Effect of N-methyl-d-aspartate receptor enhancing agents on cognition in dementia: an exploratory systematic review and meta-analysis of randomized controlled trials

Multiple N-methyl-d-aspartate (NMDA) receptor enhancing agents have had promising effects on cognition among patients with dementia. However, the results remain inconsistent. This exploratory meta-analysis investigated the effectiveness of NMDA receptor enhancing agents for cognitive function. PubMed, the Cochrane Central Register of Controlled Trials, and the Cochrane Database of Systematic Reviews were searched for randomized controlled trials (RCTs). Controlled trials assessing add-on NMDA receptor enhancing agent treatment in patients with dementia and using cognition rating scales were eligible and pooled using a random-effect model for comparisons. The standardized mean difference (SMD) was calculated in each study from the effect size; positive values indicated that NMDA receptor enhancing agent treatment improved cognitive function. Funnel plots and the I2 statistic were evaluated for statistical heterogeneity. Moderators were evaluated using meta-regression. We identified 14 RCTs with 2224 participants meeting the inclusion criteria. Add-on NMDA receptor enhancing agents had small positive significant effects on overall cognitive function among patients with dementia (SMD = 0.1002, 95% CI 0.0105–0.1900, P = 0.02860). Subgroup meta-analysis showed patients with Alzheimer’s Disease and trials using the Alzheimer Disease Assessment Scale-cognitive subscale as the primary outcome had small positive significant effects (SMD = 0.1042, 95% CI 0.0076–0.2007, P = 0.03451; SMD = 0.1267, 95% CI 0.0145–0.2388, P = 0.2686). This exploratory meta-analysis showed a very small, positive, and significant effect on overall cognition function in patients with dementia. Studies with larger samples are needed to evaluate different cognitive domains and phases of dementia.

www.nature.com/scientificreports/ Mild cognitive impairment (MCI), characterized by objective cognitive impairment, is defined as the predementia stage on the continuum of cognitive decline 8,9 . Several studies have reported that patients with MCI generally develop AD 10,11 . However, no treatment has yet been approved for MCI. Therefore, the development of alternative medications is urgently required for AD and MCI. NMDAR has been demonstrated to play a critical role in controlling synaptic plasticity and memory function 12 . Studies have reported that overactivation of NMDAR may cause neurotoxicity, especially in the late phase of AD 13,14 . NMDAR antagonists have been developed for the treatment of AD, and memantine, an uncompetitive NMDAR partial antagonist, has been approved as an antidementia medication for moderate to severe AD 15,16 ; memantine is not approved for mild AD and MCI because of unsatisfactory efficacy 17 . However, NMDAR antagonists, such as ketamine, have been demonstrated to impair spatial learning and verbal information ability in healthy humans 18 . Moreover, human studies have reported an age-related decrease in the density of NMDARs in the cerebral cortex and hippocampus 19 . These findings implicate the hypoactivation of NMDAR in cognitive impairment.
Several trials of NMDAR enhancing agents have been performed to evaluate their effect on cognitive impairment among patients with dementia. Sodium benzoate, an inhibitor of D-amino acids oxidase, can increase d-serine and thus enhance NMDAR activation 20,21 . A randomized, double-blind, placebo-controlled trial of 60 patients with early-phase AD or MCI reported a larger improvement in the AD Assessment Scale-cognitive subscale (ADAS-cog) score compared with placebo 22 . D-cycloserine, a partial agonist at the glycine site of NMDAR, has been reported to improve ADAS-cog score in patients with AD 23 . However, other trials with D-cycloserine did not report significant improvement in cognition 24,25 . Therefore, we conducted a meta-analysis on the cognitive effects of NMDAR enhancing agents in dementia.

Search strategy and inclusion criteria. Two independent authors (Chun-Hung Chang and Shaw-Ji
Chen) performed a systematic literature search from the study's inception until August 7th, 2021. We followed the drug keywords used by previous meta-analytic reviews about NMDA receptor positive modulators [26][27][28][29] . The search strategy comprised the following keywords: (Dementia OR Alzheimer*) AND (acetylcysteine OR α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid OR AMPA OR benzoate OR CX516 OR D-cycloserine OR D-serine OR glutamine OR glutamate OR glutamate carboxypeptidase 2 OR GCP2 OR glycine OR glycine transporter type 1 OR GlyT1 OR glutamate receptor ionotropic kainate OR GRIK OR kynurenine aminotransferase OR KAT OR metabotropic glutamate receptor OR mGluR OR minocycline OR N-acetyl-aspartylglutamate OR NAAG OR N-methyl-d-aspartate OR NMDA OR pregnenolone OR sarcosine) AND controlled trial 26. The authors independently evaluated the titles and abstracts of all retrieved papers for eligibility. Full-text articles of potentially eligible trials were reviewed and compiled in a list of studies for inclusion. Disagreements between the two authors regarding the inclusion of a study were settled by a third reviewer (Professor Tsai). The Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) were followed (Fig. 1, Supplementary Table S1) 30-32 . Eligibility criteria. The inclusion criteria were as follows: (a) studies on participants who received a diagnosis of dementia, (b) studies including randomized placebo-controlled trials, and (c) studies on the use of NMDAR enhancing agents as a monotherapy or adjunctive treatment to concomitant antidementia drugs. The exclusion criteria were as follows: (1) abstracts from ClinicalTrials.gov, (2) conference abstracts, (3) reviews and comments, (4) studies on treatment other than NMDAR enhancing modulators, (5) studies on a different population or outcome, and (6) trial protocols.
Outcome measures. The effects of glutamate positive modulators on cognitive deficits in patients with dementia were investigated. Overall cognitive function (primary outcome) was compared between patients receiving NMDAR enhancing agents and those receiving placebo. The common cognitive measures in dementia are the Mini-Mental State Examination (MMSE) 33 and the Alzheimer's Disease Assessment Scale-cognitive subscale (ADAS-cog) 34 . The MMSE is a cognitive test commonly used to screen for dementia and measure cognitive impairment in older adults 35 . The MMSE total score ranges from 0 to 30 (highest to lowest level of cognitive impairment). The ADAS-cog consists of 11 tasks, and its score ranges from 0 to 70 (lowest to highest level of cognitive impairment). Data extraction. Two independent reviewers (CH Chang and SJ Chen) screened and identified articles.
We recorded information including the first author, published year, number of participants, gender proportion, mean age, duration, NMDA receptor enhancing agents, and summary of neurocognitive measures (Table 1). We followed the PRISMA guidelines. The primary outcome was the difference in effect sizes between the NMDAR enhancing agent and control groups (calculated in terms of the standard mean difference [SMD] with the corresponding 95% confidence interval [CI] and P value). The primary author of a study was contacted to request the original data if these data were not available in the corresponding article. If no relevant data were reported in the article, other compatible statistical parameters (e.g., P value, sample size, or odds ratio) were used to estimate the effect size, following the Comprehensive Meta-Analysis manuals protocol and the Comprehensive Meta-Analysis website guide (https:// www. meta-analy sis. com/ downl oads/ Meta-analy sis% 20Con verti ng% 20amo ng% 20eff ect% 20siz es. pdf); the estimated effect sizes were then converted and pooled into SMDs. Variables of interest were extracted when possible, including first author, year, sample size, proportion of men, mean age, therapy duration, add-on NMDAR enhancing agents, and cognitive outcome measures. www.nature.com/scientificreports/ Quality assessment. Two authors (CH Chang and SJ Chen) evaluated the methodological quality of the included studies. The Jadad scoring system was used for randomized controlled trials (RCTs) 36 . The Jadad scoring system contains three items that assess randomization (2 points), blinding (2 points), and an account of all patients (1 point). Therefore, the score ranges from 0 to 5. A higher score indicates higher methodological quality.
Meta-analysis procedure. A random-effects model was used for the meta-analysis 37 . The SMD with 95% CI was selected to compare the ESs of our primary outcome, rather than the difference in means, because each study was presumed to use different units. Furthermore, when the SMD value was higher than 0, the effect size was defined as indicating a positive effect in the NMDAR enhancing agent group compared with the placebo group. After data from included trials were extracted, Comprehensive Meta-Analysis software version 3 (Biostat, Englewood, NJ, USA) was used to perform meta-analyses. Statistical significance was defined as a two-tailed P < 0.05. Of note, the correction of multiple comparison was not made in this study due to this was an exploratory study rather than confirmatory study.
Heterogeneity and publication bias. The I 2 statistic was used as a measure of heterogeneity 37 , which was assessed using the Cochran Q test and corresponding P value 38 . Publication bias was assessed using funnel plots 39 and Egger's regression test 40 If publication bias was discovered, Duval and Tweedie's trim-and-fill procedure, a validated model for ES estimation, was used 41 .
Sensitivity test. Studies were individually removed from the meta-analysis to verify that the results were not caused by outliers in the included studies.
Cognitive measure as the primary outcome. This meta-analysis which summarized seven trials 22 NMDAR enhancing agents. Furthermore, we performed a subgroup meta-analysis of the studies with add-on NMDAR enhancing agents to evaluate the overall cognitive effect. Nine NMDAR enhancing agents were investigated: benzoate, cycloserine, D-cycloserine, LY451395, minocycline, monosodium L-glutamate, NAC, S47445, and sarcosine. Subgroup meta-analyses for each NMDAR enhancing agent revealed small positive but nonsignificant differences in effect on overall cognitive function (Fig. 2e).

Heterogeneity and publication bias. No significant heterogeneity was observed in overall cognitive
function among the studies (Q = 3.784, df = 10, I 2 = 0.000%, P = 0.957). Egger's test revealed no significant publication bias regarding the overall cognitive effects (P = 0.4207). The funnel plots for the effect sizes of overall cognitive function are displayed in Fig. 4.

Sensitivity analysis.
The positive effect of NMDAR enhancing agents on overall cognitive function became nonsignificant when the study by Bernard 51 or Lin 22 was removed.
Methodological quality of the included studies. Among the fourteen RCTs, the average Jadad score was 3.14 with a standard deviation (SD) of 0.36 (Supplementary Table S2).

Discussion
To our knowledge, this is the first study to investigate the efficacy of NMDA enhancing agents on cognition among patients with dementia. The main findings of this analysis are that (1) NMDAR enhancing agents had small positive significant effects on overall cognitive function compared with placebo, (2) trials on patients with AD obtained small positive significant effects, (3) trials using the ADAS-cog as the primary outcome obtained small positive significant effects. These findings accord with those of most related studies. Two of the studies reported significant improvement in overall cognitive function in patients with AD 22,23  www.nature.com/scientificreports/  23 . Seven studies reported nonsignificant improvement in overall cognitive functions 24,25,43,44,48,50,51 . For example, Howard reported that patients treated with minocycline had mean MMSE score 0.1 points higher than those treated with placebo; however, the mean was nonsignificantly different (95% CI − 1.1-1.2; P = 0.90) 48 . Bernard reported that patients treated with S47445 had nonsignificant improvement in ADAS-cog score; the adjusted mean change in ADAS-cog total score from baseline to week 24 was 0.35 ± 5.53 in the placebo group, − 0.20 ± 4.91 in the 5 mg of S47445 group, − 0.74 ± 5.43 in the 15 mg of S47445 group, and − 0.05 ± 5.82 in the 50 mg of S47445 group 51 . Subgroup analyses revealed that the meta-analysis which pooled eight trials enrolling patients with AD [22][23][24][25]43,44,48,51 showed a small positive significant effect (SMD = 0.1042, 95% CI 0.0076-0.2007, P = 0.03451; Fig. 2b), while trials enrolling patients with non-AD dementia did not. This finding may imply that patients with AD may more benefit from NMDAR enhancing agents than patients with non-AD dementia. Lin and colleagues treated patients with mild AD receiving with 250-750 mg/d of sodium benzoate or placebo for 24 weeks; those treated with sodium benzoate displayed significantly greater improvement in ADAS-cog score (P = 0.0031) 22 . Their findings suggest that patients with mild AD may benefit more from NMDAR enhancing agents than patients with moderate to severe AD. However, in our meta-regression, no significant association between effect size and mean ADAS-cog was observed. Further studies are needed to investigate these findings.
Moreover, we noted that this meta-analysis summarizing seven trials 22,23,25,43,49,51 in which the ADAScog was employed as the cognitive measure revealed small positive significant effects (SMD = 0.1267, 95% CI 0.0145-0.2388, P = 0.02686), whereas two trials 44,48 using the MMSE reported nonsignificant effects (SMD = 0.0323, 95% CI − 0.1391-0.2037, P = 0.71201; Fig. 2c). The MMSE and ADAS-cog are most common cognitive measures in these trials. The MMSE is often employed to screen for dementia and measure cognitive impairment in older adults 35 . The MMSE total score ranges from 0 to 30 (highest to lowest level of cognitive impairment) while the ADAS-cog consists of 11 tasks, with a total score ranging from 0 to 70 (lowest to highest level of cognitive impairment) 53 . MMSE and ADAS-cog have shown to be relatively psychometrically poor at detecting small changes [54][55][56] . Therefore, researchers may use more sensitive instruments like Addenbrooke's Cognitive Examination 57 or Montreal Cognitive Assessment 58,59 to explore minor changes in cognition in the early stages of dementia receiving NMDAR enhancing agents.
The mechanism of the positive effect of NMDAR enhancing agents on cognition remains unclear. Metaanalytic studies have investigated NMDAR enhancing agents in patients with schizophrenia. Tsai and colleagues reported that the overall effect of NMDAR enhancing agents (glycine, D-serine, and sarcosine) on cognitive symptoms was 0.28 (95% CI 0.10-0.47, P = 0.002, 13 studies, n = 485) 29 , whereas two other meta-analytic studies did not report a significant effect on overall cognition in schizophrenia 26,28 . An updated meta-analysis enrolling 25 trials and 1951 participants revealed no significant effect of NMDA-enhancing agents on overall cognition. However, subgroup analysis suggested that NMDAR-enhancing agents may benefit young patients with schizophrenia, and NAC may have effect on working memory 27 . Human studies have reported an age-related decrease in the density of NMDARs in the cerebral cortex and hippocampus 19 . NMDAR activation improves memory function 12 . Several trials with NMDAR enhancing agents, such as benzoate and D-cycloserine, have reported promising effects, especially in early-stage AD 22,23 . www.nature.com/scientificreports/ Side effects. The side effects were mostly mild and relieved after discontinuing the agents. The side effect rate of NMDAR enhancing agents ranged from 0 to 61.1%. Minocycline was reported side effects including dermatologic symptoms (hyperpigmentation, photosensitivity, rash), gastrointestinal symptoms (diarrhea, nausea, sore mouth, vomiting), neurologic symptoms (headache, visual or auditory disturbances, dizziness), infections (oral or genital candidiasis, vaginitis, anal irritation, bacterial enteritis, staphylococcal, or Clostridium difficile). Six skin toxic effects were considered severe 48 . The most common adverse effects of the study with AMPA modulator S47445 are nasopharyngitis, blood creatine phosphokinase increased, diarrhea, fall, headache, type 2 diabetes mellitus, abdominal pain 51 . D-cycloserine was reported no side effects 23,25 . Detailed information about the adverse events addressed in the included studies is summarized in Supplementary Table S2. The attrition rate ranged from 0 to 0.49. The effect size in this exploratory meta-analysis was very small. Most side effects were mild. The benefit may weigh up against the side effects.
Implication. Cognitive impairment is a critical problem in central nerve disease such as dementia. AD is the leading cause of dementia. In 2010, the number of patients with AD was estimated at 36.6 million and is projected to double every 20 years 2 . Dementia is a considerable economic burden on both patients and society 6 . However, understanding of the complex mechanisms and treatment response remains unsatisfactory 3 . Therefore, developing alternative therapeutic approaches is crucial. Our meta-analysis revealed a positive and significant effect of NMDAR positive modulators on overall cognition among patients with dementia. Further trials with better design and larger samples are needed to explore NMDAR enhancing agents in MCI or early-phase dementia.
Strength. The present study had several strengths. First, this is the first meta-analytic study to evaluate the treatment effect of NMDAR positive modulators among patients with dementia. Second, nine compounds were included in this meta-analysis. Third, we noted that NMDAR enhancing agents may have a small positive significant effect on overall cognitive function compared with placebo.
Limitations. There are several limitations in the present study. First, this is an exploratory systematic review and meta-analysis of randomized controlled trials. We could observe related phenomena but not clarify the underlying pathophysiology because of the methodological limitations of meta-analyses. Therefore, caution should be exercised when drawing conclusions. Second, the number of participants was small in some included trials. Third, some trials were conducted over a short period (less than 6 months). Therefore, the long-term cognitive effects of NMDAR enhancing agents and the persistent effects after treatment remain unclear. Kouzuki and colleagues reported improvement in the Touch Panel-type Dementia Assessment Scale total score during a 4-week follow-up assessment 50 . Fourth, the trials used different standard cognitive tests. Fifth, nine NMDAR enhancing agents were investigated. These agents may have different mechanisms of action involving the N-methyl-d-aspartate (NMDA) system. Further studies are needed to explore the relationships between neurocognitive effects and specific NMDAR enhancing mechanisms. Some modulators may be involved in mechanisms other than the NMDA system. Sixth, the effects of the stage of dementia and concomitant antidementia drugs remain unclear.

Conclusion
The findings from this meta-analysis indicated that NMDAR enhancing agents showed a very small positive effect on overall cognitive function in patients with dementia. Further studies with a larger sample are warranted to explore the role of the NMDA system on specific cognitive domains in subgroups of patients with early-stage dementia.