The new england journal of medicine 1260 n engl j med 351;12 www.nejm.org september 16, 2004 by Schnyder et al. did not include subjects with homocysteine levels higher than 13.5 µmol per liter. Studies are needed that will test the efficacy of homocysteine-lowering vitamin regimens containing betaine instead of folate. EGFR Mutations and Sensitivity to Gefitinibto the editor: The important study by Dr. Lynch and colleagues (May 20 issue) 1 suggests that specific mutations in the epidermal growth factor receptor (EGFR) characterize a subgroup of non-smallcell lung cancers that may be highly responsive to gefitinib therapy. Do these mutations predict a greater sensitivity to chemotherapy as well? The overall objective response rate to first-line combination chemotherapy for metastatic non-small-cell lung cancer is about 20 percent. 2 Only tumors from a small cohort of patients who had a response to gefitinib were studied for the specific mutations, but all patients except one had also received prior chemotherapy. Although the authors describe Patient 6 as "representative" of the cohort, the percentage of other patients who previously had a response to chemotherapy is not reported. If the rate of response to first-line chemotherapy was high for the other patients in the cohort who had a response to gefitinib, the specific mutations may be predictive of either chemotherapy or gefitinib sensitivity, thus identifying a distinct subgroup of patients with non-smallcell lung cancer. to the editor: Lynch et al. and Paez et al. 1 report that mutations in the EGFR kinase domain in lung cancers are associated with responsiveness to gefitinib. We performed a mutational analysis of the EGFR kinase region on tumor tissue from nine patients with an event-free survival of more than 24 weeks in our phase 2 trial of gefitinib in patients with glioblastoma. 2 No mutations affecting the amino acid sequence in the kinase region were detected. However, our experience with EGFR immunolocalization in brain and lung tumors indicates that the cytoplasmic and membranous localization of wild-type EGFR and the constitutively active mutant EGFRvIII in brain tumors as compared with only membranous localization in lung tumors supports additional differences in the biology of EGFR between these tumor systems (McLendon R: personal communication). In summary, EGFR in glioblastoma did not have mutations in the kinase region, and any activity of gefitinib in glioblastoma would occur through an alternative mechanism reflective of important pathophysiological differences between glioblastomas and lung carcinomas.

Because humans are invariably exposed to complex chemical mixtures, estimating the health effects of multi-pollutant exposures is of critical concern in environmental epidemiology, and to regulatory agencies such as the U.S. Environmental Protection Agency. However, most health effects studies focus on single agents or consider simple two-way interaction models, in part because we lack the statistical methodology to more realistically capture the complexity of mixed exposures. We introduce Bayesian kernel machine regression (BKMR) as a new approach to study mixtures, in which the health outcome is regressed on a flexible function of the mixture (e.g. air pollution or toxic waste) components that is specified using a kernel function. In high-dimensional settings, a novel hierarchical variable selection approach is incorporated to identify important mixture components and account for the correlated structure of the mixture. Simulation studies demonstrate the success of BKMR in estimating the exposure-response function and in identifying the individual components of the mixture responsible for health effects. We demonstrate the features of the method through epidemiology and toxicology applications.

BackgroundHeart disease causes 45% of the deaths that occur among U.S. firefighters while they are on duty. We examined duty-specific risks of death from coronary heart disease among on-duty U.S. firefighters from 1994 to 2004. MethodsWe reviewed summaries provided by the Federal Emergency Management Agency of the deaths of all on-duty firefighters between 1994 and 2004, except for deaths associated with the September 11, 2001, terrorist attacks. Estimates of the proportions of time spent by firefighters each year performing various duties were obtained from a municipal fire department, from 17 large metropolitan fire departments, and from a national database. Odds ratios and 95% confidence intervals for death from coronary heart disease during specific duties were calculated from the ratios of the observed odds to the expected odds, with nonemergency duties as the reference category. ResultsDeaths from coronary heart disease were associated with suppressing a fire (32.1% of all such deaths), responding to an alarm (13.4%), returning from an alarm (17.4%), engaging in physical training (12.5%), responding to nonfire emergencies (9.4%), and performing nonemergency duties (15.4%). As compared with the odds of death from coronary heart disease during nonemergency duties, the odds were 12.1 to 136 times as high during fire suppression, 2.8 to 14.1 times as high during alarm response, 2.2 to 10.5 times as high during alarm return, and 2.9 to 6.6 times as high during physical training. These odds were based on three estimates of the time that firefighters spend on their duties. ConclusionsCertain emergency firefighting duties were associated with a risk of death from coronary heart disease that was markedly higher than the risk associated with nonemergency duties. Fire suppression was associated with the highest risk, which was approximately 10 to 100 times as high as that for nonemergency duties.