Introduction Dementia is a common disease

Introduction Dementia is a common disease in the elderly, and a significant source of disability, healthcare spending, and long-term care admissions. There is evidence to suggest that testosterone plays a role in cognitive functioning, and lower serum levels correlate with higher amyloid deposits (associated with Alzheimer's disease) [1]. A recent meta-analysis suggests that complete androgen deprivation (used for prostate cancer patients) may be associated with an increased risk of dementia [2]. The medications finasteride and dutasteride (5 alpha reductase inhibitors [5ARI]) are used to treat urinary symptoms from an enlarged prostate, and their mechanism of action is through a serine threonine protein kinase of dihydrotestosterone (DHT) levels. Lower DHT levels have been correlated with cognitive decline in both animal models and human studies, perhaps through a reduced ability to regulate neurosynaptic plasticity [3], [4]. Our objective was to determine if the use of 5ARIs in older men was associated with an increased risk of new onset dementia.
Methods We used multiple linked administrative databases from Ontario, Canada, which has 13 million residents with universal healthcare and medication coverage (for those >65years old). We recently described the association of suicidality and depression with the use of 5ARIs and the same methodology was utilized in this study (full methods and data-set/variable descriptions are detailed in this reference) [5]. For this analysis, men≥66years of age who had a new outpatient prescription for a 5ARI between 2003 and 2013 were matched based on numerous characteristics to men who did not use a 5ARI. In order to make these two groups as comparable as possible, we originally measured a total of 96 different covariates for the 5ARI users and non-users. These included general demographic characteristics, medical comorbidities, recent medication usage, and healthcare utilisation. We then used standardised differences≥7% to identify variables which were potentially different between 5ARI users and non-users [6]. Of the 96 variables, 44 were included in the propensity score, and patients were matched 1:1 using greedy matching based on year of index date, prior depression/self-harm or antidepressant utilisation, and the logic of the propensity score. In order to assess the risk of dementia, we used this previously described cohort, and excluded matched pairs with prior evidence of dementia based on a validated administrative data definition [7] (one or more hospital admission or two outpatient physician visits within 6months, positive predictive value 73%) and supplemented this definition with medications specific to dementia. As such, the ensuing work is presented using a subset of the original study [5]. Administrative data definitions for covariates relevant to dementia from our previous study were re-measured, and additional covariates for Down's syndrome, Parkinson's, and traumatic brain injury were added. These datasets were linked using unique encoded identifiers and analyzed at the Institute for Clinical Evaluative Sciences (ICES). Our primary outcome was a diagnosis of dementia during continuous usage of the 5ARI or within 12months of discontinuing the medication. Stratified Cox proportional hazards models accounting for matched pairs were used, and stratification of the followup time periods was used to ensure the proportionality (SAS 9.4). A subsequent analysis was done to explore possible differential effects between the two types of 5 alpha-reductase inhibitors, and an interaction term was used in the model to evaluate for significant differences. A two-sided p-value <0.05 was considered significant. This study was approved by the institutional review board at Sunnybrook Health Sciences Centre, Toronto, Canada.
Discussion To our knowledge, the association between 5ARI use for an enlarged prostate and dementia has not been investigated previously. In this population-based study of older men using 5ARIs, we did find a significant association with dementia. However, the strength of the association decreased with increased 5ARI exposure, and was non-significant among men using this medication for the longest period. These results are likely explained by lead time bias, whereby men with pre-clinical dementia sought treatment for urinary symptoms prior to being diagnosed with dementia (as has been shown to occur with a similar risk magnitude in a prior study) [8]. This is consistent with the recent results of a randomised trial which demonstrated that older men with age associated memory impairment and low serum testosterone did not have significant improvements in their memory after 1year of testosterone therapy, compared to placebo treated patients [9]. This would argue that previous studies demonstrating an association between testosterone levels and cognitive changes do not represent a cause and effect relationship. However, it is possible that 5ARIs are associated with a reduction in cognitive reserve in some elderly patients, leading to the development of clinically apparent dementia. Limitations of our study include the observational study design (with risk of differential misclassification, and residual confounding). Further research into the effects of this medication class on cognition seems to be warranted.