Association between cigarette smoking and ovarian reserve among women seeking fertility care.

Infertility and impaired fecundity among women of reproductive age remain important public health issues. Cigarette smoking in women has been associated with adverse reproductive outcomes such as poorer IVF outcome and increased adverse pregnancy outcomes [ 1 – 4 ]. However, studies of the association of smoking with ovarian reserve have yielded inconsistent results [ 1 , 2 , 5 – 15 ]. Some studies have reported significantly lower anti-Müllerian hormone (AMH) levels, a measure of ovarian reserve, among smokers [ 1 , 2 , 5 – 9 ]. Schuh-Huerta et al. however, found that serum AMH levels were significantly higher among smoking women compared to non-smoking women [ 10 ]. Other studies found no significant association between smoking and measures of ovarian reserve [ 11 – 15 ]. Previous results are difficult to reconcile as several of the studies [ 1 , 2 , 8 , 9 , 15 , 16 ] were conducted among infertile or sub-fertile populations with none of the studies controlling for polycystic ovary syndrome (PCOS). Women with PCOS diagnosis have been shown to have significantly elevated AMH levels [ 17 , 18 ]. Genetic heterogeneity is another factor that may explain the inconsistent results of previous studies on the association of smoking with ovarian reserve. N-acetyltransferase2 (NAT2) acetylator status (rapid, intermediate, or slow) has been shown to modify the association of smoking with different disease outcomes, with slow acetylators often more susceptible [ 19 – 24 ]. Taylor et al. reported reduced fecundability among current smokers who are slow acetylators [ 20 ]; however, it remains to be established whether these effects on fecundability are modulated through a reduction in ovarian reserve. No studies have assessed the effect of NAT2 acetylator status or the potential interaction with smoking on ovarian reserve. This study assessed the association of smoking with ovarian reserve in a clinical population of women seeking fertility counseling and used urinary cotinine to validate current smoking status and a smoking questionnaire to estimate cumulative lifetime exposure. In addition, this study explored genetic heterogeneity by assessing whether NAT2 acetylator status influenced the association of smoking with ovarian reserve. Analyses were further stratified by PCOS status to account for higher baseline AMH levels among women with PCOS.

Out of 261 women who were enrolled into the LOUSSI study, 54 were excluded due to incomplete records (52 were missing serum AMH levels and 2 were missing PCOS status) ( Fig 1 ). Thus, 207 women were included in the final analysis of the questionnaire data (self-reported dose data such as pack-years). For the analysis of current smoking status using cotinine measurement, 8 of the 207 were excluded as they did not provide urine for cotinine assays ( Fig 1 ). Of the remaining 199 participants, 29% were current active smokers with 55% of these reporting smoking 10 or more cigarettes per day, 21% smoking between 5 to 9 cigarettes per day and 24% reported smoking fewer than 5 cigarettes per day. Current active smokers were more likely to be of self-reported Black race (p<0.001), obese (p = 0.03) and single (67% vs. 22%, p = <0.001), compared with non-smokers ( Table 1 ). There were no significant differences between current active smokers and non-smokers for the other characteristics examined ( Table 1 ). a Women with cotinine measurements. b SD = Standard deviation We also examined the association of each potential covariate with ovarian reserve, as measured by either DOR or % change in AMH. Only age and PCOS status were significantly associated with ovarian reserve in bivariable analyses and were considered as potential confounders in the multivariable analyses. A 48% decrease (95%CI: -71.0 to -8.5) in AMH level was reported among women between ages 31 to 35 years and a 67% decrease (95%CI: -81.0 to -40.6) among women older than 35 years of age, relative to women less than 31 years old. A similar inverse association was found between age and DOR, with women between the ages of 36 to 45 years having 6 times the odds of DOR (OR = 6.2; 95% CI:1.7–22.3) compared to women 21 to 25 years of age. A PCOS diagnosis was associated with a significant increase in AMH levels (177.0%; 95% CI: 89.0–306.1) and a decrease in the odds of DOR (OR = 0.3; 95%CI: 0.1–0.7). Neither serum AMH levels nor DOR was found to be significantly associated with race, BMI, age at menarche, or parity. Current active smoking, based on cotinine measurement, was not significantly associated with DOR (OR = 0.62; 95% CI: 0.24–1.56); neither was passive smoking: (OR = 1.32; 95% CI: 0.59–3.0) ( Table 2 ). Results were similar among women without PCOS. The results were similar when assessment of current smoking was based only on self-report (OR = 1.22; 95% CI: 0.45–3.27; Non PCOS: OR = 1.87; 95% CI: 0.56–6.33). a Models adjusted for age and polycystic ovary syndrome b Models adjusted for age c P-value = 0.04 However, in the analysis of current cigarettes reported per day on the questionnaire, which was analyzed regardless of cotinine levels, current cigarette smoking was associated with an increased odds of diminished ovarian reserve among women without PCOS. On average, the odds of DOR increased by 8% for every additional cigarette currently smoked (OR = 1.08; 95% CI:1.01–1.15). There was also suggestion of a dose-response with self-reported lifetime exposure. For every 1 pack-year increase in lifetime exposure, the odds of DOR increased by 8% (OR = 1.08; 95% CI: 0.99–1.18). Results were similar but not statistically significant when analyses were limited to women with validated cotinine levels and without PCOS (Cigs/day: OR = 1.06; 95% CI: 0.99–1.14; pack-year: OR = 1.08; 95% CI: 0.99–1.18). We also examined cigarettes per day and pack-years of exposure in categories ( Table 2 ). In these models, the effects of smoking on DOR were only observed in the highest exposure groups (at least 10 cigarettes/day or more than 5 pack-years of exposure), suggesting that the observed associations may be driven by those with high exposure levels. Adjusted linear regression models for the association of current active smoking and percentage change in AMH are presented in Table 3 . There was no significant association found between current smoking, the number of cigarettes smoked, or cumulative lifetime exposure and AMH levels. a Models adjusted for age and polycystic ovary syndrome b Models adjusted for age A total of 11 NAT2 genotypes were identified. The concordance of genotype duplicates was 100% for all four SNPs. DNA extraction or NAT2 genotyping was unsuccessful in 14 participants, yielding a sample size of 155 for the NAT2 interaction analysis. Approximately 39% of women in the study were slow acetylators, 44% were intermediate, and 17% were rapid acetylators. The distribution of NAT2 acetylator status did not differ by smoking status ( Table 1 ) or self-reported race (S3 Table in S1 File ). After adjusting for age, race and PCOS status, there was no statistically significant interaction with NAT2 phenotype ( Table 4 ). Although there was a suggestion that the odds of DOR was higher among slow acetylators (OR = 1.59; 95% CI: 0.26–9.70), the sample size was insufficient to provide adequate statistical power. a Adjusted for age, race, and polycystic ovary syndrome; p-value for interaction = 0.67 b Adjusted for age and race; p-value for interaction = 0.69

Current smoking was not associated with diminished ovarian reserve. However, the trends and directions of effect observed in this study suggest that heavy smoking, or smoking for a long duration, may reduce ovarian reserve. Additional data are needed to better define the role of NAT2 polymorphisms with diminished ovarian reserve following exposure to toxins such as those found in cigarette smoke.

The Louisville Tobacco Smoke Exposure, Genetic Susceptibility, and Infertility Study (LOUSSI study) is an observational study that is focused on the effects of tobacco exposure and NAT2 acetylator status on ovarian reserve and IVF outcomes. All women 21 years and older seeking infertility treatment in the University of Louisville Reproductive Endocrinology and Infertility (REI) Division were eligible for enrollment into the study. Excluded were women with ongoing pregnancies and patients who could not communicate in English or were unable to understand and complete the informed consent and questionnaire. Women over 45 years of age at the time of initial consult visit were excluded. A total of 261 women were recruited between September 2016 and June 2018. The University of Louisville Institutional Review Board approved this study for human subjects, and all subjects provided written consent prior to participation (IRB number- 16.0063). Current smoking status was assessed based on urinary cotinine levels. Cotinine, a metabolite of nicotine, has a half-life that ranges from 7 to 40 hours and levels can be used to assess recent exposure to nicotine (past 3–5 days) [ 25 – 27 ]. Urine was collected at enrollment and stored at -80°C until time of assay. Cotinine ELISA assays (Calbiotech, Spring Valley, CA) were used be estimate urinary cotinine levels with maximum detectable level of cotinine for the assay of 100ng/mL. Urinary cotinine levels were estimated to the nearest 0.50 ng/mL. All self-reported smokers except for one had cotinine levels >100ng/ml; therefore, based on preliminary analysis examining the distribution of cotinine levels in the study population and evidence from literatures [ 25 , 28 , 29 ], the cotinine level used to discriminate smokers from non-smokers was set at ≥14 ng/mL. A “current active smoker” was defined as a person with urinary cotinine levels of ≥14 ng/mL [ 25 ]. Women with cotinine levels between 0.5–13.9 ng/mL were classified as “passive smokers” [ 25 ]. A “nonsmoker” was defined as a woman with cotinine levels <0.5 ng/mL [ 25 ]. Dose-response was assessed using response to a self-administered smoking questionnaire (SSQ). Cumulative lifetime smoking was based on pack-years smoked, calculated for current and former smokers (women with history of smoking at least 1 cigarette/week, who reported quitting more than one month prior to enrollment) by multiplying the number of packs smoked per day by the years of smoking, as reported on the questionnaire. Genomic DNA was isolated from urine samples using the ZR Urine DNA Isolation Kit™ (Zymo Research, Irvine, CA. USA) according to the manufacturer’s instructions. NAT2 alleles, haplotypes, genotypes, and deduced phenotype were determined using a NAT2 four-SNP genotype panel of rs1801279 (191G>A), rs1801280 (341T>C), rs1799930 (590G>A) and rs1799931 (857G>A). The accuracy of the four-SNP panel in determining NAT2 acetylator status is 98.4% and comparable to the seven-SNP panel [ 30 ]. The assay uses SNP-specific PCR primers and fluorogenic probes designed using Primer Express™ (Applied Biosystems, CA, USA). The presence of the four SNPs was determined using a predeveloped TaqMan ® SNP Genotyping Assay (Applied Biosystems) according to the manufacturer’s instructions. Positive and negative (no DNA template) controls were run to ensure that there was no amplification of contaminating DNA, and between 10% and 20% of samples from each plate were run in duplicate. Individuals possessing two of the NAT2 alleles associated with rapid acetylation ( NAT2*4 ) were classified as rapid acetylators; individuals possessing one of the alleles associated with rapid acetylation and one allele associated with slow acetylation ( NAT2*5 , NAT2*6 , NAT2*7 , and NAT2*14 ) were classified as intermediate acetylators; and those individuals who possessed two slow acetylation alleles ( NAT2*5 , NAT2*6 , NAT2*7 , and NAT2*14 ) were classified as slow acetylators. Ovarian reserve can be measured using serum AMH levels or antral follicle count. Serum AMH and antral follicle count (AFC) have strong and similar linear relationship with the size of the primordial follicle pool and ovarian reserve [ 31 – 34 ]. Studies have shown that AMH level and AFC are highly correlated [ 15 , 35 ]. The limited intra- and inter-cycle variation, objectivity, and potential standardization of AMH assays, makes it the preferred biomarker of ovarian reserve in women. Assessment of ovarian reserve was based on baseline serum AMH levels extracted from patients’ medical records. Women seen in the University of Louisville REI Division routinely provide a serum sample for analysis of reproductive hormones, including AMH, as part of the initial infertility workup. AMH assays were performed by Quest Diagnostics (Louisville, KY) using chemiluminescence. The limit of detection (LOD) for the assay was 0.03ng/mL; AMH values below the LOD were assigned a value of 0.03ng/mL. For this study, diminished ovarian reserve was defined as baseline serum AMH level less than 1ng/mL [ 13 , 36 ]. The effect of current active smoking on ovarian reserve was assessed by comparing current active smokers with never smokers and passive smokers. The dose of exposure for current active smoking was based on number of cigarettes smoked per day reported on the SSQ, irrespective of the urinary cotinine values. The dose-response relationship between current active smoking and ovarian reserve was modeled as both a continuous exposure variable and a categorical variable: non-smoker (0 cigarettes/day), moderate smoker (1– <10 cigarettes/day), or heavy smoker (≥ 10 cigarettes/day). Cumulative lifetime exposure was based on number of packs smoked per day and duration of smoking in years reported on the SSQ by both current active smokers and former smokers. The effect of cumulative lifetime years of active smoking was assessed with both a continuous exposure variable and categorical variable, coded as never smoker, 1–5 pack-years: or >5 pack-years. Covariates extracted from the medical record included age; race; weight and height at enrollment for calculation of body mass index (BMI); age at menarche; parity; and PCOS status. Age at enrollment was categorized as 21–25, 26–30, 31–35 and 36–45 years. Parity, defined as the number of live births, was put into 3 categories: nulliparous, one live birth, or two or more live births. PCOS status was extracted from the medical record, documented as a current diagnosis or as a previous diagnosis self-reported by the patient. BMI at enrollment was categorized as normal weight (17.5–24.9), overweight (25.0–29.9), obese (30.0–34.9), or morbidly obese (≥35). Differences in demographics and covariates between smokers and non-smokers were tested using a Chi-square/Fisher’s exact test for categorical variables, Student’s t-test for continuously normally distributed variables, and the Wilcoxon rank-sum test for continuous variables that were not normally distributed. Linear regression models using the natural logarithm (ln) of AMH as the dependent variable were generated to estimate the effects of the different smoking variables on ovarian reserve, adjusted for covariates. Percent differences and 95% confidence intervals (CIs) in serum AMH between exposure groups were calculated from the parameter estimates as follows: ([exp(β) −1] *100) and presented with corresponding 95% confidence intervals (CIs) and p-values [ 13 ]. Unconditional logistic regression models were used to estimate odds ratios (ORs) and 95% CIs for the association of smoking on DOR. Regression analyses were adjusted for variables that were significantly associated with ovarian reserve (age and PCOS). Effect modification by NAT2 genotype was assessed by including an interaction term between NAT2 and the smoking variables in the multivariable models. The rapid acetylator phenotype was combined with the intermediate acetylator phenotype to form the referent group for comparison with the slow acetylator group which has been used in other studies [ 30 , 37 – 39 ]. Interaction models were further adjusted for race, in addition to other covariates. Ethnic differences in the frequency of rapid and slow acetylator NAT2 alleles or haplotype have been reported [ 20 , 40 – 42 ]. A P-value <0.05 was considered statistically significant. All regression models were also run restricted to women without PCOS. All analyses were two-sided and conducted using Statistical Analysis Software version 9.4 (SAS Institute, Cary, NC).

(DOCX) Click here for additional data file.