LAUSR

The potential association between diabetes mellitus (DM) and Parkinson’s disease (PD) has been somewhat controversial for many years because of conflicting findings from a large number of small studies. Meta-analyses of prospective cohort studies have concluded that there is a small increased risk of ~1.34 for PD among all people diagnosed with DM. The association has been further clarified following publication of a very large study comparing rates of PD in >2 million people with DM against a reference population of >6 million people. This showed that the hazard ratio for PD was 3.81 in people with onset of DM between the ages of 25 and 44 compared to people without DM. There was a doseresponse effect in that people with onset of DM between 45 and 64 had a hazard ratio of 1.71, and those with onset between 65 and 74 had a hazard ratio of 1.40. A new prospective epidemiological study by Jeong and colleagues has confirmed this positive association. In a South Korean database including >6 million individuals, the unadjusted hazard ratio for PD among all people with DM was 2.2. They also found a consistent doseresponse relationship in that the risk of PD progressively increases according to duration and severity of impaired fasting glucose/DM. Jeong and colleagues also report on the relationship between body mass index (BMI) and the risk of PD. Given that DM is associated with increasing BMI, intuitively one might expect that there would be a similar increasing association between PD and each rising stratum of BMI; however, the exact opposite is observed. Underweight people had a higher risk of PD than normal weight, who had a higher risk of PD than obese individuals, irrespective of the presence/absence of DM. Indeed, the highest risk of PD was among underweight people who also had more than 5 years’ diagnosis of DM (hazard ratio, 3.78). Importantly, the researchers examined the possibility that confounders, such as smoking status, alcohol, hypertension, or dyslipidaemia, might be relevant, finding that the association between low BMI and PD was maintained. Although the researchers also made efforts to minimize the possible effects of reverse causation by stratifying analyses with a 3and 5-year lag before diagnosis, the preclinical phase of PD may occur up to 20 years before diagnosis, which may cause loss of initiative, changes in eating behavior, and olfactory function, which may lead to a reduced caloric intake. In a parallel study of 1.3 million adults examining the association of BMI and dementia, the researchers found that lower BMI was similarly associated with an increased risk of dementia, but only when BMI was assessed less than 10 years before diagnosis. This association was reversed when BMI was measured at 15 and 20 years before dementia diagnosis, providing some support for two separate processes—that higher BMI in middle age may be linked to an increased risk of dementia, but that preclinical metabolic changes may explain the link between lower BMI and dementia. Therefore a long follow-up period (e.g., 20 years) might be necessary to better evaluate the true risk of PD from mid-life obesity. Previous epidemiological studies have yielded conflicting results regarding the association between BMI and PD, but in general have included only a fraction of the number of participants compared with the Jeong and colleagues study. In the only other very large study (2 million participants) followed prospectively, the identical association between low BMI and increased PD risk is also reported. Supporting the possibility that low BMI may indeed have a causal relationship with PD comes from a previous Mendelian randomisation study which reported that genetic risks for higher BMI were associated with a lower risk for PD. So if these relationships are not related to bias, confounding, or reverse causation, how might they otherwise be explained? The researchers do not specifically discriminate between type 1 DM (an autoimmune © 2020 International Parkinson and Movement Disorder Society