Men’s dietary patterns in relation to infertility treatment outcomes among couples undergoing in vitro fertilization

Abstract

Purpose(s)

To evaluate the relationship of men’s dietary patterns with outcomes of in vitro fertilization (IVF).

Methods

This is a prospective cohort study including 231 couples with 407 IVF cycles, presented at an academic fertility center from April 2007 to April 2018. We assessed diet with a validated food frequency questionnaire and identified Dietary Pattern 1 and Dietary Pattern 2 using principal component analysis. We evaluated adjusted probability of IVF outcomes across the quartiles of the adherence to two dietary patterns by generalized linear mixed models.

Results

Men had a median age of 36.8 years and BMI of 26.9 kg/m². Women’s median age and BMI were 35.0 years and 23.1 kg/m², respectively. Adherence to Dietary Pattern 1 (r_Pearson=0.44) and Dietary Pattern 2 (r_Pearson=0.54) was positively correlated within couples. Adherence to Dietary Pattern 1 was positively associated with sperm concentration. A 1-unit increase in this pattern was associated with a 13.33 (0.71–25.96) million/mL higher sperm concentration. However, neither Dietary Pattern 1 nor Dietary Pattern 2 was associated with fertilization, implantation, clinical pregnancy, or live birth probabilities.

Conclusions

Data-derived dietary patterns were associated with semen quality but unrelated to the probability of successful IVF outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-021-02251-9.

Introduction

Infertility, defined as the failure to achieve a clinical pregnancy after 12 months or more of regular unprotected sexual intercourse, is a worldwide issue [1] affecting 15% of reproductive age couples [2]. Male factor is one of the most common causes of infertility and male infertility evaluation is important not only for defining infertility treatment strategies but also for men’s health itself as male infertility could be a predictor of future morbidity [3]. Contributing factors to male infertility are various: genetic factors, environmental factors such as smoking, alcohol consumption, psychological stress, substance abuse, exercise, and comorbidities including cardiovascular disease, hyperlipidemia, diabetes, and obesity [4–10].

Previous epidemiological work suggests that paternal dietary patterns associated with lower risk of cardiovascular disease and other chronic conditions [11–13], such as the Mediterranean diet and the dietary approaches to stop hypertension (DASH) diet may be positively associated with semen quality. On the other hand, dietary patterns favoring intakes of red and processed meats, animal fat, refined grains, and sweets—which have been related to a higher risk of non-communicable chronic diseases [14]—may affect negatively semen quality [15–20]. However, semen parameters are not perfectly correlated with a couple’s fertility [21, 22]. Moreover, while some evidence suggests men’s diet may influence a couple’s fertility [23–25], data on the extent to which men’s diet could impact fertility in couples trying to conceive naturally or with medical assistance remains scarce. To address this important question, we evaluated the association between adherence to men’s dietary patterns identified in the study population using a data-driven approach and outcomes of infertility treatment with in vitro fertilization (IVF). We hypothesized that couples with a male partner with greater adherence to dietary patterns consistent with diets associated with lower risk of cardiovascular disease and other chronic diseases would have higher clinical pregnancy and live birth rates during the course of infertility treatment with IVF.

Materials and methods

Study population

Couples presenting for infertility evaluation and treatment to the Massachusetts General Hospital (MGH) Fertility Center were invited to enroll in the Environment and Reproductive Health (EARTH) Study. Established in 2004, this cohort study investigates the effect of environmental and dietary factors on fertility and pregnancy outcomes; study design has been described in detail elsewhere [26]. Men (ages 18–55) and women (ages 18–45) presenting to the MGH Fertility Center planning to use their own gametes for infertility treatment are eligible to join the study. Approximately 65% of women and 45% of men who were eligible, participated in the study [26]. All study participants completed several study questionnaires which included demographics, medical, reproductive, and occupational history, and lifestyle, and underwent an anthropometric evaluation after providing written consent. Diet assessments were introduced in April 2007. Couples were encouraged but not required to join as a couple. This study included all participants who joined as a couple, the male partner completed a food frequency questionnaire (FFQ) and his female partner started at least one IVF cycle by April 2018. Among the 377 couples who joined the study during this period, 146 men did not complete diet assessments leaving 231 couples eligible for the current study. Baseline demographic and reproductive data of participants included in the study did not differ substantially from those excluded (Supplemental table 1). The study was approved by the institutional review board of both MGH and the Harvard T.H. Chan School of Public Health.

Dietary assessment and identification of dietary patterns

Diet was assessed using an extensively validated FFQ [27–35]. Participants were asked to report how often, on average, they consumed each of the 131 foods and beverages in the questionnaire during the previous year, with frequency choices ranging from “never or less than once per month” to “six or more times per day.” Individual food items were grouped into 42 pre-defined food groups based on those proposed by Hu and colleagues [36]. We used these 42 food groups to identify underlying dietary patterns in the study population using principal component analysis (PCA) with Varimax rotation in order to achieve a simpler and more interpretable structure. The number of factors retained was determined based on Eigenvalues, the scree plot, and interpretability of the resulting factors.

Clinical outcomes

The primary outcome of this study was the probability of live birth per initiated treatment cycle. Live birth was defined as the birth of a neonate at or after 24 weeks of gestation. Secondary outcomes were fertilization rate, the probability of implantation, and clinical pregnancy. Fertilization rate was defined as the number of two pronuclei embryos divided by the number of metaphase II oocytes and evaluated by mode of insemination (conventional insemination or intracytoplasmic sperm injection (ICSI)). Successful implantation was defined as an elevation of serum β-hCG level greater than 6 mIU/mL measured at approximately 2 weeks after embryo transfer. Clinical pregnancy was defined as the presence of an intrauterine gestational sac on ultrasound at 6 gestational weeks.

Semen parameters assessment

Secondary outcomes included conventional semen parameters (ejaculate volume, total sperm count, sperm concentration, total motility, progressive motility, and the percentage of sperm normal morphology). We used data from semen samples collected for diagnostic purposes as well as pre-processing data for samples collected for treatment purposes. Men provided semen specimens on-site via masturbation and completed abstinence time questionnaires. A 48-h abstinence time before sample collection was recommended. For this study, we included all semen sample data even if the abstinence time was not adhered to. Semen parameters were assessed based on the 2010 WHO manual guideline [37]. Semen samples were inspected after 30-min liquefaction on a 37 °C incubator. Ejaculate volume was calculated from sample weight, assuming a semen density of 1g/mL. Sperm concentration and motility were assessed by a computer-assisted semen analysis (CASA) system (10HTM-IVOS, Hamilton-Thorne Research, Beverly, MA) [38]. Motile spermatozoa were evaluated as total motility (progressive motility + non-progressive motility), progressive motility, non-progressive motility, and immotile sperm [37]. Total sperm count (million/ejaculate) referred to total number of spermatozoa in the entire ejaculate which was calculated by multiplying sperm concentration by ejaculated volume. Sperm morphology (% normal) was assessed on two slides per specimen (with a minimum of 200 cells assessed per slide) via a microscope (Nikon, Tokyo, Japan) with an oil-immersion ×100 magnification. Then, men were dichotomized as having normal or below normal morphology according to Strict Kruger scoring criteria [39].

Statistical analysis

Men were categorized into quartiles according to adherence to PCA-derived dietary patterns. We first examined differences in demographic, nutritional and reproductive characteristics, and semen parameters by quartile of adherence to dietary patterns using the Kruskal-Wallis test for continuous variables and Fisher exact test for categorical variables. Chi-square test was used for evaluating differences across categories of primary infertility diagnosis and initial stimulation protocol as the Fisher’s test did not run with these two variables. We investigated the relationship between two PCA-derived patterns and semen quality using linear mixed models with random intercepts, to account for multiple IVF cycles per couple, adjusting men’s age, total calorie intake, body mass index (BMI), race, smoking status, education level, and physical activity. The association of adherence to PCA-derived dietary patterns with IVF outcomes was examined using generalized linear mixed models with random intercepts, to account for multiple IVF cycles per couple, while adjusting for potential confounders. We used population marginal means to present results as probabilities and 95% confidence intervals (95% CI) adjusted for the covariates at their average levels for continuous variables and weighted average level of categorical variable in the model [40]. Tests for linear trend were conducted by modeling quartiles of adherence as a continuous variable. Confounding was evaluated using prior scientific knowledge and differences in baseline patient characteristics by dietary pattern adherence. The initial multivariable-adjusted model included terms for men’s age and total calorie intake. The second model included additional terms for men’s BMI, race, smoking status, education level, and physical activity, as well as women’s age and BMI, the couples’ primary infertility diagnosis, treatment protocol, and indicators for missing covariate data. The third model included all variables of the second model and terms for women’s adherence to the two dietary patterns, race and smoking status.

Sensitivity analyses were performed to evaluate the robustness of the findings. These included restricting all analyses to couples with complete female diet data, and to the first treatment cycle for each couple for IVF outcomes. We also conducted analyses stratified by primary infertility diagnosis (male factor vs. female and unexplained factor), IVF treatment history, and past pregnancy history. In addition, we conducted stratified analysis of previous infertility examination for the association between dietary pattern and semen quality. All analyses were performed using SAS university edition with VirtualBox version 6.1.6.

Results

A total of 231 couples who underwent 407 IVF cycles were included in the analysis. Median age of male partners at enrollment was 36.8 years (interquartile range (IQR): 33.4–40.0 years) and median BMI was 26.9 kg/m² (IQR: 24.1–29.1 kg/m²). Most men were white (89.2%) and had never smoked (66.2%) (Table 1). Women had a median baseline age and BMI of 35.0 years (IQR: 32.0–38.0 years) and 23.1 kg/m² (IQR: 21.0–25.7 kg/m²), respectively. Male factor infertility was the most common initial primary infertility diagnosis (36.8%). Two dietary patterns were identified using PCA (Fig. 1, Supplemental Table 2 and Supplemental figure 1). Dietary Pattern 1 was characterized by greater intakes of processed meats, unprocessed red meats, high-fat dairy, beer, French fries, cream soups, refined grains, pizza, snacks, and sweets. Dietary Pattern 2 was characterized by greater intakes of fruit, vegetables, legumes, soy foods, whole grains, nuts, and nut butters. Intakes of organ meats, fish, chicken, eggs, margarine, low-fat dairy, liquor, wine, tea, coffee, fruit juice, cold breakfast cereal, salad dressings, artificial sweeteners, and water did not have high loading scores on either of the identified dietary patterns (Fig. 1).

Table 1.

Baseline demographic, nutritional, and reproductive characteristics of study participants, overall and in lowest and highest quartiles of adherence to Dietary Patterns 1 and 2*

n	Total	Dietary Pattern 1			Dietary Pattern 2
		Q1	Q4	P**	Q1	Q4	P**
	231	57	58		57	58
Demographic characteristics
Age (y)	36.8 (33.4–40.0)	37.9 (34.1–40.0)	36.0 (33.6–40.5)	0.67	35.5 (33.0–39.2)	38.0 (34.0–41.9)	0.08
BMI (kg/m2)	26.9 (24.1–29.1)	26.6 (23.7–29.3)	27.5 (25.4–29.4)	0.01	27.4 (25.6–30.0)	27.1 (25.1–28.7)	0.04
Race, white	206 (89.2)	47 (82.5)	55 (94.8)	0.21	54 (94.7)	51 (87.9)	0.43
Smoking status, never smoker	153 (66.2)	44 (77.2)	34 (58.6)	0.19	36 (63.2)	40 (69.0)	0.32
Education, college or higher	183 (84.7)	48 (90.6)	46 (83.6)	0.51	37 (71.2)	47 (85.5)	0.02
Moderate-to-vigorous physical activity (min/week)	165 (60–390)	150 (60–330)	168 (47–390)	0.28	150 (30–360)	205 (72–402)	0.29
Calories (kcal/day)	1934 (1586–2384)	1530 (1225–1843)	2585 (2164–2934)	<0.001	1664 (1237–2077)	2568 (2096–2900)	<0.001
Reproductive characteristics
History of varicocele	19 (8.2)	4 (7.0)	4 (6.9)	0.95	5 (8.8)	4 (6.9)	0.57
Female partner characteristics
Age (y)	35.0 (32.0–38.0)	35.0 (32.0–39.0)	35.0 (32.0–38.0)	0.78	35.0 (32.0–38.0)	36.0 (33.0–39.0)	0.36
BMI (kg/m2)	23.1 (21.0–25.7)	22.0 (20.5–25.5)	23.9 (21.6–27.9)	0.11	23.3 (21.7–27.9)	23.1 (20.7–25.7)	0.41
Race, white	194 (84.4)	40 (70.2)	49 (84.5)	0.006	48 (84.2)	48 (82.8)	0.97
Smoking status, never smoker	166 (72.2)	41 (71.9)	43 (74.1)	0.17	39 (68.4)	44 (75.9)	0.18
Dietary Pattern 1	−0.38 (−0.87 to −0.01)	−0.88 (−1.07 to −0.34)	−0.08 (−0.40 to 0.54)	<.001	−0.36 (−0.85 to 0.07)	−0.31 (−0.86 to 0.10)	0.8
Dietary Pattern 2	−0.04 (−0.50 to 0.56)	0.10 (−0.36 to 0.75)	−0.05 (−0.42 to 0.58)	0.47	−0.44 (−0.99 to −0.15)	0.57 (−0.05 to 1.16)	<0.001
Couple-level characteristics
History of past pregnancy	86 (37.4)	22 (38.6)	19 (32.8)	0.64	24 (42.1)	21 (36.2)	0.2
Previous infertility examination	188 (83.6)	45 (81.8)	44 (78.6)	0.56	46 (82.1)	47 (83.9)	0.24
Previous infertility treatment	107 (51.7)	20 (40.0)	31 (60.8)	0.13	25 (48.1)	31 (58.5)	0.39
Primary infertility diagnosis
Male factor	85 (36.8)	21 (36.8)	23 (39.7)	0.93	20 (35.1)	17 (29.3)	0.26
Female factors
Diminished ovarian reserve	24 (10.4)	8 (14.0)	6 (10.3)		7 (12.3)	8 (13.8)
Endometriosis	14 (6.1)	4 (7.0)	2 (3.5)		5 (8.8)	4 (6.9)
Ovulatory	22 (9.5)	5 (8.8)	2 (3.5)		7 (12.3)	4 (6.9)
Tubal disease	17 (7.4)	5 (8.8)	4 (6.9)		5 (8.8)	3 (5.2)
Uterine	3 (1.3)	1 (1.8)	0 (0)		1 (1.8)	1 (1.7)
Other disease	4 (1.7)	2 (3.5)	1 (1.7)		0 (0)	4 (6.9)
Unexplained	62 (26.8)	11 (19.3)	20 (34.5)		12 (21.1)	17 (29.3)
Initial stimulation protocol
Antagonist	35 (15.2)	14 (24.6)	9 (15.5)	0.12	9 (15.8)	9 (15.5)	0.77
Flare	22 (9.5)	8 (14.0)	4 (6.9)		4 (7.0)	5 (8.6)
Luteal phase agonist	152 (65.8)	33 (57.9)	39 (67.2)		39 (68.4)	38 (65.5)
Egg donor or cryo-cycle	22 (9.5)	2 (3.5)	6 (10.3)		5 (8.8)	6 (10.3)
Initial mode of insemination; ICSI	122 (59.2)	33 (61.1)	28 (53.9)	0.85	28 (53.9)	33 (64.7)	0.71

^*Data are presented as median (interquartile range) for continuous variables or n (%) for categorical variables

^**From the Kruskal-Wallis test for continuous variables and Fisher exact test for categorical variables except for primary infertility diagnosis and in vitro fertilization treatment protocol where the Chi-square test was used

BMI body mass index, Q quartile

Fig. 1.

Principal component analysis plot with two factor loadings for food groups among 231 couples undergoing infertility treatment. Dark gray: Food groups which had factor 1 loading greater than 0.3. Light gray: Food groups which had factor 2 loading greater than 0.3. White: Food groups which had both factor 1 and factor 2 loading less than 0.3

Men’s adherence to Dietary Pattern 1 was associated with higher BMI. Men’s adherence to Dietary Pattern 2 was inversely related to BMI and positively related to educational attainment. Adherence to both patterns was positively related to higher total calorie intake. Moreover, adherence to Dietary Pattern 1 (r_Pearson=0.44) and Dietary Pattern 2 (r_Pearson=0.54) was positively correlated within couples (Table 1). Supplemental Table 3 shows the distribution of semen quality parameters among male participants. Approximately 40–60% of participants demonstrated asthenospermia according to the WHO reference limits [37].

Adherence to Dietary Pattern 1 was positively related to sperm concentration. A 1-unit increase in this pattern was associated with a 13.33 (0.71–25.96) million/mL higher sperm concentration (Table 2). This association was stronger among men who had had an infertility examination prior to joining the study (β=17.93 (3.55 to 32.30) million/mL). The association was in the opposite direction among men who had not had an infertility examination before joining the study, although sample size was limited in this group (Table 2).

Table 2.

Association between men’s adherence to Dietary Patterns 1 and 2 and semen parameters

	Estimate (95% CI)
	Pattern 1	Pattern 2
Total	N=231 men
Volume	−0.21 (−0.48 to 0.04)	−0.16 (−0.38 to 0.05)
Total sperm count	1.54 (−24.70 to 27.77)	−16.53 (−38.41 to 5.35)
Sperm concentration	13.33 (0.71 to 25.96) *	1.07 (−9.42 to 11.57)
Total motility	0.17 (−4.36 to 4.70)	−1.97 (−5.73 to 1.80)
Progressive motility	0.64 (−2.21 to 3.48)	−0.76 (−3.12 to 1.60)
Morphology	0.03 (−0.58 to 0.65)	0.11 (−0.41 to 0.62)
Past examination	N=188 men
Volume	−0.20 (−0.49 to 0.08)	−0.11 (−0.36 to 0.12)
Total sperm count	11.23 (−17.55 to 40.01)	−11.87 (−36.47 to 12.72)
Concentration	17.93 (3.55 to 32.30) *	2.98 (−9.26 to 15.22)
Total motility	1.12 (−3.67 to 5.92)	−2.24 (−6.31 to 1.83)
Progressive motility	1.15 (−1.97 to 4.27)	−1.03 (−3.68 to 1.62)
Morphology	0.40 (−0.30 to 1.09)	0.16 (−0.43 to 0.75)
Never examination	N=37 men
Volume	−0.20 (−0.84 to 0.43)	−0.36 (−0.86 to 0.14)
Total sperm count	−52.28 (−108.91 to 4.34)	−43.63 (−88.17 to 0.92)
Concentration	−15.50 (−37.52 to 6.52)	−13.02 (−30.24 to 4.21)
Total motility	−6.03 (−17.84 to 5.78)	−3.14 (−12.37 to 6.09)
Progressive motility	−3.74 (−9.82 to 2.34)	−1.82 (−6.59 to 2.94)
Morphology	−2.11 (−3.07 to −1.14) *	−0.14 (−0.95 to 0.67)

*P<0.05

There was no association between men’s adherence to two data-derived dietary patterns and fertilization rate in total cycles, stratified analyses for IVF cycles using conventional insemination and ICSI cycles (Fig. 2). There was also no discernible association of men’s adherence to either dietary pattern with probabilities of implantation, clinical pregnancy or live birth in multivariable-adjusted models without (Fig. 3) or with adjustment for women’s adherence to the same dietary patterns (Fig. 4).

Fig. 2.

Men’s adherence to Dietary Patterns 1 and 2 in relation to fertilization rate. Adjusted for men’s age, total calorie intake, BMI, race, smoking status, education level, and moderate-to-vigorous physical activity; women’s age and BMI; primary infertility and treatment protocol. Q, quartile

Fig. 3.

Men’s adherence to Dietary Patterns 1 and 2 in relation to clinical outcomes of infertility treatment with IVF (N = 231 couples, 407 cycles). Adjusted for men’s age, total calorie intake, BMI, race, smoking status, education level, and moderate-to-vigorous physical activity; women’s age and BMI; primary infertility diagnosis and treatment protocol

Fig. 4.

Men’s adherence to Dietary Patterns 1 and 2 in relation to clinical outcomes of infertility treatment with IVF after co-adjustment for women’s adherence to the same diet patterns (N = 213 couples, 367 cycles). Adjusted for men’s age, total calorie intake, BMI, race, smoking status, education level, and moderate-to-vigorous physical activity; women’s age, BMI, and adherence to Dietary Patterns 1 and 2; primary infertility diagnosis and treatment protocol. Q, quartile

In sensitivity analyses, results were consistent with the primary findings when analyses were restricted to couples with complete female diet data (Supplemental Figure 2) and to the first treatment cycle for each couple (Supplemental Figure 3). Similarly, analyses stratified by a primary infertility diagnosis (Supplemental Figure 4), past IVF treatment history (Supplemental Figure 5), or past pregnancy history (Supplemental Figure 6) showed no association between the dietary patterns and IVF outcomes either.

Discussion

We investigated the association of men’s adherence to two data-derived dietary patterns identified using a data-driven approach with intake data collected using a validated 131-item FFQ with semen quality and outcomes of infertility treatment with IVF. Despite sperm concentration being associated with one of these patterns, we found no evidence that men’s adherence to these dietary patterns had any meaningful impact on the outcome of infertility treatment with IVF. To our knowledge, this is the first study to date examining the association between men’s dietary patterns and couples’ IVF outcomes. Hence, it is important that this question is revisited in additional studies.

The finding that higher adherence to Dietary Pattern 1 was associated with higher sperm concentration stands in contrast with several observational studies suggesting that male adherence to dietary patterns linked to higher chronic disease risk could also have a negative impact on semen quality [17, 41, 42], whereas the opposite appears to be the case for adherence to diet patterns previously related to lower risk of chronic comorbidities [15–18, 41–43]. A possible reason for the observed relation and the inconsistency with previous literature may be reverse causation. As couples undergo diagnostic testing for infertility, results from these tests may influence their behavior. In this study, knowledge of results of semen analyses, and possibly other diagnostic information, may change the way in which men eat or engage in other behaviors with the goal of optimizing their fertility and this change in behavior would be expected to be differential between men with poor and men with favorable semen analyses. For example, if men who know that their semen analysis was above the WHO reference limit would not change their diet, whereas if they were informed that they had abnormal results in their semen analysis, we would expect them to change their behavior. So, any association between diet and semen quality would be more reflective of this differential change in behavior than of any true influence that diet may have on semen quality. This situation is analogous to previously described paradoxical associations in cross-sectional studies of diet with outcomes that are known to participants, such as the cross-sectional association between intake of diet sodas and BMI [44, 45]. Similarly, although there was no association between adherence to dietary patterns and sperm morphology in the entire study sample, higher adherence to Dietary Pattern 1 among men who had not been previously evaluated for infertility was inversely related to sperm morphology. This relationship is reminiscent of a previous report of processed meat intake—which has a high factor loading in Diet Pattern 1—with sperm morphology [46]. Our findings, and in particular the divergent pattern after stratified stratification by whether or not men had undergone diagnostic procedures prior to joining the study, highlight a potential peril of conducting research on behavioral determinants of semen quality in clinical populations.

Data on men’s diet and a couple’s fertility is scant, yet the literature on diet and semen quality has been interpreted as implying positive effects on a couple’s fertility. However, the evidence base to support this inference is weak. To start, semen quality is known to be a weak predictor of probability of conception both among couples attempting conception on their own as well as in couples trying to conceive with medical assistance [21, 22]. Moreover, in previous reports from this cohort, we have found that specific dietary factors related to semen quality, such as intake of processed meat, dairy, and soy foods [46–48], are unrelated to infertility treatment outcomes [23, 49, 50]. Conversely, we have found that dietary factors that have been consistently found to be unrelated to semen quality, such as intakes of alcohol and caffeine [5, 6], were, paradoxically, related to live birth rates during the course of IVF [51]. This does not mean that there are no specific nutritional factors that can impact couple fertility by improving semen quality. For example, we and others have documented positive associations between intake of fish, fish oil, or marine fatty acids and better semen quality and other markers of testicular function [52–55] and, independently, with greater fecundability [24]. Nevertheless, the discrepancies highlight the fact that improvements in semen quality do not necessarily imply improvements in IVF outcomes for couples undergoing treatment and that, therefore, the identification of male partner characteristics that impact a couple’s fertility requires the direct evaluation of IVF outcomes as the study outcome rather than relying on semen quality as a proxy outcome as has been traditional in andrology and reproductive medicine.

It is important to mention that male effects on reproduction not mediated through traditional semen quality parameters are not purely theoretical. Although the evidence base is still emerging, there is literature showing that the sperm genome and epigenome may play an important role in fertility [56, 57] and are subject to environmental modification [58, 59]. Also, while not directly examining fertility, emerging experimental data suggests that paternal environmental exposures could exert effects on pregnancy and offspring health through the sperm epigenome [60–62]. For example, folate-deficient, high-fat, or low-protein diets in males, but not females, negatively impact offspring’s metabolism through epigenetic mechanisms [60, 62]. These findings suggest that men’s diet can impact reproduction through additional mechanisms.

As discussed above, a possible interpretation of our findings is that, despite previous research relating dietary patterns to semen quality, men’s diet has no impact on a couple’s outcomes through IVF. The interpretation of a lack of effect is also in line with findings from two recent randomized trials which found no effect on live birth rate of supplementing men in infertile couples with custom micronutrient formulations [63, 64]. A related explanation for the lack of association is the fact that we studied this question among couples undergoing infertility treatment with IVF. ICSI has been the most widely utilized assisted reproductive technology over the recent years [65] and IVF/ICSI is a powerful intervention for male factor infertility and impaired semen quality (primarily on concentration and motility) that may completely offset comparatively smaller impacts of men’s diet on semen quality. If this is the case, reexamining this question among couples attempting conception without medical intervention is essential. It is also important to consider alternate interpretations. One possibility is that the data-derived dietary patterns did not capture food groups that could impact fertility. For example, fish intake was not part of either of the dietary patterns identified in this analysis, but has been previously linked to a couple’s fertility [24]. Clearly, additional work aimed at identifying male partner factors, including modifiable lifestyle factors, that could impact a couple’s fertility is necessary.

It is also important to interpret the results in light of the study’s limitations and strengths. First, we only assessed diet at baseline FFQ and hence, we were unable to document any changes in diet over time as couples underwent treatment. This could result in a dilution of the actual relationship, particularly for couples who take longer to conceive, either due to longer intervals from enrollment to first treatment cycle or more failed treatment cycles, or are never able to conceive. However, sensitivity analyses restricted to each couple’s first cycle were consistent with the primary findings. Also, this study was conducted at an academic fertility center and more than 80% of the men had already been examined before the study baseline. Therefore, it is important to consider the extent to which associations with semen quality reflect an effect of diet on spermatogenesis or the effect of being aware of one’s semen quality on subsequent dietary behaviors. Second, although we used an extensively validated diet questionnaire, measurement error is still unavoidable in questionnaire-based studies of diet. Given that diet assessment preceded treatment outcome assessment the expected effect would be an attenuation of effect estimates towards the null. Third, as mentioned above, investigating the role of men’s diet on fertility among couples undergoing IVF could completely obscure a true but modest effect of men’s diet on a couple’s fertility and therefore results cannot be generalized to couples attempting conception naturally. Last, the study population was primarily white. While this certainly limits generalizability to other racial groups, the race distribution in our study closely mirrors that of couples undergoing infertility treatment nationwide and therefore results can still inform practice.

Strengths of the study include the recruitment of both male and female partners, which is, unfortunately, still not the norm in fertility studies and allowed us to inquire about the role men’s diet may have on fertility. Moreover, it allowed us to take into consideration within-couple correlations and discordance in diet and other relevant behavioral and demographic factors that could influence the outcome of infertility treatment. For example, diet is a complex and dynamic behavior that, while correlated within couples, is not perfectly matched within couples as shown by the modest within-couple correlation for both dietary patterns. While counterintuitive at first sight, this is not unexpected. In couples where both partners work outside of the home, which is the most common arrangement for couples in our study, it is not unusual for couples to have a substantial number of eating occasions completely separate from each other, most obviously but not exclusively, lunch and any daytime snacks. The study’s prospective design with complete follow-up of clinically relevant outcomes including live birth rate, the extensive collection of key lifestyle factors in both partners, and the use of validated instruments also adds to the study’s strengths.

In brief, results from this study suggest that men’s adherence to two data-derived dietary patterns was not related to outcomes of infertility treatment with IVF in spite of an association between one of these patterns and sperm concentration. These results differ with the expanding literature suggesting that adherence to healthy dietary patterns is related to better semen quality. Given the scarcity of data on this topic, it is important that additional studies examine the role of men’s diet on fertility both in the setting of infertility treatment and among couples attempting conception without medical assistance.

Author contribution

Jorge E. Chavarro was involved in study concept and design and critical revision for important intellectual content of the manuscript and had a primary responsibility for final content. Makiko Mitsunami drafted the manuscript and analyzed data; Albert Salas-Huetos reviewed the statistical analysis; Lidia Mínguez-Alarcón contributed to the statistical analysis; Jennifer B. Ford and Irene Souter were involved in acquisition of the data; Makiko Mitsunami, Albert Salas-Huetos, Lidia Mínguez-Alarcón, Jill A. Attaman, Jennifer B. Ford, Martin Kathrins, Irene Souter, and Jorge E. Chavarro interpreted the data; all authors were involved in the critical revision of the manuscript and approved the final manuscript.

Funding

Supported by grants ES009718, ES022955, ES026648, and ES000002 from the National Institute of Environmental Health Sciences, and P30DK46200 from the National Institute of Diabetes and Digestive and Kidney Diseases.

Data availability

Data are available upon request after approval of data sharing and use agreements between institutions.

Code availability

All analyses were performed using SAS university edition with VirtualBox version 6.1.6.

Code is available upon request after approval of data sharing and use agreements between institutions.

Declarations

Ethics approval

The EARTH study was approved by the institutional review board of both MGH and the Harvard T.H. Chan School of Public Health.

Consent to participate

All participants to the EARTH study completed written consent forms.

Consent for publication

The written consent forms included the permission for publication.

Competing interests

Not aplicable for Makiko Mitsunami, Albert Salas-Huetos, Lidia Mínguez-Alarcón, Jill A. Attaman, Jennifer B. Ford, Martin Kathrins, Irene Souter.

Jorge E. Chavarro received grants from the National Institutes of Health.

Disclaimer

The funders had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript.