Effect of Carnosine or β-Alanine Supplementation on Markers of Glycemic Control and Insulin Resistance in Humans and Animals: A Systematic Review and Meta-analysis

Abstract

There is growing evidence that supplementation with carnosine, or its rate-limiting precursor β-alanine, can ameliorate aspects of metabolic dysregulation that occur in diabetes and its related conditions. The purpose of this systematic review and meta-analysis was to evaluate the effect of carnosine or β-alanine supplementation on markers of glycemic control and insulin resistance in humans and animals. We performed a systematic search of 6 electronic databases up to 31 December 2020. Primary outcomes were changes in 1) fasting glucose, 2) glycated hemoglobin (HbA1c), and 3) 2-h glucose following a glucose-tolerance test. A set of additional outcomes included fasting insulin and homeostatic model assessment of β-cell function (HOMA-β) and insulin resistance (HOMA-IR). We assessed risk of bias using the Cochrane risk of bias (RoB) 2.0 (human studies) and the Systematic Review Center for Laboratory Animal Experimentation (SYRCLE) RoB (animal studies) tools; and used the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach to assess certainty. We used Bayesian hierarchical random-effects models, with informative priors for human data and noninformative priors for animal data. Inferences were made on posterior samples generated by Hamiltonian Markov Chain Monte Carlo using 90% credible intervals (90% CrI) and calculated probabilities. Twenty studies (n = 4 human, n = 16 rodent) were included, providing data for 2 primary outcomes (fasting glucose and HbA1c) and 3 additional outcomes (fasting insulin, HOMA-β, and HOMA-IR). The model provides evidence that supplementation decreases fasting glucose [humans: mean difference (MD)0.5 = -0.95 mmol · L-1 (90% CrI: -2.1, 0.08); rodent: MD0.5 = -2.26 mmol · L-1 (90% CrI: -4.03, -0.44)], HbA1c [humans: MD0.5 = -0.91% (90% CrI: -1.46, -0.39); rodents: MD0.5 = -1.05% (90% CrI: -1.64, -0.52)], HOMA-IR [humans: standardized mean difference (SMD)0.5 = -0.41 (90% CrI: -0.82, -0.07); rodents: SMD0.5 = -0.63 (90% CrI: -1.98, 0.65)], and fasting insulin [humans: SMD0.5 = -0.41 (90% CrI: -0.77, -0.07)]. GRADE assessment showed our certainty in the effect estimate of each outcome to be moderate (human outcomes) or very low (rodent outcomes). Supplementation with carnosine or β-alanine may reduce fasting glucose, HbA1c, and HOMA-IR in humans and rodents, and fasting insulin in humans; both compounds show potential as therapeutics to improve glycemic control and insulin resistance. This review was registered at PROSPERO as CRD42020191588.

Keywords: endocrinology; histidine; metabolic health; metabolism; nutrition; obesity.

FIGURE 1

PRISMA flow diagram depicting the search and selection process. PRISMA, Preferred Reporting Items for Systematic review and Meta-Analysis.

FIGURE 2

Bayesian forest plot of meta-analysis for fasting glucose in human studies. Each interval represents posterior “shrunken” estimates based on the random-effects model fitting and borrowing information across studies to reduce uncertainty. Circles represent the median value along with 90% credible intervals. Negative values show a reduction in fasting glucose in the intervention group compared with the control group. This analysis included 172 human participants (89 intervention/83 placebo).

FIGURE 3

Bayesian forest plot of meta-analysis for fasting glucose in rodent studies. Each interval represents posterior “shrunken” estimates based on the random-effects model fitting and borrowing information across studies to reduce uncertainty. Circles represent the median value along with 90% credible intervals. Negative values show a reduction in fasting glucose in the intervention group compared with the control group. This analysis included 229 rodents (111 intervention/118 control).

FIGURE 4

Bayesian forest plot of meta-analysis for HbA1c in human studies. Each interval represents posterior “shrunken” estimates based on the random-effects model fitting and borrowing information across studies to reduce uncertainty. Circles represent the median value along with 90% credible intervals. Negative values show a reduction in HbA1c in the intervention group compared with the control group. This analysis included 134 human participants (67 intervention/67 placebo). Both studies supplemented with carnosine. HbA1c (HbA_1c), glycated hemoglobin.

FIGURE 5

Bayesian forest plot of meta-analysis for HbA1c in rodent studies. Each interval represents posterior “shrunken” estimates based on the random-effects model fitting and borrowing information across studies to reduce uncertainty. Circles represent the median value along with 90% credible intervals. Negative values show a reduction in HbA1c in the intervention group compared with the control group. This analysis included 260 rodents (127 intervention/133 control). All studies supplemented with carnosine. HbA1c (HbA_1c), glycated hemoglobin.