Male partner pesticide residue exposure from fruit and vegetable intake in relation to outcomes of infertility treatment: a prospective cohort study.

This work represents a prospective evaluation of participants from the Environment and Reproductive Health (EARTH) cohort study, a prospective cohort study established in 2004 at the Massachusetts General Hospital Fertility Center (Boston, USA) to investigate how environmental and lifestyle factors affect fertility ( 14 ). The study enrolled males aged 18–55 and females aged 18–45, seeking infertility evaluation or treatment at the Massachusetts General Hospital Fertility Center between 2004 and 2019 and planning to use their own gametes for infertility treatment were invited to participate in the study ( 14 ). Participants were followed through their infertility treatment cycles, with dietary assessments collected beginning in 2007, and outcomes tracked through ART completion, clinical pregnancy, and live birth, up to 2021 ( 14 ). The primary outcome of the EARTH study was clinical pregnancy and live birth after ART. Secondary outcomes broadly included a range of ART outcomes (e.g., fertilization, implantation, and pregnancy loss), time to pregnancy, and intermediate endpoints such as semen quality and reproductive hormone concentrations ( 14 ). The primary exclusion criteria for the EARTH study were a history of vasectomy in males or the use of exogenous hormones at enrollment ( 14 ). Participants completed a standardized health history and physical examination at enrollment. Self-administered questionnaires collected detailed demographic information (e.g., age, race [White, Black, Asian, Other], education), reproductive and medical history (e.g., parity, infertility diagnosis), and lifestyle behaviors, including multivitamin use, smoking status, and physical activity. Physical activity was quantified using a validated questionnaire ( 15 ) and expressed as minutes of moderate-to-vigorous activity per week. Diet was assessed via a 131-item food frequency questionnaire (FFQ) ( 16 , 17 ) beginning in 2007, as described in greater details below. Anthropometric data (e.g., height and weight) were measured by trained personnel to calculate body mass index (BMI). Couples were encouraged but not required to participate together. This analysis included male-female couples where the male partners’ preconception diet was assessed, and the female partner initiated at least one ART cycle between 2007 and 2021. The current analysis includes only couples undergoing ART, that is, infertility treatments involving the manipulation of both sperm and oocytes. These include in vitro fertilization (IVF), where semen processed to increase sperm concentration and motility is placed in wells containing one oocyte each, and intracytoplasmic sperm injection (ICSI), where an individual sperm is selected and injected into a mature oocyte ( 18 ). Of the 462 couples enrolled after diet assessments began, we excluded 218 couples due to (1) the male partner not completing a dietary assessment before infertility treatment or had implausible reported energy intake (4200 kcal/day; n=79), (2) the use of donor sperm (n=1), and (3) intrauterine insemination cycles (n=138). The final analysis included 244 couples (437 ART cycles) whose diet was assessed before ART outcomes and who contributed at least one ART cycle in 2007–2021 ( Figure S1 ). The Institutional Review Boards at the Massachusetts General Hospital and the Harvard TH Chan School of Public Health approved the study protocols (Protocol Number: 1999P00816). All participants provided informed written consent before enrollment. Findings are reported based on Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines ( 19 ). Participants’ usual diet over the year before enrollment was captured at study entry using a validated self-administered 131-item food frequency questionnaire ( 16 ). The FFQ used natural or standard portion sizes (e.g., one apple, a slice of bread, ½ cup of cooked vegetables) derived from diet record data and common serving conventions. Participants were asked to report average frequency of intake over the past year using nine response categories, ranging from “never or less than once per month” to “six or more times per day” ( 16 ). We derived two dietary pattern scores, Western and prudent ( 20 ), using principal components analysis, excluding intake of FV to characterize participants’ overall eating patterns. Nutrient intakes were estimated using a nutrient database derived from the US Department of Agriculture (USDA) with additional manufacturer information. For each FV item in the FFQ, we obtained pesticide residue contamination data from the USDA Pesticide Data Program (PDP) ( 21 ). The PDP is a national program initiated in 1991. The program annually tests ~10,000 samples from agricultural commodities sold in the US for the presence of ∼450 different pesticide residues. Foods sampled by the program are randomly selected without regard to whether they are conventionally grown or organic. We utilized PDP annual reports (2006–2020), aligning with the study’s diet assessment period, to calculate the pesticide residue burden score (PRBS) for each FV item as previously described ( 5 , 22 , 23 ). Briefly, three contamination measures from PDP reports were used to classify FV: (1) percentage of samples with detectable pesticides, (2) percentage exceeding tolerance levels, and (3) percentage with ≥3 detectable pesticides. FV contamination data were categorized into tertiles for each of the three contamination measures, with scores assigned as follows: 0 for the bottom tertile, 1 for the middle, and 2 for the top. The final PRBS for each food was the sum of these scores (range: 0–6). FV with a PRBS ≥4 were classified as high pesticide residue foods, while those with PRBS <4 were classified as low pesticide residue foods. Based on these criteria, 16 items were categorized as high pesticide residue and 20 as low pesticide residue ( Table S1 ). We have previously validated this PRBS score against biomarkers of pesticide exposure ( 22 , 24 ). In a nationally representative sample of 3,679 NHANES participants, we identified associations between scores—the PRBS, an organophosphate pesticide score, and an organochlorine pesticide score—and corresponding urinary or serum pesticide biomarkers ( 22 ). Likewise, in a sample of 90 men from the EARTH study, we found that the molar sum of urinary concentrations of seven urinary biomarkers—including those for organophosphates, pyrethroids, and the herbicide 2,4-dichlorophenoxyacetic acid—measured in two urine samples as well as most individual urinary biomarkers, were positively associated with intake of high pesticide residue FV ( 24 ). The primary outcome of this study was the probability of live birth. Secondary outcomes were fertilization rate, the probability of low fertilization, implantation, clinical pregnancy, total pregnancy loss, and live birth per initiated cycle. Fertilization rate was assessed 17–20 hours post-insemination, defined as the proportion of two pronuclear (2PN) zygotes/number of metaphase 2 (M2) oocytes and further evaluated by the mode of insemination (conventional insemination or intracytoplasmic sperm injection (ICSI). Low fertilization was defined as 2PN/M2 ≤50%/cycle ( 25 , 26 ). Implantation was defined as a serum beta-human chorionic gonadotropin (β-hCG) concentration >6 mIU/mL measured ~2 weeks following embryo transfer. Clinical pregnancy was defined as the visualization of an intrauterine gestational sac via ultrasound at ~6 weeks gestation. Live birth was defined as the delivery of a neonate ≥24 weeks of gestation. Pregnancy loss was defined as a spontaneous fetal demise at any point during gestation. Participants were categorized into quartiles of intake of high- and low pesticide residue FV. Baseline characteristics were reported as medians and interquartile ranges (IQR) for continuous and proportions and percentages for categorical variables. Associations between high- and low pesticide residue FV intake were assessed using Spearman correlation coefficients. Kruskal-Wallis tests (for continuous variables) and Chi-square with Fisher’s exact tests (for categorical variables) were used to compare baseline characteristics across quartiles of FV intake. We estimated the probability and 95% confidence interval (95% CI) for ART outcomes by fitting multivariable linear mixed models with binary (fertilization) or binomial (low fertilization rate, implantation, clinical pregnancy, pregnancy loss, live birth) distribution and random intercepts to account for repeated cycles. Linear trend tests were performed by modeling adherence quartiles as a continuous variable. In figures, variation is reported as standard errors for improved clarity. Confounding factors were determined based on prior knowledge and baseline differences in patient characteristics by FV intakes. We fitted three sets of models: the initial model (Model 1) adjusted for male age (continuous; year), female age (continuous; year), and male energy intake (continuous; kcal/d). Model 2 further adjusted for male body mass index (BMI; continuous; kg/m 2 ), education (categorical; college degree or higher [yes/no]), smoking status (categorical; never, current, past), and dietary patterns (continuous; Western and prudent dietary scores). Also, in Model 2, high pesticide residue FV intake was adjusted for low-pesticide FV intake (continuous; servings/day) and vice versa. Model 3 additionally adjusted for female partners’ high- and low pesticide residue FV intakes (continuous; servings/day). We used complete-case analysis in Model 1. In Models 2 and 3, missing data for covariates were addressed using median imputation for continuous variables (i.e., male BMI [n=4], female BMI [n=5]) and by including a separate missing category for a categorical variable (i.e., male education [n=18]). Since fertilization rates and live birth rates differ between ART cycles using IVF and ICSI for fertilization ( 27 , 28 ) and ICSI is often used in couples with a severe male factor or repeated treatment failure ( 29 ), we evaluated all outcomes first combining IVF and ICSI cycles and then separately for IVF and ICSI cycles. MK conducted all analyses in SAS v9.4 (SAS Institute Inc., Cary, NC) and SPSS v29.0 (IBM Corp., Armonk, NY). All statistical tests were two-sided, and p<0.05 was determined as statistically significant.

Male partners had a median age of 36.6y (IQR 33.0–39.9y) and BMI of 26.9 kg/m 2 (IQR 24.2–29.3 kg/m 2 ). Their female partners had a median age of 35.0y (IQR 32.0–38.0y) and BMI of 23.1 kg/m 2 (IQR 21.1–25.7 kg/m 2 ). Most couples were White (88.9% of males and 84.4% of females), had never smoked (66.0% of males and 73.4% of females), and had at least one partner with a college degree (78.7% of males and 86.9% of females). Participants had a median FV intake of 4.2 (2.9, 6.3) servings/day, of which 1.5 (0.9, 2.5) servings/day were high- and 2.0 (1.2, 2.8) servings/day were low pesticide residue FV ( Table 1 ). Intakes of high– and low–pesticide residue FV were positively correlated (Spearman ρ = 0.66). Similarly, male and female partner intakes of FV, overall and classified according to pesticide residue status, were modestly correlated ( Table S2 ). Age, race, multivitamin use, and physical activity were similar across quartiles of high- and low pesticide residue FV intakes. However, BMI, education, smoking, total energy intake, adherence to prudent dietary patterns, and female partner FV intake differed across the quartiles of FV intake regardless of pesticide residue status. Primary infertility diagnosis and initial treatment cycle characteristics—stimulation protocol, number of embryos transferred, embryo transfer day, and type of insemination—did not differ by FV intake, regardless of pesticide residue status ( Table 1 ). Intake of high pesticide residue FV was inversely associated, whereas intake of low pesticide residue FV was positively associated with fertilization rate ( Figure 1 and Table S3 ). After accounting for potential confounders, including female partner intake of high- and low pesticide residue FV, the multivariable-adjusted fertilization rates (95%CI) in the lowest and highest quartiles of intake were 0.78 (0.73, 0.83) vs. 0.69 (0.62, 0.75) for high pesticide residue FV ( Fig 1A ), and 0.71 (0.65, 0.77) vs. 0.81 (0.75, 0.85) for low pesticide residue FV ( Figure 1B ). This relation was more pronounced when we evaluated low fertilization rate (<50% fertilization) as an outcome. In multivariable-adjusted models, the probability of having a low fertilization cycle in the lowest and highest quartiles of intake of high pesticide residue FV was 0.04 (0.01, 0.10) vs. 0.18 (0.09, 0.34; Figure 1C ), while the corresponding figures for low pesticide residue FV intake were 0.22 (0.12, 0.36) vs. 0.02 (0.01, 0.08; Figure 1D and Table S3 ). In analyses stratified by insemination method, these relations were restricted to conventional IVF cycles and were more pronounced for low pesticide residue FV intake ( Table S4 ). High or low pesticide residue FV intake was unrelated to implantation, clinical pregnancy, and live birth in analyses considering all treatment cycles ( Table 2 ). In analyses stratified by insemination method, FV intake, regardless of pesticide residue status, remained unrelated to the probability of implantation ( Table S5 ) and clinical pregnancy ( Table S6 ) in both IVF and ICSI cycles. Nevertheless, high pesticide residue FV intake was inversely related to the probability of live birth in IVF cycles, unlike ICSI cycles ( Figure 2 and Table S7 ) . The adjusted probability of live birth in the lowest and highest quartiles of high pesticide residue FV intake were 0.58 (0.32, 0.80) vs. 0.14 (0.04, 0.38) in IVF cycles (p-trend=0.03; Figure 2A ), and 0.13 (0.02, 0.59) vs. 0.36 (0.07, 0.81) in ICSI cycles (p-trend=0.51; Figure 2C ). Intake of low pesticide residue FV was unrelated to live birth in stratified models ( Figures 2C and 2D ). Mirroring the results for live birth, intake of high pesticide residue FV was positively related to the probability of pregnancy loss in IVF cycles but not in ICSI cycles ( Table 3 ) nor analyses combining all cycles ( Table S8 ).

In this prospective cohort study of couples undergoing infertility treatment with ART, we found that male partner intake of high pesticide residue FV was associated with lower fertilization rate, lower live birth rate, and a higher rate of pregnancy loss in analyses involving all ART cycles and in analyses restricted to IVF cycles but not in analyses restricted to ICSI cycles. Nearly mirroring these findings, higher intake of low pesticide residue FV was linked to a higher fertilization rate in all ART cycles and in IVF cycles but not in ICSI cycles. These associations were independent of female partner exposure status. These observations suggest that pesticide exposure may be associated with male reproductive function beyond the previously well-documented effects on spermatogenesis. Moreover, the observation that these relations are present in IVF cycles and absent in ICSI cycles suggests that exposure to pesticide residues through diet may have an impact on sperm function or sperm-oocyte interactions that cannot be bypassed by increasing the number of motile sperm available for fertilization. Our findings should be interpreted considering the study’s strengths and limitations. First, diet was self-reported, our FFQ did not distinguish between organic and conventionally grown foods, and pesticide exposure data were sourced from PDP reports. Furthermore, PDP does not distinguish between organic and conventionally grown produce in its sampling of foods for pesticide residue assessment. Therefore, evaluating the effects of specific pesticides or the difference between organic and conventionally grown foods is not possible. However, the PRBS has been shown to accurately rank exposure to a broad range of pesticide classes measured in biological samples in a subset of participants of this study and among participants of NHANES, which is designed to be representative of the general US population ( 22 , 24 ). Further, our validation in this cohort showed that self-reported intake of organic foods assessed with a separate questionnaire was not associated with biomarkers of exposure to pesticides ( 22 ). Second, the PRBS does not differentiate between pre- and post-harvest pesticide residues or weight fungicides differently from other classes. Given that many pesticide-intensive crops include both neonicotinoids and multiple fungicides (including post-harvest applications), future refinements should consider these distinctions to improve exposure classification. Additionally, one of the scored components of the PRBS is the percentage of samples exceeding tolerance levels, which has not been established for all possible pesticide–food commodity combinations. This can result in systematically lower scores for this component. It should be kept in mind that the absence of tolerance should not be interpreted as an indication of safety, as tolerance limits themselves may not always represent toxicologically meaningful thresholds. Therefore, this aspect of the PRBS should be interpreted with caution. Third, many of our participants were White and had high levels of education. Although this demographic pattern is typical of those seeking fertility care in the US ( 30 ), our findings may not be generalizable to all races and ethnicities ( 31 ). On the other hand, the relative homogeneity of study participants minimizes confounding effects by social and economic factors and bolsters internal validity. Finally, because we analyzed a population of couples undergoing infertility treatment, the findings may not be generalizable to couples attempting conception without medical assistance. The study’s strengths included a large sample size relative to similar work in the field, a prospective design that enabled us to measure changes in ART outcomes in response to pesticide exposure and clearly define the temporal relation between them, the use of validated methods of diet assessment and pesticide residue ranking to define exposure, clinically established ART outcomes, standardized and rigorous assessment of a broad array of participants’ baseline characteristics, that allowed us to assess relevant clinical outcomes, and meticulous statistical adjustment for potential confounders. Although pesticide exposure through diet intake is prevalent, evaluating its impact on male contributions to couple’s fertility and male reproductive health in general is scarce. We are unaware of previous work on the relationship between exposure to pesticide residues from diet in males and couples’ fertility or ART outcomes. However, evidence from recent systematic reviews has provided insights into the link between environmental and occupational exposure to pesticides and disrupted male reproductive function, offering insights into possible pathways, including detrimental effects on semen quality, sperm DNA fragmentation and chromosome aneuploidy ( 32 , 33 ). Evidence from experimental animal studies indicates that exposure to pesticides, including organophosphate and pyrethroid insecticides such as diazinon ( 34 , 35 ), chlorpyrifos ( 36 ), and permethrin ( 37 , 38 ), as well as the herbicide 2,4-D ( 39 , 40 ), negatively impact male reproductive health, through pathways explored in previous studies ( 41 , 42 , 43 , 44 ). Namely, pesticides have been shown to induce oxidative stress ( 42 ), disrupt hormonal pathways, and alter the chromatin structure of sperm DNA and induce sperm aneuploidy, leading to testicular toxicity, impaired spermatogenesis, and reduced fertilization capacity ( 37 , 42 , 45 , 46 , 47 ). Pesticide’s endocrine disrupting and oxidative stress-inducing properties could extend beyond fertilization, influencing embryo quality, implantation success, and pregnancy maintenance ( 42 , 48 ). Notably, oxidative damage to sperm DNA may result in compromised peri-implantation embryo development, leading to implantation failure or early pregnancy loss ( 48 ). Additionally, endocrine-disrupting chemicals of pesticides can disrupt the hormonal environment necessary for pregnancy maintenance, increasing miscarriage risk ( 10 , 32 ). The finding that these adverse effects were more pronounced in IVF cycles and not in ICSI cycles suggests that the natural fertilization process may be more susceptible to the detrimental impacts of pesticide exposure on sperm quality. However, further research is needed to confirm these findings and elucidate the specific biological mechanisms. Our findings are consistent with our previous reports on the relation between exposure to pesticide residues through FV intake and conventional semen quality parameters. Critically, we have previously reported that intake of high pesticide residue FV is related to abnormal sperm morphology in this study population ( 6 ). On the other hand, we have also reported that low pesticide residue FV intake is positively related to total normal count and total motile count among healthy young males ( 7 ). These previous findings are relevant to the interpretation of the current findings to the extent to which sperm morphology and motility are known to predict both fertilization rate and live birth rates during infertility treatment with ART ( 49 ). Moreover, the clear difference in the association between conventional IVF and ICSI cycles further suggests that this is a sperm-mediated effect and that only the most stringent sperm selection (i.e., ICSI) may be able to negate the adverse effect that exposure to pesticides may have on a couple’s fertility. It is also important to highlight that the current findings also align with our previous study, where we found that intake of high pesticide residue FV intake was related to a lower live birth rate and a higher rate of pregnancy losses during infertility treatment with ART ( 5 ). Despite the apparent similarities between the findings of the two studies, the differences may be more informative. First, in sharp contrast with the present study, we found no relation between female high- or low pesticide residue FV intake and fertilization rate. More subtly, while we did observe lower live birth rates and higher pregnancy loss rates associated with high pesticide residue FV intake in both studies, the pattern of association differs for male and female partner exposure. Male partners’ high pesticide residue FV intake was related to lower live birth rate but unrelated to clinical pregnancy or implantation, whereas female high pesticide residue FV intake was related to all three outcomes. This suggests that exposure to pesticide residues in females may affect earlier stages of pregnancy viability, starting with implantation failures, whereas the same exposure in males may exclusively affect live birth rates by increasing the rate of clinical pregnancy losses. While the clinical outcome is ultimately similar, these differences may point to the identification of specific biological mechanisms through which pesticides impair human reproduction. While our findings are specific to couples undergoing infertility treatment, they contribute to a growing body of evidence linking dietary pesticide exposure to impaired human reproductive health. Our results build on our previous work showing associations between pesticide residues and semen quality, sperm morphology ( 6 , 7 ), and pregnancy ( 5 ) and ovarian reserve ( 4 ) outcomes, and add a novel dimension to evidence reinforcing concerns that dietary exposure to pesticides may have clinically meaningful effects on fertility. Although the direct implications for natural fertility remain uncertain—particularly because the association we report between male dietary pesticide exposure and reduced fertility treatment success appears independent of semen quality and represents a novel observation—these findings nonetheless highlight a broader concern. Specifically, they point to the possibility that cumulative environmental exposures, including dietary pesticide intake, may contribute to a gradual erosion of reproductive potential across populations. Mounting evidence suggests human reproduction has already shown long-term declining trends ( 50 , 51 ), and even small reductions in fertility-related parameters at the individual level may shift the population closer to critical inflection points where widespread reproductive challenges could emerge. In this context, our findings, if confirmed by further research, have implications extending beyond assisted reproduction and rather be viewed as a potential signal of wider reproductive vulnerability that warrants both continued investigation and proactive public health consideration. In conclusion, our findings suggest that male partners’ intake of high pesticide residue FV is associated with lower fertilization and live birth rates in couples undergoing infertility treatment involving all ART cycles and in analyses restricted to IVF cycles but not in analyses restricted to ICSI cycles. While the implications of these findings for human reproduction without medical assistance are unclear, they strongly suggest that exposure to pesticide residues through diet may impair the ability of the sperm to fertilize oocytes and affect the viability of any resulting pregnancies. Moreover, they add to the evidence that levels of exposure to pesticide residues currently considered safe by regulatory agencies may have measurable adverse health effects in humans.

Traces of pesticide residues or their metabolites have been detected in over 90% of individuals in a representative sample of US adults, with diet—particularly the consumption of conventionally grown fruits and vegetables (FV)—identified as a primary source of exposure ( 1 ). While the Environmental Protection Agency (EPA) limits pesticide levels in foods, emerging evidence has raised concerns that chronic exposure, even within these limits, may pose health risks ( 2 , 3 ). We previously reported that female intake of high pesticide residue FV was associated with a lower ovarian reserve( 4 ) and the probability of live birth in couples undergoing infertility treatment with assisted reproductive technologies (ART) ( 5 ). In males, we showed that consuming FV with high pesticide residues was associated with lower sperm count and a reduced percentage of morphologically normal sperm ( 6 ) whereas consuming FV with low pesticide residues was positively associated with total sperm count ( 7 ). On a broader scale, evidence from other studies has supported the contribution of pesticides to male infertility, secondary to azoospermia, oligospermia, increased serum levels of follicle-stimulating hormone and luteinizing hormone ( 8 ), estradiol ( 9 ), and decreased testosterone levels ( 10 , 11 ). However, it is unclear whether associations with semen quality or hormonal parameters may extend to more direct measures of couple fertility, particularly since previous studies suggest that improvements in semen quality linked to male partners’ diet quality do not necessarily translate into improved couple fertility treatment outcomes, such as higher pregnancy or live birth probabilities ( 12 , 13 ). To address this critical knowledge gap, we tested the hypothesis that male preconception intake of high pesticide residue FV negatively influences a couple’s infertility treatment outcomes.