Re: Familial clustering of Hodgkin lymphoma and multiple sclerosis.

We were intrigued by the recent report of Hjalgrim et al. (1) that showed familial clustering of multiple sclerosis and young-adult-onset Hodgkin lym-phoma in Denmark and that supported Newell ' s 1970 hypothesis (2) of a shared etiology. Their study and another report by Vineis et al. (3) indicate a familial association between multiple sclerosis and non-Hodgkin lymphoma. Given the etiologic importance of these fi ndings, we wanted to evaluate this association further in the larger population of Sweden. Because there is no national multiple sclerosis registry in Sweden, unlike the situation in Denmark, we analyzed hospital discharge diagnoses of multiple sclerosis in relatives of patients with Hodgkin lymphoma and of patients with non-Hodgkin lymphoma compared with relatives of matched control subjects taken from our ongoing studies of familial aggregation of lym-phoproliferative tumors. Using our hospital registry approach, we also analyzed our Danish data in the same manner and extended the sample to include relatives of patients with non-Hodgkin lymphoma. The Swedish Family-Cancer Database has been described previously (4 , 5). For these analyses, patients with lym-phoma, control subjects, and their relatives were linked with the Swedish Hospital Discharge Register from January 1, 1964, through December 31, 2000, which contains individual-based information on discharges from inpatient care. This register has population-based (county-wise) coverage that encompassed more than 90% of the population of Sweden after the mid-1970s and 100% since 1987. We obtained information on all discharges listing multiple sclerosis (International Classifi cation of Disease [ICD], versions 7 – 10) including dates of discharge. The Danish family data have been described previously (5) ; patients with lymphoproliferative disease, control subjects, and their relatives were linked with the Danish Hospital Discharge and Outpatient Registers from January 1, 1977, through December 31, 1999, as described above. The population of relatives that we identifi ed is essentially the same as that reported by Hjalgrim et al. (1) , which is expected because we used similar methodology to fi nd relatives of patients with Hodgkin lymphoma from the population databases. The numbers of patients, control subjects, and their relatives are shown for each population in Table 1. The outcome was a recorded diagnosis of multiple sclerosis. We tested for increased risk of multiple sclerosis in relatives by using a marginal survival model with a robust variance estimate to account for familial dependencies (6). We use relative risk (RR) to denote the hazard ratio. Risk was …