GRADE table 1
Based on the hypothesis that weight loss could improve cognition in obese or overweight individuals, Veronese et al.20 carried out a systematic review and meta-analysis aimed to investigate the effect of intentional weight loss on cognitive status in this population across observational and interventional studies. Extensive search and screening of the literature, which included several major healthcare databases, was conducted by two authors (Veronese N and Facchini S) independently from inception to 02.01.2016. The review included only studies that: included participants with a BMI of at least 25; reported only about intentional weight loss; assessed cognition through validated scales; reported at least 2 kg of weight loss (i.e. clinically significant weight loss19) in the treated group between follow-up and baseline; and included only a lifestyle/beahvioural intervention (pharmacological treatments were not included).
Data extraction was also carried out by two authors (SM and LC) independently and the results at follow-up evaluation of any cognitive tests assessed through validated scales were used as outcomes. Cognitive tests were categorised in five domains: attention; executive function; memory; motor speed; language domains.
After screening and assessment of the 1250 records obtained, seven randomised controlled trials (RCTs) were considered eligible and included in the meta-analysis.21–27 The studies included a total of 328 participants randomised to treated groups (262 in a dietary intervention group, 26 treated with physical activity, and 40 with both intervention components). In particular, the RCTs had a variety of interventions: four were based on diet alone,21,22,25,26 one on diet and physical activity,24 one on caloric restriction and physical activity,23 and one on caloric restriction or unsaturated fatty acid enhancement (in two different arms).27 Participants were followed up for a median of 20 weeks (range: 8–48). Based on the neuropsychological tests administered in each trial, the meta-analysis for each of the cognitive domains was included a sub-group of the seven RCTs selected. In the review a formal and quantitative assessment of both heterogeneity (Q2 and I2 statistics) and publication bias (Egger’s test) was conducted.
In the meta-analysis a significant improvement of the attention domain (four studies included)21,22,24,25 was reported (SMD = 0.44; 95%CI: 0.26–0.62, p < 0.0001) and the Egger’s test did not detect any publication bias. Heterogeneity, however, results high (I2= 60%). A significant improvement was also found in the memory domain (6 studies included21–23,24–26; SMD = 0.35; 95%CI: 0.12–0.57, p = 0.002). However heterogeneity was also high (I2= 64%) and significant publication bias detected (Egger’s test = 3.72 ± 0.68; p = 0.004), likely due to the inclusion of studies reporting negative findings. After a trim and fill procedure, the SMD increased to 0.61 (95%CI: 0.37–0.86). Language was the last cognitive domain to provide significant results (SMD = 0.21; 95%CI:0.05–0.37, p = 0.009). However, the highest heterogeneity rate was detected among the four studies included21,24–26 in this meta-analysis (I2= 73%) and publication bias was also detected, but in this case the trim and fill procedure did not change the results (SMD = 0.32; 95%CI:0.03–0.61). Two domains did not show any significant results: executive function and motor speed (SMD = −0.00; 95%CI: −0.30–0.37, p < 0.97; and SMD = 0.17; 95%CI: −0.14–0.48, p < 0.28, respectively). Both analysis were carried out on 2 studies (although the authors did not provide information on what specific studies were included) and although no high heterogeneity was identified (I2=41% and 12%, respectively) quantitative analysis of publication bias was not possible.
Main limitations of the studies were related to publication bias and heterogeneity (moderate to high), as well as a small sample size for the assessment of two outcomes. Furthermore, the mean duration of the intervention was relatively short and no formal assessment of dropout rates and/or adverse events was identified. The authors also reported that the effect of weight loss on cognition appeared to be not moderated by the baseline BMI, suggesting that a beneficial effect of weight loss in both overweight and obese people. Lack of studies and RCTs that the impact of weight loss on dementia and Alzheimer’s disease outcomes was identified.
Additional Evidence
The evidence (low to moderate quality), obtained from the analysis of the systematic review, indicates a small, but nonetheless significant, beneficial effect of lifestyle interventions aimed at weight reduction, in both overweight and obese people, on cognition in the attention, memory, and language domains, in particular.
The evidence included in GRADE is partially confirmed by an older (2011) systematic review and meta-analysis, published by Siervo and colleagues,28 aimed at assessing the effect of intentional weight loss reported on cognitive function in overweight and obese people. Twelve trials (seven randomised and five non-randomised) were included in this study. Key inclusion criteria were: 1. statistically significant and intentional weight loss greater than 2 kg (considered as clinically meaningful) and likely association with improvements in metabolic and vascular functions; and 2. reported assessment of cognitive function before and after weight loss through standardised and validated neuropsychological tests. A small size significant effect of weight loss was found for memory (SMD 0.13, 95% CI 0.00–0.26, P = 0.04) and attention/executive functioning (SMD 0.14, 95% CI 0.01–0.27, P < 0.001). However, the association between weight loss and cognitive improvements was identified only in obese but not in overweight individuals. The quality of the evidence was mostly limited by heterogeneity and publication bias both formally assessed with standardised tests.
In addition to this, a body of observational evidence generally supports a role for overweight and obesity in increasing the risk of cognitive impairment, and highlight age-based difference on such effect.
In 2014, Prickett et al. examined the relationship between obesity and cognitive function in a systematic review of cross-sectional and/or prospective studies.29 The review included studies on adults between 18 and 65 years of age, with a BMI of at least 30, with concurrent assessment of cognitive function. Evidence from the 17 studies that were identified and included showed a significant association between obesity and cognitive impairment across almost all the cognitive domains investigated (complex attention, verbal and visual memory, decision making). However the quality of the evidence was hampered by methodological limitations identified in the studies considered (e.g. matching or handling of confounders, variability in the study design, use of appropriate comparison groups, incomplete investigation of the cognitive domains) as well as publication bias due to challenges in publishing non-significant results.
On the following year, Xu and colleagues published a Meta-analysis on risk and protective factors for Alzheimer’s disease (AD).30 PubMed and the Cochrane database of systematic reviews were systematically searched from inception to July 2014 for cohort studies and retrospective case–control studies reporting on risk factors for AD and dementia. Studies were included if: they reported original data concerning odds ratio (OR) or risk ratio (RR) of AD using a longitudinal cohort study or retrospective case–control study design; the study population was representative of the general population and; modifiable risk factors were included. A total of 323 papers were included in the meta-analysis. Concerning BMI, Grade I evidence indicated that its influences the risk of AD are complex and depend on age: high BMI in mid-life would increase the risk of the disease while high BMI in late life would be protective.
Pedditizzi et al. (2016) conducted a systematic review of epidemiological longitudinal studies, published from inception since September 201431 that reported on incidence of AD/dementia and cognitive function, as well as data related to overweight and obesity. The search was conducted on a range of relevant databases and studies had at least 2 years follow up and included an assessment of incident dementia and 21 studies met the selection criteria. The meta-analysis, which included 13 studies, showed that obesity below the age of 65 was associated with a higher risk of dementia (RR 1.41, 95% CI 1.20–1.66), but the opposite was seen in those aged 65 and over (RR 0.83, CI: 0.74–0.94).
In the same year, Lafortune et al carried out a rapid systematic review on the midlife risk factors associated with dementia.32 Longitudinal cohort studies were searched in several relevant databases starting from 2000 and 164 were included in the qualitative synthesis. Weight change/weight cycling was one of the risk factors considered, but the authors identified only limited evidence suggesting that weight changes (in both directions) in midlife is associated with an increase of dementia.
Finally, in 2017 Hersi et al. published a systematic review and qualitative synthesis of risk factors associated with progression to AD.33 The authors searched for both primary observational studies and systematic reviews. Eleven systematic reviews and six primary studies that reported on the association between obesity and body mass index (BMI) with risk of AD, were identified. Overall the evidence from the synthesis of the included publications was inconclusive, although differences based on age were identified.
