After removal of duplicates, the initial search yield 101 unique articles from PubMed, Scopus, Embase and Web of Science (Fig. 1). Of these, 81 were found to be irrelevant reports after screening the title and abstract. Among the remaining 20 full-text articles, a total of 8 met eligibility criteria and included in this review.

Further search in ClinicalTrials.gov found another 16 NMN trials (6 completed; 6 recruiting, 1 active but not recruiting, 2 not yet recruiting, 1 unknown status) examining the effect of NMN on at least one marker of glucose control or lipid profile as outcome measures (Supplementary Table 1). However, none of the completed trials reported their results in the registry and thus we were not able to include any in this review.

All 8 eligible studies were double-blinded, randomised and controlled trials, among which 2 trials [23, 24] had randomised design with mutiple intervention groups (Table 1). All trials were reported during 2021–2023, and the majority were conducted in Asia. The funding source for these trials was summarised in Supplementry Table 2.

All included trials were small in scale, with sample size ranging from 14 to 108, and a total of 342 participants (49% female) took part. The trials were mainly carried out among middle-aged adults except 3 RCTs recruited older subjects with mean age over 60 years [15, 23, 25]. Most trials recruited subjects with normal baseline glucose and lipid profiles except 1 study recruited prediabetic patients. [15] 4 RCTs were conducted among subjects with mean BMI of 25 or above. [15, 23, 24, 26]

NMN supplements were orally administrated in all trials. Duration of supplementation ranged from 14 days to 12 weeks, with 6 trials [15, 24,25,26,27,28] having intervention more than 8 weeks. The dosage of NMN ranged from 250 mg/d to 2000 mg/d; 250 mg/d was the most commonly used dosage and was employed in 4 studies. [15, 25, 27, 28] The largest dosage of 2000 mg/d was tested in Pencina et al.,2023 for 14 days. [23]

The majority of included trials tested fasting glucose [15, 23, 25,26,27,28,29] and/or lipid profile. [23,24,25,26,27,28,29] Some studies also evaluated the effect of NMN on fasting insulin [15, 26, 28, 29], HOMA-IR [24,25,26, 28] and HbA1c [15, 25, 28].

Based on an adapted version of Rob2 implemented specifically for parallel-group trials with individual randomization (Fig. 2), 5 studies had a low risk of bias. [24, 25, 27,28,29] The other 3 studies had some concerns about the overall risk of bias due to lacking information of allocation concealment in the randomization process [15, 23, 26] or analysis plan for the selection of the reported result [23].

Risk of bias assessment in the randomised parallel-arm studies. A. Risk of bias summary. B. Quality assessment percentages of all included studies

Five out of the 8 RCTs reported an increase in blood NAD + level following NMN supplementation. [23,24,25,26,27] The meta-analyses indicated no significant benefit of NMN on fasting glucose, fasting insulin, HbA1c and lipid profile. There was a marginally significant reduction on HOMA-IR by 0.27 (95% CI -0.01 to 0.55; p = 0.06; I2 = 0%) based on 3 studies (Fig. 3). However this effect on HOMA-IR became not signficant after excluding the trial by Huang 2022 [26] whose mean difference in HOMA-IR was attributed to a deterioration of HOMA-IR in placebo group, instead of an improvement in the supplementation group. Included studies had low heterogeneity (I2 = 0%) for almost all outcome measures except for fasting insulin and HbA1c with I2 > 50%.

Forest plot of the effect of NMN supplements on markers of glucose and lipid metabolism

Okabe et la., 2022, [28] presented their results in plots only thus its data was not included in the meta-analyses. It reported null difference in the related biomarkers between NMN group and placebo group from a small sample of 30 subjects, therefore including its data in the meta-analysis would unlikely affect the combined results. Given none of meta-analyses included findings from 10 or more trials, we are unable to examine from a funnel plot to explore publication bias.

Reports of adverse events in the included trials were summarised. (Supplementary Table 3) Three trials reported no adverse events. No trial reported safety concerns of NMN supplementation, however there seems to be a tendency of more adverse events, regardless of their relevance to the intervention, reported in NMN groups with dosage 1000 mg/d or higher. [23, 29] Additional meta-analyses found no statistically significant difference in BMI, blood pressure and liver function (AST, ALT and ALP levels) in 6 studies [15, 24,25,26,27, 29] between the NMN intervention groups (250–1250 mg/d) and the controls. (Supplementary Fig. 1).