Effect of carnosine or beta-alanine supplementation therapy for prediabetes or type 2 diabetes mellitus: a systematic review and meta-analysis of randomized controlled trials

Abstract

Background

Carnosine and beta-alanine (β-alanine) have shown potential in the management of chronic conditions, including metabolic disorders. However, their therapeutic efficacy in individuals with type 2 diabetes mellitus (T2DM) and prediabetes remains inconclusive due to heterogeneity in clinical trial results and limited synthesis of human evidence.

Objective

This systematic review and meta-analysis aim to evaluate the effects of carnosine and β-alanine supplementation on patients with prediabetes and T2DM.

Methods

We searched PubMed, Cochrane Library, Web of Science, and Embase from inception to 9 October 2024 for randomized controlled trials that compared carnosine or β-alanine supplementation to placebo in prediabetic and diabetic populations. The quality of evidence was appraised using the Jadad scale, and the risk of bias was assessed using the Cochrane Risk of Bias tool. Data were analyzed using RevMan and Stata, employing fixed-effects models and I-V methods.

Results

Eight trials met the inclusion criteria, totaling 377 participants. Our analysis indicated that supplementation significantly reduced fasting blood glucose (FBG) (SMD: -0.53; 95% CI: -0.75 to -0.31; p < 0.00001) and hemoglobin A1c (HbA1c) levels (SMD:-0.36; 95% CI:-0.59 to -0.12; p = 0.003) compared to placebo. No significant effects were observed on body mass index (BMI), fasting insulin. low-density lipoprotein cholesterol (LDL-c) or high-density lipoprotein cholesterol (HDL-c), but a lowering effect was observed in total cholesterol (TC). Notably, Homeostasis Model Assessment of Beta-cell Function (HOMA-β) values were improved, suggesting enhanced β-cell function, while changes in homeostasis model assessment of insulin resistance (HOMA-IR) did not reach statistical significance.

Conclusions

Carnosine and β-alanine supplementation show potential as adjunct therapies for improving FBG, HbA1c and HOMA-β in prediabetes and T2DM. Further rigorous studies are warranted to establish optimal dosage, treatment duration, and long-term efficacy in clinical practice.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12902-025-02016-w.

Introduction

Type 2 diabetes mellitus (T2DM) has become a critical global public health challenge, with projections indicating that its prevalence will reach 12.2% by 2045, affecting approximately 783.2 million individuals worldwide [1]. T2DM accounts for over one million deaths annually, ranking ninth in global mortality [2]. A hallmark of T2DM is impaired glycemic control accompanied by insulin resistance, often initially presenting as impaired fasting glucose or glucose intolerance, commonly referred to as prediabetes [3]. Previous research showed that T2DM is largely preventable and potentially reversible with early detection and management. However, evidence consistently indicates a rising global incidence of diabetes [4]. Adopting and maintaining novel lifestyle modifications can significantly reduce the incidence of T2DM [5]. However, implementing and maintaining such lifestyle changes over the long term remains challenging. Therefore, searching for new long-term and effective therapies has become a priority to improve glycemic control and help prevent or delay disease progression.

In recent years, natural or herbal dietary supplements have drawn growing interest as potential interventions for diabetes prevention and management [6]. For instance, okra has been shown to reduce fasting blood glucose (FBG), though it has no effect on HbA1c [7]. Carnosine, a naturally occurring histidine-containing dipeptide, is abundant in skeletal muscle and present in lesser quantities in other excitable tissues [8]. Carnosine has demonstrated potential in the treatment of various chronic diseases [9]. Previous studies have indicated that carnosine can reverse the glucolipotoxic inhibition of insulin secretion and enhance glucose uptake in skeletal muscle cells [10]. This effect is significant for improving glycemic control and aiding in the prevention or delay of disease progression. β-alanine, as a precursor to carnosine, can elevate tissue carnosine levels, potentially enhancing its effects on glucose metabolism [11]. The synergistic action of these two compounds offers new potential for adjunctive therapy in diabetes, particularly in reducing oxidative stress and potentially improving insulin sensitivity [12–14]. However, despite their biochemical potential, the therapeutic efficacy of carnosine and β-alanine in patients with diabetes and prediabetes remains inconclusive, with inconsistent results from existing randomized controlled trials (RCTs) [15, 16].

Previous studies involving carnosine and β-alanine supplementation have shown mixed outcomes, in part due to the inclusion of animal models, which may limit the applicability of findings to human physiology [8]. While animal studies provide valuable mechanistic insights, species differences necessitate a focused evaluation of human data to generate robust clinical recommendations.

To address these gaps, this study conducts a systematic review and meta-analysis of human RCTs to evaluate the therapeutic effects of carnosine and β-alanine on glycemic control in prediabetes and T2DM, aiming to provide an evidence-based foundation for their use in clinical practice. By excluding non-human studies, this analysis seeks to clarify the clinical utility of these supplements and contribute to the development of evidence-based interventions in diabetes management.

Methods

Our meta-analysis was carried out and written based on the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines [17] and was registered at PROSPERO (ID: registration number: CRD42024610407).

Selection criteria

Studies with the following criteria were selected for meta-analysis: (1) enrolled persons with prediabetes (fasting blood glucose (FBG): 6.1–6.9 mmol/L, 2 h oral glucose tolerance test (2hOGTT): 7.8–11.0 mmol/L, hemoglobin A1c (HbA1c): 5.7% 6.4%) [18] or T2DM (FBG ≥ 7.0mmol/L, HbA1c ≥ 6.5%, or random blood glucose ≥ 11.1mmol/L) [19]; (2) compared carnosine or β-alanine supplementation therapy and placebo therapy; (3) reported outcome of interest; (4) they were RCTs.

Information sources and search strategy

We searched four databases (PubMed, Cochrane Library, Web of Science, and Embase) from inception to 17 May 2024, using search terms “Type 2 diabetes mellitus”, “prediabetes”, “carnosine”and “β-alanine”. No language restrictions were applied. All the articles mentioned are accessible in full text. The detailed search strategy is provided in the Table S1. All the references were managed by Endnote X9.

Study selection

All titles and abstracts retrieved by the systematic search were independently screened for potential eligibility by two investigators (NL and XY). Then they screened the full text of potentially relevant trials. Disagreements regarding eligibility were resolved through discussion and negotiation with corresponding author (YW).

Data extraction

Two investigators (NL and JL) independently extracted relevant data from the included studies, including details such as: (1) study information: first author, publication year, country, sample size; (2) patient characteristics: mean ages, sex, BMI, type of patients, diabetes duration, intervention and follow up; (3) outcome data. For accuracy, a third reviewer (YW) checked the extracted data for any discrepancies. Disagreements among investigators were resolved through collaborative discussions to reach a consensus.

Quality of evidence appraisal and and risk of bias

Given that only RCTs were included in our study, we employed the Jadad scale to assess the quality of evidence, focusing on four key aspects: (1) Generation of Random Sequences; (2) Allocation Concealment; (3) Blinding; (4) Withdrawal. Evidence quality for each outcome was scored from 0 (inappropriate) to 2 (appropriate), with 1 indicating unclear evidence [20]. Two investigators (NL and XY) independently assessed the risk of bias using the Cochrane Risk of Bias tool, evaluating each study across seven domains: random sequence generation and allocation concealment (selection bias), blinding of study participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias. They resolved any disagreements by discussion and consensus or by consulting a third investigator (YW) [21]. Studies with more than two high-risk domains were rated as moderate risk, and those with more than four as high risk. Additionally, two investigators evaluated evidence quality for primary and secondary outcomes using the Grading of Recommendation, Assessment, Development and Evaluation (GRADE) [22].

Data analysis

We carried out the statistical analysis using RevMan (version 5.4), Stata (version 15.0) and Stata (version 18.0). The inverse-variance (I-V) method was used to calculate the standard mean difference (SMD) and 95% confidence intervals (CIs). A fixed-effects model was selected for continuous outcomes. In this study, the p-value < 0.05 was considered statistical significance. For assessing heterogeneity, we used the I [2] statistics, with I [2] was classified as low (0–50%), moderate (51–75%) or high (> 75%) for assess heterogeneity [23]. We performed subgroup analysis to search heterogeneity sources and explore the optimal treatment dosage and duration. We conducted a sensitivity analysis to assess the robustness of the study. The Begg’s test and Egger’s test were used for evaluation of publication bias [24]. To assess the potential impact of publication bias, a sensitivity analysis was performed using the Trim-and-Fill method to supplement presumed unpublished studies and to adjust the combined effect size.

Trial sequential analysis

We conducted trial sequential analysis (TSA version 0.9.5.10 Beta) to avert type I error rates by estimating the required information size and adjusting the statistical significance threshold. We used a two-sided trial sequential analysis to control a type I error at 5% and a power of 80%.

Results

Search and selection of studies

Our initial search yielded a total of 2363 articles, of which only 8 studies met the inclusion criteria after comprehensive screening [15, 16, 25–30]. Figure 1 shows the detailed study selection flow diagram. Seven RCTs included 377 patients (187 undergoing carnosine supplementation therapy and 190 undergoing placebo therapy). Table 1 outlines the primary characteristics of each included trial. From Table 1, we can presents that all studies had a high quality of evidence as evaluated by Jadad Scale.

Fig. 1.

Search strategy and final included and excluded studies

Table 1.

Characteristics of each trial incorporated

Study	Year	Country	Number, n	Male, n	Mean age, years	Type of patients	Diabetes duration,years	BMI, kg/m2	Intervention	Follow up	Jadad score
Liu	2015	France	26/26	NR	NR	Prediabetes	NR	31.4±3.1/31.6±4.5	Carnosine	4 months	5
Nealon	2016	Australia	6/4	5/4	62±4.6/66±6.4	T2DM	NR	30.92±2.53/35.19±8.50	β--alanine	28 days	5
Houjeghani	2017	Iran	22/22	10/12	43.0±7.6/40.4±5.1	T2DM	4.5±2.0/4.2±2.2	29.1±5.3/28.3±4.6	Carnosine	12 weeks	6
Elbarbary	2017	Egypt	43/42	20/23	12.4±3.4/13.3±3.8	T2DM and diabetic nephropathy	7.3±2.4/6.7±2.1	NR	Carnosine	12 weeks	7
Siriwattanasit	2021	Thailand	20/20	11/4	55.6±4.8/57.0±6.9	T2DM and diabetic nephropathy	10.5±6.5/13.0±8.8	30.3±5.6/28.4±5.2	Carnosine	12 weeks	6
Saadati	2023	Switzerland	20/23	14/16	53.7±11.4/52±19.7	Prediabetes or T2DM	NR	30.54±4.74/ 28.29±3.66	Carnosine	14 weeks	6
Hariharan	2024	Australia	20/23	14/16	53.8/52	Prediabetes or T2DM	NR	29.8±4.9/28.5±3.7	Carnosine	14 weeks	6
Mahitab	2025	Egypt	30/30	7/7	70.33±9.38/70.50±9.43	T2DM and diabetic neuropathy	16.97±7.25/17.17±7.09	29.48±4.48/29.17±4.29	Carnosine	12 weeks	6

Abbreviation: BMI Body mass index., T2DM Type 2 diabetes mellitus

Values are all given as carnosine or beta-alanine supplementation therapy/control group

Outcomes with respect to carnosine or β-alanine supplementation as compared with placebo in each study are presented in Table 2.

Table 2.

Comparison of outcomes between carnosine or β-alanine supplementation therapy and placebo therapy

Study	FBG(mmol/L)	HbA1c(%)	BMI(kg/m2)	Fasting insulin(uU/mL)	HOMA-IR	HOMA-β	TC(mg/dL)	LDL-C(mg/dL)	HDL-C(mg/dL)
Liu	5.9(0.6)/6.2(0.8)	5.99(0.47)/6.12(0.5)	31.8(3.2)/31.9(4.7)	9.9(3.9)/9(4.5)	1.3(0.5)/1.3(0.6)	81.7(19.8)/68.4(18.4)	210(40)/220(40)	140(50)/145(30)	50(10)/50(20)
Nealon	7.14(1.63)/9.06(3.81)	NR	31.02(2.63)/36.8(9.57)	9.14(2.19)/13(6.44)	NR	NR	NR	NR	NR
Houjeghani	7.05(1.17)/7.72(1.74)	5.8(0.6)/6.1(0.8)	29(5.3)/28.3(4.4)	1.8(3.8)/1.6(3.7)	1.5(0.7)/1.4(0.7)	23.5(12.1)/20.2(12.5)	159.5(33.9)/151.5(30)	86.2(28.2)/77.5(24.5)	46.7(6.6)/51.3(18)
Elbarbary	7.0(1.93)/8.44(2.38)	7.4(1.3)/8.3(2.4)	NR	NR	NR	NR	169.1(31.2)/188.6(32.7)	NR	52.2(7.5)/41.6(6.9)
Siriwattanasit	8.44(2.64)/8.71(3.89)	7.48(1.3)/7.86	30.12(5.6)/28.18(5)	NR	NR	NR	147.6(35.1)/166.43(51.6)	91.5(35.2)/102.59(48.1)	46.89(9.2)/51.17(13.2)
Saadati	NR	NR	NR	NR	NR	NR	204.6(27.5)/199.6(32.9)	132.2(22.0)/117.6 (21.3)	42.54(8.89)/44.5(10.05)
Hariharan	10.6(4.1)/11.9(3)	6.7(1.1)/6.72(1.14)	30.2(4.8)/28.4(3.6)	9.1(4.6)/10.3(6.8)	2.7(1.6)/2.52(1.35)	140.1(217.8)/86.2(108.4)	NR	NR	NR
Mahitab	6.53(1.57)/8.60(2.58)	6.83(0.66)/7.37(0.95)	NR	NR	NR	NR	145.47(50.33)/155.27(44.31)	NR	NR

Values are all given as carnosine or beta-alanine supplementation therapy/control group

Abbreviation: FBG Fasting blood glucose, HbA1c Hemoglobin A1c, BMI Body mass index, HOMA-IR Homeostasis Model Assessment of Insulin Resistance, HOMA-β Homeostasis Model Assessment of β-cell Function, TC Total Cholesterol, LDL-c Low density lipoprotein cholesterol, HDL-C High density lipoprotein cholesterol

Quality of evidence appraisal and risk of bias

Risk-of-bias assessments are illustrated in Fig. 2. Study quality appraisal indicated that studies were of variable quality and that all trials are at a low risk of bias. Supplementary Fig. 1 offers GRADE summary findings for all outcomes.

Fig. 2.

(A) Risk of bias summary; (B) Risk of bias graph

Primary outcome

FBG

Seven [15, 16, 25–28, 30] trials reported the levels of FBG. A fixed-effects model demonstrated that carnosine or β-alanine supplementation significantly reduced FBG levels in individuals with T2DM compared to the placebo group (SMD: −0.53; 95% CI: −0.75 to −0.31; p < 0.00001; I² = 0%) (Fig. 3A).

Fig. 3.

Forest plots for (A) FBG, (B) HbA1c. Abbreviation: FBG Fasting blood glucose, HbA1c Hemoglobin A1c

HbA1c

Six [15, 16, 25, 26, 29, 30] trials were conducted to evaluate changes in the levels of HbA1c. A fixed effect model showed that carnosine or β-alanine supplementation therapy resulted in a more significant reduction in HbA1c levels SMD:−0.36; 95% CI:−0.59 to −0.12; p = 0.003; I² = 0%) than did the placebo treatment (Fig. 3B).

Secondary outcomes

BMI

Five [16, 25–27, 29] trials reported BMI. The pooled results showed that there was no significant difference in BMI between two groups (SMD:0.16; 95% CI:−0.13 to 0.45; p = 0.28; I² = 0%) (Fig. 4A).

Fig. 4.

Forest plots for (A) BMI, (B) Fasting insulin, (C) HOMA-IR, (D) HOMA-β. Abbreviation: BMI Body mass index, HOMA-IR Homeostasis Model Assessment of Insulin Resistance, HOMA-β Homeostasis Model Assessment of β-cell Function

Fasting insulin

Four [16, 25–27] trials evaluated fasting insulin levels. Fixed-effect modeling showed no statistically significant difference in fasting insulin between the intervention and control groups(SMD:0.07; 95% CI:−0.25 to 0.40; p = 0.66; I² = 0%) (Fig. 4B).

HOMA-IR and HOMA-β

Three [16, 25, 26] trials reported on HOMA-IR and HOMA-β. The fixed-effect model indicated no statistically significant reduction in HOMA-IR (SMD: 0.08; 95% CI: −0.25 to 0.41; p = 0.63; I² = 0%). However, pooled results revealed that HOMA-β values were elevated in the supplementation group compared to placebo (SMD: 0.43; 95% CI: 0.09–0.77; p = 0.01; I² = 0%) (Fig. 4C and D).

Lipid metabolism

The impact of carnosine or β-alanine supplementation on lipid metabolism was analyzed in four [16, 26, 28, 29] and six [15, 16, 26, 28–30] trials respectively. Fixed-effect model results indicated no significant reduction in low-density lipoprotein cholesterol (LDL-c), or high-density cholesterol (HDL-c) levels, though trends were observed. Specifically, the SMD for LDL-c, and HDL-c were 0.14 (95% CI: −0.16 to 0.44; p = 0.35; I² = 46%), and 0.24 (95% CI: −0.01 to 0.50; p = 0.06; I² = 88%), respectively (Fig. 5B and C). Notably, a lowering effect of carnosine or β-alanine supplementation therapy was observed in total cholesterol (TC) (SMD:−0.23; 95% CI:−0.45 to −0.01; p = 0.04; I² = 32%). (Fig. 5A). For HDL-C, our sensitivity analysis revealed that the study by Elbarbary et al. may be a source of heterogeneity (Figure S2). However, when we removing this study, the results showed no statistically significant reduction in HDL-C (SMD: −0.21; 95% CI: −0.51 to 0.08; p = 0.16; I² = 0%).

Fig. 5.

Forest plots for (A) TC, (B) LDL-C, (C) HDL-C. Abbreviation: TC Total Cholesterol, LDL-c Low density lipoprotein cholesterol, HDL-C High density lipoprotein cholesterol

Subgroup analysis

We conducted subgroup analysis on the HDL-C based on type of patients. However, we did not identify the sources of heterogeneity (Figure S3). Additionally, we conducted subgroup analyses of FBG, HbA1c, and HOMA-β in terms of dosage and treatment duration, in order to explore the optimal treatment dosage and duration. Interestingly, we found significant improvements in FBG and HbA1c with > 1000 mg/day for therapeutic doses, but with ≤ 1000 mg/day for HOMA-β. Similarly, for treatment duration, ≤ 12 weeks was associated with significant improvements in FBG and HbA1c, but > 12 weeks was associated with significant improvements in HOMA-β (Figures S5 and S6).

Publication bias

Based on Begg’s test, publication bias was not indicated in the included studies. However, our Egger’s test indicated the presence of some publication bias regarding fasting insulin and HDL-C (Table S2). Further investigation using the trim-and-fill test showed that this publishing bias did not affect the estimations (Figure S4).

Trial sequential analysis

Figure 6 shows that the Z-curve for FBG and HbA1c surpasses the boundary set by the trial sequence analysis, indicating that the accumulated data reaching the expected value, additional tests are not required to make a conclusive positive statement.

Fig. 6.

Calculation of optimum sample size. Red vertical line indicates optimum same size; blue line indicates Z-curve

Discussion

In this meta-analysis, we found that significant improvements favoring carnosine or β-alanine supplementation therapy were detected for FBG and HbA1c. However, no statistically significant effects were observed for HOMA-IR, BMI, fasting insulin, and lipid metabolism did not show statistical significance. Interestingly, HOMA-β values, a marker of β-cell function, were higher in the supplementation group, suggesting an enhancement in pancreatic β-cell responsiveness.

The improvement in FBG and HbA1c aligns with several proposed biochemical mechanisms by which carnosine and β-alanine may exert therapeutic effects. β-cell dysfunction is one of the key pathological features of T2DM and prediabetes, and β-cell damage is usually closely related to oxidative stress, inflammatory response and advanced glycation end-products (AGEs) accumulation. Carnosine is known for its antioxidant properties, which allow it to neutralize reactive oxygen species (ROS) and reduce oxidative stress [31–33]. This antioxidative capacity may protect pancreatic β-cells from damage, thereby supporting insulin secretion and responsiveness [34]. Furthermore, by inhibiting the formation of AGEs, carnosine can not only prevent protein cross-linking and protect beta cells from direct glycation damage, but also reduce the inflammatory response triggered by AGEs, thereby enhancing the function of pancreatic beta cells [35, 36]which could contribute to improved glucose metabolism in T2DM and prediabetes populations. In addition to antioxidative effects, carnosine and β-alanine may enhance glucose uptake in peripheral tissues [10]. Carnosine has been shown to improve the functionality of the glucose transporter 4 (GLUT4) in muscle cells, facilitating insulin-stimulated glucose uptake [37]. This effect is particularly relevant to skeletal muscle, which is responsible for the majority of glucose clearance from the blood [38]. β-alanine, as a precursor to carnosine, augments tissue carnosine levels and potentially amplifies these effects, thereby supporting enhanced glucose utilization and improved glycemic control [11]. Although we observed an improvement in β-cell function in this study, this effect did not extend to lipid metabolism or other indices of insulin resistance. This suggests that the intervention studied may primarily target the improvement of β-cell function, with a relatively limited impact on broader metabolic processes. Considering these results, we propose that while carnosine and β-alanine may be beneficial in modulating specific mechanisms related to β-cell function and glucose metabolism regulation, achieving comprehensive metabolic improvements may require additional interventions that act synergistically across multiple metabolic pathways. Additionally, although Egger’s tests identified possible publication bias for fasting insulin and HDL-c, the trim-and-fill analysis did not meaningfully alter the pooled effect sizes. This suggests that while selective reporting cannot be entirely excluded, its influence on our main conclusions is likely limited.

Our findings suggest that carnosine and β-alanine supplementation may offer promising adjunctive therapies for improving glycemic control in individuals with T2DM and prediabetes, particularly for individuals in the early stages of the disease or those at high risk of progression to diabetes. Although the improvements in FBG, HbA1c, and HOMA-β reported in our analysis were statistically modest, emerging evidence suggests that even small changes in these glycemic markers may carry meaningful clinical significance, particularly in the context of early-stage or personalized diabetes care. FBG and HbA1c are both independently and jointly important predictors of diabetes risk and complications, and their integration into prediction models has demonstrated high accuracy and clinical utility [39]. In patients with prediabetes, small changes in FBG and HbA1c have been associated with a decreased risk of developing T2DM. These modest improvements can help stabilize blood sugar levels, reduce the long-term risk of diabetes-related complications, and enhance quality of life by improving overall metabolic control. Furthermore, HOMA-β, a marker of pancreatic β-cell function, was improved in the supplementation group. This suggests that carnosine and β-alanine may help preserve or even modestly enhance β-cell function, which is critical for maintaining proper insulin secretion and glucose regulation [40]. These findings underscore the potential role of carnosine and β-alanine as adjunctive interventions capable of contributing to long-term metabolic benefits, particularly when introduced early in the disease course. However, the lack of significant changes in BMI, fasting insulin, and lipid profiles may be due to the short intervention durations (4–14 weeks) and small sample sizes (several studies with fewer than 20 participants per arm) in the included trials, which likely limited the ability to detect changes in these outcomes [41]. Additionally, baseline heterogeneity in insulin resistance and lipid profiles across studies may have diminished the power to detect improvements in these metabolic parameters [42]. These discrepancies may limit the generalizability of the findings and complicate clinical interpretation, underscoring the need for further investigation with more homogenous study populations. Our subgroup analyses provided a detailed breakdown of the optimal treatment doses and durations. We found that doses greater than 1000 mg/day resulted in significant improvements in both FBG and HbA1c, while improvements in HOMA-β were observed with doses of ≤ 1000 mg/day. Similarly, treatment durations of ≤ 12 weeks were associated with significant improvements in FBG and HbA1c, whereas treatment durations exceeding 12 weeks led to greater benefits for HOMA-β. These findings suggest that different doses and durations may be more effective for specific metabolic outcomes, with higher doses and shorter durations improving glycaemic control (FBG and HbA1c) and longer durations improving β-cell function (HOMA-β). Based on these results, future studies should refine the gradient of treatment doses and durations to determine the optimal regimen for each outcome. Specifically, exploring more precise dose ranges and treatment periods could help identify the most effective balance for improving both glycaemic control and β-cell function. Given the minimal effects on BMI, HOMA-IR, fasting insulin, and most lipid parameters, these supplements should not be considered standalone metabolic treatments. Instead, they may be better positioned as adjunctive therapies, supporting glycemic regulation within a broader diabetes management strategy that includes lifestyle interventions and, when necessary, pharmacologic agents.

Several limitations in our study should be acknowledged. First, the number of RCTs included in this meta-analysis was limited, which may affect the robustness and generalizability of our conclusions. Only eight studies were included, with a total of 377 participants, several of which had sample sizes below 20 participants per arm. Small sample sizes reduce the statistical power, particularly for secondary outcomes such as BMI, fasting insulin, and lipid indices. This limitation may have contributed to the lack of significant findings for these secondary outcomes. Moreover, small trials are more susceptible to random errors and less likely to detect clinically relevant effects, particularly in secondary outcomes that require larger sample sizes to detect modest changes. Consequently, our meta-analytic estimates for BMI, fasting insulin, HOMA-IR, and lipid indices are under-powered, and the lack of significant findings should be interpreted cautiously. Larger, more adequately powered studies are required to confirm these results and provide reliable conclusions for these outcomes. Second, there was considerable heterogeneity across the included studies regarding supplementation dosages, intervention durations, and patient characteristics, such as baseline glycemic status and comorbidities. This variability may have contributed to outcome heterogeneity and limits the generalizability of our findings. Future studies should consider adopting standardized protocols to ensure consistency in dosages and treatment durations. Specifically, the variations in dosages (e.g., ≤ 1000 mg/day vs. >1000 mg/day), treatment durations (e.g., 4–14 weeks), and participant characteristics (e.g., baseline insulin resistance, lipid profiles, and comorbidities) could have resulted in differential effects, complicating the interpretation of the results. For instance, higher doses (> 1000 mg/day) were found to improve FBG and HbA1c, while smaller doses (≤ 1000 mg/day) seemed to primarily enhance β-cell function (HOMA-β). Furthermore, shorter treatment durations (≤ 12 weeks) were associated with better glycemic control, whereas longer durations (> 12 weeks) were more effective for improving β-cell function. This heterogeneity underscores the necessity for future studies to adopt standardized protocols that ensure consistency in dosages, treatment durations, and participant inclusion criteria. Stratification of participants based on glycemic risk, baseline conditions, and comorbidities should be prioritized to ensure a more homogeneous study population. By tailoring interventions to the most appropriate subgroups, researchers can better identify the populations that will benefit the most, thus enhancing the overall clinical applicability of the findings. This approach would also reduce potential confounding factors and increase the statistical power to detect significant effects, leading to more reliable and interpretable results. Additionally, standardizing these parameters will help address the existing heterogeneity and improve the robustness of conclusions drawn from future research. Third, the majority of studies included small sample sizes, which likely limited the ability to detect significant effects on secondary outcomes such as BMI, fasting insulin, and lipid profiles. The limited power of these studies makes it difficult to draw definitive conclusions for these parameters. Fourth, publication bias remains a potential limitation, as indicated by Egger’s test for fasting insulin and HDL-C. Although the trim-and-fill analysis suggested that publication bias had minimal impact on the pooled effect sizes, the potential for publication or selective reporting bias still exists. This is an inherent limitation of meta-analyses that rely on published data, where smaller studies or negative results may be underreported. Thus, caution is needed in interpreting the results for these secondary outcomes. Finally, while our analysis was restricted to human RCTs, the relatively short follow-up durations in many trials may not capture the long-term effects of carnosine and β-alanine on metabolic health in T2DM and prediabetic populations. Future research should prioritize longer follow-up durations to better evaluate the sustained impact of supplementation.

Conclusion

In summary, while carnosine and β-alanine supplementation show promise for improving glycemic outcomes in T2DM and prediabetes, further research is necessary to explore their full therapeutic potential, optimal dosing strategies, and target populations. Addressing these limitations in future studies can provide a more comprehensive understanding of how these supplements may fit into an evidence-based approach to diabetes prevention and management. Specifically, refining the dosage ranges, treatment durations, and understanding the underlying mechanisms of action will help pinpoint the most effective therapeutic regimens. Furthermore, larger and well-powered multicenter trials are urgently needed to improve the generalizability and robustness of the conclusions. Moreover, future research should prioritize longer follow-up durations to better evaluate the sustained impact of supplementation. Such trials will not only enhance the precision of results but also allow for a more informed evaluation of the long-term effects of carnosine and β-alanine supplementation, ensuring that these supplements can be optimally integrated into diabetes care regimens alongside lifestyle interventions and pharmacologic treatments.

Authors’ contributions

NL and YW originated and designed the study. NL, JL, MW, XH, ZF, XH, JY, HL, WW, GX and YZ conducted the literature search, data extraction, data analysis and interpretation and drafting of the manuscript. XY conducted the data interpretation and methodology. NL and ZS conducted critical revision of article and final approval. All authors contributed to the writing and revisions. All authors read and approved the final manuscript.

Funding

All authors were supported by the Natural Science Foundation of Sichuan Province (NO.2025ZNSFSC0743, NO.2024NSFSC1618, NO.2024YFFK0287, No.2024NSFSC0657); the Key Research and Development Program of Chengdu Economic Development Zone New Economy and Science and Technology Bureau (NO.2024LQRD0048); Academic Degree and Postgraduate Education Reform Project of Sichuan Province (NO.CDJGY2024006); Longquanyi Talent Program, Class C Innovative Talent Program Special Funding; Chengdu Medical Research Project (2022291, 2023618), Key Projects of Chengdu University School of Clinical Medicine and Affiliated Hospital (Y202202), Chengdu University Research Initiation Programme (2081923030). Achievements of the provincial college Student Entrepreneurship Training Program of Chengdu University (S202511079022X).

Data availability

The data supporting the findings of this study are derived from previously published articles. Further inquiries regarding the data can be directed to the corresponding authors.

Author information

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Clinical trial number

Not applicable.