Serum cholesterol concentration and risk of brain cancerBMJ 1995; 310 doi: https://doi.org/10.1136/bmj.310.6976.367 (Published 04 February 1995) Cite this as: BMJ 1995;310:367
- Correspondence to: Dr Herrinton.
- Accepted 17 October 1994
Two recent epidemiological studies have reported a positive relation between cholesterol concentration and brain cancer. In one cholesterol concentration was measured at the time of diagnosis,1 and in the other mortality was observed five or more years after the measurement.2 In both studies brain cancer could have influenced cholesterol concentration. The Kaiser Permanente medical care programme obtained serum cholesterol measurements in roughly 160000 members as part of multiphasic health examinations conducted during 1964–72. We examined the risk of developing malignant brain cancer five or more years after the cholesterol measurement to evaluate further their relation.
Subjects, methods, and results
The participants were members of the Northern California programme who were aged 15 years and over and had completed a multiphasic health examination during 1964–72.3 Follow up began five years after the date of the cholesterol measurement, so that prevalent cancers would be excluded, and ended on the date of diagnosis of a malignant brain tumour (code 191 of the ninth revision of the International Classification of Diseases), the date of termination of membership, or 31 December 1991. Cancer diagnoses were identified by the Northern California Cancer Center, which manages a population based registry that has participated in the United States National Cancer Institute's surveillance, epidemiology, and end results programme since 1973, and by review of the computerised hospital discharge records of the Kaiser Permanente medical care programme to confirm the diagnoses of primary malignant brain tumours. Five controls, individually matched for age (within five years), year of examination, and sex, were selected for each case. The case-control design provided valid estimates of the relation between serum cholesterol concentration and risk of malignant brain cancer and was easier and less costly to conduct than a cohort study. Both cases and controls were required to be members of the Kaiser Permanente medical care programme at the time of diagnosis of cancer in the case.
We considered men and women separately because they differ in terms of their risks of developing brain cancer and the distributions of their age specific cholesterol concentrations. Serum cholesterol concentration, available for 91% of those who received a multiphasic health examination, was coded into six categories containing roughly equal numbers of male and female subjects. To test for a trend cholesterol concentration was left as a continuous variable. Conditional logistic regression was used to compute the odds ratio and 95% confidence interval, with examination of age, race, educational attainment, marital status, history of cigarette smoking, and alcohol intake in the year before the examination as potentially confounding variables. These variables had been considered in previous studies of the hypothesis.3
The case group consisted of 85 women and 123 men. The proportion of cases and controls in each category of serum cholesterol concentration was similar in both sexes, and there was little association between cholesterol concentration and subsequent risk of developing malignant brain cancer (table). Women with serum cholesterol values greater than 258 mg/dl (6.67 mmol/l) were at a somewhat reduced risk relative to women with lower values, but a trend in risk was not apparent (P value for trend 0.10).
As in two earlier studies in which the incidence4 and mortality5 of brain cancer were examined several years after serum cholesterol concentration was measured we found little evidence for a relation between cholesterol value and subsequent brain cancer. Our case group was restricted to people with malignant tumours. A relation with subsequent benign tumours may exist, but a more likely explanation for the difference between our results and the positive associations reported earlier stems from the timing of the serum cholesterol measurement. Further study of an aetiological role of cholesterol in the development of brain cancer should be conducted in populations in which cholesterol concentration has been measured before the onset of the disease.
We thank Ms Donna Wells for programming assistance. This study was supported by grant No R35 CA49761 from the US National Cancer Institute.