Electronic Letters to:

Endocrine Care: Glenn D. Braunstein, B. Delia Johnson, Frank Z. Stanczyk, Vera Bittner, Sarah L. Berga, Leslee Shaw, T. Keta Hodgson, Maura Paul-Labrador, Ricardo Azziz, and C. Noel Bairey Merz Relations between Endogenous Androgens and Estrogens in Postmenopausal Women with Suspected Ischemic Heart Disease J Clin Endocrinol Metab 2008; 93: 4268-4275 [Abstract] [Full text] [PDF]

eLetters: Submit a response to this article

Electronic letters published:

Response to letter of Gerald B. Phillips

Glenn D. Braunstein, B. Delia Johnson and C. Noel Bairey Merz   (28 July 2009)

Re: Relations between Endogenous Androgens and Estrogens in Postmenopausal Women with Suspected Isch

Gerald B. Phillips   (13 July 2009)

Response to letter of Gerald B. Phillips 28 July 2009
Glenn D. Braunstein, Physician Cedars-Sinai Medical Center, B. Delia Johnson and C. Noel Bairey Merz Send letter to journal: Re: Response to letter of Gerald B. Phillips braunstein{at}cshs.org Glenn D. Braunstein, et al.	Based upon the observation of Dr. Phillips that the T/E2 ratio in his studies correlated more significantly with cholesterol, glucose, and insulin than did the reciprocal ratio (E2/T), we reanalyzed our data in a similar fashion after adjusting for age and BMI. The Spearman correlations (P values) for T/E2 and cholesterol, glucose, and insulin were -0.16 (0.011), 0.13 (0.036), and 0.07 (0.31), respectively. The correlations for E2/T were 0.16 (0.011), -0.13 (0.036), and -0.07 (0.31), respectively. Since T/E2 and E2/T are inverses to each other, the correlations of these ratios with other blood measures are identical except for the sign. Therefore, we are puzzled how Dr. Phillips was able to find that T/E2 correlated more significantly than E2/T with these other analytes. We analyzed the associations of these ratios with CAD and found that they were not mirror images of each other due to the fact that the odds ratios reflect log likelihoods rather than simple linear combinations. The odds ratio for Free T/E2 and CAD adjusted for age and BMI was 2.53 (95% CI 0.96-6.66; P = 0.06), while that for E2/Free T was 0.87 (95% CI 0.77 -0.98; P = 0.02). Thus, using the ratios did not offer a benefit in modeling prediction of CAD. Second, we cannot confirm Dr. Phillip's findings that Free T/E2 was more strongly associated with CAD than was E2/Free T. Potential explanations include random effects, study population differences, such as the overall health of the population, or other variables that would impact hormone levels, such as reporting error, supplement use, and nutrition.
Re: Relations between Endogenous Androgens and Estrogens in Postmenopausal Women with Suspected Isch 13 July 2009
Gerald B. Phillips, Professor of Medicine Department of Medicine, Columbia University, St. Luke's-Roosevelt Hospital Center, New York, NY Send letter to journal: Re: Re: Relations between Endogenous Androgens and Estrogens in Postmenopausal Women with Suspected Isch gbp1{at}columbia.edu Gerald B. Phillips	Braunstein et al. (1) point out that because androgens are precursors of estrogens and their levels in serum correlate positively, they should be mutually controlled for in epidemiological studies. After controlling for estradiol (E2), they observed an association (positive) of total testosterone and free testosterone (FT) with the presence of coronary artery disease (CAD) in postmenopausal women. They cite our observation of a positive association of FT with degree of CAD in postmenopausal women, which we demonstrated in a multiple regression model that included both FT and E2 (2). If I remove E2 from our model, the P < 0.008 for the FT-CAD association becomes P = 0.015. Thus, as observed by Braunstein et al., inclusion of E2 strengthened the FT-CAD association. Because E2 and FT related to CAD in opposite directions, we had tried substituting E2/FT for E2 and FT, which gave a P = 0.027. Shortly after publishing the paper, I found to my surprise that FT/E2 in the model associated differently, and more significantly (P = 0.002), with CAD than did E2/T. Results were similar after excluding from the group of 60 the 11 women who had ever had a myocardial infarction. Unpublished data on postmenopausal women from a subsequent study from our laboratory (3) showed that on controlling for age and BMI, T/E2 correlated more significantly (positively) with cholesterol, glucose, and insulin than did E2/T. In men, we found that E2/T correlated more significantly (positively) than T/E2 with insulin, controlled for age and BMI, in all five of the studies from this laboratory in which these correlations were determined (4). References 1. Braunstein GD, Johnson BD, Stanczyk FZ, Bittner V, Berga SL, Shaw L, Hodgson TK, Paul-Labrador M, Azziz R, Merz CNB 2008 Relations between endogenous androgens and estrogens in postmenopausal women with suspected ischemic heart disease. J Clin Endocrinol Metab 93:4268-4275 2. Phillips GB, Pinkernell BH, Jing TY 1997 Relationship between serum sex hormones and coronary artery disease in postmenopausal women. Arterioscler Thromb Vasc Biol 17:695-701 3. Phillips GB, Jing T, Heymsfield SB 2008 Does insulin resistance, visceral adiposity, or a sex hormone alteration underlie the metabolic syndrome? Studies in women. Metabolism 57:838-844 4. Phillips GB, Jing T, Heymsfield SB 2003 Relationships in men of sex hormones, insulin, adiposity, and risk factors for myocardial infarction. Metabolism 52:784-790