Introduction
The periconceptional period is a critical window of exposure that can impact fetal growth and development.Reference Louis, Cooney, Lynch and Handal 1 Maternal smoking has been associated with adverse reproductive health effects on the offspring, including reduced fecundity in malesReference Juul, Almstrup and Andersson 2 and females.Reference Jensen, Joffe and Scheike 3 A novel endpoint in humans that has been proposed as an indicator of sexual development in utero is the 2nd (index) to 4th (ring) finger digit ratio (2D:4D). The 2D:4D ratio reflects sexual differentiation early in life, probably toward the end of the first trimester of pregnancy and the beginning of the second trimester,Reference Malas, Dogan, Evcil and Desdicioglu 4 – Reference Galis, Ten Broek, Van Dongen and Wijnaendts 6 with males having relatively shorter 2nd than 4th digit lengths. For this reason, the 2D:4D has been proposed as an endpoint for the organizational effects of prenatal androgens in the body, and as such, lower 2D:4Ds could be potential indicators of greater androgen exposure during fetal development.Reference Manning, Scutt, Wilson and Lewis-Jones 7
In the framework of the Developmental Origins of Health and Disease, as the 2D:4D consistently differs by sex, and under the hypothesis that this difference is related to in utero androgen levels, several studies have evaluated the association between 2D:4D as a marker of prenatal androgen exposure and different reproductive outcomes across the life course. As we have previously described,Reference Velez, Arbuckle, Monnier and Fraser 8 several cross-sectional studies have reported low sperm counts, decreased testicular volume and impaired hormonal status in men having higher digit ratios (i.e. implying low androgen exposure during fetal life).Reference Oh, Kim, Yoon, Kim and Kim 9 , Reference Klimek, Galbarczyk, Nenko, Alvarado and Jasienska 10 However, other studies have failed to replicate the associations between 2D:4D and hormone levelsReference Honekopp, Bartholdt, Beier and Liebert 11 or a number of male reproductive endpoints including semen quality.Reference Firman, Simmons, Cummins and Matson 12 , Reference Bang, Carlsen and Holm 13
In females, two retrospective studies in adult women reported associations between lower digit ratios (i.e. implying greater androgen exposure during fetal development) and delayed age at menarche,Reference Manning and Fink 14 , Reference Matchock 15 whereas a prospective cohort in pre-menarcheal girls aged 5–12 years found lower digit ratios as predictors of early age at menarche.Reference Oberg and Villamor 16 In addition, lower female digit ratios have been associated with low offspring counts,Reference Manning and Fink 17 congenital adrenal hyperplasiaReference Oswiecimska, Ksiazek and Sygulla 18 and polycystic ovary syndrome,Reference Cattrall, Vollenhoven and Weston 19 although this finding was challenged by Lujan et al.Reference Lujan, Podolski, Chizen, Lehotay and Pierson 20 , Reference Lujan, Bloski, Chizen, Lehotay and Pierson 21 In relation to fecundity, one study has reported a longer time to pregnancy (TTP) at a higher 2D:4D in males.Reference Auger and Eustache 22 However, we did not find an association between female 2D:4D and TTP in The Maternal-Infant Research on Environmental Chemicals (MIREC) Study.Reference Velez, Arbuckle, Monnier and Fraser 8
Smoking has a recognized detrimental impact on fertility, in both male and female.Reference Buck Louis 23 , Reference Baird and Wilcox 24 Furthermore, maternal smoking during pregnancy has been associated with adverse reproductive outcomes in offspring,Reference Ernst, Kristensen and Toft 25 including impaired fecundity.Reference Weinberg, Wilcox and Baird 26 The causal mechanism by which smoking adversely affects fertility is still the subject of active research. In animals, prenatal nicotine exposure has been related to increased testosterone levels in adolescent female offspring rats and female ovine fetuses, with more modest increases in the male fetuses.Reference Smith, Cloak, Poland, Torday and Ross 27 In humans, maternal smoking during pregnancy has been positively associated with concentrations of maternal testosterone.Reference Kandel and Udry 28
Thus, if the adverse effect of maternal smoking on the offspring’s subsequent fertility acts through the homeostasis of fetal androgens, and under the hypothesis that 2D:4D is an indicator of the fetal concentrations of androgens, then 2D:4D should differ among children whose mothers smoked during the periconceptional period compared with those whose mothers did not smoke. Only one cross-sectional study has assessed the impact of self-reported maternal smoking during pregnancy on 2D:4D,Reference Rizwan, Manning and Brabin 29 reporting lower digit ratios in boys prenatally exposed to smoking, although this association was not observed in females.
The purpose of this study was to evaluate the effect of maternal periconceptional smoking on the 2D:4D as a potential sensitive endocrine endpoint in children within the MIREC cohort, a Canadian pregnancy and birth cohort.
Material and methods
Population and study design
The MIREC Study is a pregnancy cohort conducted in 10 cities across Canada between 2008 and 2011.Reference Arbuckle, Fraser and Fisher 30 Approximately 2000 women attending prenatal clinics were recruited during the first trimester of pregnancy and followed up to 10 weeks postpartum. Three subsequent follow-up studies have been conducted for the longitudinal assessment of child health and development:Reference Arbuckle, Fraser and Fisher 30 The MIREC-ID Study (MIREC: infant development to 6 months), MIREC-CD3 (MIREC: child development at the age of 3 years) and MIREC-CD Plus (MIREC: growth and development to 5 years).
The present analysis is part of MIREC-CD Plus, a follow-up study in a subsample of some 800 MIREC children. Baseline demographic and lifestyle characteristics of the women were obtained from data collected at the first trimester visit, including gestational age at recruitment, and maternal age, education, ethnicity, household income, and pre-pregnancy body mass index (BMI) gestational age at delivery, birth weight and gender were obtained from the maternal delivery chart.
One component of MIREC-CD Plus was the measurement of the index (2nd, 2D) and ring (4th, 4D) digit lengths in mothers and their children to calculate the 2D:4D ratio. As we recently described,Reference Velez, Arbuckle, Monnier and Fraser 8 digital pictures of the ventral surface of both hands were obtained using a 16-GB iPod touch (Apple Inc., CA, USA), placed in a device specially designed to standardize this procedure. Measurements of the 2nd and 4th digit lengths were conducted by a single experienced research assistant with the aid of Adobe Photoshop 8 software (Adobe Systems Incorporated, San Jose, CA, USA). The lengths of the 2nd and 4th fingers of each hand were measured using mouse-controlled calipers. Calipers were positioned midline along the finger’s basal crease and expanded to the edge of the finger tip. 2D:4D was calculated as the ratio of the 2nd and 4th fingers of each hand. In order to assess intra-observer reliability, each measurement was performed twice after a period of 2 weeks, and the intraclass correlation coefficients (ICCs) were calculated in a convenience subsample of 80 mother and children pairs (Table 1). The reliability of the two repeated digit length measures performed by the same examiner was very high, resulting in ICCs higher than 0.97 for the finger lengths with narrow 95% confidence intervals.Reference Velez, Arbuckle, Monnier and Fraser 8
Table 1 Intraclass correlation coefficient (ICC) subsample (n=80)
CI, confidence interval.
A total of 808 follow-up visits were conducted in MIREC-CD Plus. In total, 24 children did not have their anthropometric measurements taken, including the hand pictures. The hand pictures of two children were excluded because of visible malformations of their fingers (i.e. syndactyly); six children refused to have the pictures taken; eight children were excluded due to technical problems related to their pictures (i.e. poor quality, failed uploading or transfer from the study site to the study coordinating center). A total of 19 pairs of twins were excluded from the analysis. This resulted in a sample of 749 children. The MIREC Study was approved by the Research Ethics Board of Health Canada, the research ethics committee of the coordinating center of Ste-Justine’s Hospital in Montreal, and the academic and hospital ethics committees of the study sites across Canada. All the participants signed informed consent forms.
Operational definition of variables
The exposure of interest was self-reported periconceptional maternal smoking (i.e. 1 year before through the first trimester of pregnancy), assessed at the first visit of the MIREC Study (6–14 weeks of gestational age). In MIREC, women who had smoked <100 cigarettes over their lifetime (about four packs) were considered as ‘never’ smokers. Women who had smoked >100 cigarettes over their lifetime but stopped smoking before pregnancy, and at the time of the interview were not smoking at all, were considered as ‘former’ smokers. Women who had smoked >100 cigarettes over their lifetime, but stopped when they knew they were pregnant, and when interviewed were not smoking at all, were considered as ‘quit during pregnancy’. Women, who at interview reported smoking daily or occasionally, were considered as ‘current’ smokers. For the present analysis, women that quit during pregnancy or 1 year before and current smokers were considered as the exposure of interest (periconceptional maternal smoking), as there is evidence suggesting that in former smokers who quit, testosterone levels start to decline only after 1 year of smoking cessation.Reference Brand, Chan and Dowsett 31 The outcome was the 2D:4D of the children. Potential confounders were maternal 2D:4D, ethnicity, education, pre-pregnancy BMI and children’s age. The analysis was stratified by sex.
Statistical analysis
The reliability of the digit length measurement was evaluated by calculating the ICC using ANOVA analysis (Table 1). The descriptive phase of analysis included assessment of the distributions of the baseline characteristics of the mothers and their children. To assess the impact of maternal smoking during the first trimester of pregnancy on the children’s 2D:4D, mean differences were calculated by an unpaired t-test. To account for potential confounders, multiple linear regression analyses were conducted. Statistical analysis was performed using STATA 10.0 (Stata Corporation, College Station, TX, USA).
Results
The mean maternal age at recruitment was 33.45 (s.d. 4.71) years, with a median gestational age of 12 weeks (Table 2). Most of the women were whites, two-thirds had a university degree, more than half had a prior pregnancy with a live birth and 14.5% were obese before the index pregnancy. In all, 82 women (11.8%) smoked during the periconceptional period (i.e. 1 year before through the first trimester of pregnancy). The mean maternal 2D:4D was similar in both hands (right 0.985±0.036 and left 0.978±0.036). Ethnicity was associated with female 2D:4D, with black mothers having smaller ratios in both hands. Education was significantly associated with maternal 2D:4D in the right hand, with women who had high school or less education having a smaller ratio.
Table 2 Population baseline characteristics and its association with children’s 2nd (index) to 4th (ring) finger digit ratio (2D:4D)
BMI, body mass index.
a Linear regression.
b ANOVA F test.
The mean age of the children at follow-up was 3.53 (s.d. 0.87) years. The mean gestational age at birth was 39 (s.d. 1.7) weeks, with a mean birth weight of 3464 (s.d. 512) g. The gender distribution was similar, as represented by 369 girls (50.7%) and 380 boys (49.3%). As shown in Table 3, boys had lower mean 2D:4Ds compared with girls. The mean 2D:4D did not differ among children whose mothers had smoked during the periconceptional period compared with those whose mothers did not smoke, irrespective of gender (Table 3). The same was observed in the multiple linear regression analyses (Table 4), where the only factors associated with the child’s 2D:4D were their age and maternal 2D:4D.
Table 3 Mean 2nd (index) to 4th (ring) finger digit ratio in children according to gender and maternal periconceptional smoking
a t-Test.
Table 4 AdjustedFootnote a mean changes (β) for the 2nd (index) to 4th (ring) finger digit ratio (2D:4D) in children per maternal periconceptional smoking status
CI, confidence interval.
a Adjusted by maternal 2D:4D, ethnicity, education, pre-pregnancy body mass index and children’s age.
Discussion
Exposure to maternal smoking during the periconceptional period failed to show any effect on the 2D:4D, independent of gender. Instead, we found that age and maternal 2D:4D were strong determinants of the children’s 2D:4D.
In line with the evidence, we were able to demonstrate that boys have lower mean 2D:4D than girls and that this effect is stronger in the right hand.Reference Manning, Scutt, Wilson and Lewis-Jones 7 , Reference Honekopp and Watson 32 In spite of this, we were unable to demonstrate a significant effect of self-reported maternal smoking during the periconceptional period on the offspring’s 2D:4D. We are not aware of any previous study assessing prospectively the effect of self-reported maternal periconceptional smoking on the offspring’s 2D:4D, although one cross-sectional study by Rizwan et al.Reference Rizwan, Manning and Brabin 29 evaluated the effect of self-reported maternal smoking during pregnancy on the 2D:4D in children in primary school. The study reported a significant association between maternal smoking during pregnancy and low mean right 2D:4D in male offspring only. Certain aspects of our respective studies may account for the different results. First, our exposure was maternal periconceptional smoking (i.e. 1 year before through the first trimester of pregnancy), while in Rizwan et al.,Reference Rizwan, Manning and Brabin 29 smoking throughout pregnancy was the exposure of interest. Currently, there are reports from rodent studies that the 2D:4D is fixed during a narrow window of embryonic development,Reference Zheng and Cohn 33 and human studies suggest that this occurs toward the end of the first trimester and the beginning of the second trimester.Reference Malas, Dogan, Evcil and Desdicioglu 4 – Reference Galis, Ten Broek, Van Dongen and Wijnaendts 6 We are not aware of studies that have quantified testosterone levels after smoking cessation in pregnant women; however, a study in postmenopausal women found that after quitting smoking, testosterone levels remained stable for 1 year, showing a decline only after 1 year of cessation.Reference Brand, Chan and Dowsett 31 Thus, we are assuming that by using self-reported maternal periconceptional smoking as our exposure of interest, we are capturing exposure to testosterone as consequence of smoking, toward the end of the first trimester and the beginning of the second trimester. Unfortunately, the concentrations of sex steroid hormones during pregnancy have not been measured in women from the MIREC Study, which would have provided an objective measure of testosterone exposure during pregnancy according to periconceptional maternal smoking, a limitation of the present study. Second, the children in Rizwan et al. study were older (mean age 8.3, s.d. 1.7 years) than those in our study. One study using long-term serial data to evaluate the validity of finger length ratios as markers for prenatal testosterone action suggests that the digit ratio increases with age,Reference McIntyre, Ellison, Lieberman, Demerath and Towne 34 and indeed in our multiple regression analyses, the child’s age was significantly associated with 2D:4D. Third, although the prevalence of maternal smoking was low, it was comparable in both studies (12% in MIREC study v. 13% in Rizwan et al. study). Nonetheless, the doses of exposure (i.e. number of cigarettes per day, puffing behaviors) may have differed between the studies, information that was not available in the Rizwan et al. study. In general, our study had a power of 72% to detect mean differences in 2D:4D of 0.2 (s.d. 0.05), with 95% confidence intervals (α=0.05), and the power would be lower to test a dose–response effect. In addition, other factors could not be ruled out such as the heritability of the 2D:4D.Reference Manning, Barley and Walton 35 In fact, the crude β coefficients (right 0.26 and left 0.18 in girls; right 0.29 and left 0.10 in boys) and the adjusted β coefficients (Table 4) of the respective maternal index of each hand were significantly associated with the children’s 2D:4D. Rizwan et al. recognized that they could not rule out that smoking, high fetal testosterone and low 2D:4D are associated through a relationship of risk-taking behavior,Reference Rizwan, Manning and Brabin 29 and that there are studies suggesting the heritability of the 2D:4D.Reference Manning, Barley and Walton 35 Smoking, a trait of risk-taking behavior, is partially heritable,Reference Vink and Boomsma 36 raising the possibility that more than a proxy of in utero androgen levels, 2D:4D could be a heritable trait.
On the other hand, it has been reported in a large international sample that smoking in women is linked to high 2D:4D but that alcohol consumption is associated with low 2D:4D.Reference Manning and Fink 37 As parental 2D:4D is correlated with their children’s 2D:4D, in order to consider smoking and its effect on the fetus it might be argued that maternal alcohol consumption needed to be considered in our model. Nonetheless, the low prevalence (3.3%) of alcohol consumption throughout pregnancy (⩾1 drink/week) in our populationReference Colapinto, Arbuckle, Dubois and Fraser 38 would have resulted in a low statistical power to test this effect. Moreover, as smoking and alcohol consumption are highly correlated, we consider that including both variables in our adjusted model might result in over adjustment.
The method used to measure digit length could also account for the differences among studies. Most of the studies have measured finger lengths by a physical method, whereas we used digital imaging of the hands. This could explain why our mean 2D:4Ds (Table 3) were lower than those reported in Rizwan et al. study where direct measures were used,Reference Rizwan, Manning and Brabin 29 not only in girls, but also in boys. There are studies reporting that photocopies, an indirect method to measure 2D:4D, result in lower digit ratios (2D:4D) than direct finger measurements,Reference Manning, Fink, Neave and Caswell 39 and concerns about the accuracy of indirect in comparison to direct methods, as recently reviewed.Reference Ribeiro, Neave, Morais and Manning 40 However, other studies support that computer-assisted measurement of the 2D:4D are more accurate than physical measurement, photocopies or printed scanned images measured with Vernier calipers.Reference Allaway, Bloski, Pierson and Lujan 41 Although we were not able to compare our indirect method with a direct measurement, we are confident on the reproducibility of our measures based on the high ICCs shown in Table 1.
Another possibility is that the association between prenatal smoking and fertility outcomes later in life may have a causal pathway independent of androgen exposure. In women, given that the primordial ovarian follicular pool is established in utero, prenatal maternal smoking might have an impact on the ovarian reserve of the daughters. However, human studies have failed to demonstrate associations between prenatal smoking exposure and objective markers of ovarian reserve in the daughters, such as the Anti-Müllerian Hormone and the Antral Follicle Count.Reference Fraser, McNally and Sattar 42 , Reference Hart, Sloboda and Doherty 43 In men, while studies focused on reproductive hormones as outcome of perinatal maternal smoking are inconsistent, some evidence suggest that maternal cigarette smoking during pregnancy has adverse effects on sperm concentration and total sperm count of the sons, as recently reviewed.Reference Hakonsen, Ernst and Ramlau-Hansen 44
In summary, we were able to demonstrate a sexual dimorphic trait with boys having lower ratios than girls. As well, we found that the child’s age and maternal 2D:4D were strong determinants of the children’s 2D:4D. However, we did not find an impact of maternal periconceptional smoking on children’s 2D:4D, although the smoking prevalence was low. From a public health perspective, our study failed to demonstrate the utility of the 2D:4D as an objective marker of prenatal smoking, questioning its utility as an epidemiological tool to decrease bias inherent to the ascertainment of prenatal smoking through questionnaires (e.g. recall bias, social desirability bias).
Acknowledgments
This work would not have been possible without the generous collaboration of the MIREC participants (i.e. women, partners and children). Special thanks to Stéphanie Bastien and Nicole Lupien for their assistance during the 2D:4D component, and to Gabriel Abad who designed and produced the hand scan devices and measured the digit lengths. The authors acknowledge the contributions of the MIREC Study Group, especially the site investigators and their staff. The authors are grateful to the reviewers for their helpful comments and suggestions.
Financial Support
The MIREC Study was funded by Health Canada’s Chemicals Management Plan, the Canadian Institute of Health Research (CIHR) and the Ontario Ministry of the Environment. MIREC-CD Plus is funded by Health Canada’s Chemicals Management Plan Research Fund. The 2D:4D component was funded by a research grant from the CIHR-Quebec Training Network in Perinatal Research (QTNPR). M.P.V. was supported by a CIHR Fellowship Award and a QTNPR scholarship. P.M. is supported by the Research Institute of the McGill University Health Centre. W.F. is supported by a CIHR Canada Research Chair.
Authors’ Contributions
M.P.V., T.A., P.M. and W.F. were all involved in the conception and design of the study. M.P.V. carried out analysis and interpretation of data, in addition to drafting the manuscript. T.A. and W.F. are the co-principal investigators of the MIREC Study, and along with P.M., contributed to data interpretation and review of the manuscript.
Conflicts of Interest
None.
Ethical Standards
The authors assert that all procedures contributing to this work comply with the Helsinki Declaration of 1975, as revised in 2008, and has been approved by the Research Ethics Board at Health Canada and the research ethics committee at the coordinating centre at Ste-Justine’s Hospital in Montreal, as well as the participating academic and hospital ethics committees across Canada.
