||5R03CA115234-02 Interpret this number
||University Of Southern California
||Risk Factors for Non Hodgkin's Lymphoma in Twins
DESCRIPTION (provided by applicant): The incidence of non-Hodgkin's lymphoma (NHL) is increasing over time and little is known about risk factors for non-HIV NHL. There is some suggestion that alcohol in moderation may be protective, but this finding is not consistent across studies. Over 10 studies have examined the effect of cigarette smoking on NHL risk and although some find an increased risk associated with a particular subtype (follicular), many other studies find no association. Both alcohol and tobacco are behaviors that can be manipulated, so if either is related to risk, there are important public health consequences. On the other hand, these factors are also highly associated with socioeconomic status and therefore subject to confounding. Twin pairs in whom one twin developed NHL and the other did not offer ideal subjects with which to study the effect of tobacco and alcohol on risk. We developed and maintain two twin registries. One is a volunteer registry of twins with cancer and chronic disease (International Twin Study, with 17,000 twin pairs), and the other is a population-based registry (California Twin Study) with 136,000 pairs. We have received completed questionnaires from 208 pairs of NHL affected twins from the International Twin Study, and 136 pairs from the California Twin Program, for a total of 344 disease-discordant twin pairs. We propose a case-control study in which the case is the twin with NHL and the control is the unaffected twin. Comparisons will include absolute and relative difference between twins with respect to alcohol and tobacco use, body mass index, cumulative number of x-rays, and history of allergy or asthma. There are 50-180 twin pairs discordant for each of these exposures, which will allow us 80% power to detect and odds ratio as low as 0.6 and as high as 1.8. A matched case-control analysis using SAS will be performed using multivariate logistic regression.