The Relationship between Cancer and Alzheimer's Disease

To the Editor:—We write to describe the relationship, found at autopsy, between cancer and Alzheimer's disease. It is surprising that the autopsy relationship between two such common and well studied diseases has been so little examined before. This may be because, in recent years, the study of Alzheimer's disease has been based on brain-limited autopsies. The only published autopsy series in which both conditions were defined is that of Corsellis.1 He does not comment on the relationship, but his results show that subjects with Alzheimer's disease had significantly less cancer than a control group with functional psychosis. We examined data from 210 consecutive autopsies at the Willard Psychiatric Center over a 10-year period. In 64 cases there was definite Alzheimer's disease, meeting Kachaturian criteria.2 In 61 cases no Alzheimer's disease at all could be found. In the remaining cases there was slight or moderate Alzheimer's disease. We compared the prevalence of cancer in the definite Alzheimer's cases with the prevalence in those with no Alzheimer's disease and found the following results: Patients with Alzheimer's disease (n = 64). Mean age 83.2 (SD 9.5) 38 female, 26 male. Six had cancer contributing to death (stomach, colon, cecum, bladder, prostate, small cell cancer of lung). Patients without Alzheimer's disease (n = 61). Mean age 74 (SD 12.6) 12 female, 49 male. Fifteen had cancer contributing to death (stomach, two colon, three prostate, three of lung (squamous, adenocarcinoma, anaplastic), pancreas, melanoma, leukemia, lymphoma, glioblastoma, anaplastic metastases to liver and lungs with undiagnosed primary). The difference in prevalence was significant (Chi-square with Yates correction 3.97, P < 0.05). It would be premature to speculate on biological reasons for the relationship. This trend for those with Alzheimer's disease to have less cancer may be due to demographic variables. For example, Alzheimer's disease largely affects elderly females, who have a very low prevalence of smoking. Confirmation of a true inverse relationship between these two conditions would, however, be interesting, and we should be glad to hear if any other readers of JAGS have studied this.