Reduced folate carrier 80A→G polymorphism, plasma folate, and risk of placental abruption

Abstract

Folate deficiency and maternal smoking are strong risk factors for placental abruption. We assessed whether the reduced folate carrier [NM_194255.1: c.80A→G (i.e., p.His27Arg)] (RFC-1) polymorphism was associated with placental abruption, and evaluated if maternal smoking modified the association between plasma folate and abruption. Data were derived from the New Jersey-Placental Abruption Study—a multicenter, case-control study of placental abruption (2002–2007). Maternal DNA was assayed for the RFC-1 c.80A→G polymorphism using a PCR-dependent diagnostic test. Maternal folate (nmol/l) was assessed from maternal plasma, collected immediately following delivery. Due to assay limitations, folate levels at ≥60 nmol/l were truncated at 60 nmol/l. Therefore, case–control differences in folate were assessed from censored log-normal regression models following adjustment for potential confounders. Distribution of the mutant allele (G) of the RFC-1 c.80A→G polymorphism was similar between cases (52.3%; n = 196) and controls (50.5%; n = 191), as was the homozygous mutant (G/G) genotype (OR 1.1, 95% CI 0.6–2.2). In a sub-sample of 136 cases and 140 controls, maternal plasma folate levels (mean ± standard error) corrected for assay detection limits were similar between placental abruption cases (63.6 ± 5.1 nmol/l) and controls (58.3 ± 4.7 nmol/l; P = 0.270), and maternal smoking did not modify this relationship (interaction P = 0.169). We did not detect any association between the RFC-1 c.80A→G polymorphism and placental abruption, nor was an association between plasma folate and abruption risk evident. These findings may be the consequence of high prevalence of prenatal multivitamin and folate supplementation in this population (over 80%). It is therefore not surprising that folate deficiency may be rare and that the RFC-1 c.80A→G polymorphism is less biologically significant for placental abruption.

Appendices

Appendix 1

Investigators currently participating or who have been previously involved in the New Jersey-Placental Abruption Study include Cande V. Ananth, PhD, MPH (principal investigator), Darios Getahun, MD, MPH, Neela Srinivas, MD, MPH, Celeste DeMarco, RN, BSN, Denise Elsasser, MPH, Yu-Ling Lai, RN, Shelby Pitts, RN (Division of Epidemiology and Biostatistics), John C. Smulian, MD, MPH, Wendy L. Kinzler, MD, Morgan R. Peltier, PhD, and Marian Lake, RN, MPH (Division of Maternal-Fetal Medicine) from the Department of Obstetrics, Gynecology, and Reproductive Sciences; Claire Philipp, MD from the Department of Medicine, all at UMDNJ-Robert Wood Johnson Medical School, New Brunswick, NJ; and George G. Rhoads, MD, MPH from the Department of Epidemiology and Dirk F. Moore, PhD from the Department of Biostatistics at UMDNJ-School of Public Health, Piscataway, NJ.

Other investigators who were involved with the study included Jacques Genest, MD (McGill University, Montreal, Canada), Susan Shen-Schwarz, MD (Department of Pathology, Saint Peter’s University Hospital, New Brunswick, NJ); and Vinay Prasad, MD (Department of Pediatric Pathology, Nationwide Children’s Hospital, Ohio State University, Columbus, OH).

Appendix 2

We model the distribution of the folate level using a lognormal distribution. Let Y _i ^* denote maternal plasma folate (nmol/l) concentration for the ith subject (1 ≤ i ≤ n). The probability density function is given by

$$ f(y^{*} ;\mu ,\sigma ) = \frac{1}{{y^{*} \sigma \sqrt {2\pi } }}\exp \left[ { - \frac{1}{{2\sigma ^{2} }}\left( {\ln y^{*} - \mu } \right)^{2} } \right]. $$

Since plasma folate concentrations exceeding 60 nmol/l were censored at that value, the likelihood function (L _i ) for estimating the lognormal parameters is a product of terms of the form:

$$ L_{i} (\mu _{i} ;y_{i} ) = f(\mu _{i} ;y_{i} )^{{1 - c_{i} }} S(\mu _{i} ;y_{i} )^{{c_{i} }} $$

where

$$ S(y;\mu ,\sigma ) = \int\limits_{y}^{\infty } {f(v;\mu ,\sigma )} {\text{d}}v $$

is the probability that the concentration y exceeds 60, c _i is a censoring indicator, coded as

$$ c_{i} = \left\{ {\begin{array}{*{20}c} {0\,} & {{\text{if}}\;y^{*} < 60} \\ {1\,} & {{\text{if}}\;y^{*} \ge 60} \\ \end{array} } \right. $$

and y _i  = min (y _i ^*, 60). Covariates may be accommodated in a regression model framework by expressing the lognormal mean folate as follows:

$$ E(\log Y_{i}^{*} ) = \beta _{0} + \beta _{1} ({\text{CASE}}_{i} ) + \beta _{2} ({\text{SMK}}) + \beta _{3} ({\text{CASE}} \times {\text{SMK}}) + \sum\limits_{{k = 4}}^{p} {\beta _{k} (X_{{ki}} )} $$

where CASE refers to case–control status (coded as 0 and 1 for controls and cases, respectively), SMK to maternal smoking status, coded as 0 for non-smokers and 1 for smokers, and X _k to the set of confounders. Maximum likelihood estimates of the parameters may be obtained using the “survreg” function in the “survival” package of the R statistical system (R Development Core Team 2007).