Bertazzi et al. Respond to Smith and Lopipero

Abstract

Received for publication February 6, 2001, and accepted for publication February 21, 2001. Abbreviation: TCDD, 2,3,7,8-tetrachlorodibenzo-p-dioxin. 1 Department of Occupational and Environmental Health, EPOCA, Research Center for Occupational, Clinical and Environmental Epidemiology, University of Milan, Milan, Italy. 2 Unit of Epidemiology, Department of Occupational Health and Safety, Istituti Clinici di Perfezionamento, Milan, Italy. 3 Regional Epidemiological Office, Health Directorate, Regione Lombardia, Milan, Italy. Reprint requests to Dr. Pier Alberto Bertazzi, EPOCA, Epidemiology Research Center, Department of Occupational and Environmental Health, University of Milan, 8 Via San Barnaba, 20122 Milano, Italy (e-mail: pieralberto.bertazzi@unimi.it). We thank Smith and Lopipero for their invited commentary (1), which we found appropriate and useful for interpreting the latest Seveso, Italy, findings (2). Human exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) is widespread (3). Epidemiologic studies are essential to assessing the risk linked to the environmental contamination, and consistency in the findings enhances confidence in their results. It is not surprising that, on average, exposure to TCDD in Seveso was lower than in occupational cohorts. In the former case, the chemical release was limited in time, and its content eventually was diluted over an extended outdoor area (4). Still, the average TCDD blood concentration in the Seveso population was two orders of magnitude higher than the reported background environmental level (1). Thus, the finding of increased cancer mortality cannot be considered totally unexpected. An increase in all-cancer and lung cancer mortality, consistent with what was observed in occupational cohorts, was found in men in zones A and B after 15 years. Smith and Lopipero (1) suggest that smoking might be the actual explanation. This is a possibility, and we intend to further explore smoking habits in our index and referent populations. However, if the increased risk was due to smoking, one would expect high rate ratios for the whole study period, not only 15–20 years since dioxin exposure. With regard to concomitant potentially smoking-related findings, several other studies linked increased cardiovascular mortality to dioxin exposure, sometimes with a dose-related pattern (5–7) and use of internal reference groups (6, 7), thus leaving little room for confounding arguments. In addition, other smoking-related cancers (e.g., larynx, esophagus, pancreas, and bladder) were not elevated in Seveso males after 15 years (2). To be a confounder, smoking should be distributed unevenly among zones. Independent surveys assessed smoking habits in 1986–1987 in random samples of the population of Desio (352 subjects), one of the cities affected by the Seveso accident, and in the population in a nearby district outside the study area (466 subjects), quite similar to our reference population. Distribution of the subjects by smoking category in the two samples was as follows: never smokers, 76 (22 percent) in Desio versus 104 (22 percent) in the outside area; former smokers, 107 (30 percent) versus 154 (33 percent); current smokers of 20 cigarettes/day, 40 (11 percent) versus 36 (8 percent) (8).This, although limited, set of data suggests that geographic variation in smoking habits might be negligible across subareas in the region including the accidentally contaminated territory. Although we admit that with so many data “there is a little bit of something for everybody” (1, p. 1046), results for lymphatic and hemopoietic neoplasms clearly do stand out. The pattern suggests a systematic difference in mortality from these causes between the TCDD-exposed and the local reference population. Cancer incidence data for the first postaccident period support these findings (9). The nonHodgkin’s lymphoma increase became significant after 15 years (five observed; rate ratio 2.8, 95 percent confidence interval: 1,1, 7.0). An increased occurrence of nonHodgkin’s lymphoma also has been observed in occupational cohorts (6, 10, 11). On the basis of our interpretation, the 20-year mortality results do show that the Seveso population experienced, consistent with other exposed populations, a modest excess of all-cancer and respiratory cancer mortality long after initial exposure and a moderately increased occurrence of lymphatic and hematopoietic cancer. This pattern is different from what might have been expected on the basis of animal experiments. A limited number of persons probably had extremely high levels of exposure, and susceptibility factors might also have played a role.