Osteoporosis and dementia: establishing a link

Patients with dementia often fall and frequently fracture. Alzheimer’s Disease (AD) is the most prevalent form of dementia, and patients with AD often have profound changes in body composition as the disease progresses¹, including the observed association between AD, osteoporosis, and greater fracture risk^{2, 3}. In addition, fracture has been shown to be an independent risk factor for dementia⁴. While patients with osteoporosis and dementia have increased risk for morbidity and mortality, those suffering with the two diseases may experience even greater consequences. For instance, the risk of hip fracture in those with dementia is increased by up to three-fold⁵, and for those with end-stage dementia that sustain a hip fracture, six-month mortality was found to be 55% compared with 12% for cognitively intact patients (adjusted hazard ratio, 5.8)⁶. Thus dementia patients are not only at increased risk of fracture, but also potentially at increased risk of mortality should they sustain a fracture. The reasons for the clinical association between dementia and osteoporosis may extend beyond their increased incidence with aging. Common risk factors apart from age include physical activity, vitamin D, and sarcopenia⁵, and symptomatic treatments for AD can increase the risk for falls. Nevertheless, the current evidence establishing a causal link between the two diseases of aging has been limited, to a large extent, to cross-sectional studies, and there are a paucity of data implicating a common underlying biology.

In this issue of the Journal, Bliuc et al. describe their study of the longitudinal association between cognitive decline and rate of bone loss. Their study included 1,741 women and 620 men 65 years and older from the Canadian Multicentre Osteoporosis Study (CaMos), and had two aims: 1) to find the longitudinal relationship between cognitive decline and rate of bone loss, and 2) to determine if the rate of cognitive decline measured at the beginning of the study predicted future fracture risk. Cognitive decline was measured using Mini Mental State Examination (MMSE) scores, and the rate of bone loss was measured by the change in femoral neck DXA-derived BMD, with each measure annualized based on at least two measurements. They conducted a ten year study with an additional five years of fracture follow-up, and models were adjusted for many covariates include age, demographic variables, education, comorbidities, lifestyle factors, and Medical Outcome Study Short Form-36 components. The authors showed a significant association of decline in cognitive score with bone loss in women in multivariate adjusted models with an estimated 6.5% (CI 3.2%, 9.9%) decline in femoral neck BMD for each percentage point decline in MMSE from baseline. The association was smaller and not statistically significant in men. Additional significant covariates included age, height, activity level, prior fracture, and comorbidities. The authors then measured the risk of osteoporotic fracture for women with a cognitive decline of at least 3 points on the MMSE scale in the first five years of the study compared to women with minimal cognitive decline. Women with a 3 point or more decline experienced an increased risk of osteoporotic fracture with a hazard ratio of 1.61 (95% CI, 1.11– 2.34) in a multivariable adjusted model. This analysis was not performed in men due to insufficient sample size.

The strong association between bone loss and dementia in women unexplained by a comprehensive set of confounders raises the possibility of an underlying biologic mechanism linking the two diseases. Estrogen exposure may be an important common factor as it has been linked to cognitive decline through either endogenous exposure (natural or surgical menopausal age) or treatment with exogenous hormone replacement therapy^{7, 8}, and has well known associations with bone health^{9, 10}. Indeed, studies often consider BMD as a measure of lifetime exposure to estrogen in measuring its impact on dementia and brain function^{11, 12}. In addition to estrogen, immunological factors may contribute to both bone loss¹³ and dementia¹⁴. In addition, bone-derived proteins and cells may influence AD progression¹⁵, with AD progression associated with serum levels of the bone turnover markers osteopontin, osteocalcin, and sclerostin¹⁵, and AD patients and patients with mild cognitive impairment (MCI) reporting higher levels of osteopontin¹⁶ that correlate with reduced BMD and cognitive decline^{17, 18}. Also, AD or dementia-related genes may impact bone or other peripheral systems. For instance, Bliuc et al. reference a study in which gene expression of the AD-related genes APP and BACE1 in the bone of hip fracture patients is positively associated with bone resorption and negatively associated with bone loss surrogate measures¹⁹. In addition, TREM2 is mutated in Nasu-Hakola disease, causing low bone density.^{20, 21} This gene is also known to regulate osteoclastogenesis²², and is important in AD, especially in the microglia^{23, 24}. The gene FNDC5, or Irisin which results through cleavage of the protein precursor, has been shown to have exercise-mediated effects on bone^25–27 and to have important effects in AD^28–30. Finally, the gene alpha synuclein (SNCA), which has well known effects in Lewy Body Dementia³¹ and AD^{32, 33}, has also shown effects in bone³⁴ and in glucose regulation³⁵.

Genetic approaches may be able to identify candidates for causal mechanisms and test causality. The shared genetic component (h²) between femoral neck BMD and AD has been estimated to be 13% (standard error = 9.4%)³⁶ as calculated using LD score regression³⁷. In a recent GWAS study of estradiol levels in men and women, a number of loci were discovered, and a Mendelian Randomization (MR) analysis was used to find evidence for a causal effect of estradiol on increased BMD in both men and women³⁸. A logical next step would be to examine the association of estradiol with dementia using MR with these loci, or alternatively to perform a multivariable MR with both conditions. These initial studies into genetic or molecular mechanisms underlying the brain bone connection should stimulate new research to understand potential shared biology.

The clinical implications of these findings are important. As noted in the Bliuc study, only 5% of hip fracture patients with dementia receive appropriate treatment for osteoporosis compared with treatment rates of about 30% in the general community³⁹. Thus paradoxically, those at highest risk of morbidity are undertreated. Better epidemiological data on the interactions between osteoporosis and dementia will also help clinicians to develop more specific clinical management regimes for both conditions. Increased surveillance for dementia symptoms may be warranted for osteoporosis patients if the evidence base becomes strong enough to support such an association.

The strengths of the Bliuc study are that it is a multi-site study across Canada with longitudinal measurements of exposures and outcomes, and comprehensive adjustment for important confounders. Second, the results of this study are consistent with the smaller previous studies, thereby reinforcing the association^{12, 40, 41}. However, there are some limitations. For example, the sample size of men was too limited to draw inferences about dementia and bone loss in males. Importantly, there were no details about race and ethnicity. African Americans have a greater prevalence of AD than whites⁴², hence any conclusions from this study must be tempered by the absence of data in this and other underserved populations. Finally, cause and effect cannot be ascertained from these association studies.

Nevertheless, the study by Bliuc et al. takes us one step further in understanding the connection between bone loss and dementia. Given the high prevalence of each condition and the increasing rates in our aging population, particularly among underserved populations, further research is warranted into the shared etiology of these diseases and into clinical management workflows that include joint management of fracture and dementia risks for appropriate patient subgroups.