Soy-based infant formula feeding and menstrual pain in a cohort of women aged 23–35 years

Abstract

STUDY QUESTION

Is soy formula feeding during infancy associated with menstrual pain in reproductive-age women?

SUMMARY ANSWER

Our data suggest that soy formula feeding during infancy is associated with several indicators of severe menstrual pain in reproductive-age women.

WHAT IS KNOWN ALREADY

A prior study observed greater severity of menstrual pain in young women who as infants participated in feeding studies and were assigned to soy-based formula feeding.

STUDY DESIGN, SIZE, DURATION

We used data from the Study of Environment, Lifestyle & Fibroids (SELF), a cohort of 1696 African-American women ages 23–35 years at enrollment.

PARTICIPANTS/MATERIALS, SETTINGS, METHODS

Data on infant soy formula feeding was ascertained by self-administered questionnaire for 1553 participants, with 89% of participants receiving assistance from their mothers. Information on menstrual pain indicators was collected by web- and telephone-interview. We estimated the relative risk (RR) and 95% confidence interval (CI) using log-binomial regression, or log-multinomial regression, adjusting for participant age and maternal education.

MAIN RESULTS AND THE ROLE OF CHANCE

Women ever fed soy formula as infants were more likely than unexposed women to report ever use of hormonal contraception for menstrual pain (RR 1.4, CI: 1.1–1.9) and moderate/severe menstrual discomfort/pain with ‘most periods’, but not ‘every period’, during early adulthood (ages 18–22 when not using hormonal contraception) (RR 1.5, CI: 1.1–2.0).

LIMITATIONS, REASONS FOR CAUTION

We relied on retrospective recall to ascertain infant exposure to soy formula feeding and data on menstrual pain indicators.

WIDER IMPLICATIONS OF THE FINDINGS

Our observations add to the growing body of literature from animal and human studies on the reproductive health consequences of early-life exposure to soy formula.

STUDY FUNDING/COMPETING INTEREST(S)

This research was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences and, in part, by funds allocated for health research by the American Recovery and Reinvestment Act. This research was also supported by grant K99NR017191 (KU). None of the authors has a conflict of interest.

TRIAL REGISTRATION NUMBER

Not applicable.

Introduction

Menstrual pain (dysmenorrhea) is characterized by menstrual bleeding-related abdominal pain, cramping or backache (Harlow and Ephross, 1995). It is considered the most common menstrual complaint, with population-based studies reporting a prevalence of 60% among reproductive-age women (Zondervan et al., 2001; Burnett et al., 2005). Menstrual pain can have a substantial impact on quality of life, affecting school performance, work productivity and relationships [as reviewed by (Iacovides et al., 2015)].

Menstrual pain is believed to be triggered by the overproduction of endometrial prostaglandins in response to the withdrawal of progesterone at the onset of menses, leading to increased uterine contractility, hypoxia and pain (Dawood, 1987). The development of the reproductive system components involved in menstrual pain pathophysiology, the hypothalamus–pituitary–ovarian (HPO)-axis and the uterus, begins in utero, with the activation of the HPO-axis and uterine tissue differentiation continuing after birth, during infancy (Valdes-Dapena, 1973; Winter et al., 1975, 1976; Kuiri-Hanninen et al., 2011, 2013). Given that infancy is a critical window of development and that steroid hormone signaling is central to the development of the reproductive system [as reviewed by (Ho et al., 2017)], it is plausible that disruption from an exogenous estrogenic exposure in infancy may result in persistent reproductive system changes, including menstrual pain in adulthood.

One such infant exogenous estrogenic exposure is soy-based formula. Soy formula contains phytoestrogens, predominantly the isoflavones genistein and daidzein, that are structurally similar to estradiol (Woods, 2003). Although these phytoestrogens are less potent than estradiol with respect to estrogen receptor binding (as reviewed in (Jefferson et al., 2012) and (McCarver et al., 2011)), they appear to have a higher binding affinity than several man-made estrogenic endocrine disrupting chemicals, including Bisphenol A (Kuiper et al., 1997, 1998; Mueller et al., 2003). Exposure to these phytoestrogens in soy formula can be high, particularly when soy formula is the exclusive source of nutrition (McCarver et al., 2011); the concentration of genistein in the urine of soy formula-fed infants was shown to be 500 times that of cow’s milk formula-fed infants (Cao et al., 2009). Vaginal tissue alterations indicative of an estrogenic response have been observed in infant girls exclusively fed soy formula. (Adgent et al., 2018).

A follow-up study of primarily white, young US adults who as infants were assigned to soy-based or cow’s milk-based formulas in infant feeding studies reported an increased risk of greater menstrual pain severity in adulthood with infant soy formula feeding (Strom et al., 2001). However, the authors cautioned that this association may be due to chance given the many outcomes evaluated in that study (Strom et al., 2001). Since the publication of that study, mounting data from animal studies support the disruptive effects of early postnatal genistein exposure on reproductive system development that persist into adulthood (Faber and Hughes, 1993; Nagao et al., 2001; Jefferson et al., 2005, 2011, 2012; Bateman and Patisaul, 2008; Tang et al., 2008; Cimafranca et al., 2010; Greathouse et al., 2012). Hence, the purpose of the present study was to further evaluate the association between soy formula feeding and menstrual pain in a cohort of young, African-American women.

Materials and Methods

Study population

For the present analyses, we used enrollment data (years 2010–2012) from the National Institute of Environmental Health Sciences (NIEHS) Study of Environment, Lifestyle & Fibroids (SELF). As previously described, SELF is an ongoing prospective cohort study being conducted among 1696 African-American women residing in the Detroit, Michigan area, ages 23–35 years at enrollment (Baird et al., 2015). SELF was designed to identify risk factors for fibroid incidence and growth, and as such, the primary study eligibility criteria included having an intact uterus and no prior diagnosis of uterine fibroids.

Ethical approval

Both the institutional review boards at NIEHS and at the collaborating healthcare system for this project, the Henry Ford Health System, approved the conduct of SELF and each participant provided informed consent before study enrollment.

Exposure ascertainment

At enrollment, study participants were asked to complete and return by postal mail the Early-Life Questionnaire. This questionnaire was designed so that the study participant could interview her mother to obtain information on a range of early-life exposures. If her mother was not available, the participant was instructed to obtain assistance from other relatives or her mother’s friends. The questionnaire was completed and returned for 94% (n = 1598) of SELF participants.

The Early-Life Questionnaire included questions on soy formula feeding during infancy, including whether the participant was ever fed soy formula (yes, no), the duration of soy formula feeding (<1 month, 1–3 months, 4–6 months, >6 months) and whether the feeding was initiated within the first two months after birth (yes, no). Ever feeding of soy formula was reported as ‘don’t know’ and missing for 37 and 8 participants, respectively. This resulted in data on soy formula feeding being available for 1553 SELF participants (198 ever fed soy formula, 1355 never fed soy formula), among whom 89% had the assistance of mothers when completing the questionnaire.

Outcome ascertainment

Medication and hormonal contraceptive use for menstrual pain

Participants also completed a computer-assisted telephone-interview (CATI) at study enrollment; they were asked whether they had ever taken any prescription or over-the-counter medication to treat or prevent menstrual cramps, pelvic pain, or discomfort (yes, no), the age at which medication was first used, and whether the participant was still taking medication to treat or prevent menstrual cramps, pelvic pain, or discomfort at least for some periods (yes, no). The use of any prescription or over-the-counter medication could include a wide array of medications, including hormonal contraception; hence, we hereinafter refer to this outcome as ever (or current) use of medication for menstrual pain. Given that dysmenorrhea can appear within months of menarche (Dawood, 1990) and occurs with ovulatory cycles (which are established within 5 years of menarche for most women) (Metcalf et al., 1983) we created a variable considering timing of first use of medication in relation to menarche (no use, first use <5 years after menarche, first use ≥5 years after menarche).

Participants were also asked about ever use of birth control pills, hormonal implant such as Norplant or Implanon, hormonal patch, hormonal vaginal ring, hormone shots like Depo-Provera, and intrauterine device (IUD) (yes, no). For each of these methods, participants were asked age at first use and current use (yes, no). They were also asked reasons for use including ever use for menstrual pain (yes, no). We used these data to create the following summary variables: ever use of a hormonal contraception for menstrual pain (yes, no), and, if hormonal contraception was ever used for menstrual pain, timing of first use (no use, first use <5 after menarche, first use ≥5 years after menarche), and current use (yes, no).

Frequency of moderate/severe menstrual pain, ages 18–22 years

The CATI also included a question about the frequency of moderate or severe menstrual discomfort or pain when the participant was aged 18–22 years (rarely/never, occasional periods, most periods, every period); participants were instructed to report for only those times when they were not using a hormonal contraceptive. In the cohort of 1553 participants with data on soy formula feeding, 275 participants were not asked about the frequency of moderate/severe menstrual pain when ages 18–22 years because they were using hormonal contraception the entire time when ages 18–22 (n = 251), missing information on hormonal contraceptive use when ages 18–22 years (n = 3), or reported no menstrual periods when ages 18–22 years (n = 21). One participant reported ‘don’t know’ to the question. Data on this outcome were available for 1277 (82%) participants.

Data collection on participant and maternal characteristics

Characteristics of the participant as an infant and that of her mother when she was pregnant with the participant were primarily collected using the Early-Life Questionnaire. Additional data on the characteristics of the participant were collected during the enrollment study questionnaire and interview activities. Measurement of participant height and weight were conducted at the clinic visit (Baird et al., 2015).

Statistical analyses

We used log-binomial regression to estimate the relative risk (RR) and 95% confidence intervals (CI) for the association between soy formula feeding (ever, never) and binary menstrual pain outcomes. For menstrual pain outcomes that had more than two categories, we used log-multinomial regression (Leigh Blizzard, 12/19/2014, Stata command documentation; Blizzard and Hosmer, 2007). Similar to log-binomial regression, log-multinomial regression directly estimates the RR and 95% CI, with one outcome category not being fitted in the modeling process (Blizzard and Hosmer, 2007); we always chose the outcome category representing never use of medication for menstrual pain or lowest frequency of moderate/severe menstrual pain as the non-fitted outcome category.

All the analyses were adjusted for participant age (continuous) and the highest educational level of mother or primary caregiver when the participant was age 10 years (≤high school or GED; some college or associate/technical degree; bachelor’s or higher degree). The analyses were conducted using Stata version 14.2 (StataCorp, College Station, TX) with α = 0.05 as the level of significance.

We conducted three sensitivity analyses. First, we repeated the analyses after employing a stricter definition of soy formula feeding that required a feeding duration of at least four months and use beginning within two months of birth (versus never fed soy formula). This stricter definition aimed to capture the period when soy formula would be the primary source of nutrition; the American Academy of Pediatrics in 1980 (around the time the SELF cohort was born) recommended that solid foods not be introduced until 4–6 months of age (AAP, 1980). Second, we restricted the study population to those whose mothers helped with the completion of the Early-Life Questionnaire. Third, we repeated the analyses additionally adjusting for household income during the participant’s childhood and ever being breastfed as an infant.

We conducted an exploratory analysis evaluating infant soy formula feeding in relation to impact of menstrual pain on quality of life in the 12 months before study interview (described in the Supplemental Materials). These analyses were restricted to the subset of participants (n = 422) who had a menstrual period in the year before the study interview, were not using hormonal birth control or medication that alters menses in the year before the last menstrual period or currently, and were not currently taking any medication to treat or prevent menstrual pain.

Results

Participant characteristics for our study sample of 1553 study participants by soy formula feeding exposure have been previously published (Upson et al., 2016) and are described in Supplementary Tables SI and SII.

Medication and hormonal contraceptive use for menstrual pain

We did not observe an association between soy formula feeding and ever or current medication use for menstrual pain (Table I). However, when we considered the timing of first medication use after menarche, women fed soy formula as infants were 20% more likely than unexposed women to use medication for menstrual pain within five years of menarche (RR 1.2, 95% CI: 1.0–1.3). We observed that soy formula feeding was associated with a 40% increased risk of ever use of hormonal contraception for menstrual pain (RR 1.4, 95% CI: 1.1–1.9); the magnitude of the association was stronger if hormonal contraception (if ever used for menstrual pain) was first used within 5 years of menarche (RR 1.7, 95% CI: 1.0–2.7). Soy formula feeding was also associated with current use of hormonal contraception (if ever used for menstrual pain), although the confidence interval included the null (RR 1.6, 95% CI: 0.9–2.7).

Table I.

Relative risk (RR) and 95% CI for the association between soy formula feeding and medication and hormonal contraceptive use for menstrual pain, Study of Environment, Lifestyle & Fibroids (SELF), 2010–2012.

	Soy formula feeding
Characteristics	Ever	Never	Age-adjusted	MV-adjusted
	n = 198	n = 1355
	n (%)	n (%)	RR (95% CI)	RR (95% CI)a
Ever use of medication for menstrual painb
No	55 (28)	400 (30)
Yes	143 (72)	955 (70)	1.0 (0.9–1.1)	1.0 (0.9–1.1)
First use <5 years after menarche	108 (55)	616 (46)	1.2 (1.0–1.3)	1.2 (1.0–1.3)
First use ≥5 years after menarche	35 (18)	330 (25)	0.7 (0.5–1.0)	0.7 (0.5–1.0)
Missing	0	9
Current use of medication for menstrual painc
No	103 (52)	752 (56)
Yes	95 (48)	603 (45)	1.1 (0.9–1.3)	1.1 (0.9–1.3)
Ever use of hormonal contraception for menstrual paind
No	156 (79)	1158 (85)
Yes	42 (21)	197 (15)	1.4 (1.1–1.9)	1.4 (1.1–1.9)
First use <5 years after menarchee	19 (10)	72 (5)	1.7 (1.1–2.8)	1.7 (1.0–2.7)
First use ≥5 years after menarchee	23 (12)	124 (9)	1.3 (0.8–1.9)	1.3 (0.8–1.9)
Missing	0	1
Current use of hormonal contraception for menstrual paind,e
No	183 (92)	1294 (96)
Yes	15 (8)	61 (5)	1.6 (0.9–2.8)	1.6 (0.9–2.7)

^aAdjusted for participant age and maternal education.

^bMedication use includes any prescription or over-the-counter medication to treat or prevent menstrual cramps, pelvic pain or discomfort.

^cDefined as current use of medication to treat or prevent menstrual cramps, pelvic pain or discomfort at least for some periods.

^dHormonal contraceptive methods include birth control pills, hormonal implant such as Norplant or Implanon, hormonal patch, hormonal vaginal ring, hormone shots like Depo-Provera, and intrauterine device (IUD). Participants reported menstrual pain as reason for use of the specific hormonal contraceptive method; the reason for use specifically at first use or current use was not collected.

^eIf hormonal contraception was ever used for menstrual pain.

Frequency of moderate/severe menstrual pain, ages 18–22 years

Among participants who reported having menstrual periods and not using hormonal birth control all the time when ages 18–22 years (n = 1277), women fed soy formula as infants were 50% more likely than unexposed women to report experiencing moderate or severe menstrual pain with ‘most periods’ (RR 1.5, 95% CI: 1.1–2.0), although no association was observed with ‘every period’ (Table II).

Table II.

Relative risk (RR) and 95% CI for the association between soy formula feeding and frequency of moderate/severe menstrual pain when ages 18–22 years^a, Study of Environment, Lifestyle & Fibroids (SELF), 2010–2012 (N = 1277).

	Soy formula feeding
Characteristics	Ever	Never	Age-adjusted	MV-adjusted
	n = 163	n = 1114
	n (%)	n (%)	RR (95% CI)	RR (95% CI)b
Frequency of moderate/severe menstrual pain
Rarely or never	45 (28)	380 (34)
Occasional periods	25 (15)	226 (20)	0.8 (0.5–1.1)	0.8 (0.5–1.1)
Most periods	44 (27)	197 (18)	1.5 (1.2–2.0)	1.5 (1.1–2.0)
Every period	49 (30)	311 (28)	1.1 (0.8–1.4)	1.1 (0.8–1.4)

^aAmong 1277 participants who reported having menstrual periods during times which they did not use hormonal contraception when ages 18–22 years.

^bAdjusted for participant age and maternal education.

Sensitivity analyses

In our sensitivity analyses employing a stricter definition of soy formula feeding exposure (excluding 111 participants fed soy formula <4 months or first fed soy formula after two months of birth) and restricting the study population to those whose mother assisted with the completion of the Early-Life Questionnaire, the magnitude of the association was stronger for the outcomes of ever and current use of hormonal contraception for menstrual pain and frequency of moderate/severe menstrual pain when ages 18–22 years (Supplementary Tables SIII and SIV). When we repeated the main analyses, additionally adjusting for household income during the participant’s childhood and ever being breastfed as an infant, the estimates of the associations remained essentially the same (Supplementary Table SV).

Exploratory analyses

Among the subset of participants experiencing menses and not using medication that alter menses or treat or prevent menstrual pain (n = 422), soy formula feeding was associated with ‘some/a lot’ of interference with activities of daily living due to menstrual cramps/discomfort in the 12 months before study interview (RR 1.8, CI: 1.1–3.1) (Supplementary Table SVI). Our data also suggested an association between soy formula feeding and any days missed work or stopped home activities in the past year due to menstrual cramps/discomfort (RR 1.5, CI: 0.9–2.6).

Discussion

In the entire cohort, we observed that soy formula feeding in infancy was associated with ever use of hormonal contraception for menstrual pain, particularly hormonal contraceptive use (if ever used for menstrual pain) started within 5 years of menarche. However, we did not observe an association with ever or current use of any medication for menstrual pain. Since menstrual pain is highly prevalent (>60%) (Zondervan et al., 2001; Burnett et al., 2005; Iacovides et al., 2015), and over-the-counter medications are readily available, the outcome of any medication use for menstrual pain may capture a range of menstrual pain severity. In contrast, use of hormonal contraception requires a prescription from a healthcare provider, and probably reflects more severe menstrual pain; this may have improved our ability to detect an association.

Among 82% of participants who reported having menstrual periods and not using hormonal birth control all the time when ages 18–22 years, we observed that soy formula feeding was associated with a greater frequency of moderate/severe menstrual pain with ‘most’ periods during early adulthood; this association was observed despite the potential exclusion of participants with the severest problems with menstrual pain.

Our results are consistent with the association observed in a follow-up study of young adults ages 20–34 years who as infants had participated in formula feeding studies (Strom et al., 2001). Participants fed soy formula as infants, from age 9 days to 4 months of age (Fomon et al., 1971, 1986), reported greater severity of menstrual pain (‘extremely painful’ vs. ‘mildly painful/not painful’ menstrual discomfort, unadjusted relative risk 1.77, 95% CI: 1.04–3.00) compared to those fed cow’s milk-based formula.

Mounting data from animal studies support the disruptive effects of early postnatal genistein exposure on uterine and HPO-axis development and function in adulthood, the reproductive components involved in the pathophysiology of menstrual pain (Faber and Hughes, 1993; Nagao et al., 2001; Jefferson et al., 2005, 2011, 2012; Bateman and Patisaul, 2008; Tang et al., 2008; Cimafranca et al., 2010; Greathouse et al., 2012). Subsequent epigenetic studies have identified pathways by which early postnatal administration of genistein, at doses relevant to those of soy formula-fed infants, results in the early-life developmental reprogramming of gene expression involved in uterine development and adult uterine function that persists into adulthood (Tang et al., 2008; Jefferson et al., 2011, 2012; Greathouse et al., 2012).

The present analyses may have been limited by several aspects related to the ascertainment of data on soy formula feeding. First, we relied on the retrospective report of soy formula feeding. However, the error in exposure reporting was likely minimized by the assistance of the participant’s mother in the completion of the Early-Life Questionnaire for most participants (89%). This is supported by the similar prevalence of soy formula feeding from around the time of SELF cohort was born (~11%, estimated from reported data) (Fomon, 1987) to that in the present analyses (13%). Second, data on non-soy formula feeding and the feeding of solid foods during infancy was not collected in SELF; this information would have helped us to better characterize exposure to soy formula. Third, data were not available on the reasons for feeding soy formula. While we are not able to rule out confounding by indication in SELF, this was addressed in the prior study by Strom et al. (2001) that observed an association between soy formula feeding and menstrual pain; in that study, infants were assigned a specific formula feeding.

Just like the study by Strom et al. (2001), our study was also limited by relying on recall of menstrual pain indicators. However, misclassification of menstrual pain indicators may be minor; the one previous study to compare retrospectively- and prospectively-collected menstrual pain data reported that a single question on menstrual pain frequency was predictive of the proportion of cycles with menstrual discomfort collected by daily diary for at least four cycles (Jukic et al., 2008).

We also did not have data on menstrual pain characteristics, including the location, quality, timing in relation to the menstrual cycle, age at onset, duration and co-occurring symptoms. This information would have allowed us to investigate soy formula feeding in relation to primary and secondary dysmenorrhea, respectively defined as menstrual pain in the absence and presence of other pelvic pathology (Burnett and Lemyre, 2017). Given that primary and secondary dysmenorrhea may involve distinct disease entities with different etiologic pathways, future research disaggregating the outcome of dysmenorrhea could inform the biologic mechanisms involved in the development of menstrual pain after exposure to soy formula during infancy.

Our study had several strengths. First, this study contributes new data on the association between infant soy formula feeding and indicators of menstrual pain in a cohort of young, African-American women. The consistency of our results to that of the prior study conducted in a study population of primarily young, white women (Strom et al., 2001) suggests that our findings may be applicable to women in general. Second, the use of novel indicators regarding a woman’s prior experience with menstrual pain allowed us to evaluate the relationship between soy formula feeding and menstrual pain among the entire SELF cohort. We were able to investigate this relationship without the subject selection issues that arise when evaluating only women who are menstruating and not receiving pharmacologic treatment that could obscure the assessment of menstrual pain (excluding women who may be substantially affected by menstrual pain).

In conclusion, our data suggest that soy formula feeding during infancy is associated with menstrual pain in adulthood. Our observations add to the growing body of literature that collectively lend support to the reproductive health consequences of early-life exposure to soy formula.

Authors’ roles

DDB conceived, designed, and acquired data in the SELF study and KU substantially contributed to the conception and design of the present study. KU conducted the statistical analyses and drafted the manuscript. All authors (KU, MAA, GW, DDB) provided input on the interpretation of data and manuscript revisions critical for important intellectual content. All authors (KU, MAA, GW, DDB) approved the final manuscript version for publication.

Funding

This research was supported by the Intramural Research Program of the National Institutes of Health (NIH), National Institute of Environmental Health Sciences and, in part, by funds allocated for health research by the American Recovery and Reinvestment Act. This research was also supported by NIH grant K99NR017191 (KU) from the National Institute of Nursing Research.

Conflict of interest

None of the authors have a conflict of interest.