserves to be stated with numbers.
Diabetic foot ulcer
Diabetic foot ulcers affect up to a third of people with diabetes during their lifetime, with roughly 2 million amputations annually worldwide [15]. MSCs act on wound healing through angiogenesis (VEGF and other pro-angiogenic factors), reduction of chronic inflammation, and extracellular matrix remodelling [16][17][18].
A 2026 systematic review and meta-analysis of 49 randomised controlled trials quantified it [20]:
- MSC-based therapies: overall healing SMD 0.46 (95% CI 0.24–0.69); time to closure SMD −1.04 (95% CI −1.28 to −0.80)
A 2026 cross-sectional appraisal of nine existing meta-analyses, assessed by AMSTAR (scores 7–10, median 9) and Oxford levels of evidence, found the most methodologically robust of them consistently showed improved ulcer healing rates, with no clear evidence of increased serious adverse events [19]. Its authors caution that the primary studies overlap and are heterogeneous.
⚠️ The comparison that matters, and that is usually omitted. The same 49-trial analysis assessed several modalities side by side, and negative-pressure wound therapy, bioengineered skin substitutes, extracorporeal shockwave therapy and platelet-rich plasma all also showed statistically significant improvements in healing and time to closure [20]. Negative-pressure wound therapy accelerated closure with an SMD of −0.94 (95% CI −1.07 to −0.80).
Stem cell therapy works for diabetic foot ulcer. So do several established, cheaper, widely available treatments. Anyone weighing this decision should be told both.
Diabetic peripheral neuropathy
Diabetic peripheral neuropathy affects around half of people with diabetes and is a major risk factor for ulceration and amputation. A 2024 systematic review and meta-analysis screened 5,431 records and included seven controlled trials [21]:
All four improved significantly. Adverse effects were limited to pain and swelling at injection sites, resolving within days.
⚠️ Route matters, and this is the key qualifier. The cells were bone-marrow mononuclear cells and umbilical-cord MSCs, "administered mainly via intramuscular transplantation" — injected into the muscle of the affected limb. These results are not evidence for intravenous infusion, which distributes cells systemically and predominantly to the lungs. A result obtained by one route is not transferable to another.
What Is Approved
No stem cell therapy is approved for diabetes — not by the FDA, not by the EMA, for either type. Any treatment offered is either part of a registered clinical trial or is unapproved.
For completeness: one mesenchymal stromal cell product does now have FDA approval — Ryoncil (remestemcel-L-rknd), approved December 2024 [27]. It is bone-marrow-derived, and it is licensed for steroid-refractory acute graft-versus-host disease in children, not for diabetes or any diabetic complication. Its existence shows the regulatory pathway is passable; it does not extend to this indication.
Limitations, Stated Plainly
Type 1 diabetes
- Stem cell-derived islet therapy is investigational, requires lifelong immunosuppression, and has recorded deaths in early-phase study [4]
- The best insulin-independence data belong to haematopoietic, not mesenchymal, cells [14]
- Cord blood produced uniformly negative results [14]
- Long-term durability of transplanted cells is unknown
Type 2 diabetes
- Glycaemic effects are real but modest, and long-term durability is unestablished
- The best trial is single-centre, 91 patients, 48 weeks, one population [11]
- MSC therapy does not address the metabolic and lifestyle drivers of insulin resistance
- Where a meta-analysis reports only a within-group change from baseline, that is not evidence of effect [13]
Across both
- Wide variability in cell quality between sources and manufacturing protocols
- Most follow-up is 12 months or less
- No standardised protocol, dose, route or schedule
- No cost-effectiveness data
Questions Worth Asking
- "Which type of diabetes is the evidence for?" — the single most important question on this subject.
- "Which cells, and by what route?" — CD34+ haematopoietic, mesenchymal, cord blood and cord tissue have produced different results, and intramuscular results do not transfer to intravenous.
- "Is this a registered trial with ethical approval, or an unapproved treatment?"
- "What is the evidence for my specific complication?" — foot ulcer, neuropathy, retinopathy and nephropathy each have separate evidence bases.
- "What established treatments should I try first?" — for foot ulcer in particular, several have comparable evidence.
- "Who manages my insulin and glucose during and after this?" — the answer should be a diabetes specialist.
The Bottom Line
For type 1 diabetes, stem cell-derived islets have done something genuinely new: restored measurable insulin production in people who had none [4]. That achievement currently costs lifelong immunosuppression, and the published series records two deaths among fourteen participants. It is not a treatment yet.
For type 2 diabetes, umbilical-cord MSCs produce a real but modest improvement in glycaemic control — an HbA1c reduction of about one percentage point across 334 patients [12], and 20% versus 4.55% reaching a composite endpoint in the best placebo-controlled trial [11]. That is worth knowing and is not a cure.
For complications, the evidence is strongest and most immediately relevant — particularly diabetic foot ulcer, where MSC therapy significantly improves healing [20], alongside several other modalities that do the same.
None of it is approved. None of it replaces insulin, glucose monitoring, medication, diet or specialist care. And no reader should change any part of their diabetes management on the strength of a web page.