The effect of genetic variation in estrogen transportation and metabolism on the severity of menopause symptoms: a study from the RIGHT 10K cohort

Abstract

Objective:

The severity of menopause-related symptoms varies considerably among women. The determinants of this variation are incompletely understood. The aim of this study was to assess the association between genetic variation in estrogen metabolism and transport pathways and the severity of menopause symptoms.

Methods:

This was a cross-sectional study of 60 peri- and postmenopausal women in the Mayo Clinic RIGHT study (which involved sequencing of genes involved in drug metabolism and transport), who had also been evaluated in the Women’s Health Clinic at Mayo Clinic in Rochester, MN. All participants completed the Menopause Rating Scale (MRS) for assessment of menopause symptoms, including hot flashes. The association between severity of menopause symptoms and the variation in genes encoding 8 enzymes and transporters involved in estrogen metabolism was evaluated.

Results:

Lower CYP3A4 activity and higher COMT activity were associated with lower severity of somatic menopause symptoms (p=0.04 and 0.06, respectively). These associations did not persist after adjustment for hormone therapy use. No differences in MRS scores or hot flash severity were noted among other genetic variant groups. Age at natural menopause was not affected by variation in the genes studied.

Conclusion:

The current study did not show an association between genetic variation in estrogen metabolism and transport pathways and the severity of menopause symptoms. Further studies with larger sample sizes may be required to understand this potentially complex association.

INTRODUCTION

Hot flashes and night sweats (vasomotor symptoms, VMS) are the most common symptom of the menopause transition, affecting up to three-fourths of women in this phase of life.¹ They are severe in intensity in about one-third of affected women, and can significantly impact mood, sleep, sexual function and quality of life.^2-5 In addition to these more immediate consequences of VMS, there is increasing recognition that VMS are predictive of disease risk in women, including risk of cardiovascular disease,^1,6 low bone density and fractures,^7,8 and breast cancer.⁹ A hot flash is no longer considered “just a hot flash.”¹⁰ Therefore, it is important to understand the mechanistic pathways underlying VMS and the factors that control the variation in the frequency and severity of VMS in order to offer women individualized counseling and management.

The physiology underlying VMS is poorly understood, but likely relates to a complex interplay between sex steroids, gonadotropins, and neural pathways, both central and peripheral.^6,11-13 A profound decline in estrogen levels is the hormonal hallmark of menopause¹⁴ and is thought to be responsible for several menopause-related symptoms. Even though estrogen levels do not directly correlate with the severity of VMS and other menopause symptoms, differences in the rate of decline in estrogen levels likely contribute to the variation in symptom severity, at least in part.¹³

There is significant genetic variation in the pathways controlling estrogen transport and metabolism (Figure 1). Genes that encode the cytochrome P450 (CYP)1A2, CYP3A, catechol-O-methyl transferase (COMT), sulfotransferase (SULT)1A1, and UDP-glucuronosyl transferase (UGT) enzymes, as well as the SLCO1B1 gene that encodes the OAT1P1 transporter, are highly polymorphic and are known to impact the metabolism and transport of both endogenous and exogenous substrates, such as medications. Specifically, estrone and estradiol can be metabolized by CYP1A2 and CYP3A enzymes to hydroxylated estrogen, which in turn can be methylated by COMT.^15-17 These estrogen metabolites may undergo glucuronidation and/or sulfate conjugation mediated by UGT enzymes or SULT enzymes, respectively, and be eliminated.^18,19 Therefore, higher activity of these enzymes would be expected to potentially lead to lower overall levels of estrogen and metabolites. However, sulfate-conjugated estrogens may also represent an inactive storage pool that can be re-activated through deconjugation by sulfatases (STs).²⁰ Therefore, the impact of genetic variation in SULT enzymes, such as SULT1A1, on estrogen concentrations may be more nuanced. Finally, the OATP1B1 transporter facilitates movement of estrogen into cells.²¹ Higher efficiency of OATP1B1-mediated estrogen transport may increase the estrogen movement into hepatocytes for metabolism and elimination, but also may allow for a greater availability of estrogen for sulfate conjugation and increase the storage pool.²²

Figure 1:

Estrogen metabolism pathways. Estrogen (blue E) enters the cell through the OATP1B1 transporter, which is encoded by SLCO1B1. Reduced function of this transporter may lead to decreased estrogen metabolism. Once inside of the cell, the estrogen may be sulfate conjugated by sulfotransferases (green SULTs symbol) into a sulfated form that comprises a storage pool (blue E with green S symbol). The sulfate group may be removed by sulfotransferases (green STs). Alternatively, estrogen may undergo cytochrome P450-mediated metabolism by CYP1A1, CYP1A2, and/or CYP3A isoforms including CYP3A4, CYP3A5, and CYP3A7 (purple symbols), resulting in an oxidized form (E with purple OH). This form of estrogen may be further metabolized by catechol-O-methyltransferase (COMT, red symbol), which adds a methyl group (red Me symbol). These forms may be sulfate conjugated and/or glucuronide conjugated by uridine 5’-diphospho-glucuronosyltransferases (UGTs, yellow symbol) and eliminated from the cell as estrogen conjugates (blue symbol with “conjugates”). Aside from the SULTs, decreased activity of all these enzymes may decrease estrogen metabolism and excretion.

There is limited evidence to suggest that the variation in estrogen metabolism and transport can affect the response to hormone therapy (HT) in menopausal women and even influence the timing of menopause onset.^23,24 However, it is not known whether variation in the estrogen transport and metabolism pathways influence trajectory of estrogen levels or the course of VMS and other symptoms in women during the menopause transition. It is possible that genetic variation in estrogen transport and metabolism results in a variable hormonal milieu during the menopause transition, which may explain differences in symptom onset, frequency, and severity. The primary objective of this study was to expand beyond our previous studies^23,24 to include additional genes and to improve our understanding of the influence of interindividual variation in estrogen metabolism and transport on the age at menopause and the severity of menopause symptoms, particularly VMS.

METHODS

Study design and participants

This was a cross-sectional study conducted at Mayo Clinic in Rochester, MN. It included patients who overlapped in two Mayo Clinic databases- the RIGHT Protocol Study (The Right Drug, Right Dose, Right Time: Using Genomic Data to Individualize Treatment Protocol)^25,26 and the Data Registry on Experiences of Aging, Menopause and Sexuality (DREAMS).²⁷ The RIGHT Protocol Study enrolled 11,098 patients and ultimately tested 10,030 patients previously seen at the Mayo Clinic who had provided DNA samples to the Mayo Clinic Biobank for future study. These participants, of whom 6,688 were females, had also consented to the use of their medical records for research and underwent sequencing of 77 pharmacogenes.²⁵ DREAMS includes health information obtained from women seen for subspecialty menopause and sexual health consultations in the Women’s Health Clinic (WHC) at Mayo Clinic in Rochester, MN. The information collected at the time of these consultations included demographic data and questionnaire data that assessed menopause symptoms, including VMS, depression, anxiety, sleep, sexual function, and partner satisfaction, among others. For the current study, we included midlife women from the DREAMS registry who overlapped with the RIGHT Protocol Study and were peri- or postmenopausal (as ascertained during the women’s health clinical visit) (Figure 2). These women had consented to the use of their records for research and were seen in the WHC between 2015 and 2017. The study was approved by the Mayo Clinic Institutional Review Board.

Figure 2:

Flowchart depicting recruitment strategy of the study

Measures

Menopause Symptoms-

These were assessed using the Menopause Rating Scale (MRS). This self-reported questionnaire consists of 11 items covering somatic, psychological, and urogenital domains as experienced by a woman in the preceding 4 weeks. Somatic symptoms include hot flashes, heart discomfort, sleep problems, physical and mental exhaustion, and joint/muscular discomfort. The psychological domain includes depressive mood, irritability, and anxiety. Urogenital symptoms include sexual problems, bladder problems, and dryness of the vagina. Each item is scored on scale from 0 to 4 for severity (0 = none; 1 = mild; 2 = moderate; 3 = severe; 4 = very severe) with total scores ranging from 0-44. The higher the composite score, the higher the menopause symptom burden.²⁸ In addition to total and domain specific MRS scores, the hot flash severity for each participant was recorded based on the information provided in the MRS.

Variants in estrogen metabolizing enzymes and transporters

(Figure 1)- Next generation sequencing was performed using the Pharmacogenomics Research Network (PGRN)-Seq v3 custom capture that included 77 pharmacogenes.^25,29,30 The exon and intron-exon boundaries were included for each gene along with clinically relevant intronic and promoter variants. A specialized software algorithm, CNVAR v1.0, developed at Mayo Clinic was used to analyze CYP2D6. Copy number variation was included when evaluating CYP2D6, but was not available for other genes, including SULT1A1. A subset of the 77 genes (COMT, CYP1A2, CYP3A4, CYP3A5, CYP3A7, SLCO1B1, SULT1A1, and UGT1A1) were evaluated for this study. Sequence data were used to determine haplotypes (star alleles) and predicted phenotypes using the methodology utilized in the Mayo Clinic Personalized Genomics Laboratory.³¹ Phenotype ranges (e.g., intermediate to normal metabolizer) were used to reflect uncertainty in phenotype. Rare variants were curated by laboratory directors using the American College of Medical Genetics/Association for Molecular Pathology³² guidelines for variant interpretation with modifications, and variants predicted to impact the phenotype were reported and included in the analysis.³¹ For data analysis, given our small cohort size, results for each gene were categorized as resulting in predicted phenotypes with higher enzyme or transporter activity or a phenotype with a lower activity. The phenotype and genotype groupings along with example alleles are summarized in Table 1.

Table 1.

Groupings for each gene according to the phenotype categories. Example genotypes are indicated in parentheses.

Gene	Higher Activity Phenotype Group	Lower Activity Phenotype Group
CYP1A2	Rapid metabolizers (*1/*1F, *1F/*1F)	Normal metabolizers (*1/*1) Intermediate to ultrarapid metabolizers (*1F/loss-of-function variant)
CYP3A4	Normal metabolizers (*1/*1)	Intermediate (*22/*22) Intermediate to normal metabolizers (*1/*22)
CYP3A5	Normal metabolizers (*1/*1) Intermediate metabolizers (*1/*3)	Poor metabolizers (*3/*3)
CYP3A7	Normal metabolizers (*1C/*1C) Intermediate metabolizers (*1/*1C)	Poor metabolizers (*1/*1)
UGT1A1	Normal metabolizers (*1/*1, also known as TA6/TA6) Intermediate metabolizers (*1/*28 also known as TA6/TA7, *1/*6)	Poor metabolizers (*28/*28 also known as TA7/TA7, individuals with a no function variant) Poor to intermediate metabolizers (*6/*28)
SLCO1B1	Normal function (c.521TT)	Reduced function (c.521T>C heterozygotes and homozygotes)
COMT	Normal metabolizers (c.472GG [p.158 Val/Val]) Intermediate metabolizers (c.472G>A heterozygotes [p.Val158Met heterozygotes, Val/Met])	Poor metabolizers (c.472G>A homozygotes [p.Val158Met homozygotes, Met/Met])
SULT1A1	Normal metabolizers (*1/*1) Intermediate to normal metabolizers (heterozygotes for a variant of uncertain significance) Intermediate metabolizers (*1/*2)	Poor metabolizers (*2/*2 or no coverage suggesting homozygous deletion)

Covariates

Additional information relating to age, body mass index (BMI), race/ethnicity, educational status (high school graduate or lower, some college education, 4-year college graduate, or postgraduate), employment status (employed, full time homemaker, retired or other), age at natural menopause, current HT use, and relationship status was gathered from the clinical intake form and the vital signs recorded at the time of the clinical visit. Participants’ depression and anxiety scores were collected at the time of the clinical visit using the Patient Health Questionnaire-9 (PHQ-9)³³ and Generalized Anxiety Disorder-7 (GAD-7)³⁴ scales, respectively.

Statistical analyses

Data are summarized using means and standard deviations (SD) or medians and interquartile ranges (IQR) for continuous variables, and frequencies and percentages for categorical variables. Linear regression was used to assess the association between genetic variations and the outcomes of age at natural menopause, MRS total score, and MRS domain scores. Logistic regression was used to assess the association between genetic variations and the outcomes of severe/very severe hot flashes and MRS scores in the third or fourth quartiles. All models were also adjusted for HT use, because the use of HT can later the menopause experience. An overall “pathway” score was created where each enzyme or transporter was assigned a score for each participant ranging from 1 point for a poor metabolizer or SLCO1B1 homozygous c.521T>C, 2 points for an intermediate metabolizer or SLCO1B1 heterozygote, 3 points for a normal metabolizer or SLCO1B1 wild-type, and 4 points for genotypes associated with increased metabolism (e.g. CYP1A2 rapid metabolizer). The points were added for an overall score for each participant. Linear regression was used to assess the association between the overall pathway score and the severity of menopause symptoms. Analyses were performed using SAS version 9.4 software (SAS Institute, Inc.; Cary, NC). All tests were two-sided, and p-values ≤ 0.05 were considered statistically significant.

RESULTS

Sixty women were included in the current study; 11 were perimenopausal (average age 47.9±6.4 years) and 49 were postmenopausal (average age 61.2±8.2 years). For the postmenopausal group, the average age at menopause was 50.5±4.6 years, with the median duration since menopause of 8.6 years (IQR 4.8, 17.5). Participant characteristics are summarized in Table 2. The majority were white, educated (some college education or higher), employed and non-smokers. Forty-five percent reported current HT use. The median MRS score was 9.5 (IQR 5-14), with 20% of the women reporting severe or very severe VMS.

Table 2.

Participant Characteristics

	Total (N=60)
Menopause status
Perimenopausal	11 (18.3%)
Postmenopausal (natural menopause)	49 (81.7%)
Age at menopause (years), Mean (SD)	50.5 (4.6)
Race
White	55 (91.7%)
Mixed	4 (6.7%)
Unknown/Missing	1 (1.7%)
Partner status
Divorced	7 (11.7%)
Married	44 (73.3%)
Partnership	1 (1.7%)
Single	4 (6.7%)
Widowed	4 (6.7%)
Education
Some high school/High school/GED	5 (8.3%)
Some college or 2-year degree	20 (33.3%)
4-year college degree	10 (16.7%)
Post graduate studies	24 (40.0%)
Unknown	1 (1.7%)
Employment
Employed	37 (61.7%)
Full time homemaker	4 (6.7%)
Retired	16 (26.7%)
Work disabled	1 (1.7%)
Other	2 (3.3%)
Smoking Status
Never smoked	51 (85.0%)
Former smoker	9 (15.0%)
Hormone therapy use	27 (45.0%)
MRS total, Median (IQR)	9.5 (5, 14)
MRS psychological symptoms, median (IQR)	2 (0, 5)
MRS somatic symptoms, median (IQR)	4 (2, 6)
MRS urogenital symptoms, median (IQR)	3 (1, 5.5)
MRS – hot flashes
None	18 (30.0%)
Mild	16 (26.7%)
Moderate	14 (23.3%)
Severe	10 (16.7%)
Very severe	2 (3.3%)
GAD-7 score
Missing	4
< 5	48 (85.7%)
≥ 5	8 (14.3%)
PHQ-9 score
Missing	3
< 5	40 (70.2%)
≥ 5	17 (29.8%)
CYP1A2 phenotype
Lower activity phenotype	5 (8.3%)
Higher activity phenotype	55 (91.7%)
CYP3A4 phenotype
Lower activity phenotype	6 (10.0%)
Higher activity phenotype	54 (90.0%)
CYP3A5 phenotype
Higher activity phenotype	6 (10.0%)
Lower activity phenotype	54 (90.0%)
CYP3A7 phenotype
Higher activity phenotype	4 (6.7%)
Lower activity phenotype	56 (93.3%)
UGT1A1 phenotype
Missing	8
Higher activity phenotype	43 (82.7%)
Lower activity phenotype	9 (17.3%)
SLCO1B1 phenotype
Missing	8
Lower activity phenotype	17 (32.7%)
Higher activity phenotype	35 (67.3%)
COMT phenotype
Higher activity phenotype	48 (80.0%)
Lower activity phenotype	12 (20.0%)
SULT1A1 phenotype
Higher activity phenotype	55 (91.7%)
Lower activity phenotype	5 (8.3%)

The associations between the genotypes and age at menopause, MRS total scores, MRS domain scores, MRS score quartiles, reported hot flash severity, and the depression and anxiety scores are shown in Tables 3 and 4. Lower CYP3A4 activity and higher COMT activity were both associated with lower MRS scores in the somatic domain, but the association did not persist after adjustment for HT use. There were no other significant differences in the other outcome variables between the groups with higher or lower predicted enzyme/transporter activity, both before and after adjustment for HT use (Tables 3-6).

Table 3.

Associations between phenotypes and age at natural menopause, MRS^a total score, and MRS domain scores (unadjusted)

	Age at menopauseb		MRS total		MRS psychological		MRS somatic		MRS urogenital
	Difference (95% CI)	p	Difference (95% CI)	p	Difference (95% CI)	p	Difference (95% CI)	p	Difference (95% CI)	p
CYP1A2c	1.08 (−5.63, 7.80)	0.75	−2.15 (−7.92, 3.63)	0.46	−1.20 (−4.09, 1.69)	0.41	−1.04 (−3.33, 1.26)	0.37	0.09 (−2.54, 2.72)	0.95
CYP3A4 c	1.41 (−3.44, 6.26)	0.56	−0.41 (−5.75, 4.94)	0.88	0.85 (−1.82, 3.52)	0.53	−2.20 (−4.25, −0.16)	0.035	0.94 (−1.47, 3.35)	0.44
CYP3A5 c	−1.29 (−6.83, 4.25)	0.64	0.78 (−4.56, 6.12)	0.77	1.19 (−1.48, 3.85)	0.38	−1.50 (−3.59, 0.59)	0.16	1.09 (−1.31, 3.50)	0.37
CYP3A7 c	1.88 (−2.95, 6.72)	0.44	0.13 (−6.30, 6.55)	0.97	1.23 (−1.97, 4.44)	0.44	−0.11 (−2.67, 2.45)	0.93	−1.00 (−3.90, 1.90)	0.49
UGT1A1 c	−2.25 (−5.64, 1.14)	0.19	−1.40 (−5.69, 2.89)	0.52	−0.96 (−3.10, 1.18)	0.37	0.90 (−0.80, 2.60)	0.29	−1.34 (−3.26, 0.58)	0.17
SLCO1B1 c	1.57 (−3.96, 7.10)	0.57	1.44 (−5.91, 8.79)	0.70	0.95 (−2.73, 4.63)	0.61	−1.25 (−4.16, 1.66)	0.40	1.74 (−1.57, 5.04)	0.30
COMT c	0.11 (−3.35, 3.57)	0.95	2.67 (−1.28, 6.61)	0.18	−0.40 (−2.40, 1.61)	0.69	1.48 (−0.07, 3.03)	0.06	1.58 (−0.19, 3.35)	0.08
SULT1A1 c	−2.33 (−6.68, 2.03)	0.29	−1.71 (−7.49, 4.08)	0.56	−0.76 (−3.67, 2.14)	0.60	−0.16 (−2.47, 2.14)	0.89	−0.78 (−3.40, 1.84)	0.55

^aMenopause Rating Scale

^bOnly among 49 post-menopausal women with natural menopause

^cEstimates are for lower activity vs higher activity phenotype group (as defined in Table 1)

Table 4.

Associations between phenotypes and severe/very severe hot flashes, and MRS^a total and domain scores being in the top two quartiles (vs in the lower two quartiles) (unadjusted)

	Severe/very severe hot flashesc		MRS total (Q3/Q4 vs Q1/Q2)		MRS psychological (Q3/Q4 vs Q1/Q2)		MRS somatic (Q3/Q4 vs Q1/Q2)		MRS urogenital (Q3/Q4 vs Q1/Q2)
	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p
CYP1A2b	0.32 (0.01-8.03)	0.49	0.64 (0.10-4.15)	0.64	0.86 (0.13-5.57)	0.88	0.35 (0.04-3.32)	0.36	0.86 (0.13-5.57)	0.88
CYP3A4b	0.26 (0.01-6.24)	0.41	1.00 (0.19-5.40)	0.99	2.91 (0.49-17.29)	0.24	0.27 (0.03-2.47)	0.25	2.91 (0.49-17.29)	0.24
CYP3A5b	0.45 0.07, 2.86)	0.40	1.00 (0.19-5.26)	0.99	1.59 (0.27-9.09)	0.60	0.64 (0.12-3.45)	0.60	1.59 (0.27-9.09)	0.60
CYP3A7b	2.50 (0.09-100.0)	0.58	1.00 (0.13-7.69)	0.99	2.44 (0.24-25.00)	0.46	2.08 (0.20-20.00)	0.53	0.75 (0.10-5.56)	0.78
UGT1A1b	1.96 (0.42-9.09)	0.39	1.00 (0.26-3.85)	0.99	0.50 (0.12-2.17)	0.36	1.00 (0.25-4.00)	0.99	0.50 (0.12-2.17)	0.36
SLCO1B1b	0.48 (0.04-5.88)	0.56	2.08 (0.18-25.00)	0.56	1.56 (0.13-16.67)	0.72	0.31 (0.03-3.70)	0.36	1.56 (0.13-16.67)	0.72
COMTb	0.76 (0.14-4.00)	0.75	2.38 (0.63-9.09)	0.20	0.92 (0.26-3.33)	0.90	4.00 (1.04-14.29)	0.043	2.13 (0.59-7.69)	0.25
SULT1A1b	0.32 (0.01-7.69)	0.49	0.64 (0.10-4.17)	0.64	0.86 (0.13-5.56)	0.88	1.00 (0.15-6.67)	0.99	0.86 (0.13-5.56)	0.88

^aMenopause Rating Scale

^bEstimates are for lower activity vs higher activity phenotype group (as defined in Table 1)

^cCompared to none/mild/moderate

Table 6.

Associations between phenotypes and severe/very severe hot flashes, and MRS^a total and domain scores being in the top two quartiles (compared to the lower two quartiles), all adjusted for hormone therapy use.

	Severe/very severe hot flashesc		MRS total (Q3/Q4 vs Q1/Q2)		MRS psychological (Q3/Q4 vs Q1/Q2)		MRS somatic (Q3/Q4 vs Q1/Q2)		MRS urogenital (Q3/Q4 vs Q1/Q2)
	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p
CYP1A2b	0.26 (0.01-7.20)	0.43	0.59 (0.09-4.04)	0.59	0.84 (0.13-5.49)	0.85	0.33 (0.03-3.23)	0.34	0.82 (0.12-5.49)	0.84
CYP3A4b	0.44 (0.02-11.71)	0.62	1.57 (0.26-9.44)	0.62	4.04 (0.62-26.44)	0.15	0.31 (0.03-3.01)	0.32	4.80 (0.71-32.47)	0.11
CYP3A5b	0.67 (0.10-4.55)	0.68	1.37 (0.24-7.69)	0.72	1.89 (0.31-11.11)	0.49	0.74 (0.13-4.17)	0.73	2.08 (0.34-12.50)	0.43
CYP3A7b	2.63 (0.08-100.0)	0.58	0.94 (0.12-7.69)	0.96	2.38 (0.23-25.00)	0.47	2.04 (0.20-20.00)	0.55	0.71 (0.09-5.56)	0.75
UGT1A1b	2.33 (0.45-12.50)	0.31	1.06 (0.26-4.34)	0.93	0.51 (0.12-2.22)	0.37	1.03 (0.25-4.17)	0.97	0.52 (0.12-2.27)	0.38
SLCO1B1b	0.54 (0.04-7.14)	0.64	2.44 (0.20-33.33)	0.49	1.67 (0.22-1.75)	0.69	0.33 (0.03-3.85)	0.38	1.72 (0.14-20.00)	0.66
COMTb	0.51 (0.09-2.94)	0.45	1.96 (0.50-7.69)	0.34	0.81 (0.22-3.03)	0.75	3.70 (0.95-14.29)	0.06	1.85 (0.49-6.67)	0.36
SULT1A1b	0.26 (0.01-7.14)	0.43	0.59 (0.09-4.00)	0.59	0.84 (0.13-5.56)	0.85	0.97 (0.15-6.25)	0.98	0.82 (0.12-5.56)	0.84

^aMenopause Rating Scale

^bEstimates are for lower activity vs higher activity phenotype group (as defined in Table 1)

^cCompared to none/mild/moderate

In addition to the single gene analyses, an overall pathway score was analyzed. For the pathway score, each enzyme/transporter phenotype was assigned a score based on expected impact on estrogen metabolism and the scores for each gene were added together such that a higher score would theoretically be associated with a higher degree of estrogen metabolism. For example, a woman who was a CYP1A2 rapid metabolizer (4 points), CYP3A4 normal metabolizer (3 points), CYP3A5 normal metabolizer (3 points), CYP3A7 normal metabolizer (3 points), UGT1A1 normal metabolizer (3 points), SLCO1B1 wild-type (c.521 TT, 3 points), COMT wild-type (Val/Val, 3 points), and SULT1A1 normal metabolizer (3 points) would have a score of 25. In contrast, a CYP1A2 normal metabolizer (3 points), CYP3A4 intermediate metabolizer (2 points), CYP3A7 poor metabolizer (1 point), UGT1A1 poor metabolizer (1 point), SLCO1B1 c.521T>C homozygote (1 point), COMT Met/Met (p.Val158Met homozygote, 1 point), and SULT1A1 poor metabolizer (1 point) would have a pathway score of 10. However, there was no significant association between the pathway score and any of the clinical variables evaluated (all p>0.05, data not shown).

DISCUSSION

These results demonstrate that variations in estrogen metabolism pathways may not contribute to interindividual variation in the severity of menopause symptoms or to the variability in age at natural menopause. However, this is a small study, and due to the complexity of the estrogen pathways and involvement of multiple genes, each variable may have a small contribution such that a significantly larger cohort would be required to understand the impact of variation in estrogen metabolism. Similarly, it may not be accurate to weight these variations in metabolism equally as some may play a proportionally greater role in the overall effect than others. Previous studies by our group and others have revealed associations between some of the genes studied and age at menopause as well as menopause symptoms.^23,24 We were not able to replicate those findings in this study, potentially due to the limited sizes of our current and previous cohorts. In addition, we were not able to include CNV for SULT1A1 in the current study. However, it is also important to recognize that these are complex phenotypes that likely involve the interplay between multiple genetic and non-genetic variables.

There is a significant variability in VMS severity and frequency among women from different racial and ethnic backgrounds, with African American women reporting the most severe VMS, followed by Hispanic and non-Hispanic white women, and Asian women reporting the least severe symptoms. These differences persist despite adjustment for factors like BMI and smoking. The racial and ethnic differences in VMS severity may be a result of underlying genetic architecture, but more likely reflect the complex interplay between genes and environment (e.g. epigenetics).

Additionally, the rate of decline in estrogen levels appears to contribute to the severity of symptoms as evidenced by the more severe symptoms experienced by women who undergo removal or ablation of the ovaries, or medical ovarian suppression as part of breast cancer therapies, in comparison to women experiencing natural menopause.^{13, 35, 36} Based on these observations, it is likely that the rate of decline in the estrogen level plays an important role in determining the severity of VMS, despite the lack of a direct link between estrogen levels and the intensity of VMS in perimenopausal women.¹³ Therefore, variation in estrogen metabolism pathways could conceivably impact the severity of VMS during the menopause transition by accentuating the pre-existing variability in estrogen levels that is characteristic of this reproductive stage.

Few studies have previously evaluated the impact of genetic variability on VMS. A recent metaanalysis including 18 studies found significant associations between VMS and 14 of the 26 genes assessed.⁴⁰ These included variants in the estrogen metabolism genes (COMT, CYP1A1, CYP1A2, CYP1B1, CYP3A4, CYP19A1 and SULT1A1). CYP1B1 was the most frequently studied gene, but its variants did not consistently correlate with VMS severity across studies. Most variants were reported in a single study only, limiting generalizability of results. For those variants that were reported in more than one study, the results were often conflicting. There was also significant heterogeneity among the included studies, particularly with respect to VMS measurement, and majority of the studies had small sample sizes. The authors concluded that genetic variation may play a role in VMS pathophysiology, but future larger studies were needed to confirm and extend their findings.

Although this study included 8 genes involved in estrogen metabolism and transport, estrogen-related pathways are complex and involve additional genes that were out of the scope of the current study. The attempt to create a “pathway score” was limited by insufficient data in the existing literature and the uncertainty regarding the relative contribution of each gene to the overall estrogen metabolism phenotype. Furthermore, it was difficult to incorporate SULT1A1 and SLCO1B1 due to their potential contribution to not only estrogen metabolism, but also to the inactive storage pool.

Although there may be a hereditary component to the menopause experience, based on the previous studies,^23,24,41 it is becoming increasingly clear that the genetic contribution is not monogenic and likely not even oligogenic. Further studies are required to understand the complex interplay between genetics and menopause phenotypes, and ultimately to better manage menopause symptoms. However, given the polygenic pathophysiology and the need to consider environmental and epigenetic influences, several challenges must be overcome. First, large cohorts will be required to evaluate the data and to potentially generate a polygenic score to represent the overall genetic contribution to estrogen metabolism. As more patients undergo genetic sequencing on either a research or clinical basis, this barrier may be overcome in the future. However, the second requirement is that the menopause experience of individual women be fully characterized to have granular and specific phenotype data for study. Outside of specialized women’s health clinics, this may be the more challenging obstacle in the study of impact of genetic variability on the menopause experience.

Further studies including multivariable analyses and artificial intelligence-based approaches should be considered to better understand the interplay between these genetic variants. A composite polygenic risk score based on multiple genetic variants as well as individual characteristics could be useful for predicting menopause symptom severity. Future studies should include assessment of additional genes and the contribution of individual characteristics including race/ethnicity, BMI, and other factors that may contribute to epigenetic changes, such as childhood adversity and other social determinants of health in the composite score with the goal of developing individualized risk prediction models for menopause symptoms as well as proactive treatment strategies.

Acknowledging its limitations, this work represents a step forward toward better understanding of the influence of genetic variations on the experience of menopause and may help inform future study designs.

Strengths:

The participants had reliable documentation of menopause-related symptoms. While the overall RIGHT10K cohort included 6,688 women, menopause symptoms are typically not well-documented in the medical record, which limits the ability to study these important phenotypes. Multiple genetic variants of interest were studied.

Limitations:

The most significant limitation of this study was its small sample size, diminishing the power. The cohort also demonstrated limited racial and ethnic diversity. Common genetic variants in drug metabolizing enzymes are present at different frequencies among populations, and it is certainly conceivable that the genetic variants have a differential impact on the menopause experience based on environmental or epigenetic factors that may differentially impact different racial or ethnic groups. We only studied women experiencing natural menopause. Future studies should consider expanding the study population to women experiencing premature menopause. Genome-wide association studies were not performed, which limits the ability to detect the impact of unknown genetic variants on the menopause experience.

Table 5.

Associations between phenotypes and age at natural menopause, MRS^a total score, and MRS domain scores, all adjusted for hormone therapy use

	Age at menopausec		MRS total		MRS psychological		MRS somatic		MRS urogenital
	Difference (95% CI)	p	Difference (95% CI)	p	Difference (95% CI)	p	Difference (95% CI)	p	Difference (95% CI)	p
CYP1A2b	1.08 (−5.71, 7.88)	0.75	−2.29 (−7.98, 3.40)	0.42	−1.22 (−4.13, 1.69)	0.41	−1.10 (−3.34, 1.13)	0.33	0.03 (−2.57, 2.64)	0.98
CYP3A4b	1.75 (−3.42, 6.92)	0.50	0.76 (−4.69, 6.21)	0.78	1.07 (−1.70, 3.85)	0.44	−1.81 (−3.90, 0.29)	0.09	1.49 (−0.96, 3.94)	0.23
CYP3A5b	−1.22 (−7.03, 4.59)	0.67	1.66 (−3.69, 7.01)	0.54	1.35 (−1.38, 4.07)	0.33	−1.16 (−3.25, 0.94)	0.27	1.48 (−0.94, 3.89)	0.23
CYP3A7b	1.88 (−3.01, 6.77)	0.44	−0.02 (−6.36, 6.33)	0.99	1.21 (−2.02, 4.45)	0.46	−0.17 (−2.67, 2.33)	0.89	−1.06 (−3.93, 1.82)	0.46
UGT1A1b	−2.24 (−5.68, 1.21)	0.20	−1.25 (−5.48, 2.99)	0.56	−0.94 (−3.10, 1.22)	0.39	0.97 (−0.68, 2.63)	0.24	−1.28 (−3.18, 0.63)	0.18
SLCO1B1b	1.63 (−3.99, 7.25)	0.56	1.77 (−5.48, 9.02)	0.63	0.99 (−2.72, 4.71)	0.59	−1.10 (−3.95, 1.75)	0.44	1.87 (−1.40, 5.14)	0.26
COMTb	0.06 (−3.49, 3.60)	0.97	2.10 (−1.90, 6.09)	0.30	−0.51 (−2.57, 1.56)	0.63	1.22 (−0.33, 2.78)	0.12	1.38 (−0.42, 3.18)	0.13
SULT1A1b	−2.36 (−6.77, 2.05)	0.29	−1.85 (−7.56, 3.85)	0.52	−0.78 (−3.71, 2.14)	0.59	−0.23 (−2.49, 2.03)	0.84	−0.84 (−3.44, 1.76)	0.52

^aMenopause Rating Scale

^bEstimates are for lower activity vs higher activity phenotype group (as defined in Table 1)

^cOnly among 49 post-menopausal women with natural menopause

Highlights.

• Genetic variation in estrogen metabolism may affect menopause symptoms.

• Lower CYP3A4 activity and higher COMT activity were associated with less severe menopause symptoms in the present study.

• No association was found between other genetic variants and menopause symptom severity.

• Further study is required to understand the complex association between genetic variation in estrogen metabolism and transport pathways and the severity of menopause symptoms.

Abbreviations

CYP

Cytochrome P450

COMT

Catechol-O-methyl transferase

VMS

Vasomotor symptoms

SULT

Sulfotransferase

UGT

UDP-glucuronosyl transferase

SLCO

Solute carrier organic anion transporter family member

OATP

Organic anion transporting polypeptide

ST

Sulfatase

HT

Hormone therapy

RIGHT Protocol Study

The Right Drug, Right Dose, Right Time: Using Genomic Data to Individualize Treatment Protocol

DREAMS

Data Registry on Experiences of Aging, Menopause and Sexuality

WHC

Women’s Health Clinic

MRS

Menopause Rating Scale

BMI

Body mass index

PHQ-9

Patient Health Questionnaire-9 questionnaire

GAD-7

Generalized Anxiety Disorder-7 questionnaire