American Journal of Epidemiology Advance Access published online on July 2, 2007
American Journal of Epidemiology, doi:10.1093/aje/kwm200
| ||||||||||||||||||||||||||||||||||||||||||||||||||
RE: "META- AND POOLED ANALYSIS OF GSTT1 AND LUNG CANCER: A HUGE-GSEC REVIEW"
We thank Dr. Weiss (1) for pointing out a discrepancy between the results reported in table 9 of our paper (2) and what is known about the positive association between asbestos exposure and lung cancer.
|
We carefully checked our original analysis of the data used to produce the second half of table 9 and found that this section of the table (which refers to the association between occupational exposure to asbestos, GSTT1 deletion, and lung cancer) is in error because of incorrect use of the data set.
In the original analysis, subjects included in the cells labeled as being without occupational asbestos exposure were actually subjects without occupational exposure in general but for whom specific information on asbestos exposure was not collected. When the analysis was restricted to those nine studies that did collect information on asbestos exposure, we found that asbestos exposure is indeed significantly associated with lung cancer in subjects carrying GSTT1, and less so in exposed subjects carrying the GSTT1 deletion, suggesting a negative (but not significant) interaction between asbestos exposure and GSTT1 deletion. The corrected table is shown below.
The authors and the Journal regret the error.
References
- Weiss NS. Meta- and pooled analysis of GSTT1 and lung cancer: a HuGE-GSEC review. (Letter). Am J Epidemiol (2007) 165:1462.
[Free Full Text] - Raimondi S, Paracchini V, Autrup H, et al. Meta- and pooled analysis of GSTT1 and lung cancer: a HuGE-GSEC review. Am J Epidemiol (2006) 164:1027–42.
[Abstract/Free Full Text]
EDITORIAL NOTE
Weiss (1) recently pointed out an interesting methodological issue regarding the pooled analysis of the glutathione S-transferase theta 1 (GSTT1) gene and lung cancer conducted by Raimondi et al. (2). The authors concluded that "GSTT1 appeared to modulate occupational-related lung cancer, at least for asbestos exposure" (2, p. 1039). Weiss pointed out, however, that if the original data (table 9) were examined more closely, they would seem to indicate that asbestos exposure is associated with a reduction in risk of lung cancer. Given that the opposite results have been obtained in many prior studies, Weiss concluded that it is unlikely that Raimondi et al.'s observations pertaining to asbestos are valid. He also suggested that the finding of gene-environment interaction becomes questionable as a result. In the published response to the letter by Weiss, Emanuela Taioli did not address this point (3). Subsequently, the author redid the analysis and found an analytic error, outlined in the erratum above.The editors would like to point out the importance of continued scrutiny of the Journal's articles by authors, reviewers, and readers. When authors report interaction but the average association ("main effect") seems implausible, it should raise a red flag regarding the validity of the results. We are thankful for the persistence of an astute reader in uncovering fixable errors in epidemiologic analysis and inference.
References
- Weiss NS. Meta- and pooled analysis of GSTT1 and lung cancer: a HuGE-GSEC review. (Letter). Am J Epidemiol (2007) 165:1462.
[Free Full Text] - Raimondi S, Paracchini V, Autrup H, et al. Meta- and pooled analysis of GSTT1 and lung cancer: a HuGE-GSEC review. Am J Epidemiol (2006) 164:1027–42.
[Abstract/Free Full Text] - Taioli E. An author replies. (Letter). Am J Epidemiol (2007) 165:1462–3.
[Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||