Reduced cancer incidence in Huntington's disease: record linkage study clue to an evolutionary trade‐off?

To the Editor : Neurodegenerative disorders probably arose long before printed records, but poor preservation of neural tissue limits paleoneurobiology. Given the typical emergence of such disorders in post-reproductive life, they might represent an ‘unintended’ consequence of increased life expectancy. However, it is possible that there are associated survival advantages that operate during the fertile years, which might then be linked to evolutionary persistence. CAG trinucleotide-repeat disorders result in a series of glutamine residues that form a polyglutamine (polyQ) tract. PolyQ diseases are associated with neurodegenerative syndromes, including Huntington’s disease (HD), X-linked spinobulbar muscular atrophy (SBMA, Kennedy’s disease) and some forms of spinocerebellar ataxia (SCA). A study in Denmark found a lower-than-expected incidence of cancer among HD patients. The authors postulated a possible apoptotic effect of the polyQ tract in the mutant protein huntingtin expressed widely in tissues (1). A Swedish study confirmed a reduction in cancer more widely among those with polyQ diseases (2). We used two large hospital record databases to study this. The Oxford Record Linkage Study (ORLS) research group has built a file of all-England linked national hospital episode statistics (LHES), including linkage to mortality data, spanning 1999–2010 (population 52 million). The original ORLS spans 1963–1998 but covers a much smaller population. Both datasets include anonymised statistical data on all National Health Service hospital admissions, including day cases, and all deaths in the populations they cover. Analysis of both datasets was approved by the English Central Office for Research Ethics Committees (reference 04/Q2006/176). We constructed cohorts of records of people coded with HD (ICD codes 331.0, 333.4, G10+F02.2, in the 8th, 9th and 10th revisions; 4865 people in the LHES cohort), SBMA (codes 335.2, G12.1, available in the 9th and 10th revisions only; 599 people) or hereditary ataxia (HA, codes 332, 334, G11, 8th to 10th revisions; 9084 people), using their first recorded admission for each disease. HA was used as the search term because of the absence of specific codes for SCA [as in (2)]. We constructed comparison cohorts, used in a similar study of cancer in relation to motor neuron disease (3). We calculated rate ratios of subsequent cancer, as measures of relative risk, comparing cancer in the HD, SBMA and HA cohorts with the reference cohorts (3). Our results for the all-England population showed a significantly reduced incidence of cancer in HD, but not for SBMA or HA (Table 1). The rate ratio for cancer in people with HD in the all-England cohort was 0.71 (95% confidence interval 0.61–0.83). Subdivision of the analysis by time between admission for all polyQ diseases and cancer showed elevated cancer risks in the first year after polyQ admission, the most likely explanation being surveillance bias, i.e. the detection of cancer because the patient was under care for the polyQ condition. Excluding cancer cases within the first year, cancer in people with HD remained significantly low, with a rate ratio of 0.53 (0.42–0.65). Numbers for individual cancers were small but the overall low rate ratio seemed to result from the sum of lows for several cancers, most of which were individually non-significant. Numbers in the ORLS were very small but the rate ratio for HD (but not for SBMA or HA) was also low at 0.50 (0.14–1.27, based on four observed and eight expected cases). Notwithstanding generic caveats about studies based on routinely collected administrative data (2, 3), we provide additional evidence that HD is associated with a reduced incidence of cancer. The absence of an effect in SBMA and HA [despite using the same ICD codes as (2)] calls into question a generic polyQ effect, possibly indicating additional factors unique to HD. A potentially modifying role for the CAG repeat length also warrants investigation. The reduction in cancer incidence in HD was seen despite the observation that smoking is more common in presymptomatic HD gene carriers (4). As well as implications for novel cancer prevention strategies, this finding may inform understanding of the evolutionary persistence of HD (5), and prompt wider consideration of similar ‘trade-offs’ between evolutionary gain and the risk of neurodegenerative diseases.