Higher Coffee Consumption Is Associated With Reduced Cerebral Gray Matter Volume: A Mendelian Randomization Study - PubMed
This two-sample MR study showed that genetically predicted higher coffee consumption is causally associated with reduced gray matter volume of the brain.
pubmed.ncbi.nlm.nih.gov
Abstract
Background
Recently published two-sample Mendelian randomization (MR) studies showed that genetically predicted coffee consumption may be associated with increased risk of Alzheimer’s disease and intracerebral hemorrhage but associated with a decreased risk of small vessel ischemic stroke. We aimed to investigate the effects of genetically predicted coffee consumption on magnetic resonance imaging (MRI) markers of cerebral small vessel disease and brain volume using the two-sample MR method.Methods
Twelve single nucleotide polymorphisms (SNPs) in up to 375,833 individuals were used as genetic instruments for cups consumed per day of coffee. Another four SNPs from an independent sample were used to perform the replication analysis. Three SNPs in up to 45,821 individuals were used as genetic instruments for high coffee consumption vs. low/no coffee consumption.Results
Mendelian randomization analysis showed that coffee consumption (cups/day) was inversely associated with gray matter volume. Replication analysis and multivariable analyses after adjusting for other risk factors confirmed the effect. High coffee consumption was also suggestively associated with decreased gray matter volume compared with low/no coffee consumption. All analyses did not find an effect of coffee consumption on other outcomes including white matter hyperintensity volume, mean diffusivity, fractional anisotropy, brain microbleed, total brain volume, white matter volume, and hippocampus volume.Conclusion
This two-sample MR study showed that genetically predicted higher coffee consumption is causally associated with reduced gray matter volume of the brain.Introduction
Coffee is one of the leading consumed beverages worldwide. A study showed nearly half of the interviewed samples consumed at least one cup of coffee a day even during pregnancy (1). Therefore, the effects of coffee consumption on health outcomes are of interest and of importance.A previous review of 201 meta-analyses of observational studies concluded that coffee consumption was more often associated with benefit than harm (2). Harmful associations were only found for the fracture in women and for the fetus (2). As for stroke, beneficial effects were also observed in observational studies (3, 4). A recent review found a significant inverse association between coffee consumption and risk of stroke (3). However, the effects of dementia were inconclusive (5, 6).
Since inferences from observational studies are often biased by confounding factors, a new method of two-sample Mendelian randomization (MR) has been widely used to assess the causality (7). Recently published studies using the two-sample MR method showed that genetically predicted coffee consumption may be associated with increased risk of Alzheimer’s disease and intracerebral hemorrhage but with a decreased risk of small vessel ischemic stroke (8, 9).
It has been well established that magnetic resonance imaging (MRI) markers of cerebral small vessel disease are associated with an increased risk of stroke, cognitive impairment, and dementia (10–14). In addition, brain volume is one of the surrogate measures of brain reserve, a concept that relates to cognitive function and Alzheimer’s disease (15). To provide more evidence regarding the effects of coffee consumption on cerebrovascular disease and cognitive function, we further investigated the effects of genetically predicted coffee consumption on MRI markers of cerebral small vessel disease and brain volume using the two-sample MR method.
[...]
Discussion
This two-sample MR study showed that cups of coffee consumed per day were inversely associated with the gray matter volume of the brain. In addition, high coffee consumption was suggestively associated with decreased gray matter volume compared with low/no coffee consumption. We did not find evidence that genetically predicted coffee consumption was associated with WMH volume, mean diffusivity, fractional anisotropy, BMBs, total brain volume, white matter volume, or hippocampus volume.A recently published observational study using data from the UK Biobank showed that coffee consumption was not associated with WMH volume, which was the same as the results of our study (31). However, the study showed that coffee consumption was inversely and significantly associated with total brain volume, gray matter volume, white matter volume, and hippocampal volume. Using genetic data from the UK Biobank with more MRI data samples, our MR study confirmed the causal effect of coffee consumption on gray matter volume. However, coffee consumption showed no causal effect on total brain volume, white matter volume, and hippocampal volume. Although our study showed coffee consumption trended toward a decrease in the total brain volume, this trend was mainly attributed to the effect of coffee consumption on the gray matter because the effect on the white matter was very close to null. Our finding was supported by a recently published small double-blind crossover randomized trial (32). This trial included 20 healthy male habitual caffeine consumers aged 18–35 years. This trial showed that 9-day administration of caffeine (3 doses of 150 mg/day) was associated with decreased gray matter volume compared with the placebo. For the association between coffee consumption and risk of cognitive disorders and Alzheimer’s disease, observational studies showed inconclusive results. An early dose-response meta-analysis showed a “J-shaped” association with the lowest risk of incident cognitive disorders at a daily consumption level of 1–2 cups of coffee (33). However, a subsequent dose-response meta-analysis did not support the non-linear relationship and did not find an association between coffee consumption and increased risk of overall dementia or Alzheimer’s disease (6). A meta-analysis of one-sample MR studies based on two genetic variants showed no effect of coffee intake on global cognition or memory (34). However, a recently published two-sample MR study showed that coffee consumption was associated with increased risk of Alzheimer’s disease, but the effect attenuated to non-significant when adjusted for drinks per week. Our study showed that coffee consumption can only affect gray matter volume and does not have an effect on hippocampal volume. In addition, we did not find evidence that coffee consumption was associated with MRI markers of cerebral small vessel disease. These may explain why the effects of coffee consumption on cognitive disorders and Alzheimer’s disease were not very robust in previous studies.
Although observational studies showed that coffee consumption was associated with a decreased risk of cardiovascular disease (including stroke) and cardiovascular mortality (2, 3), the effects were not fully supported by two-sample MR studies. Two two-sample MR studies showed that genetic predisposition to higher coffee consumption was not associated with any of the studied cardiovascular outcomes, except for a suggestive positive association for intracerebral hemorrhage after adjusting for body mass index and smoking (35, 36). Another two-sample MR study confirmed the harmful effects of coffee consumption on an intracerebral hemorrhage (8). However, this study and another two-sample MR study also showed weak suggestive evidence for a potential beneficial effect of coffee consumption on small vessel ischemic stroke (8, 9). Since small vessel ischemic stroke and intracerebral hemorrhage both belong to cerebral small vessel disease and shared similar risk factors (e.g., hypertension), it is interesting that coffee consumption showed totally opposite effects for them. Nevertheless, we did not find evidence that coffee consumption was associated with MRI markers of small vessel disease including WMH volume, mean diffusivity, fractional anisotropy, or BMBs. Therefore, taking into account both the weak suggestive evidence in previous studies and the findings of our study, the potential beneficial effect of coffee consumption on small vessel ischemic stroke found in previous studies should be interpreted with care, at least before being validated in an independent sample.
It has been suggested that the main components of coffee (i.e., caffeine and polyphenolic acids) may have antioxidant effects, anti-inflammatory effects, as well as neuroprotective effects through blocking the adenosine receptor (4, 37). In addition, in vivo studies showed that treatment with caffeine could prevent Aβ-induced cognitive impairment in mice (38, 39). The antioxidant and anti-inflammatory effects have been suggested to account for the potential protective effect of coffee consumption on the risk of small vessel ischemic stroke (9). However, intake of caffeine may temporarily increase the blood pressure in individuals with hypertension, which may lead to an increased risk of intracranial hemorrhage in hypertensive individuals (8, 40). A recently published study confirmed that coffee consumption is the main trigger factor for intracranial hemorrhage (41). Although caffeine could reinstate the adenosine-inhibited synaptic excitatory signaling in human neurons, it also could attenuate synaptic long-term potentials (42, 43). The impact of caffeine on long-term potentials might be partially responsible for the effects of coffee consumption on gray matter volume (31, 32).
This study has several limitations. First, there were substantial sample overlaps between primary data of cups per day and some outcomes. However, a recent study showed that main two-sample MR methods can be safely used for one single large dataset than from large biobanks such as the UK Biobank. In addition, we calculated the bias from sample overlap. The results showed that biases from sample overlap were negligible. Furthermore, replication analysis using an independent sample with no sample overlap supported the findings. Second, although primary analysis and replication analysis all showed coffee consumption (cups/day) was associated with decreased gray matter volume, the magnitude of the effects was quite different. This may be caused by the differences in included samples and analysis protocols. Third, although we included exposures of both cups per day and high coffee consumption vs. low/no coffee consumption, we could not analyze the potential non-linear relationship using summary data. Fourth, in the included studies, all types of coffee were treated the same to estimate total coffee consumption. Therefore, we could not perform the analysis stratified by types of coffee. It is uncertain whether different types of coffee have different influences for gray volume or not. Fifth, the duration of coffee consumption is certainly varied among individuals. For example, women are more likely to abstain from coffee during pregnancy and breastfeeding. However, the duration of coffee consumption was not adjusted in the GWASs of coffee consumption. Sixth, the power of the test was relatively low for some outcome measures. Finally, the samples included were mainly European ancestry individuals, which limits the generalizability of our findings for other populations.