Estrogen plus Progestin and Colorectal Cancer in Postmenopausal Women
http://www.100md.com
《新英格兰医药杂志》
To the Editor: Chlebowski and colleagues (March 4 issue)1 claim that "short-term use" of progestin and estrogen reduces the risk of invasive colorectal cancer. For a valid conclusion regarding prevention, we need to know how many of the 115 women with invasive cancer were currently taking hormones and the total duration of use for any reason.
Most of the women in the placebo group had taken hormones: 10.7 percent started taking them during the trial, 25.6 percent had taken them previously (12.3 percent of whom had taken them for 10 years or more), and 42.5 percent had previously used oral contraceptives. At baseline, 84 more women in the placebo group than in the hormone group had relatives with colorectal cancer. Sixty-four of the 72 women in the placebo group who had invasive cancer were over the age of 59 years, 55 had a family history of colorectal cancer, and 17 had previously used postmenopausal hormones. The annualized rates of cancer associated with these characteristics in the placebo group were roughly double those in the hormone group, in which there were 43 cases of invasive cancer. Since hormone use declines with age, bias could explain the result. In addition, the disease may have been diagnosed at a more advanced stage in the hormone group than in the placebo group because progestin and estrogen use can thicken and dilate endometrial blood vessels, and similar changes in systemic blood vessels may hasten the spread of tumors.2
Ellen C.G. Grant, M.B., Ch.B.
20 Coombe Ridings
Kingston-upon-Thames KT2 7JU, United Kingdom
ellencggrant@onetel.net.uk
References
Chlebowski RT, Wactawski-Wende J, Ritenbaugh C, et al. Estrogen plus progestin and colorectal cancer in postmenopausal women. N Engl J Med 2004;350:991-1004.
Grant ECG. Relation between headaches from oral contraceptives and development of endometrial arterioles. Br Med J 1968;2:402-405.
To the Editor: Chlebowski et al. present seminal work showing a significant reduction in the rate of colorectal cancers in women who received postmenopausal hormone therapy as compared with women who received placebo. The numbers of regional, metastatic, and noninvasive tumors were similar in the two groups. There was a pronounced effect with respect to localized cancers: of the 8506 hormone-treated women, only 10 had localized cancer, whereas of the 8102 women in the placebo group, 36 had localized cancer. The question to be addressed is whether the carcinomas detected in the period after, for example, year 3 might be new carcinomas, which are found predominantly in the placebo group.
Kathrin Machens, V.M.D.
Karin Schmidt-Gollwitzer, M.D.
Schering
13342 Berlin, Germany
kathrin.machens@schering.de
Editor's note: Schering (Berlin) produces preparations of postmenopausal hormones.
To the Editor: Chlebowski and colleagues' updated analysis of the Women's Health Initiative (WHI) data affirms the effectiveness of postmenopausal hormone therapy in the prevention of colorectal cancer. After an average follow-up of 5.6 years, the risk of colorectal cancer was reduced by 44 percent among women who received hormones as compared with those who received placebo.
The authors conclude that the colorectal cancers in women who took hormones were diagnosed at a more advanced stage than were those in women in the placebo group. However, the data do not support this interpretation. In fact, the absolute number of advanced tumors in the hormone group was slightly reduced (7 per 10,000 woman-years, vs. 8 per 10,000 woman-years in the placebo group), and the absolute number of localized tumors was greatly reduced in the hormone group (2 per 10,000 woman-years, vs. 8 per 10,000 woman-years in the placebo group). The authors' interpretation is also implausible, given that the growth characteristics and histologic features of the tumors did not differ between the two groups. Therefore, the interpretation of the data should be subject to further debate.
Rolf Schürmann, M.D., Ph.D.
Maureen Cronin, M.D.
Jens U. Meyer, M.D., Ph.D.
Schering
13342 Berlin, Germany
Editor's note: Schering (Berlin) produces preparations of postmenopausal hormones.
The authors reply: Dr. Grant raises the issue of how prior hormone therapy may have introduced potential bias. In the randomized trial, the rates of prior hormone therapy in the estrogen-plus-progestin and placebo groups were very similar. Only 3.1 percent of the placebo group (not the 12.3 percent claimed) had used hormones during menopause for 10 or more years, and the exposure in the estrogen-plus-progestin group was identical. A sensitivity analysis in which data were censored six months after the first nonadherence identified a hazard ratio for colorectal cancer of 0.52 (95 percent confidence interval, 0.32 to 0.85). A model that adjusted for the duration of prior postmenopausal hormone therapy yielded a hazard ratio for colorectal cancer of 0.56 (95 percent confidence interval, 0.32 to 0.82). Therefore, prior exposure to postmenopausal hormone therapy was similar in the two groups, and the hazard ratio for colorectal cancer did not change after adjustment for prior hormone exposure and adherence or nonadherence to the study medication.
Drs. Machens and Schmidt-Gollwitzer suggest that the effect of estrogen plus progestin on the risk of colorectal cancer is seen largely during and after year 4. The hazard ratios (and 95 percent confidence intervals) for colorectal cancer in the estrogen-plus-progestin group, according to the study year, are 0.64 (0.26 to 1.56) for year 1, 0.95 (0.39 to 2.28) for year 2, 0.43 (0.13 to 1.41) for year 3, 0.46 (0.22 to 0.99) for year 4, 0.52 (0.18 to 1.56) for year 5, and 0.43 (0.16 to 1.16) for year 6. The P value for the trend of the effect on the risk of colorectal cancer over time is 0.34. Thus, there is no evidence that the effect of the hormone on colorectal cancer changes over time.
Our original statement that the proportion of colorectal cancers diagnosed at an advanced stage was higher in the hormone group than in the placebo group is correct. That said, we agree with Dr. Schürmann and colleagues that a major effect of estrogen and progestin on the risk of colorectal cancer was a reduction in the absolute number of localized cancers (hazard ratio for localized cancer, 0.26; 95 percent confidence interval, 0.13 to 0.53). However, the effect on regional tumors (hazard ratio, 0.76; 95 percent confidence interval, 0.44 to 1.30) and metastatic tumors (hazard ratio, 1.54; 95 percent confidence interval, 0.50 to 4.71) is less clear. However, as previously stated, even within the category of regional or metastatic disease, the cancers in the hormone group were associated with a greater number of positive nodes than were the cancers in the placebo group (3.6±4.2 vs. 1.6±2.1 nodes, P=0.012), and the number of deaths was very similar in the hormone and placebo groups (nine and eight, respectively). The biology underlining these findings and their clinical implications are unclear.
Rowan T. Chlebowski, M.D., Ph.D.
Harbor–UCLA Research and Education Institute
Torrance, CA 90502
rchlebowski@rei.edu
Rebecca J. Rodabough, M.S.
Fred Hutchinson Cancer Research Center
Seattle, WA 98109
Jean Wactawski-Wende, Ph.D.
University at Buffalo
Buffalo, NY 14214
Most of the women in the placebo group had taken hormones: 10.7 percent started taking them during the trial, 25.6 percent had taken them previously (12.3 percent of whom had taken them for 10 years or more), and 42.5 percent had previously used oral contraceptives. At baseline, 84 more women in the placebo group than in the hormone group had relatives with colorectal cancer. Sixty-four of the 72 women in the placebo group who had invasive cancer were over the age of 59 years, 55 had a family history of colorectal cancer, and 17 had previously used postmenopausal hormones. The annualized rates of cancer associated with these characteristics in the placebo group were roughly double those in the hormone group, in which there were 43 cases of invasive cancer. Since hormone use declines with age, bias could explain the result. In addition, the disease may have been diagnosed at a more advanced stage in the hormone group than in the placebo group because progestin and estrogen use can thicken and dilate endometrial blood vessels, and similar changes in systemic blood vessels may hasten the spread of tumors.2
Ellen C.G. Grant, M.B., Ch.B.
20 Coombe Ridings
Kingston-upon-Thames KT2 7JU, United Kingdom
ellencggrant@onetel.net.uk
References
Chlebowski RT, Wactawski-Wende J, Ritenbaugh C, et al. Estrogen plus progestin and colorectal cancer in postmenopausal women. N Engl J Med 2004;350:991-1004.
Grant ECG. Relation between headaches from oral contraceptives and development of endometrial arterioles. Br Med J 1968;2:402-405.
To the Editor: Chlebowski et al. present seminal work showing a significant reduction in the rate of colorectal cancers in women who received postmenopausal hormone therapy as compared with women who received placebo. The numbers of regional, metastatic, and noninvasive tumors were similar in the two groups. There was a pronounced effect with respect to localized cancers: of the 8506 hormone-treated women, only 10 had localized cancer, whereas of the 8102 women in the placebo group, 36 had localized cancer. The question to be addressed is whether the carcinomas detected in the period after, for example, year 3 might be new carcinomas, which are found predominantly in the placebo group.
Kathrin Machens, V.M.D.
Karin Schmidt-Gollwitzer, M.D.
Schering
13342 Berlin, Germany
kathrin.machens@schering.de
Editor's note: Schering (Berlin) produces preparations of postmenopausal hormones.
To the Editor: Chlebowski and colleagues' updated analysis of the Women's Health Initiative (WHI) data affirms the effectiveness of postmenopausal hormone therapy in the prevention of colorectal cancer. After an average follow-up of 5.6 years, the risk of colorectal cancer was reduced by 44 percent among women who received hormones as compared with those who received placebo.
The authors conclude that the colorectal cancers in women who took hormones were diagnosed at a more advanced stage than were those in women in the placebo group. However, the data do not support this interpretation. In fact, the absolute number of advanced tumors in the hormone group was slightly reduced (7 per 10,000 woman-years, vs. 8 per 10,000 woman-years in the placebo group), and the absolute number of localized tumors was greatly reduced in the hormone group (2 per 10,000 woman-years, vs. 8 per 10,000 woman-years in the placebo group). The authors' interpretation is also implausible, given that the growth characteristics and histologic features of the tumors did not differ between the two groups. Therefore, the interpretation of the data should be subject to further debate.
Rolf Schürmann, M.D., Ph.D.
Maureen Cronin, M.D.
Jens U. Meyer, M.D., Ph.D.
Schering
13342 Berlin, Germany
Editor's note: Schering (Berlin) produces preparations of postmenopausal hormones.
The authors reply: Dr. Grant raises the issue of how prior hormone therapy may have introduced potential bias. In the randomized trial, the rates of prior hormone therapy in the estrogen-plus-progestin and placebo groups were very similar. Only 3.1 percent of the placebo group (not the 12.3 percent claimed) had used hormones during menopause for 10 or more years, and the exposure in the estrogen-plus-progestin group was identical. A sensitivity analysis in which data were censored six months after the first nonadherence identified a hazard ratio for colorectal cancer of 0.52 (95 percent confidence interval, 0.32 to 0.85). A model that adjusted for the duration of prior postmenopausal hormone therapy yielded a hazard ratio for colorectal cancer of 0.56 (95 percent confidence interval, 0.32 to 0.82). Therefore, prior exposure to postmenopausal hormone therapy was similar in the two groups, and the hazard ratio for colorectal cancer did not change after adjustment for prior hormone exposure and adherence or nonadherence to the study medication.
Drs. Machens and Schmidt-Gollwitzer suggest that the effect of estrogen plus progestin on the risk of colorectal cancer is seen largely during and after year 4. The hazard ratios (and 95 percent confidence intervals) for colorectal cancer in the estrogen-plus-progestin group, according to the study year, are 0.64 (0.26 to 1.56) for year 1, 0.95 (0.39 to 2.28) for year 2, 0.43 (0.13 to 1.41) for year 3, 0.46 (0.22 to 0.99) for year 4, 0.52 (0.18 to 1.56) for year 5, and 0.43 (0.16 to 1.16) for year 6. The P value for the trend of the effect on the risk of colorectal cancer over time is 0.34. Thus, there is no evidence that the effect of the hormone on colorectal cancer changes over time.
Our original statement that the proportion of colorectal cancers diagnosed at an advanced stage was higher in the hormone group than in the placebo group is correct. That said, we agree with Dr. Schürmann and colleagues that a major effect of estrogen and progestin on the risk of colorectal cancer was a reduction in the absolute number of localized cancers (hazard ratio for localized cancer, 0.26; 95 percent confidence interval, 0.13 to 0.53). However, the effect on regional tumors (hazard ratio, 0.76; 95 percent confidence interval, 0.44 to 1.30) and metastatic tumors (hazard ratio, 1.54; 95 percent confidence interval, 0.50 to 4.71) is less clear. However, as previously stated, even within the category of regional or metastatic disease, the cancers in the hormone group were associated with a greater number of positive nodes than were the cancers in the placebo group (3.6±4.2 vs. 1.6±2.1 nodes, P=0.012), and the number of deaths was very similar in the hormone and placebo groups (nine and eight, respectively). The biology underlining these findings and their clinical implications are unclear.
Rowan T. Chlebowski, M.D., Ph.D.
Harbor–UCLA Research and Education Institute
Torrance, CA 90502
rchlebowski@rei.edu
Rebecca J. Rodabough, M.S.
Fred Hutchinson Cancer Research Center
Seattle, WA 98109
Jean Wactawski-Wende, Ph.D.
University at Buffalo
Buffalo, NY 14214