Efficacy and safety of stem cell therapy in patients with Diabetes Mellitus – a systematic review and meta-analysis

Abstract

Background

To support the development of a national guideline on stem cell therapy, the Department of Health Research, India, commissioned this systematic review to evaluate the efficacy and safety of various stem cell types in patients with type 1 and type 2 diabetes mellitus (DM), focusing on patient-important outcomes.

Methods

Following PRISMA guidelines, a literature search was conducted in PubMed, Embase, Web of Science, and Cochrane databases from inception to August 30, 2024. Critical outcomes for type 1 DM included insulin-free periods, hypoglycemic episodes, quality of life, and serious adverse events. For type 2 DM, outcomes included HbA1c, insulin requirements at 6, 12, and 24 months, and serious adverse events. Meta-analyses used random- or fixed-effects models based on heterogeneity (Chi-square test and I²). Risk of bias was assessed using the Cochrane Risk of Bias Tool 2.0, and evidence certainty was evaluated with GRADE.

Results

The search identified 11,026 articles, of which 20 randomized controlled trials (RCTs) were included, encompassing 427 and 351 patients in the intervention and control groups, respectively, with follow-ups ranging from 3 to 96 months. Predominantly studied therapies included mesenchymal and bone marrow mononuclear stem cells. In type 1 DM, stem cell therapy showed no significant improvement in quality of life [MD: 3.15% (95% CI: -0.80 to 7.10); 2 trials, n = 63 participants; I² = 0%; GRADE: very low certainty] or reduction in hypoglycemic episodes [RR: 0.90 (95% CI: 0.56 to 1.45); 3 trials, 68 participants; I² – 0%; GRADE: very low. In type 2 DM, stem cell therapy significantly reduced insulin requirements at 6, 12, and 24 months, with MDs in IU/day of -14.42 (95% CI: -24.25 to -4.59); 6 trials, n = 167 participants; I² = 91.64%; GRADE: low certainty; -17.79 (95% CI: -26.39 to -9.18); 6 trials, n = 212 participants; I² = 70.96%; GRADE: low certainty; and -35.73 (95% CI: -40.82 to -30.64); 1 trial, n = 61 participants; I² = NA; GRADE: very low certainty, respectively, with a low certainty of evidence.

Conclusion

Stem cell therapy did not achieve an insulin-free state or improved quality of life in type 1 DM patients. However, it reduced insulin requirements by 14–36 units over 6–24 months in type 2 DM patients, without significant glycemic control. Larger, high-quality RCTs with extended follow-ups are essential to determine the therapeutic potential of stem cell therapy in diabetes mellitus.

Systematic review registration

PROSPERO ID: CRD42023451602.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13643-025-03054-0.

Background

Diabetes Mellitus (DM) has become a critical health concern, reaching epidemic proportions in India. According to the recent Indian Council of Medical Research–India Diabetes (ICMR-INDIAB) study, the prevalence of diabetes in urban areas was measured as high as 11.2% in urban areas and 5.2% in rural areas [1]. High disease burden and the need for prolonged treatment of diabetes with oral hypoglycemic agents and/or insulin therapy, carrying the risk of hypoglycemic events, pose challenges in the management of diabetes. Cellular therapy with islet cell transplantation or multipotent stem cells offers alternative treatment avenues as they have the potential to reduce prolonged dependence on drugs or insulin. However, the translation of stem cell therapy from bench to bedside needs to be supported by data on the clinical effectiveness and safety of stem cell therapy in diabetic patients.

Previously published systematic reviews and meta-analyses have shown that stem cell therapy is efficacious with respect to insulin dose reduction, increase in c-peptide levels in type 1 DM, and reduction in HbA1c levels in type 2 DM [2–4]. However, previous reviews are limited by the inclusion of study designs that are likely to have a high risk of bias, such as non-randomized controlled designs and case reports, which questions the validity of results and lack of consideration of patient-important outcomes such as insulin-free period, quality of life, hypoglycemic episodes and serious adverse events [2, 3]. Given the rapid advancements in stem cell research and the increasing number of studies published in recent years, it is crucial to consolidate the latest evidence on stem cell therapy for diabetes mellitus. Moreover, as the burden of diabetes continues to rise, a timely synthesis of updated data is essential to guide clinical practice and inform policy decisions for more effective treatment strategies. In this context, this systematic review is conceived to assess the efficacy of diverse types of stem cells studied in randomized controlled trial settings to investigate their clinical effectiveness and safety in patients with type 1 and 2 diabetes mellitus with patient-important outcomes.

Methods

We conducted this systematic review as per the methods of the Cochrane Handbook for Systematic Reviews for Intervention and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines [5, 6]. A complete protocol is available at PROSPERO with registration number CRD42023451602. PICO characteristics of this systematic review are summarized in Table 1.

Table 1.

PICO Characteristics of review question assessing efficacy and safety of stem cells

Population	Patients with type 1 or 2 Diabetes mellitus (DM)
Intervention	Any stem cell and product derived from stem cells
Comparison	Usual diabetic care with or without placebo administration
Outcome	Critical outcome for type 1 DM: Insulin free period, hypoglycemic episodes, quality of life and serious adverse events
	Critical outcome for type 2 DM: HbA1c and insulin requirement at 6, 12 and 24 months, and serious adverse events
	Important outcomes: C-peptide levels, fasting blood glucose, HbA1C levels, episodes of diabetic ketoacidosis and end organ dysfunction
Study design	Randomized controlled trials

Eligibility criteria

Inclusion criteria of this review include – (i) studies involving patients with diabetes mellitus, including type 1 and 2 or maturity-onset diabetes of the young (MODY) or pancreatic diabetes, with or without complications, (ii) studies reporting use of any of stem cell or its derived product, as therapy either as a standalone intervention or as add-on therapy in the intervention arm, (iii) studies having the comparator arm receiving usual diabetic care with or without placebo administration, (iv) studies reporting atleast one of the following critical outcomes such as insulin free period, hypoglycemic episodes, quality of life, HbA1c and insulin requirement at any of the timepoints—6, 12 and 24 months, and/or, safety outcomes as serious adverse events, mortality or any tumour formation, and (v) studies with randomized controlled trial (RCT) design were considered without any language restriction. Exclusion criteria included animal studies, studies conducted among people with gestational diabetes or diabetes insipidus, studies assessing stem cell therapy to cure diabetic foot/wounds and those based on pancreatic or islet cells organ transplantation as intervention.

Search strategy

Literature search was conducted in the Medline database through PubMed, Embase, Web of Science and Cochrane database for peer-reviewed articles using search terms related to ‘diabetes Mellitus’, ‘stem cell’ and ‘randomized controlled trial’ without any filters. The detailed search strategy adopted in each database has been listed in Appendix A (Supplementary file).

Study selection and data extraction

Literature retrieved from databases was imported into the Rayyan online platform, and duplicate records were removed [7]. Two authors (M.S and R.H.) independently screened the titles and abstracts initially and later the full-text articles using the predefined inclusion and exclusion criteria. The discrepancies in the selection of articles between the two authors were resolved by the third author (B.S.B). Full-text articles excluded at this stage are listed in Appendix B, along with reasons for exclusion. Two authors (M.S and R.H.) independently extracted data using a piloted structured data extraction form in concurrence with the third reviewer (B.S.B.). Data extraction was done in Google sheet.

Risk of bias assessment

Cochrane Risk of Bias Tool 2.0 was used to assess the risk of bias in the randomized controlled studies included in the systematic review for each outcome [8]. The tool consisted of five domains and was used to assess studies for risk of bias during randomization process, deviation in delivering the intended intervention, missing outcome data, bias in measurement of outcome and selective reporting of outcomes, under each outcome. Domain-wise assessment was considered to classify the study overall into ‘low risk’, ‘some concerns and ‘high risk’ of bias for each outcome.

Synthesis of evidence (meta-analysis)

Raw data extracted from the studies were harmonised before analysis, using the standard procedure suggested by the Cochrane methods. The characteristics of the included studies for the systematic review were tabulated. For trials that reported outcomes at baseline and endline of the study, the effect estimates reported during endline assessment in the intervention and control groups were considered for meta-analysis. Meta-analysis was carried out separately for studies that included type 1 and 2 diabetes mellitus patients. The mean difference and 95% confidence interval (CI) were used to pool continuous outcomes such as quality of life, insulin requirement, HbA1C, fasting blood glucose and fasting C-peptide levels. Risk ratio and 95% CI were used to pool hypoglycemic episodes. Meta-analysis was performed using ‘Stata Statistical Software: Release 17, College Station, TX: StataCorp LP’.

Heterogeneity assessment and subgroup analysis

Statistical heterogeneity was assessed using the Chi-Squared test (P < 0.05) and the I² statistic. The source of heterogeneity in studies was explored using subgroup analysis, including place of study, type of stem cell administered and route of administration. If I² is greater than 50%, pooled analysis was performed based on the random effects model using the DerSimonian and Laird approach. A fixed effects model using the Mantel–Haenszel method was used if I² is less than 50%.

Publication bias

Assessment of publication bias for the meta-analysis was planned per protocol using a funnel plot when the analysis pooled results from more than 10 studies.

GRADE assessment

Quality of evidence for the body of evidence was assessed as per GRADE methodology, considering the risk of bias, imprecision, inconsistency, indirectness and publication bias [9]. Based on this, evidence was rated as high, moderate, low and very low for critical and important outcomes listed in the review.

Role of funding source

There is no funding source for conducting this systematic review.

Results

The electronic search of included databases yielded 11,026 potential research articles, of which 1,778 duplicates were removed. After title and abstract screening, 135 articles were included for full-text screening. After the full-text screening, twenty RCTs were eligible for inclusion in the systematic review and meta-analysis. The PRISMA flow diagram showing the selection process of studies is summarized in Fig. 1.

Fig. 1.

Study selection (PRISMA) flowchart

Characteristics of the included studies

The characteristics of the twenty included studies are presented in Table 2. The studies were published from 2005 to 2023 and were conducted in Argentina, Australia, China, Ireland, Italy, the United Kingdom, Sweden, India, and the United States, with sample sizes ranging from 15 to 91 patients. The number of studies that included patients with type 1 diabetes mellitus (T1DM) [10–16], type 2 diabetes mellitus (T2DM) [17–25] and both [26–28] was 7, 9 and 3, respectively. Mirzaei M et al, has not specified type of DM patients that were included in the study and hence this study is included only in the qualitative summary [29]. These patients had an average diabetes history ranging from recent onset to 19.6 years. The follow-up period ranged from 3 to 96 months. When considering the design of these studies, six were open-label randomized controlled trials, six were single-blind, seven were double-blind, and one was a triple-blind randomized controlled trial. Intervention reported in studies includes – six studies using mesenchymal stem cells, five studies using bone marrow mononuclear cells, three using mesenchymal stromal cells, two studies reporting a combination of mesenchymal stem cells and bone marrow mononuclear cells, one reporting umbilical cord blood, two studies each using hematopoietic stem cells and mesenchymal precursor cells, respectively. Thirteen studies used saline as a placebo in the control arm, while other studies used saline (11 studies), usual diabetic care (4 studies), human serum albumin and DMSO (1 study), CryostarCS10 fluid (1 study), prostaglandin E1 (1 study) and unspecified placebo/control in two studies.

Table 2.

Characteristics of the study populations in the included studies

Name of the study	Country of study	Type of DM	Type of stem cells	Stem cell count	Route of administration	No of patients SC group/control group	Mean age of SC group/control group (years)	Mean duration of DM of SC group/control group (years)	Total follow up period (months)	Outcomes included in the study
Perico N et al., 2023 [17]	Ireland, Italy and United Kingdom	T2DM	allo BM Mesenchymal stromal cell	80 million cells	IV	12/4	69/59	> 3/> 3	18	FBG, HbA1C, adverse events
Carlsson PO et al., 2023 [10]	Sweden	T1DM	allo WJ Mesenchymal stromal cell	200 million cells	IV	10/5	31/31	1/1	12	HbA1C, FBG, Insulin requirement, fasting c-peptide & MMTT AUC c-peptide
Izadi M et al., 2022 [12]	Iran	T1DM	auto BM MSC	1 million cells/kg	IV	11/10	10.27/11.5	NM/NM	12	FBG, HbA1C, C-peptide, endogenous insulin, hypoglycemic episodes, adverse events and QoL
Ghodsi M et al., 2012 [26]	Iran	Both	allo HSC	35—55 million cells	IV	28/28	34.97/32.08	5.13/5.07	12	FBG, HbA1C, fasting C-peptide, insulin free period
Zang L et al., 2022 [18]	China	T2DM	allo UC MSC	1 × 10 million cells/kg	IV	45/46	50/50.45	11.44/11.70	12	FBG, reduction in insulin requirement, insulin free period, HbA1C, fasting C-peptide
Wu Z et al., 2022 [13]	China	T1DM	allo UC Mesenchymal stromal cell + auto BM MNC	UC-MSC: 1.1 million cells/kg + autologous BM-MNC: 106.8 million cells/kg	IA	21/21	33.5/36	16/15	96	FBG, incidence of end-organ dysfunction, HbA1C, fasting C-peptide, insulin dose requirement, hypoglycemic episodes and adverse events
Mirzaei M et al., 2021 [29]	Iran	NM	auto MSC	50—60 million cells	corpus cavernosum of penis	10/10	63.8/65.6	10.4/10.1	6	FBG, HbA1C
Estrada EJ et al., 2019 [19]	Argentina	T2DM	auto BM MNC	NM	IA	13/10	59/59	NM/NM	12	FBG, C-peptide, insulin requirement, HbA1C
Sood V et al., 2017 [20]	India	T2DM	auto BM MNC	6.88 ± 2.30 × 10^8	IA & IV	21/7	53.85/55.7	15.58/19.6	6	Fasting C-peptide, reduction in insulin dose, HbA1C
Packham DK et al., 2016 [21]	Australia	T2DM	allo MPC	150—300 million cells	IV	20/20	67.65/74.8	NM/NM	3	HbA1C & adverse events
Hu J et al., 2016 [22]	China	T2DM	allo WJ MSC	1 million cells/kg	IV	31/30	52.43/53.21	8.93/8.3	36	HbA1C, FBG, fasting C-peptide, insulin requirement, incidence of diabetic complications
Cai J et al., 2016 [16]	China	T1DM	allo UC Mesenchymal stromal cell + auto BM MNC	UC-MSC: 1.1 million cells kg + autologous BM-MNC: 106.8 million cells/kg	IA	21/21	18.29/20.38	9.2/7	12	HbA1C, FBG, daily insulin requirements, C-peptide, AUC C-peptide, adverse events and QoL
Esfahani EN et al., 2015 [27]	Iran	Both	allo HSC	35—55 million cells	IV	56/56	34.97/32.08	5.13/5.07	36	Incidence of retinopathy, neuropathy, ischemic heart disease, nephropathy and adverse events
Bhansali A et al., 2014 [23]	India	T2DM	auto BM MNC	3.2 (2.4—5.6) million cells	IA & IV	13/13	51/54	12/20	12	HbA1C, insulin requirement, stimulated C-peptide
Haller MJ et al., 2013 [14]	USA	T1DM	UC blood	NM	IV	10/5	7.2/6.6	NM/NM	12	HbA1C, peak C-peptide, AUC C-peptide, insulin dose, safety
Hu J et al., 2013 [15]	China	T1DM	allo WJ MSC	2.6 ± 1.2 × 10^7 (1.5—3.2 × 10^7)	IV	15/14	17.6/18.2	NM/NM	21	FBG, HbA1c, C-peptide and C-peptide/glucose ratio. insulin free period, adverse events
Huang P et al., 2005 [28]	China	Both	auto BM MNC	3 X 10^9	IM	14/14	71.1/70.9	12.9/11.6	3	FBG
Bhansali S et al., 2016 [24]	India	T2DM	I: auto BM MSC II: BM MNC	1 × 10^9/kg & 1 × 10^9	IA	20/10	47.5/53.5	14/14	12	HbA1c, insulin requirement, fasting c-peptide & AUC c-peptide
Skyler JS et al., 2015 [25]	USA	T2DM	allo MPC	0.3—2 million cells/kg	IV	45/16	56.7/58.7	10.2/9.8	3	HbA1c, hypoglycemic episodes, fasting plasma insulin, C-peptide
Carlsson PO et al., 2015 [11]	Sweden	T1DM	auto BM Mesenchymal stromal cell	2.1–3.6 million cells	IV	11/11	24/27	NM/NM	12	HbA1c, insulin dose requirement, C-peptide

Abbreviations: BM-HSCs bone marrow hematopoietic stem cells, BM-MNCs bone marrow mononuclear cells, DM diabetes mellitus, FBG Fasting Blood Glucose, IA Intraarterial, IM Intramuscular, IV Intravenous, MPC Mesenchymal Precursor Cells, MSCs mesenchymal stem cells, NM not mentioned, PB-MNC Peripheral blood mononuclear cells, T1DM type 1 diabetes mellitus, T2DM type 2 diabetes mellitus, UC MSC Umbilical cord mesenchymal stem cells, WJ MSC Wharton's jelly mesenchymal stem cells

Risk of bias

Risk of bias assessment using the ROB2 tool identified that most of the studies, 15 (75%), included in the review had an overall high risk of bias. This result is influenced by the randomization domain, wherein 12 (60%) trials had some concerns, and 4 (20%) had serious concerns with random sequence generation and/or allocation concealment. Other domains, including deviations from intended interventions, missing outcome data and outcome measurement, observed low risk of bias predominantly in 11 (55%), 13 (65%), and 15 studies, respectively (Fig. 2 and Appendix C).

Fig. 2.

Risk of bias assessment for studies included in the systematic review as per ROB2 tool

Efficacy and safety parameters of stem cell therapy in type 1 DM

Critical outcomes

Insulin-free period

A narrative synthesis of three studies reporting insulin-free period showed that three out of 37 patients in the intervention arm became insulin-free during the course of the study, while none of the patients in the control group became insulin-free [10, 15, 26]. [Appendix-E, supplementary material].

Quality of life

Meta-analysis of two studies reporting quality of life (QOL) using the SF-36 scale showed that QOL scores increased by three percentages among stem cell recipients compared to the control arm during 12 months of follow-up and this difference this was not statistically significant [MD: 3.15% (95% CI: −0.80 to 7.10); 2 trials, n = 63 participants; I² = 0%; GRADE: very low certainty]. [Fig. 3 and Appendix-E, supplementary material].

Fig. 3.

Meta-analysis of quality of life scores among type 1 diabetes mellitus

Hypoglycemic episodes

Meta-analysis of three studies reporting hypoglycemic episodes reported a pooled risk ratio of [RR: 0.90 (95% CI: 0.56 to 1.45); 3 trials, 68 participants; I² – 0%; GRADE: very low], where a 10% risk reduction of hypoglycemic episodes is observed among stem cell recipients compared to those patients receiving usual diabetic care, although this was not statistically significant [12, 13, 22, 25] [Fig. 4, and Appendix-E, supplementary material].

Fig. 4.

Meta-analysis of hypoglycemic episodes among patients with type 1 diabetes mellitus

Serious adverse events

Information on serious adverse events was available from seven studies, of which five reported nil serious adverse events following stem cell administration during the varying follow-up period of 3 to 36 months [10, 11, 14, 15, 17, 27, 28]. Carlsson PO et al., 2023 reported one serious adverse event (SAE) (i.e.) pregnancy, which was deemed unrelated to the investigational product by the study team [10]. It is important to note that Esfahani et al. used fetal hematopoietic stem cells as an intervention and observed a case of transitional meningioma in the intervention arm, and considered it as SAE related to stem cell infusion [27]. [Appendix-E, supplementary material].

Important outcomes

Other important outcomes among type 1 DM patients considered for meta-analysis include insulin requirement (U/Kg/day), fasting blood glucose, HbA1c and fasting c-peptide levels (Table 2). Pooled estimates of insulin requirement at 6 and 12 months did not show a statistically significant reduction among the stem cell group compared to the control with an MD of −0.07 (95% CI: −0.34 to 0.20); 2 trials, n = 57 participants; I² – 84.28%; GRADE: very low and −0.03 (95% CI: −0.25 to 0.17); 4 trials, n = 96 participants; I² – 96.42%; GRADE: very low respectively. Similarly, there was no significant reduction in fasting blood glucose levels observed among patients who received stem cell intervention. On the other hand, a meta-analysis of HbA1c percentage observed a significant reduction in HbA1c levels following stem cell therapy only at 6 months, with a pooled mean difference of −0.57 (95% CI: −1.06 to −0.09); 3 trials, n = 87 participants; I² = 40.62%; GRADE: low certainty. In line with this, it was also observed that stem cell recipients had higher fasting C-peptide levels of 0.08 nmol/L compared to control patients at 6 months in type 1 DM patients (Table 2). Further, subgroup analysis showed the possibility of achieving glycaemic control through infusion of a combination of mesenchymal stromal cells and bone marrow mononuclear cells, with a statistically significant improvement in HbA1C and C-peptide levels at 6 months [Fig S5b and S7b].

Efficacy and safety parameters of stem cell therapy in type 2 DM

Insulin requirement (U/day)

A significant reduction in insulin requirement was observed in patients following stem cell therapy at 6, 12 and 24 months of follow-up with a low certainty of evidence. Pooled mean difference of insulin requirement in IU/day was −14.42 (95% CI: −24.25 to −4.59); 6 trials, n = 167 participants; I² = 91.64%; GRADE: low certainty, −17.79 (95% CI: −26.39 to −9.18); 6 trials, n = 212 participants; I² = 70.96%; GRADE: low certainty; and −35.73 (95% CI: −40.82 to −30.64); 1 trial, n = 61 participants; I² = NA; GRADE: very low certaintyat 6, 12 and 24 months respectively, with an observable pattern of a steady reduction in insulin dose over the time points (Table 2, Fig. 5, S9a, S9b, S10, S11 and Appendix-E, supplementary material).

Fig. 5.

Meta-analysis of insulin requirement among type 2 DM at 12 months

HbA1C percentage

Pooled analysis of HbA1C percentage reduction in type 2 DM patients at 6, 12 and 24 months showed a pooled mean difference of −0.08% (95% CI: −0.53 to 0.36);, 8 studies, n = 207 participants; I² = 93.2%; GRADE: very low certainty; −0.20% [95% CI: −0.72 to 0.32]; 8 studies, n = 252 participants; I² = 77.03%; GRADE: very low certainty and −1.60% (95% CI: −2.16 to −1.04); 1 study, n = 61 participants; I² = NA; GRADE: low certainty, respectively with an observable pattern of steady improvement in HbA1C over the time points [Table 3, Fig. 6, S15a, S15b, S16, S17 and and Appendix-E, supplementary material].

Table 3.

Summary of findings of efficacy parameters of stem cell therapy in patients with diabetes mellitus

Outcome(s)	Time points (in months)	No. of participants (No. of studies)	Pooled estimates (MD) (with 95% CI), I2	Anticipated absolute effects		Certainity of the evidence (GRADE)
				Risk with [usual care]	Risk difference [Stem cell therapy]
Type 1 Diabetes Mellitus
Quality of lifea	12	63 (2 RCTs)	3.15 (−0.80 to 7.10), 0%	-	MD 3.15↑ (0.80↓ to 7.10↑)	⨁◯◯◯a
Hypoglycemic episodesa	12	68 (3 RCTs)	RR: 0.90 (0.56 to 1.45), 0%	44%	Risk: 44↓ per 1000 (194↓ to 199↑)	⨁◯◯◯a
Insulin requirementb (U/Kg/day)	6	57 (2 RCTs)	−0.07 (−0.34 to 0.20), 84.28%	-	MD 0.07↓ (0.34↓ to 0.20↑)	⨁◯◯◯a
	12	96 (4 RCTs)	−0.03 (−0.25 to 0.17), 96.42%	-	MD 0.03↓ (0.25↓ to 0.17 ↑)	⨁◯◯◯a
	24	-	-	-	-	-
Fasting blood glucoseb (mg/dl)	6	90 (2 RCTs)	−22.49 (−63.66 to 18.68), 46.88%	-	MD 22.49↓ (63.66↓ to 18.68 ↑)	⨁◯◯◯a
	12	107 (4 RCTs)	−1.27 (−41.07 to 38.52), 78.05%	-	MD 1.27↓ (41.07↓ to 38.52↑)	⨁◯◯◯a
	24	-	-	-
HbA1Cb (percentage)	6	87 (3 RCTs)	−0.57 (−1.06 to −0.09), 40.62%	-	MD 0.57↓ (1.06↓ to 0.09↓)	⨁⨁◯◯b
	12	140 (6 RCTs)	−0.52 (−1.09 to 0.04), 88.08%	-	MD 0.52↓ (1.09↓ to 0.04↑)	⨁◯◯◯a
	24	-	-	-	-
Fasting C-peptideb (nmol/L)	6	72 (2 RCTs)	0.08 (0.07 to 0.09), 0%	-	MD 0.08↑ (0.07↑ to 0.09↑)	⨁⨁◯◯b
	12	111 (4 RCTs)	0.11 (−0.12 to 0.34), 99.79%	-	MD 0.11↑ (0.12↓ to 0.34↑)	⨁◯◯◯a
	24	-	-	-
Type 2 Diabetes Mellitus
Insulin requirementa (U/day)	6	167 (6 RCTs)	−14.42 (−24.25 to −4.59), 91.64%	-	MD 14.42↓ (24.25↓to 4.59↓)	⨁⨁◯◯b
	12	212 (6 RCTs)	−17.79 (−26.39 to −9.18), 70.96%	-	MD 17.79↓ (26.39↓ to 9.18↓)	⨁⨁◯◯b
	24	61 (1 RCT)	−35.73 (−40.82 to −30.64), NA	-	MD 35.73↓ (40.82↓to 30.64↓)	⨁◯◯◯a
Fasting blood glucoseb (mg/dl)	6	179 (7 RCTs)	−8.34 (−29.47 to 12.80), 91.26%	-	MD 8.34↓ (29.47↓to 12.80↑)	⨁◯◯◯a
	12	179 (7 RCTs)	−5.81 (−22.31 to 10.69), 87.59%	-	MD 5.81↓ (22.31↓to 10.69↑)	⨁◯◯◯a
	24	61 (1 RCT)	−11.65 (−19.53 to −3.77), NA	-	MD 11.65↓ (19.53↓to 3.77↓)	⨁⨁◯◯b
HbA1Ca (percentage)	6	207 (8 RCTs)	−0.08 (−0.53 to 0.36), 93.15%	-	MD 0.08↓ (0.53↓to 0.36↑)	⨁◯◯◯a
	12	252 (8 RCTs)	−0.20 (−0.72 to 0.32), 77.03%	-	MD 0.20↓ (0.72↓to 0.32↑)	⨁◯◯◯a
	24	61 (1 RCT)	−1.60 (−2.16 to −1.04), NA	-	MD 1.60↓ (2.16↓to 1.04↓)	⨁⨁◯◯b
Fasting C-peptideb (nmol/L)	6	167 (6 RCTs)	0.18 (0.04 to 0.32), 81.41%	-	MD 0.18↑ (0.04↑ to 0.32↑)	⨁⨁◯◯b
	12	212 (6 RCTs)	0.17 (−0.01 to 0.36), 83.76%	-	MD 0.17↑ (0.01↓ to 0.36↑)	⨁◯◯◯a
	24	61 (1 RCT)	0.25 (0.19 to 0.31), NA	-	MD 0.25↑ (0.19↑ to 0.31↑)	⨁⨁◯◯b

^aCertainty of evidence downgraded for risk of bias of included studies by two levels, imprecision by two levels and heterogeneity by 1 level

^bCertainty of evidence downgraded for risk of bias of included studies by two levels

Fig. 6.

Forest plot of HbA1C percentage among type 2 DM at 12 months

Serious adverse events

Information on serious adverse events was qualitatively synthesized from seven studies, where no. of SAEs ranged between zero and eleven in the stem cell group compared to the control group with SAEs of 0–2 events during follow-up of 3–18 months [17, 21, 24, 25, 28]. (Appendix-E, supplementary material).

Important outcomes

Other important outcomes considered for pooled analysis among type 2 DM patients include fasting blood glucose and fasting C-peptide levels. Pooled analysis of fasting blood glucose from 7 RCTs each at 6- and 12-months following stem cell therapy did not show a significant reduction in glucose levels and was graded as a very low certainty of evidence. On the contrary, stem cell therapy was observed to have increased fasting C-peptide levels significantly, with pooled mean difference estimates of 0.18 (95% CI: 0.04–0.32); 6 studies, n = 167 participants; I² = 81.41%; GRADE: low certainty and 0.25 (95% CI: 0.19–0.31); 1 study, n = 61 participants; I² = NA; GRADE: very low certainty at 6 and 24 months [Table 3, Fig S12a, S12b, S13a, S13b, S14a, S18a, S18b and S20].

Discussion

This systematic review summarized the efficacy of stem cell therapy in patients with DM. Among T1DM patients, stem cell therapy did not result in significant improvement in quality of life or a reduction in hypoglycemic episodes. While the combination therapy of mesenchymal stem cells and bone marrow mononuclear cells infusion showed significant improvement in c-peptide and HbA1c at 6 months, the effect size was noted to be trivial. Most importantly, a case of transitional meningioma was reported eight months followed by the infusion of fetal stem cells in a type 1 DM patient. With regard to T2DM, stem cell infusion was found to result in a steady significant reduction in insulin dose requirement during 6—24 months of follow-up, with no SAE related to the intervention being reported.

To our knowledge, this systematic review provides a comprehensive summary of the effect of diverse stem cell types on DM patients, based on RCTs, evaluating critical efficacy and safety outcomes. Previously published systematic reviews have focused on a narrower subset of studies, either by including only a specific type of stem cell therapy (e.g., mesenchymal stromal cells) or restricting their analysis to either type 1 or 2 diabetes [4, 30, 31]. Our review is more comprehensive in scope, as it includes a broader range of stem cell therapies (e.g., mesenchymal stromal cells, bone marrow mononuclear cells) and covers both Type 1 and Type 2 diabetes. This could lead to differences in the pooled effect estimates as we incorporate a more diverse set of interventions and patient populations. In addition, the current review examines stem cell efficacy at multiple timepoints between 6–24 months, thereby assessing short- and long-term effects. This more detailed and time-segmented analysis could have led to discrepancies in efficacy estimates of stem cell therapy compared to what was reported in the previous systematic reviews [4, 30, 31].

We observed that there were no systematic reviews reporting data on critical outcomes for type 1 DM, such as hypoglycemic episodes and quality of life, which thus limited comparisons on these outcomes from published literature. With regard to insulin free period, our review summarized that only 8% of intervention arm participants belonging to three RCTs became insulin free. This estimate appears lesser compared to the recent review by Madani et al., which reported 26.5% of type 1 DM patients becoming insulin free [3]. This could be an overestimate by Madani et al. due to the inclusion of a range of study designs, such as randomized, non-randomized and descriptive studies in their review [3]. Our meta-analysis showed statistically significant improvement, although the effect size is trivial, in HbA1c and c-peptide levels among type 1 DM patients at six months. However, statistical significance for HbA1c and c-peptide levels was lost in subsequent time points at 12 and 24 months. These results are comparable with Sun et al. and Madani et al., systematic reviews that were based on RCTs and NRCTs, showing a trivial improvement in HbA1c and c-peptide levels in a short-term period of 6–12 months [3, 30]. Of note, combined stem cell administration with mesenchymal stromal cells and bone marrow mononuclear cells was observed to improve C-peptide and HbA1c levels at 6 and 12 months in type 1 DM patients, as is observed in Madani et al., which summarized the effects of combined administration of hematopoietic stem cells and mesenchymal stem cells [3].

Our meta-analysis showed that stem cell therapy resulted in a steady decrease in insulin requirement among type 2 DM patients by 14–35 units from their baseline insulin requirement, ranging between 40–60 units over 6–24 months post-intervention. Ranjbaran et al. in their meta-analysis based on three studies, reported an insignificant reduction in insulin dose by 1.45 units following mesenchymal stem cell therapy [2]. This difference between the estimates can be due to the restriction of mesenchymal stem cell therapy alone in Ranjbaran et al. compared to our meta-analysis, which considered various stem cell therapies together [2]. Interestingly, subgroup analysis in our meta-analysis showed that bone marrow mononuclear stem cell therapy has shown significant insulin dose reduction with a relatively larger effect size than mesenchymal stem cell therapy. Although stem cell therapy resulted in a significant insulin dose requirement, it is important to note that overall certainty of evidence ranges between low and very low. Although stem cell therapy significantly reduced insulin requirements (by ~ 15–35 IU/day), this result should be interpreted in context. For a typical Type 2 patient (baseline ~ 40–60 IU/day), a 30 IU/day reduction represents a larger proportional change, suggesting improved β-cell function or insulin sensitivity. However, because glycemic control (HbA1c) did not improve to the same degree, the clinical impact of insulin sparing may be modest. While promising, we have clarified that these findings need cautious interpretation regarding patient outcomes.

Also, we acknowledge significant heterogeneity in insulin-requirement outcomes. This heterogeneity likely reflects clinical differences among studies, including patient factors (age, baseline insulin dose, duration of diabetes) and variations in stem cell interventions (cell type, source tissue, dose, and culture/preparation methods). For example, patients with longer-standing diabetes or older age tended to show smaller insulin-dose reductions, suggesting these variables influence response. A similar review has noted that heterogeneity can arise from non-uniform patient baselines and treatment protocols [32].

In case of HbA1c reduction among T2DM patients, our meta-analysis at six and twelve months did not show a statistically significant HbA1c, which is in contrast with previous meta-analyses that observed significant HbA1c reduction with mesenchymal stem cell therapy [2, 4]. Further, subgroup analysis in our meta-analysis showed mesenchymal stem cell therapy has shown a statistically significant reduction in HbA1c levels by 0.6–1.1 per cent, comparable to a recent review by He et al. that reported pooled HbA1c levels of 0.87 per cent [4].

While our review provides insights into short- and medium-term outcomes (up to 24 months), future studies should focus on the long-term durability of stem cell therapy effects in diabetes, including sustained HbA1C reduction, insulin independence, and the prevention of diabetes-related complications over periods exceeding 5–10 years.

Our study has many strengths, the first one being the inclusion of randomized controlled trials, which strengthens the reliability of the findings due to the high level of evidence these study designs provide. Additionally, the broad search strategy, which covered a wide range of studies from different geographical locations, enhances the generalizability of the results. Importantly, using four major databases (PubMed, Cochrane Library, Scopus, and Embase) ensured a comprehensive search, minimizing the risk of missing relevant studies. Another key strength is using the GRADE system to assess the certainty of the evidence. This transparent grading approach allows for an objective evaluation of the robustness of the outcomes, highlighting areas with strong evidence and identifying where limitations may exist.

However, there are a few limitations. Our review included diverse populations (e.g. pediatric and adult patients with mean ages from ~ 10 to ~ 60 years), but many trials excluded patients with advanced complications or severe comorbidities. This may limit generalizability to those subgroups. We noted high heterogeneity between studies, as evidenced by the wider confidence intervals in several meta-analyses, even after factoring in types of stem cells and route of administration of stem cells in subgroup analysis. This implies the possibility of clinical and methodological heterogeneity between the studies due to varying patient characteristics and intervention delivery procedures. Furthermore, many of the included studies had small sample sizes, limiting the generalizability of these findings. Most trials had short follow-up periods (median < 12 months; the longest was 96 months), which limits confidence in the long-term durability of effects. These findings emphasise the need for longer-term RCTs to confirm sustained efficacy. The risk of bias in individual studies, especially in allocation concealment and blinding, was another limitation, as these methodological flaws may affect the validity of the results. Lastly, the evidence quality was often downgraded due to imprecision, indicating the need for evaluating stem cells in a larger trial context.

This study raises several potential controversies, particularly surrounding the interpretation of the findings and the limitations of the current evidence base. One major controversy lies in the varying types of stem cell interventions used across the included studies, which makes it difficult to compare results directly. While the meta-analysis attempts to pool data, the heterogeneity in the type, source, and application methods of stem cells introduces complexity and may result in difficulty in interpreting the efficacy of these treatments.

Additionally, the study highlights the short-term benefits of stem cell interventions, but the lack of long-term follow-up data remains a point of contention. Critics may argue that without robust long-term data, the clinical utility of stem cell therapy cannot be fully assessed, especially considering the potential risks or unknown long-term effects. This uncertainty may lead to hesitancy in adopting these therapies into clinical practice.

Finally, the ethical considerations surrounding stem cell therapy may add another layer of controversy, particularly in terms of sourcing stem cells and ensuring equitable access to these treatments [33]. Some stakeholders may raise concerns about the commercialization and ethical boundaries of stem cell research, particularly in cases where treatments are still considered experimental [34].

The lack of standardization in stem cell sourcing, preparation, dosage, and administration routes across studies limits the ability to draw definitive conclusions [35]. Future research should aim to develop and validate standardized protocols to enhance comparability across trials and improve clinical translation.

Conclusion

This systematic review summarized that stem cell therapy did not achieve an insulin-free period or a better quality of life in type 1 diabetes mellitus patients. In type 2 diabetes patients, stem cell therapy observed a reduction in insulin requirement for these patients by 14–36 units in 6–24 months of follow-up, although significant glycemic control was not achieved. Further evaluation of the efficacy of stem cells from well-conducted large randomized trials with longer follow-ups is required to consider their therapeutic potential in diabetes mellitus.

Abbreviations

BM-HSCs

Bone Marrow Hematopoietic Stem Cells

BM-MNCs

Bone Marrow Mononuclear Cells

CI

Confidence Interval

DM

Diabetes Mellitus

FBG

Fasting Blood Glucose

GRADE

Grading of Recommendations Assessment, Development, and Evaluation

HbA1c

Hemoglobin A1c

I²

Statistical Measure of Heterogeneity

IA

Intraarterial

ICMR-INDIAB

Indian Council of Medical Research–India Diabetes

IM

Intramuscular

IV

Intravenous

MD

Mean Difference

mg/dL

Milligrams per Deciliter

MODY

Maturity-Onset Diabetes of the Young

MPC

Mesenchymal Precursor Cells

MSC

Mesenchymal Stem Cells

NA

Not Applicable

NM

Not Mentioned

nmol/L

Nanomoles per Liter

NRCTs

Non-Randomized Controlled Trials

PB-MNC

Peripheral Blood Mononuclear Cells

PICO

Population, Intervention, Comparison, Outcome

PROSPERO

International Prospective Register of Systematic Reviews

QOL

Quality of Life

RCTs

Randomized Controlled Trials

ROB2

Risk of Bias 2 Tool

RR

Risk Ratio

SAE

Serious Adverse Events

SF-36

Short Form-36 (a health survey scale)

T1DM

Type 1 Diabetes Mellitus

T2DM

Type 2 Diabetes Mellitus

U/Kg/day

Units per Kilogram per Day

UC MSC

Umbilical Cord Mesenchymal Stem Cells

WJ MSC

Wharton’s Jelly Mesenchymal Stem Cells

Authors’ contributions

Review question for this systematic review was conceived at DHR Secretariat for developing clinical practice guidelines. MS and RH wrote the review protocol, conducted search in databases, screened articles and completed data extraction. MS conducted the data analysis. MS and RH wrote the first draft of the manuscript. BBS provided critical inputs in designing the review, database search, screening, data analysis and manuscript preparation. All authors had full access to review data and had the final responsibility for the decision to submit for publication.

Funding

There is no funding involved in conduct of this review.

Data availability

Data extraction sheet will be made available upon the request to the corresponding authors.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.