Rose G. Radin, Sunni L. Mumford, Robert M. Silver, Laurie L. Lesher, Noya Galai, David Faraggi, Jean Wactawski-Wende, Janet M. Townsend, Anne M. Lynch, Hyagriv N. Simhan, Lindsey A. Sjaarda, Neil J. Perkins, Shvetha M. Zarek, Karen C. Schliep, Enrique F. Schisterman
Submitter: Tom Leibson | tom.leibson@sickkids.ca
Authors: Tom Leibson, Gali Pariente, and Shinya Ito
The Hospital for Sick Children, Toronto, ON
Published December 11, 2015
The recent report (Radin 20151) on role of low dose aspirin on sex ratios at birth is very interesting. While EAGeR trial data2 and following analysis are compelling, we had concern regarding the position endowed to inflammation in authors’ unifying hypothesis. hsCRP plays a pivotal role in establishment of inflammatory status which determines effect of low dose aspirin on gender ratio. Yet, two doubts regarding its ability to support such reaching conclusions are worthy of discussion:
1. What exactly does this marker inform on reproductive tract associated systemic inflammation? As previously published, hsCRP could become elevated in various conditions and associations. Human studies have identified obesity3 and hypertension4, for example, as possible confounders.
2. What is the relation between the concentrations of hsCRP and the level of inflammation in this context? Clinical usage of this marker warrants great caution as it may be perceived as a continuous variable, when in fact it does not provide more data than does a dichotomous variable. Risk stratification using hsCRP is feasible5; however, assignment of hsCRP concentrations to corresponding categories of inflammation severity and further to risk categories of interest should be optimally derived from a model describing the relationship among these parameters as shown for other conditions5.
Our opinion is that inflammation could indeed be a factor accounting for the reported significant change in sex ratios at birth, but as published, we find the indirect laboratory support for this notion potentially misleading. Future studies on this important topic are encouraged to actively seek and control for potential confounders, and categorize serum hsCRP levels with utmost care.
References
1. Radin, R. G. et al. Sex ratio following preconception low-dose aspirin in women with prior pregnancy loss. J. Clin. Invest. 125, 3619–3626 (2015).
2. Schisterman, E. F. et al. Preconception low-dose aspirin and pregnancy outcomes: results from the EAGeR randomised trial. Lancet Lond. Engl. 384, 29–36 (2014).
3. Sørensen, C. J. et al. Combined oral contraception and obesity are strong predictors of low-grade inflammation in healthy individuals: results from the Danish Blood Donor Study (DBDS). PloS One 9, e88196 (2014).
4. Fukutomi, M., Hoshide, S., Eguchi, K., Watanabe, T. & Kario, K. Low-grade inflammation and ambulatory blood pressure response to antihypertensive treatment: the ALPHABET study. Am. J. Hypertens. 26, 784–792 (2013).
5. Cook, N. R., Buring, J. E. & Ridker, P. M. The effect of including C-reactive protein in cardiovascular risk prediction models for women. Ann. Intern. Med. 145, 21–29 (2006).
Submitter: Rose G. Radin | rose.radin@nih.gov
Authors: Rose G. Radin, Lindsey A. Sjaarda, and Enrique F. Schisterman
Epidemiology Branch, Division of Intramural Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development
Published December 11, 2015
We appreciate the comments by Leibson et al. on our recent study of sex ratio following preconception low-dose aspirin (LDA) in the EAGeR Trial (1). This response addresses their specific questions on bias from confounding and the choice of tertiles to analyze high-sensitivity C-reactive protein (hsCRP) hsCRP.
As in any large randomized trial, the balance of confounders with respect to LDA is ensured by randomization, as was demonstrated by the balance of baseline characteristics by treatment assignment (2). Furthermore, baseline characteristics including BMI and blood pressure were balanced by treatment assignment even after stratifying by lower, middle, and higher hsCRP. We agree there are many predictors of hsCRP, and we expect all of them to be balanced across treatment groups by randomizations. Therefore, it should mitigate concerns about confounding.
Risk stratification for sex ratio was not our objective. Rather, the aim was to elucidate the role of inflammation that depends on the gender of the embryo. We categorized hsCRP into tertiles in order to evaluate a potential dose-response effect of inflammation on male-live birth with analytical efficiency. However, the conclusions did not depend on the cut-point for hsCRP. We also analyzed continuous, log-transformed hsCRP in relation to the percent of live births that were male, stratified by treatment assignment. As log hsCRP increased, the percent male decreased among the placebo group, but not in the LDA group (Figure 1). At higher values of log hsCRP, the percent male was significantly lower in the placebo group: the percent male in the placebo group, (the solid red line), was lower than the lower 95% confidence limit for the percent male in the treatment group (the lower dashed blue line). Congruently, the percent male in the LDA group, (the solid blue line) was higher than the upper 95% confidence limit for the percent male in the placebo group (the upper dashed red line).
Figure 1. Restricted Cubic Spline Regression of log-transformed preconception hsCRP and probability of a male among live births, stratified by treatment group. The estimated percent male in the LDA group is shown by the solid blue line with dashed blue lines for the 95% confidence interval. The estimated percent male in the placebo group is shown by the solid red line with dashed red lines for the 95% confidence interval. Log-binomial regression curve was fit by using the SAS macro program GLMCURV9 (E. Hertzmark, R. Li, B. Hong, D. Spiegelman. Channing Laboratory, Boston, MA). The model was adjusted for selection of live births using longitudinal inverse-probability weighting. The maximum of the range shown, hsCRP=7.4 mg/L, was the 92nd percentile of hsCRP. For comparison, a horizontal line is placed at 51% male.
We agree with Leibson et al. that there is a need for further evaluation of the role of systemic inflammation and LDA on reproductive outcomes and encourage our scientific colleagues to conduct additional investigations to inform the prevention of pregnancy loss and complications.
References
1. Radin, R.G., et al. Sex ratio following preconception low-dose aspirin in women with prior pregnancy loss. J Clin Invest 2015; 125(9):3619-3626.
2. Schisterman, E.F., et al. Preconception low-dose aspirin and pregnancy outcomes: results from the EAGeR randomised trial. Lancet 2014; 384(9937):29-36.
3. Schisterman, E.F., et al. A randomised trial to evaluate the effects of low-dose aspirin in gestation and reproduction: design and baseline characteristics. Paediatr Perinat Epidemiol 2013; 27(6):598-609.