Transitioning to the Menopausal Transition: A Scoping Review of Research on the Late Reproductive Stage in Reproductive Aging

Abstract

Importance and Objective:

In 2001 Staging Reproductive Aging Workshop conferees described the late reproductive stage (LRS) of reproductive aging as preceding the onset of the menopausal transition, yet there has been little attention to this aspect of reproductive aging. The aim of this scoping review was to examine scientific publications characterizing the LRS to map what is known about this stage with particular focus on reproductive endocrine patterns, menstrual cycle changes, and symptoms.

Methods:

The initial search strategy included PubMed and CINAHL searches for the phrase “late reproductive stage” (LRS) and “human”. Given a low yield of research articles, a second stage used “late reproductive age” (LRA) as a search term. These strategies yielded 9 and 26 research articles, respectively. Publications meeting inclusion criteria (data-based research studies, focus on LRS or LRA and hormonal patterns, menstrual characteristics, and symptoms) published in English were reviewed by co-investigators. Excluded studies were related to specific diseases, such as cardiovascular disease, and treatment studies. Data were summarized using qualitative methods. In order to ensure adequate coverage of published research we expanded our review to a third phase in which we identified longitudinal studies of the menopausal transition.

Discussion and Conclusions:

Studies of the LRS focused on: symptoms (anxiety and mood symptoms, bladder symptoms, urinary incontinence, urinary frequency and nocturia) and associated factors, such as endocrine levels and gene polymorphisms; symptom clusters women experienced during the LRS; cognitive function testing results; changing patterns of physiology such as cytokines and chemokines, lipids, hormone patterns/levels; and association of lifestyle factors such as smoking with hormone levels and symptoms. The LRA search yielded a preponderance of studies of reproductive hormones (such as AMH) and menstrual cycle patterns. Remaining studies focused on symptoms, gene variants, health-related behaviors and approaches to classifying menstrual cycles. Longitudinal studies revealed reports of symptoms as well as attempts to classify the progression from the reproductive years to the menopausal transition. Study of the LRS has not been systematic and the limited number and scope of completed studies have yet to contribute a clear and complete picture of the LRS. In some, LRS provided a comparison stage against which to evaluate menopausal transition hormonal and cycle patterns and symptoms. Harmonizing the results of studies of the LRS and LRA is essential to understand more completely women’s experiences of the LRS and to allow clinicians to provide better support for women during this time. The LRS also represents an ideal inflection point to promote lifestyle choices that could alter the trajectories of chronic diseases that arise in the fifth, sixth and seventh decades of women’s lives.

In order to develop a shared understanding of midlife women’s progression from menstruating regularly to postmenopause, researchers and clinicians sought to develop a common framework to align research and discourse on reproductive aging. In 2001 Staging Reproductive Aging Workshop (STRAW) conferees proposed three overarching phases through which women progressed from their reproductive years to postmenopause: reproductive, menopausal transition, and postmenopause (1).

STRAW conferees further refined these three phases, resulting in 7 stages of reproductive aging numbered with reference to the final menstrual period as time zero (0). The reproductive years were divided into three sub-stages: termed early, peak, and late reproductive stages, noted as −5, −4, and −3 respectively. The menopausal transition (MT) was divided into two sub-stages: called early and late menopausal transition stages as −2 and −1 respectively. The final menstrual period at time 0 is followed by post menopause, sub-divided as early postmenopause (+1) and late postmenopause (+2).

The mid- and late reproductive stages (STRAW −4 and −3) were characterized by regular menstrual cycles and increasing FSH levels in −3. Entry into the early menopausal transition, stage −2, requires persistent variation in menstrual cycle length of 7 or more days in consecutive cycles.

The 2001 Staging Reproductive Aging Workshop (STRAW) (1) established criteria for the LRS as having increasing FSH levels with regular menstrual cycles contrasting with the increasing cyclic irregularity of the MT. In 2011 STRAW+10 conferees modified definitions and criteria for the LRS, stipulating it as the time when women’s fecundability begins to wane and they may notice “subtle changes in menstrual flow/length” (2). They did not attribute other descriptive characteristics to this stage.

STRAW + 10 conferees recommended subdividing the LRS into two substages (2). During the first substage (−3b) menstrual cycles remained regular, without changes in length or early follicular phase FSH levels. During the second substage (−3a) which immediately precedes the early menopausal transition (ET), subtle changes in menstrual cycle characteristics occur, typically shorter cycles and changes in menstrual flow. In −3a biomarker changes occur: FSH levels rise and become more variable during the early follicular phase, and AMH and inhibin B levels and antral follicle counts are low (2). Subtle menstrual cycle changes associated with stage −3a resemble those described by Mitchell in 2000 that occurred prior to the break in cyclicity associated with the STRAW early menopausal transition stage (3).

The research community used STRAW as a way to harmonize research findings, while professional organizations, such as the North American Menopause Society, and educational resources such as UpToDate have encouraged clinicians to incorporate the STRAW criteria in their practice. Although the STRAW criteria now guide staging of the early and late menopausal transition in many studies, researchers have directed little attention to events that precede the early menopausal transition stage, termed the late reproductive stage (LRS).

Reports from women indicate that when they express concern to their health care providers about small changes in their bleeding patterns, menstrual cycle length, or related symptoms prior to noticing significant (i.e. >=7 day) differences in the length of adjacent cycles, they are often advised these are “not related to menopause”. Not surprisingly, some women reporting these experiences express dissatisfaction with health care, believing that providers invalidate or minimize their concerns (personal communication, Coslov 2019). Health care providers familiar with the STRAW stages of the menopausal transition may be adhering to the criterion of persistent 7-day variability between the length of adjacent cycles as a marker for the early transition stage, but may be unaware of the slight but noticeable changes in cycle characteristics or symptoms occurring prior to the onset of the early MT stage.

Currently there are limited studies describing women’s experiences and health care providers’ views of women’s experiences, such as bleeding patterns and symptoms in the years before cycle irregularity is obvious. The purpose of this scoping review is to identify published peer reviewed research on the late reproductive stage (LRS) in order to create a picture of what is known about LRS, including its definition and bleeding, hormone, and symptom patterns.

Methods

A scoping review was selected to guide this inquiry. This is an appropriate choice when the aim is to map the body of existing literature in a topical area, to identify the scope of evidence, concepts and definitions guiding existing research as well as knowledge gaps, and/or research methods used in published studies to date (4). Results of these reviews can guide future systematic reviews and/or future research.

Sampling

We started by employing a PubMed search for articles published after 2000 in which the phrase “late reproductive stage” (LRS) and “human” were used as search terms. We found only 14 papers, 4 of which were results of the 2011 Staging Reproductive Aging Workshop published simultaneously in 4 journals (1,5,6,7). One article was a review of the neuroendocrine physiology of the early and late menopause (8). Articles reporting research findings identified in this search (n=9) are listed in Table 1. Searching PsychInfo and CINAHL did not identify additional published research reports. (See Figure 1 for sampling of these studies.)

Table 1.

Research Reports Identified in Initial PubMed Search of “Late Reproductive Stage” Meeting Review Criteria (N=9)

Author (year)/ Study Name/Country	Purpose	Study Design and Methods	Results	Definitions of Late Reproductive Stage
Cray et al 2012 (10) Seattle Midlife Women’s Health Study (SWMHS) USA	Identify symptom clusters that characterize women’s experiences through the late reproductive (LR) stage, menopausal transition (MT) and early postmenopause (PM)	Subset of SMWHS participants studied longitudinally while progressing from LR to early postmenopause (PM) provided symptom data (22 symptoms) between 1990 and 2005 in health diaries with ratings on 6857 occasions. Multilevel latent class analysis used to identify classes using scores for hot flashes and factors for sleep, cognitive, mood, pain, and tension	Three classes identified. Class 1 – low severity levels for all symptoms; Class 2 - low severity hot flashes and moderate severity levels of all other symptoms; and Class 3 - high severity hot flashes with lower severity of all other symptoms. LR stage used as referent in comparison to early transition (ET), late transition (LT) and PM. Being in Class 3 relative to class 1 was more likely during the ET or LT stages or early PM than LR. Being in Class 2 relative to class 1 was more likely during the ET and LT vs LR stage.	Used data from menstrual calendars to classify women using STRAW criteria for LR, ET, LT, and PM Stages.
Cray et al 2013 (11) SMWHS USA	Explore how symptoms clustered over LR stage, early and late MT stages, and early PM and whether factors identified in one stage would predict those of the next stage	Subset of SMWHS participants studied longitudinally while progressing from LR to early PM stages provided symptom data (22 symptoms) between 1990 and 2005 in health diaries Principal components analysis used to extract factors for each stage	For late reproductive stage 3 factors identified: Factor 1 included depressed mood, difficulty concentrating, tired, forgetful, irritable, nervous, panic, mood changes, crying, and tension. Factor 2 included breast pain, problem falling asleep, early awakening, hot flash, weight gain, vaginal dryness, and wake up during night; factor structure unique to LR stage. Factor 3 included backache, cold sweats, headache, and joint ache symptoms. LR stage Factor 1 (mood/cognitive) symptoms predicted the same for ET stage; Factor 2 and Factor 3 predicted pain factor and somatic factor during ET respectively.	Used data from menstrual calendars to classify women using STRAW criteria for LR, ET, LT, and PM Stages
Tani et al 2013 (16) Tokushima, Japan	Clarify changes in circulating cytokines and chemokines in women during the menopausal transition using classification based on STRAW criteria	554 women recruited from Tokushima University Hospital OB/GYN service studied using cross-sectional design. Sample divided into 7 stages: mid-reproductive (MR) stage, LR, ET, LT, very early PM, early PM and late PM Levels of IL-1B, IL-5, IL-6, IL-7, IL-8, IL-10 TNF-a, MIP-1B, and MCP-1 measured	Serum IL-8 concentrations in PM > MR or LR stage and ET or LT. MCP-1 in LT and PM > MR, LR, ET. MCP-1 level correlated with FSH level during MT. IL-8 high PM, MCP-1 high during MT and may be sensitive to hormonal changes and involved in development of estrogen deficiency diseases.	Used menstrual regularity and FSH levels to stage women in 7 stages: MR, LR, ET, LT, very early PM, early PM and late PM (elevated FSH defined as >10 mIU/ml). Methods for determining regularity not described.
Woods and Mitchell, 2013 (12) SMWHS USA	Determine effects of urinary incontinence (stress – SUI and urge - UUI) on mood, perceptions of self, attitudes toward midlife, and consequences for daily living.	Subset of SMWHS participants (n=299) studied longitudinally provided data from LR, ET, LT, and early PM from menstrual calendars, annual questionnaires, and symptom diaries. Multilevel modeling used to test models of SUI and UUI. .	Both SUI and UUI were associated with lower self esteem and mastery. Effects of either SUI or UUI on mood symptoms, attitudes toward aging and toward menopause, perceived health, and consequences for daily living were not significant.	Used data from menstrual calendars to classify women using STRAW criteria for LR, ET, MT, and PM.
Weber et al, 2013 (9) Rochester Investigation of Cognition across Menopause (RICAM) USA	Determine if cognitive function differs across stages of reproductive aging and if hormones or menopausal symptoms predict cognition during perimenopause	Midlife women (n=117) in LR (34), ET (28), LT (41) or early PM (14), mean age 48.7 years, completed neuropsychological battery of tests, rated menopausal symptoms, and provided serum tested for estradiol, FSH. Study conducted in 3 waves with a subset of women followed longitudinally with overlap across waves (n=14 waves 1 and 2, n=37 waves 2 and 3). For 100 women prospective menstrual cycle data following baseline visit were available.	Women performed worse during PM than LR and LT on verbal learning, verbal memory, and motor function and worse than women in LT on attention/working memory tasks	MT stage classified based on self-reported bleeding patterns using questions about bleeding patterns, date of last menses, average length of current and prior cycles for past 12 months prior to study entry as baseline. Monthly calendars were used to track menstrual cycles for length and variability and date of FMP. For discrepancies between self-report and 6 months of calendar data, staging based on bleeding criteria and FSH and estradiol levels. For 17 women with no prospective data, staging used self-report of bleeding patterns only.
Jones et al, 2016 (17) USA UCSF Women’s Health Study	Identify factors associated with bladder symptoms in LR and MT for midlife women	Women (mean age 48, 44–54 years) (N=158) provided anthropometric data, menstrual cycle length and symptoms, urine samples analyzed for FSH, and self-reported health and depressive symptom ratings. Women were followed at 6 month intervals after study entry. Bladder symptoms assessed with self-report questionnaire measures during years 6 and 7 of the study.	Most common bladder symptoms were nocturia and urinary incontinence. Incontinence more prevalent in LR stage than MT. Women in LR stage more likely to report nocturia than MT. African American women less likely to report incontinence than European Americans and Latinas.	LR or MT coded using menstrual regularity and FSH trends based on STRAW criteria
Jurcsak et al, 2016 a (13) West Pomerania Study, Poland	Determine whether anxiety and mood disorders in LR stage women were related to serotonin transporter and MAO A gene polymorphisms	Healthy midlife (mean age 42+/−4.5 years) Polish women studied cross-sectionally in LR stage (n=345) completed the State-Trait Anxiety Inventory and UWIST Mood Adjective Checklist	No relationship between the 5-HTT and MAO-A gene polymorphisms and severity of anxiety and mood disorders in healthy women in LR stage.	Women classified as in LR stage based on AMH and FSH levels. Normal FSH levels were follicular levels, 3.5–12.5 mIU/ml. Mean FSH level was 6.4 mIU /ml, IQR 3.5 and mean AMH l was 1.33ng/ml, IQR 2.34. Mean estradiol 80 pg.ml, IQR 96.1.
Jurczak et al, 2016 b (14) West Pomerania Study, Poland	Analyze relationships between AMH and AMH receptor type 2 genotypes, hormone levels and menstrual cycle in Polish women in late reproductive stage	Healthy Polish women (N=345, mean age 42+/−4.5 years) provided bodyweight, height, serum samples analyzed for AMH, FSH, and E2 and genetic samples from DNA analyzed for AMH rs10407022) and AMHR2 (rs2002555 and rs11170547) polymorphisms	No relationships between FSH, E2, AMH and gene variants of AMH and AMHR2 in LR stage in healthy women	Women were classified as being in the LR stage based on results of FSH and AMH testing. (See Jurczak et al, 2016 a above)
Szkup et al, 2018 (15) West Pomeria Study, Poland	Analyze lipid and hormone metabolism, body mass index, and age in LR stage women in relation to smoking	Healthy LR stage women (mean age 42+/−4.5 years, n=345) living in Poland, 13 % smokers provided blood samples for lipid studies, hormone levels during the early follicular phase and completed the Primary Care Evaluation of Mental Disorders (PRIME-MD) and provided blood samples assayed for total, HDL and LDL cholesterol, triglycerides and had assessment of BMI	Smokers had significantly < l HDL, > LDL and triglyceride and E2 and FSH. Among non-smokers age predicted increased total cholesterol, LDL, triglycerides, and AMH and BMI predicted lower HDL and triglyceride levels. Among smokers age associated with > levels of E2 and < AMH; BMI associated with lower HDL.	Women were classified as being in LR stage based on their levels of AMH and FSH. FSH levels included as “normal” were follicular levels, 3.5–12.5 mIU/ml referencing STRAW+10 criteria.

Figure 1.

Flow Diagram for Search using “Late Reproductive Stage”

A second related search used the terms “late reproductive age” (LRA) and “women“. This yielded 70 citations. Abstracts published in English and reporting original research findings were reviewed. They reported studies of healthy women, based on cross-sections of populations of late reproductive age women, and emphasized the ways in which late reproductive stage and age were related, including efforts to develop staging systems and biomarkers associated with LRS and LRA. Studies of populations of women who were diagnosed with health problems, such as diabetes or polycystic ovary syndrome, were excluded as were treatment trials for such diseases or assisted reproductive technology. The studies identified from abstracts and included in this section of the review (n=26) are summarized in Table 2. Of the 70 abstracts screened, 44 were excluded from further review for the following reasons: 4 were not in English, 6 were reviews, commentaries or erratum notices, 17 focused on diseases and 16 focused on therapies for diseases among this population, and 2 focused on assay development. (See Figure 2 for sampling of studies)

Table 2.

Research Reports Identified in PubMed Search of “Late Reproductive Age” Meeting Review Criteria (N=26)

Author (year)/ Study Name/ Country	Purpose	Study Design and Methods	Results	Definitions of Late Reproductive Age
Manson (2001) (18) Penn Ovarian Aging Study (POAS) USA	Evaluate racial differences in reproducibility of hormone levels over time (E2, DHEAS, FSH and testosterone) adjusting for covariates	Healthy lLR age community-based sample including African American (N=202) and White (N=208) women. Hormone levels assessed early follicular phase, 4 times over 9 months. Excluded if pregnant or breast-feeding; using hormone therapy, hormonal contraceptives, psychotropic medications, history of illness potentially comprising hormonal function, history of alcohol or drug abuse in past year, did not speak English.	African American women had significantly lower E2 and DHEAS levels with increasing age and significantly decreased levels of E2 and increased DHEAS with increasing BMI. No racial differences in reproducibility of hormone measures.	Women between 35 and 47 years having menstrual cycles every 22–35 days for prior 3 months, intact ovaries and uteri.
Hollander et al, 2001 (35) POAS USA	Estimate prevalence of perceived poor sleep in women 35–49 years of age and correlate sleep quality with gonadal steroids and predictors of poor sleep	Black (n=218) and White women (n=218) 37–49 years of age with regular menstrual cycles identified through random digit dialing for longitudinal study of ovarian aging correlates. Data obtained at 4 assessment periods over 2-year interval, collected between days 1 and 6 of menstrual cycle. Women rated sleep quality with St. Mary’s Hospital sleep questionnaire at each assessment period and provided blood sample for hormone levels (E2, FSH, LH, testosterone, DHEAS) and clinical, behavioral, and demographic variables.	17% reported poor sleep at each assessment period. Hot flashes, higher anxiety levels, higher depression levels, greater caffeine consumption, and lower estradiol levels in women 45–49 were significant correlates of poor sleep.	35–49 years of age, regular menstrual cycles.
Cramer et al, 2002 (19) Harvard Study of Moods and Cycles USA	Assess FSH and estradiol as markers for ovarian reserve in women of LR age	Women 36–45 years of age with spontaneous periods (n=406 non-depressed) measured at entry, 6 and 12 months later.	FSH and LH increased over 12 months, but considerable variation in FSH and estradiol, often from cycle to cycle. Age, smoking, and shorter cycle length during adolescence associated with > FSH. Older age and < BMI associated with > LH and more ovulatory cycles associated with > estradiol levels.	Women 36–45 years of age with spontaneous periods
Freeman et al, 2003 (38) POAS USA	Determine psychometric properties of a brief menopause symptom checklist	African American and White women 38–52 years (N=350) completed a 12 item menopause symptom list (MSL)	MSL items demonstrated internal consistency. 3 factors identified: psychological, somatic, and vasomotor (VMS). Multivariate analysis indicated symptoms increased over time and factors were differentially associated with menopausal status. Psychological symptoms (irritability, anxiety, sad, mood swings, poor concentration, memory, trouble sleeping, headaches) > in premenopause and ET, < in LT and PM groups. VMS > in ET and LT-postmenopausal groups. Somatic symptoms (urine leaks, vaginal dryness, decreased interest in sex) > with time, not menopausal status. Psychological symptoms correlated with depressed mood and anxiety measures	Premenopausal - regular menstrual cycles in 22–35 day range; ET - within participant change of 7 days or more for at least 2 cycles; LT - 3–11 months amenorrhea; PM - 12 months of amenorrhea/no hysterectomy.
Freeman et al, 2005 (20) POAS USA	Characterize relationship between follicular phase hormone levels and menstrual bleeding pattern in approach to menopause; identify racial differences in hormone levels; determine independent contributions of menstrual status, race, age, BMI and smoking to hormone levels	Women 35–47 years of age with regular menstrual cycles provided blood samples twice in each of 7 assessment periods during days 1–6 of cycle, interviewed about menstrual dates and daily symptom reports, anthropometric measures and standardized questionnaires at each assessment.	Mean E2, FHS, inhibin B and LH differentially associated with menstrual status groups defined by bleeding pattern changes. Changes in hormone levels occurred before missed menstrual cycles for inhibin B, FHS, and LH. All interacted with menstrual status and BMI. African American women had lower levels of E2 and LH compared to White women. Race, menstrual status, and BMI interaction significant for E2. African American women had lower E2 levels until PM when E2 levels were > in those with BMI >25 and BMI >30.	Menstrual status defined: premenopause – regular menstrual cycles in 22–35 day range; late premenopausel – change in cycle length >7 days in one menstrual cycle compared to baseline; ET – change in cycle length >7days in 2 or more cycles; LT – 3–11 months of amenorrhea; postmenopausal - >12 months of amenorrhea, no hysterectomy.
LaMarca et al, 2005 (21) (Italy)	Determine AMH levels in “menopausal” (PM) women, confirm source of AMH in cycling women and disappearance after removal of source	Women (n=47) in late reproductive age (n=24, mean age 44 years); menopausal women (n=14, mean age 56 years) and regularly cycling women undergoing surgical menopause (n=9, mean age 43 years) provided blood samples assayed for AMH	AMH undetectable in 13/14 postmenopausal but only 2/24 LR age women. AMH undetectable in women having oophorectomy 3–5 days after surgery.	Women defined as menopausal (meaning PM) if they were mean age 56 years had FSH >40 IU/L, mean time from last menses 18 months (12–29 mo range) and not taking hormone therapy. LR age women had regular menstrual cycles, not taking hormones and mean age 44 years
Phillips et al, 2007 (35) Harvard Study of Moods and Cycles USA	Determine association between alcohol consumption and onset of perimenopause	Women (n=502) aged 36–45 years self-reported perimenopausal symptoms over 5 years. Alcohol consumption measured with food frequency questionnaire at enrollment	No association of total alcohol and time to perimenopause. Inverse association between red wine consumption and perimenopause onset, especially among never smokers. Positive association observed with liquor consumption. No association for white wine or beer regardless of smoking.	Onset of perimenopause defined as absolute change of 7 days or more in menstrual cycle length relative to baseline; change in menstrual flow by two flow categories or more (e.g. light or moderately light to moderately heavy or heavy) and absolute change in duration lasting at least 2 days relative to baseline; or periods of amenorrhea for 3 months or more.
Freeman et al, 2007 (22) POAS USA	Describe AMH levels in healthy LR age women and differences in AMH in obese vs non-obese women	Women (n=122) selected from POAS cohort to compare BMI, menopausal status, age, and race effects on AMH	AMH levels 65% lower in obese women vs non-obese. AMH levels lower in menopausal transition vs premenopause and lower in women > 40 vs 35–39 yrs. BMI significantly associated with AMH, lower in obese women in multivariable model controlling for menopausal status, age, race, and cycle day.	Definitions adapted from STRAW Premenopausal – regular cycles in 22–35 day range; Late premenopausal – change in cycle length >7 days in either direction compared to personal baseline value; ET – change in cycle length > 7 days observed for at least 2 cycles; LT – 3–11 months amenorrhea; PM - >12 months amenorrhea
Hale et al, 2007 (27) Australia	Describe endocrine features associated with STRAW stages	Healthy women 21–35 years and 45–55 years provided 3 blood samples per week for single menstrual cycle. MR age (n=21), LR age (n=16) and ET (n=16) and LT (n=23)	Anovulatory cycles identified, most in LT stage. Ovulatory cycle FSH, LH and E2 levels > with progression to LT stage and < mean luteal phase progesterone. Early cycle inhibin B < across STRAW stages, undetectable during longer ovulatory and anovulatory cycles in MT. AMH decreased markedly across STRAW stages.	STRAW (2001) stages −5 and −4 and 21–35 years with regular menstrual cycles= MR age; women 45–55 classified by STRAW criteria for −4 and −3 = LR age with regular menstrual cycles; ET stage with variable length cycles with 7 days difference in consecutive cycles STRAW −2; LT with one intermenstrual interval >60 days as STRAW −1.
Su et al, 2008 (32) POAS USA	Examine association between obesity and serum and ultrasound measures of ovarian reserve in LR age women	Healthy women aged 40–52 (n=36) with BMI <25 or BMI > 30. Provided early follicular phase blood samples, anthropometric measurement and transvaginal ultrasound. Serum AMH, inhibin B, estradiol, FSH and ultrasound ovarian volume and antral follicle counts determined.	Antral follicle count and ovarian volumes did not differ between groups. AMH and inhibin B < in obese women, FSH and E2 did not differ. AFC significantly related to AMH, but not inhibin B. Inhibin B secreted from gonadotropin-dependent recruited follicles may be less consistent measure of ovarian reserve pool.	No staging system reported
Robertson et al, 2008 (24) Australia	Characterize menstrual cycles in LR age women and in MT based on serum hormone levels	Women 45–55 years (n=55) compared with women 21 to 35 years of age (n=21) with respect to serum levels of estradiol, progesterone FSH, LH, inhibin A and B, and AMH	3 types of ovulatory cycles identified: Type 1 - similar hormone levels to women aged 21–35 years except for 20-fold lower AMH levels. Type 2 cycles - increased FSH, decreased inhibin B, and increased FSH to inhibin B ratios, with normal estradiol and progesterone levels. Type 3 cycles - similar to Type 2 with lower luteal phase progesterone and increased LH. FSH to inhibin B ratios and AMH early indicators of MT. Type 1 cycles present in STRAW −4, −2, and −1; type 2 in STRAW −3, −2, and −1; Type 3 in STRAW −2 and −3 and anovulatory in STRAW −1 only.	STRAW stages −5 and −4 and 21–35 years with regular menstrual cycles = MR age; women 45–55 classified by STRAW criteria for −4 and −3 = LR age with regular menstrual cycles; ET stage - variable length cycles with 7 days difference in consecutive cycles STRAW −2; LT stage with one intermenstrual interval >60 days as STRAW −1.
Tehrani et al, 2009 (23) Tehran Lipid and Glucose Study Iran	Assess capability of single measurement of AMH to predict menopausal status in LR-aged women	Women aged 40–50 years with regular menstrual cycles (n=147) assessed 3 times at 3 year intervals and serum AMH measured	AMH predicted not reaching menopause within 6 years with 88% accuracy. AMH over .39 ng/ml had positive predictive value of .90 and negative predictive value of .76 for not becoming postmenopausal. Results were similar for women 40–44 and 45–49 years	No staging system reported; used WHO definition of menopause = absence of menses for more than 12 months.
Hale et al, 2009 (28) Australia	Explore causes of erratic changes in E2, within-cycle secretion patterns of E2, progesterone, FSH, LH, inhibin A and B explored	Blood samples obtained 3X/week over 1 1/3 menstrual cycles from women 21 to 55 years (n=77), classified as MR age (n=21), LR age (n=16), ET (n=17) and LT (n=23)	11/29 ET and LT ovulatory cycles exhibited a second increase in E2 during mid and late luteal phases that continued to peak during subsequent menstrual phase. Second E2 rise was superimposed on existing ovulatory cycle during the luteal and menstrual phases, termed a luteal out of phase (LOOP) cycle. Compared to typical ovulatory cycles, LOOP cycles had lower luteal phase progesterone, higher early cycle FSH and lower inhibin B and were associated with abnormally short (<21 da) or long (>40 da) cycle length.	STRAW stages −5 and −4 and 21–35 years with regular menstrual cycles= MR age; women 45–55 classified by STRAW criteria for −4 and −3 = LR age with regular menstrual cycles; ET stage with variable length cycles with 7 days difference in consecutive cycles STRAW −2; LT stage with one intermenstrual interval >60 days as STRAW −1.
Plante et al, 2010 (36) USA	Determine extent to which smoking influences ovarian reserve as measured by AMH levels	Community sample (n=284) aged 38–50 years completed questionnaire with detailed smoking history and provided serum AMH assayed from days 2,3,4 of menstrual cycle	Geometric mean AMH levels were 6.7, 2.7 and 1.3 pM for women 38–42, 43–45, and 46–50 yrs respectively. Current but not past smokers had 44% lower AMH levels. Smoking impact on AMH not dose dependent and prenatal exposure to smoking had no effect on AMH. Results suggest effect of smoking on depletion of antral but not primordial follicles.	No classification system for reproductive aging stages nor definition of LR age reported.
Hale et al, 2010 (29) Australia	Measure menstrual blood loss before and during MT and explore relationships of blood loss to menstrual cycle irregularity and reproductive hormone levels	Menstrual blood loss measurements performed in 2 consecutive cycles in 77 healthy women 21–55 years (MR age (n=21), LR age (n=17), ET (n=16) and LT (n=23). Estradiol, progesterone, FSH, LH, and inhibins measured 3 times per week from start of one menses to end of subsequent menses. Women collected all used pads and tampons during the entire menstrual period at the beginning of cycle 1 through cycle 2. Blood samples collected 3 X weekly	Median and range of menstrual blood loss increased from MR age to LT stage, but was significantly > in ovulatory cycles in the late MT group. Highest levels of menstrual blood loss were in LT group with ovulatory cycles with abnormally high E2 levels and disturbed E2 secretion patterns, e.g. LOOP cycles. Menstrual blood loss for LR age women was similar to that for MR age.	STRAW stages −5 and −4 and 21–35 years with regular menstrual cycles = MR age; women 45–55 classified by STRAW criteria for −4 and −3 = LR age with regular menstrual cycles; ET with variable length cycles with 7 days difference in consecutive cycles STRAW −2; LT stage with one intermenstrual interval >60 days as STRAW −1.
Robertson et al, 2011 (24) Australia	Assess pattern of serum AMH across normal ovulatory menstrual cycle in women in LR age when ovarian follicle reserve and AMH are reduced	Serum AMH levels measured with enzyme linked immunosorbent assay across ovulatory menstrual cycle from women in MR (n=18) and LR (n=43) age, including MT. Serum LH, FSH, estradiol, progesterone, inhibin A, inhibin B, and AMH measured	No intracycle variation in AMH level in MR women nor in 33% of women with normal ovulatory cycles in LR age. Two-fold decrease in AMH in 11 cycles and 4-fold increase in AMH in 10 cycles observed between follicular and luteal phase. AMH below detection level in 8 ovulatory cycles. Separate patterns of AMH detected in follicular and luteal phases of ovulatory menstrual cycles.	MR age women 21–35 years with regular menstrual cycles provided control for women 45–55 years of age with variable cycle characteristics. Older women classified using STRAW criteria: STRAW stage −3 (LR age with regular menstrual cycles); STRAW −2 ET with variable length cycles with difference of >7 days in consecutive cycles; Straw stage −1 (LT with at least one intermenstrual interval of >60 days)
Freeman et al, 2012 (25) POAS USA	Evaluate predictive value of AMH levels in determining median time to menopause and predictive ability of AMH vs FHS and inhibin b	401 LR age women participated and were followed for 14 years (1996–2010)	Premenopausal women at baseline (mean 41 yrs) median AMH of .68 ng/ml at baseline; AMH predicted time to menopause and age improved prediction. AMH stronger predictor of time to menopause than FSH or inhibin b.	Not defined. LR years used as descriptor
Butts et al, 2012 (39) POAS USA	Investigate relationship between smoking and hot flash occurrence as function of genetic variation in sex steroid-metabolizing enzymes	POAS participants median age 50.6 years, 47% African American (n=137) and 53% European American (n=128), provided blood samples analyzed for SNPs: COMTVal158Met (rs4680), CYP1A2*1F (rs762551), CYP1B1*4 (Asn452Ser,rs1800440), CYP1B1*3 (Leu432Val, rs1056836), and CYP3A4*1B (rs2740574) and rated hot flashes (occurrence, frequency, severity in last month)	European-American smokers with COMTVal158Met double-variant increased odds of hot flashes vs nonsmokers; European American heavy smokers with COMTVal158Met double variant carriers had more frequent moderate or severe hot flashes vs nonsmokers; European-American CYP 1B1*3 double-variant carriers more frequent moderate or severe hot flashes than non-smokers. African American smokers single variant CYP 1A2 carriers more likely to report hot flashes than nonsmoker carriers.	Used staging system for reproductive aging validated in earlier reports. Premenopausal = regular menstrual cycles in 22–35 day range; late premenopausal= change in cycle length of 7 days in either direction compared with personal baseline at enrollment and observed at one cycle in study; ET = change in cycle length of 7 days and observed for 2+ cycles in study; LT = 3–11 months amenorrhea with no hysterectomy; PM =12 months amenorrhea with no hysterectomy. Categories were collapsed into premenopausal, transition, and postmenopausal.
Freeman et al, 2012 (26) POAS USA	Determine rate of change of AMH in LR years and association with time to menopause	293 women with 2 measures of AMH evaluated using survival analysis to menopause	Rate of AMH change was strong independent predictor of time to menopause (no bleeding for 12 months) after adjusting for AMH baseline, age, and smoking. Fast rate of AMH change associated with shorter time to menopause. Significant interaction of AMH rate of change and age, such that faster decrease in AMH associated with increased risk of menopause in women 35–39 years of age; less dramatic but significant associations in women 40–44 and 45–48 years.	Not defined
Vanden Brink et al, 2013 (31) Canada	Test hypothesis that major follicular wave dynamics differ in women with age	58 women of MR age (18–35 yrs), LR age (LRA 36–44 yrs), and advanced reproductive age (ARA - 45–55 yrs) participated in ultrasonographic counts of number and diameters of all follicles every 2–3 days for one complete interovulatory interval (IOI). Antral follicle count and major follicular wave dynamics during IOI were compared among age groups	1 or 2 major follicular waves observed during the IOI in all women. Presence of luteal phase dominant follicles and follicular phase dominant follicles during IOI did not differ with age. LPDFs emerged earlier relative to ovulation, grew longer, and developed to larger diameter in advanced reproductive age group versus MR and LR age groups. Tendency for greater prevalence of poly-ovulation as women aged, e.g. ARA > MRA.	LR age defined as 36–44 years
Freeman et al, 2014 (37) POAS USA	Identify within-woman changes in depressive symptoms over a 14 period around time of menopause, and associations of a history of depression with symptom patterns and rate of change in reproductive hormones as predictors of depressive symptoms	203 generally healthy women premenopausal, cohort of LR age women followed for 14 years completed CESD depression scale	Risk of high CESD scored decreased from 10 years before menopause to 8 years postmenopause. Risk of depressive symptoms greater before and lower after the FMP. Women with history of depression 13 times greater overall and 8 times greater postmenpause vs women without depression history. Women first experiencing depressive symptoms approaching menopause had declining risk after FMP. Risk of depressive symptoms PM decreased by 35% for each unit increase in rate of FSH change before FMP.	LR age not defined
Butts et al, 2014 (40) POAS USA	Evaluate associations between variations in genes involved in the metabolism of environmental chemicals and steroid hormones and risk of menopause in smokers	410 women of late-reproductive-age provided leukocyte samples for SNPS: COMTVal158Met, CYP1B1*4Asn453Ser, CYP1B1*3Leu432Val, CYP3A4*1B and smoking	Interactions between smoking and SNPS (CYP3A4*1B, CYP1B1*3 in European-American women were associated with greater risk of menopause	LR age not defined. Time to menopause was outcome assessed.
Zheng et al, 2015 (36) Study of Women’s Health Across the Nation (SWAN) USA	Evaluate patterns in actigraphy-defined sleep measures across the menstrual cycle	163 women (58 African American, 78 White, 27 Chinese) of LR age (mean 51.5 years) provided daily measures of sleep (efficiency and total sleep time) and movement during sleep using wrist actigraphy across a menstrual cycle or 35 days, whichever was shorter	Sleep efficiency percentage declined across menstrual cycle, but was pronounced in the premenstrual week. Sleep efficiency declined by 5% total sleep time was 25 minutes less from the 3rd to the 4th week of the cycle	LR age not defined
Jurczak et al, 2015 (42) West Pomerania Study Poland	Analyze influence of genetic and hormonal factors on cincidence of depressive symptoms in LR age women	347 healthy LR age women, mean age 42 years completed Beck Depression Inventory, provided samples for FSH and AMH, PCR and genetic testing	s/s genotype of 5-HTTLPR polymorphism and 3/3/ genotype of 30-bp VNTR polymorphism in MAO-A promoter region did not contribute to depressive symptoms; AMH not related to depressive symptoms	LR age determined using STRAW+10 criteria for −3b stage: menstrual cycles regular and FSH and AMH level low
Ruth et al, 2019 (41) United Kingdom	Explore the role of genetic variation in determining AMH levels in women of late reproductive age	LR age women (median age 44 −48 years) at blood draw (N=3344) had data included in genome-wide meta-analysis	Gene variant rs16991615 in MCM8 associated with age at menopause in earlier studies reached genome-wide significance. Genetically-predicted age at menopause was associated with lower AMH levels in age-adjusted analyses (0.18 SD per one-year earlier age at menopause.	Not defined LR age defined as median age between 44–48 across samples
Jurczak et al, 2019 (43) West Pomerania Study Poland	Analyze stress-coping styles and personality traits in healthy LR age women with regard to genetic factors	345 healthy LR age women, mean age 42 years, completed Coping inventory for Stressful Situations, Neuroticism-Extraversion-Openness Five-Factor Inventory, genetic testing for 30-bp VNTR polymorphism in MAO-A promoter region	Personality traits correlated with choice of task-oriented coping and emotion-oriented coping. No direct influence of genetic factors on choice of stress-coping style.	LR age classified based on FSH and AMH levels

Figure 2.

Flow Diagram for Search using “Late Reproductive Age”

Given our limited results, we devised an additional strategy: searching within reports of longitudinal studies of women during the menopausal transition published shortly before and after the introduction of the term LRS was introduced and defined by the STRAW workshop (1). We located studies that followed cohorts of midlife women over the course of the menopausal transition, some of which reported data using the STRAW criteria. This process led to the identification of additional articles from longitudinal studies, including the Melbourne Women’s Midlife Health Project (MWMHP), Penn Ovarian Aging Study (POAS), the Study of Women and Health Across the Nation (SWAN), the Seattle Midlife Women’s Health Study (SMWHS), Massachusetts Women’s Health Study (MWHS) among others. In searching for publications related to these studies, we used the search terms: “late reproductive stage” and “STRAW criteria”, “stages of reproductive aging” and symptoms (e.g. hot flashes, night sweats, depressed mood, anxiety, pain, and cognitive, vaginal, urinary and somatic symptoms); menstrual bleeding patterns; and women’s experiences of menopause.

Data Extraction

Once articles were identified from these approaches, they were reviewed by two investigators to insure eligibility for inclusion in the review. In cases of disagreement, the investigators discussed their inclusion until unanimity was achieved.

Definitions provided in the STRAW reports (1,2) were considered prior to reviewing the articles identified by the searches. Data were extracted from the articles reviewed with particular emphasis on the definitions of LRS and related classification criteria used to guide the research. In addition, the study name, purpose, methods (population, design, measures, and analysis) and results were summarized. (See Tables 1 and 2). Authors discussed and concurred on the data extraction entries in the Tables as well as the identification of gaps in existing literature about the late reproductive stage and recommendations for future research.

Results

1. Publications Referencing the “Late Reproductive Stage”

We identified 9 unique publications focusing on the “late reproductive stage” (LRS) in the initial search using various search engines (See Table 1). Three papers were based on data from the Seattle Midlife Women’s Health study (SMWHS) (10–12), three from the West Pomerania study (13–15), and remaining papers included one each from Tokyshima University Hospital (16), Rochester Investigation of Cognition across Menopause (RICAM) (9), and the UCSF Women’s Health Study (17). Studies were conducted in the US (n=5), Poland (n=3), and Japan (1).

1.1. Topics Studied in relation to the Late Reproductive Stage.

The identified research reported relationships between menstrual cycle patterns and hormone levels (14); hormone patterns/levels and genotypes (14); hormone or cycle patterns and other physiologic correlates (16); hormone patterns and lipids (15); lifestyle patterns and reproductive hormones, e.g. hormone metabolism and smoking (15); cognitive function during stages of reproductive aging (9); symptoms and gene polymorphisms, (13); symptoms and lifestyle factors associated with bladder symptoms and consequences of urinary incontinence (17); and clustering of symptoms during the LRS as well as the early and late MT and early PM stages (10,11).

1.2. Symptoms and Correlates.

Studies of LRS focused on a variety of symptoms and their correlates. Weber and colleagues investigated differences in cognitive function across stages of reproductive aging, finding that women in the LRS from the Rochester Investigation of Cognition across Menopause (RICAM) Cohort performed better than PM women on verbal learning, verbal memory, and motor function (9). Jones and colleagues examined urinary incontinence symptoms, noting that incontinence and nocturia were more prevalent in the LRS than during the MT and also identified racial/ethnic differences in symptom experiences (17). A second study examined effects of urinary incontinence on mood, perceptions of oneself, attitudes toward midlife, and consequences for daily living and demonstrated both SUI and UUI were associated with lower self-esteem and sense of mastery (12). Two reports addressed symptom clusters (hot flashes, sleep, cognitive, mood, pain and tension) women experienced during the LRS and MT (10). Types of symptoms women experienced in the LRS (mood/cognitive symptoms, pain and somatic symptoms) predicted occurrence of the same symptom groups during the early menopausal transition stage (11).

1.3. Gene Variants, Hormonal Changes, and Late Reproductive Stage Changes

Results from the West Pomerania Study revealed that anxiety and mood disorders were not associated with 5-HTT and MAO gene variants among these women during the LRS (13). Investigators also found no relationships between AMH and AMHR2 gene variants and FSH, estradiol, and AMH during the LRS. (14). Two studies focused on physiological factors other than reproductive hormones during the LRS. A single study investigated relationships of circulating cytokines, MCP-1 and chemokines, in the LRS versus postmenopause (PM) (16). Lipids, triglyceride, estradiol and FSH levels were compared for smokers vs non-smokers in this population (15).

2. Publications Referencing “Late Reproductive Age”

In our second search phrase we employed “late reproductive age” (LRA). This revealed 26 publications meeting inclusion criteria. (See Figure 2 and Table 2). Notably in these articles, investigators used a variety of age ranges and other criteria to define LRA. Some used age categories, such as from 33–49 years, others required that women report having regular menstrual cycles and some incorporated elements of the STRAW stages.

Topics addressed in articles meeting our LRS inclusion criteria focused on a wide variety of subjects (see Table 2). The largest proportion of these focused on AMH (n=10) and AMH and hormones in relation to menstrual cycle patterns (n=19). Studies were conducted in the US (n=15), Australia (n=5), Poland (n=2), and Italy, Iran, Canada, and UK (1 from each country).

2.1. Hormone Patterns during Late Reproductive Age.

In the period since STRAW 2001, investigators have searched for biomarkers to predict menopause i.e. the final menstrual period or time to menopause. In addition to ovarian steroids and pituitary hormones (FSH, LH) (18, 19), researchers explored patterns of inhibins (A and B) and anti-Mullerian hormone (AMH). Examples include Freeman and colleagues’ (20) efforts to identify changes in levels of inhibin B, FSH and LH occurring before women experienced skipped menstrual cycles. Also, the POAS investigators identified reproductive endocrine differences between African American and European American women.

La Marca and colleagues (21) found that AMH was undetectable in only 2 of 24 LRA women (defined as having a mean age of 44 years, regular menstrual cycles, and not using hormone therapy), compared to 13 of 14 postmenopausal women and 9 of 9 women tested 3–5 days after oophorectomy. Freeman and colleagues (22) identified that AMH levels were 65% lower in obese compared to non-obese women, in women in the menopausal transition compared to premenopause (defined as having regular cycles in 22–35 day range), and in women older than 40 years compared to those 35–39 years of age. Tehrani and colleagues (23) found that among women 40–49 years of age with regular cycles, AMH levels over .39 ng/ml had positive predictive value of .9 and negative predictive value of .76 for identifying women not reaching menopause within the next 6 years.

Robertson and colleagues (24) assessed patterns of serum AMH across ovulatory menstrual cycles among LRA women (45–55 years of age with variable cycle characteristics). In contrast to mid-reproductive age women who experienced no intracycle variation in AMH, women of LRA experienced 3 distinct patterns of AMH: no change across the cycle; increase within the follicular phase; and increase within the luteal phase. These findings suggested AMH variations reflect patterns of developing antral follicles. As anticipated, AMH levels were below detection levels in anovulatory cycles.

Freeman and colleagues (25) evaluated the predictive value of AMH levels for median time to menopause, comparing AMH to FSH and inhibin B. AMH levels predicted time to menopause and age improved the prediction. AMH was a stronger predictor of time to menopause than FSH or inhibin B. In a later evaluation Freeman and colleagues (26) found that rate of AMH change was a strong independent predictor of time to menopause after adjusting for AMH baseline, age, and smoking. Taken together, these studies offer support for declining AMH levels as a biomarker identifying progression to menopause especially when measures are standardized to menstrual cycle phase, e.g. early follicular phase.

2.2. Menstrual Cycle Patterns and Stages of Reproductive Aging.

Some investigators have studied LRA women integrating the STRAW staging system as a framework. Hale and colleagues (27) described endocrine patterns based on blood samples women provided 3 times per week for a single menstrual cycle. They compared ovulatory to anovulatory cycles and associated STRAW stages with age. The four age groups were: women 21–35 years of age, termed “Mid-reproductive age” (MRA), comparable to STRAW Stages −4 and −5 (n=21); women 45–55 years of age, LRA with regular menstrual cycles, somewhat but not precisely comparable to STRAW −3 (n=16); “Early Menopausal Transition”, STRAW −2, women with consecutive cycle length differing by more than 7 days (n=16); and “Late Menopausal Transition,” STRAW −1, for those with at least one intermenstrual interval of 60 days or longer (n=23). Distinct hormonal profiles were identified for each group. Ovulatory cycle length and mean follicular phase length were shorter among the LRA than MRA group. FSH levels were higher in the LRA than the MRA group and AMH decreased from MRA to LRA as did early cycle inhibin B.

Hale and colleagues (28) also identified a menstrual cycle pattern in which estradiol levels were erratic, including a second increase in estradiol during mid and late luteal phases that continued to peak during a subsequent menstrual phase, which they termed luteal out of phase (LOOP) cycles. These cycles were associated with abnormally short (<21 days) or long (>40 days) cycle length, and most likely to be observed during the MT. None occurred among MRA or LRA cycles.

Menstrual blood loss varies across the lifespan but has not been characterized strictly in relation to cycles and hormone levels until recently. Hale and colleagues (29) found that menstrual blood loss measurements from 2 consecutive cycles in women in MRA, LRA (45–55 years of age in STRAW −4 and −3), early menopausal transition, and late menopausal transition varied, increasing from the mid-reproductive age to the late menopausal transition stage. Menstrual blood loss was significantly greater in ovulatory cycles in the late MT group. Highest levels of blood loss were observed in the late MT group in ovulatory cycles with abnormally high E2 levels, such as in the LOOP cycles. Blood loss was similar in mid- and late reproductive age groups and lower than during the MT stages.

Investigators have attempted to create or elaborate on staging systems based on both menstrual cycle attributes such as cycle regularity, length, amount of flow, and hormonal patterns linking to or approximating the STRAW criteria. As one example, Robertson and colleagues (30) characterized menstrual cycles in LRA women based on serum hormone levels in women 45–55 years of age. They identified 3 types of cycles including:

Type 1 cycles – similar to cycles of younger MRA women, with 20-fold lower AMH levels;

Type 2 cycles – increased FSH, decreased inhibin B and increased FSH to inhibin B ratios, and lower AMH levels than type 1 cycles;

Type 3 cycles similar to Type 2 with lower luteal phase progesterone and increased LH, and lower AMH levels than seen in type 1 cycles.

Type 1 cycles were present in 100% of STRAW stage −4 (peak reproductive) cycles, but also in stage −2, and −1; Type 2 in STRAW stages −3, −2, and −1; Type 3 in STRAW −2 and −1; and anovulatory cycles in STRAW −1 only. LRA groups may experience cycles resembling those of women in STRAW MR or LR stages.

2.3. Ovarian Reserve.

Further efforts to understand reproductive aging included a study of follicular wave dynamics in women of mid- and late reproductive age (18–35 and 36–44 respectively) compared with women of advanced reproductive age (45–55 years) who participated in ultrasonographic counts of the number and diameters of all follicles every 2–3 days for one complete ovulatory cycle. Vanden Brink and colleagues (31) found luteal phase dominant follicles emerged earlier relative to ovulation, grew longer, and developed to larger diameter in the advanced reproductive age group vs middle and LRA groups. There was a tendency for greater prevalence of polyovulation as women aged. Su and colleagues (32) examined the association between obesity and serum and ultrasound measures of ovarian follicle reserve in LRA women 40–52 years of age. AMH and inhibin B were lower in obese women, but FSH and E2 did not differ. Antral follicle counts were associated with AMH, but not with inhibin B.

2.4. Health-related Behavior (Smoking and Alcohol Use) and Hormone Patterns.

Health-related behaviors, such as smoking and alcohol use, affect reproductive hormonal patterns. Phillips and colleagues (33) studied women 36–45 years of age, measuring alcohol consumption with a food frequency questionnaire and subsequent experience of perimenopause (absolute change of 7 days or more in menstrual cycle length relative to baseline in contrast to STRAW criterion of persistent changes between contiguous cycles) over a 5 year period. Type of alcohol intake had differential effects on perimenopause onset: red wine consumption was associated with delayed onset of perimenopause, and liquor consumption was related to earlier onset of perimenopause. Plante et al (34) explored the extent to which smoking influenced ovarian reserve and therefore reproductive aging stage. Current smokers had 44% lower AMH levels than their counterparts who did not smoke.

2.5. Symptoms during Late Reproductive Age.

Although investigators have examined symptoms during the MT in several cohorts, few have documented symptoms during LRA. Hollander and colleagues (35) estimated the prevalence of perceived poor sleep in women 35–49 years of age with regular menstrual cycles and correlated sleep quality with gonadal steroids and other predictors. Both African American and White women enrolled in the POAS participated in assessments on 4 occasions over a 2-year interval conducted on days 1–6 of the menstrual cycle to coincide with endocrine sampling. They rated sleep quality at each assessment and provided blood samples assayed for E2, FSH, LH, T, and DHEAS. Seventeen percent reported poor sleep at each assessment period. Hot flashes, higher anxiety levels, higher depression levels, greater caffeine consumption, and lower E2 levels in women 45–49 years of age were significantly correlated with poor sleep. Zheng and colleagues (36) evaluated actigraphy-defined sleep patterns across the menstrual cycle in women (African American, Chinese, and White) of late reproductive age (mean 51.5 years). Sleep efficiency declined by 5% for LRA women and total sleep time was 25 minutes less from the 3^rd to the 4^th week of the cycle (premenstrual week).

Freeman and colleagues also administered the validated Menopause Symptom List (MSL) to a cohort of African American and White women aged 38–52 years from the POAS (37,38). The MSL includes three dimensions (psychological, somatic, and vasomotor). Psychological symptoms (irritability, anxiety, feeling sad, mood swings, poor concentration/memory, trouble sleeping, headaches) were most prevalent among women in the premenopausal group, which most clearly overlaps with LRS. Vasomotor and somatic symptoms were more prevalent during the MT and PM.

2.6. Genes and Late Reproductive Age Experiences and Characteristics.

Butts et al. (39) investigated the relationship between smoking and hot flashes as a function of genetic variation in sex steroid-metabolizing enzymes. They studied POAS participants with median age 50.6 years, with nearly half African American and half European American. They (40) also evaluated associations between variants in genes involved in metabolism of environmental chemicals and steroid hormones and risk of menopause in women of LRA. They found interaction effects between smoking and CYP3A4*1B and CYP1B1*3 in European American women associated with increasing risk of menopause (indicated by 12 months of amenorrhea).

Ruth and colleagues (41) explored the role of genetic variation in determining AMH levels in 3344 women of late productive age (44–48 years at blood draw) from populations included in genome-wide meta-analysis, identifying a gene variant (rs16991615) from those in/around the AMH gene associated with age at menopause. Genetically predicted age at menopause was associated with lower AMH levels in age-adjusted analysis, providing genetic support for use of AMH as a marker of ovarian reserve.

Jurczak and colleagues (42) studied healthy late reproductive age women (mean age 42 years), exploring the relationship between the 5-HTTLPR s/s polymorphism and VNTR polymorphism in the MAO-A promoter region and symptoms of depression. Neither polymorphism was related to depressive symptoms, nor was AMH.

Jurczak and colleagues (43) also studied stress-coping styles in healthy late reproductive age women in relationship to a 30-bp VNTR polymorphism in MAO-A promoter region and personality traits. There were no genetic factors identified influencing stress-coping style.

3. Extended Search Results

In order to ensure adequate coverage of published research we expanded our review to a third phase in which we identified longitudinal studies of the menopausal transition. A search of references cited in LRS and LRA publications as well as of published literature from 2000 to 2019 based on selected longitudinal study names was conducted. During the decade preceding the first STRAW conference in 2001, longitudinal studies of the menopausal transition and women’s health were launched in the US as well as in other countries. Among these were the Massachusetts Women’s Health Study (MWHS), the Melbourne Midlife Women’s Health Project (MWMHP), Seattle Midlife Women’s Health Study (SMWHS), Penn Ovarian Aging Study (POAS), and Study of Women’s Health Across the Nation (SWAN). Our review of these studies focused on conceptualization of stages of reproductive aging, with specific mention of the late reproductive stage (LRS) or late reproductive age (LRA); descriptions of endocrine levels and variability during LRS/LRA; and symptoms and related experiences women reported during the LRS/LRA.

3.1. Conceptualizing Stages of Reproductive Aging.

Our third search identified efforts to define stages of reproductive aging using indicators such as menstrual cycle characteristics (cycle length and regularity), hormonal patterns, and more recently combinations of hormonal and menstrual cycle characteristics. Studies reported prior to publication of the STRAW criteria relied typically on women’s retrospective histories of menses, although some used data from daily diary recordings or menstrual calendars.

Use of the term LRS did not appear until after publication of the STRAW 2001 criteria (1). Later some investigators attempted to incorporate the STRAW stages in their research, and other investigators differentiated women by age groups by focusing on those 35–55 or a subset of women in this age range. Identifying data definitively associated with the LRS in these studies was impossible.

3.2. Methods for Staging Reproductive Aging before and after STRAW.

In the decade preceding and following the first STRAW conference held in 2001 investigators employed a variety of approaches to staging reproductive aging (see Figure 3 which compares staging systems including those used for the Massachusetts Women’s Health Study (MWHS), Melbourne Midlife Women’s Health Project (MMWHP), SMWHS, SWAN, POAS, and FREEDOM). Although there were overlapping stages among the staging systems used in these studies, for example, early menopausal transition was defined similarly across studies, all stages did not correspond precisely. In particular, the LRS was neither consistently defined nor even identified in some studies. Immediately prior to the STRAW conference, Mitchell and colleagues (3) proposed a system for staging the menopausal transition based on data participants contributed in menstrual calendars, interviews and questionnaires. Three stages of the menopausal transition were proposed: “the early MT” was characterized by self-reported changes in flow amount, flow duration, and/or cycle length since age 35 in the presence of regular menstrual cycles for one calendar year; a middle MT stage corresponded to the STRAW definition of the early MT and a late MT stage was defined by persistent amenorrhea for 60 days or more. Mitchell has also characterized a pretransition stage as absence of self-reported changes in flow amount, flow duration, cycle length or cycle regularity since age 35. Because data required for identifying the early MT stage using the Mitchell criteria were not widely available to other investigators, the SMWHS investigators adopted the STRAW criteria for publications following 2001 (44).

Figure 3.

Comparison of Approaches to Staging Reproductive Aging (2000 – Present) with Emphasis on Late Reproductive Stage Using Women’s Observations as Anchor. Adapted with permission from Women Living Better Website August 2020 https://womenlivingbetter.org/menopause-related-changes/

Mansfield and colleagues (45) used data provided by participants contributing to the TREMIN Trust database (based on original work by Treloar and colleagues) to identify patterns of menstrual cycle changes during the 5 years preceding the final menstrual period based on menstrual calendars women kept for up to 12 years before they reached menopause. Mansfield and colleagues found that from one year to the next women experienced regularity followed by changing cycle length (Treloar and colleagues had originally reported a shortening of cycles as an early sign and lengthening of the cycle as a later sign of the menopausal transition) followed by menopause, although women sometimes reverted from changing cycle length to regular cycle length prior to menopause. These findings demonstrated the variety in women’s reproductive aging trajectories and exposed an aspect of the challenge of creating universally applicable criteria for staging reproductive aging.

Initially the SWAN investigators used questions about bleeding characteristics including timing, duration, and intensity as a basis for categorizing women aged 42–52 years as premenopausal, early perimenopausal, late perimenopausal and postmenopausal (46). Women’s cycles were considered premenopausal if menses had occurred in the past 3 months with no decrease in predictability, early perimenopausal if menses occurred in the past 3 months but had become less predictable. Using this approach, women in either the premenopausal or early perimenopausal stage might meet STRAW criteria for LRS (See Figure 3).

Gracia and colleagues (47) compared definitions of menopausal status between SWAN, STRAW, and the POAS with respect to hormone levels, including inhibin B, FSH, LH, and estradiol. The PENN-5 system distinguished the stages of reproductive aging that immediately preceded the beginning of the menopausal transition, which they termed premenopause and late premenopause: premenopause was indicated by regular cycles and no change in cycle length and late premenopause by a difference of 7 days or more in cycle length. Using the Penn-5 system, inhibin B decreased significantly from premenopause to late premenopause while FSH increased during this same period. Estradiol levels did not change. Changes in inhibin B and FSH levels were similar using criteria for early menopausal transition stage (STRAW) and early perimenopause (SWAN). LH values were not significantly different in premenopause compared to early perimenopause (SWAN), premenopause compared to early menopausal transition (STRAW) nor for the PENN-5 premenopause compared to late premenopause. These analyses illustrate differences in hormonal values when using different criteria for staging the menopausal transition. The PENN-5 system revealed differences in inhibin B and FSH levels that were not evident using either the SWAN or the STRAW staging criteria. Nonetheless, cycles from the STRAW LR stage could have been classified as either premenopausal or late premenopausal using the POAS staging system.

Harlow led a collaborative of investigators from the TREMIN, MWMHS, SMWHS, and SWAN in the ReSTAGE study to evaluate the STRAW recommendations for staging reproductive aging (48–51). The criterion for entry into the early MT, the upper boundary for the LRS, was validated at a persistent difference of 7 or more days in the length of consecutive cycles. However, the lower boundary of the LRS was not addressed.

3.3. Reproductive Hormone Patterns and Stages of Reproductive Aging

Since publication of the STRAW criteria (1), investigators for several cohort studies have focused on hormonal levels and their variability during the menopausal transition, with only a few reports focusing on the LRS. Of note is that while the STRAW criteria were being validated, groups of investigators were studying endocrine changes in women’s menstrual cycles as they approached menopause. Some used years prior to menopause and some used age as a marker but did not relate their observations to the STRAW stages. However, some were attempting to use the STRAW stages to classify women’s cycles and/or exploring variations on the STRAW Stages in addition to the Penn5 staging system described earlier (47).

In addition to work by Hales’ (27), Robertson’s (30), and Freeman and Gracia’s (47) teams presented in Section 2, Miro and colleagues (52) proposed 5 stages of reproductive aging based on unique endocrine data from women enrolled in the FREEDOM study in Great Britain. A cohort of 112 women collected daily urine samples for 6–18 months and recorded menstrual periods. Over 36,000 samples were analyzed for FSH, LH, estrone-3-glucuronide (EG), and pregnandiol-3-glucuronide (PdG). Women were recruited to represent different reproductive ages as a basis for developing a classification of five sequential endocrine stages of reproductive aging. Women in stage 1 (mean age 41 years, range 30–50) had regular menstrual cycles with evidence of ovulatory cycles, and low LH and FSH levels (mean day 1–5) FSH levels of less than 5 IU/L. Women in stage 2a had regular cycles with initial FSH 5–10 IU/L; in 2b initial FSH was > 10 IU/L and modest rises in LH and shorter cycle length. Stage 2a resembled STRAW LRS and Stage 2b STRAW early menopausal transition. Women in stage 3 had irregular cycles, with appearance of “delayed-response cycles” in which there was a delay in the first day of a sustained rise in estrogen (termed estrogen takeoff) following the FSH rise. Women in stage 4, had acyclical ovarian activity with no evidence of ovulation and luteinization; and in stage 5, ovarian quiescence and persistently raised gonadotropin levels were seen.

The shortening of the follicular phase and elevated FSH seen in Miro’s Stage 2a may correspond to the events of the LRS denoted by STRAW +10 criteria that included subtle irregularity of less than 7 days in cycle length and variable FSH levels. Onset of menstrual irregularity marked the boundary between stages 2 and 3. Stages 3a and 3b resembled STRAW early and late MT respectively. Stage 4 was characterized by lack of ovulatory cycles and the end of ovarian activity marked progression from stages 4 to stage 5.

3.4. Ovarian Reserve.

Since the STRAW+10 staging system was published, investigators have attempted to validate stages using observations of follicular types and their development. Hansen and colleagues examined follicle counts determined by ultrasound and simultaneous serum FSH, AMH, estradiol and inhibin B levels in a small cohort. They identified significant differences in each of the 4 STRAW stages studied (53). Specifically, they found that during STRAW stage −3 (LRS) there was a decrease in primordial follicles, non-growing follicles and antral follicle counts with increasing FSH and decreasing AMH and inhibin B levels.

3.5. Symptoms and Related Experiences.

Surprisingly few reports from large cohort studies focus on women’s experiences of symptoms sometimes attributed to menopause during the late reproductive stage. Freeman’s analyses of the POAS data provide prevalence estimates in stages defined as premenopause and late premenopause. These stages occur prior to the early menopausal transition (defined by POAS as a persistent 7-day difference in cycle length compared to baseline cycle length that occurred for at least 2 cycles). Of interest is that approximately 35% of women reported hot flashes during the premenopausal and 37% during the late premenopausal stages. Less than half described them as moderate or severe. Although approximately 40% reported poor sleep, less than half of these women reported moderate or severe sleep problems. The most commonly reported symptoms during the premenopausal and late premenopausal stages were aches, depressed mood, and poor sleep (54). Approximately 56% of women reported depression symptoms during the late premenopausal stage, but less than half this number described these as moderate or severe (54). The proportion of women reporting headache, irritability, mood swings, and difficulty concentrating was similar in premenopause, late premenopause and early MT. The proportion of women reporting anxiety rose from the premenopause and late premenopause to the early menopausal transition stage, and then declined in the late MT and postmenopause (55).

Participants in the MWMHP project classified in “premenopause” were more likely to report breast tenderness or soreness and urine control problems during this period than during early perimenopause (56). SMWHS participants reported a variety of symptoms became more frequent or severe as they progressed from the LRS to the MT stages: hot flashes (57) became more severe from LRS to the transition stages and PM. Depressed mood became more severe from the LRS to the MT, then decreased significantly during early PM (58). In contrast, memory problems (Memory Functioning questionnaire) occurred more frequently during the LRS and early MT vs late MT, including forgetting where they put things, recent phone numbers, things others had told them, and what they were doing (59). None of these was considered serious. In contrast, women rated night-time awakening (60), back pain (61) and decreased sexual desire (62) as less severe during the LRS compared to the MT stages and postmenopause.

Discussion

Following the STRAW 2001 conference, a number of investigators studied the newly established stages of reproductive aging. Most focused on the MT stages (STRAW −2 and −1) with their clearly delineated entry criteria, e.g. a persistent break in cyclicity of 7 days or more in consecutive cycles. There was little emphasis on the LRS, with little attention to either the identifying markers of entry to this stage or characterizing associated menstrual cycle changes or symptoms. Some investigators used LRS as a comparison for the MT.

Our searches revealed a very limited number of research reports about the LRS, itself. LRS may have attracted less attention than the early and late MT stages for a variety of reasons. First, methods of determining the stages of reproductive aging have been varied and inconsistent, especially with respect to the LRS; currently there is no clear marker of entry to LRS. Second, the primary concern of early efforts to develop a staging system was to clearly identify the boundaries of the MT. Most investigators leading early longitudinal studies of the MT recruited women who had begun noticing changes in their cycles, such as menstrual irregularity, to increase the likelihood of observing the final menstrual period (FMP). STRAW conferees did not include LRS in the MT but viewed it as a “phase with declining fertility.” Third, data related to some of the LRS characteristics of subtle menstrual cycle changes such as shorter cycle length or changes in flow were not collected from many cohorts. Given the nature of questions related to LRS symptoms, most designs were descriptive and correlational (3). Finally, menopause researchers have had limited interest in the late reproductive stage.

Because the first component of this scoping review using the term “late reproductive stage” (LRS) yielded only a few papers a second search was conducted using the terms “late reproductive age” (LRA) and women. Use of the search term LRA identified several different cohorts as seen in Table 2. In some studies, LRA indicated a group of women defined by age, typically spanning 35–55 years, who were experiencing certain types of menstrual cycles. Some research groups such as Hale’s used the STRAW criteria (cyclic differences of >=7 days) in conjunction with age groups to identify the early MT. Given the differences in definitions used to demarcate LRS and LRA, it is challenging to harmonize the data from these two groups of studies. Although onset of the early MT stage as defined by STRAW+10 was specified in many studies of LRA women as an upper boundary for the LRS, identifying the lower boundary for LRS has been difficult due to its reliance on subtle menstrual cycle changes often not measured.

Studies of LRA emphasized reproductive hormonal and menstrual cycle changes noted during this portion of women’s lifespan. Research on hormonal patterns during LRA points to the utility of AMH as a predictor of the FMP and its variability over the MT stages, as well as how it is modified by personal characteristics such as obesity and chronological age. Both the level and trajectory of AMH are promising predictive markers of FMP. Hale’s team found that LRA women experienced shorter cycles, higher FSH, and lower AMH levels than those seen in younger women. However, without more precise criteria it is impossible to determine which LRA cycles should be designated as indicating LRS. More sensitive and specific assays for AMH could support research focusing on defining entry to and characterization of the LRS on the basis of AMH levels. (63,64).

Hale’s team incorporated STRAW stages in their examination of groups of women classified according to age, e.g. mid-reproductive and late-reproductive age, super-imposing STRAW criteria for Stages −3 (Late Reproductive), −2 (Early Transition) and −1 (Late Transition) stages on age categories (27). This research program has contributed to understanding relationships between changing menstrual characteristics, and reproductive endocrine evolution as women transition from LRA/LRS to the MT stages (STRAW −2 and −1). Robertson’s (30) identification of types of menstrual cycles indicated they were not specific to a stage of reproductive aging, rather appearing during several of the STRAW stages. For example, Type 2 cycles (similar to LRS cycles with decreasing AMH, increasing FSH, decreasing inhibin B) appeared in STRAW stages −4, −2, and −1. The linkage between cycle type and stages was not specific for women in the LRS, but the addition of more sensitive AMH measures may contribute to a clearer picture of the LRS and more refined classification system.

Investigators also examined ovarian reserve, including changes in follicular wave dynamics in menstrual cycles of women in mid-, late-, and advanced reproductive age. This work suggested that dominant follicles in the follicular and luteal phase did not differ with age, but luteal phase dominant follicles emerged earlier relative to ovulation and developed to larger diameter in older women (31). In contrast, Hansen and colleagues (53) who studied STRAW stages and ovarian reserve found a decrease in primordial follicle counts, non-growing follicle counts, and antral follicle counts along with increasing FSH levels and decreasing AMH and inhibin B levels among women in STRAW Stage −3 (LRS). In addition to predicting the approach of menopause, ovarian reserve is a correlated with fertility of women in the late reproductive years, and this is another area of research opportunity.

Studies of LRA women substantiated the importance of considering health-related behaviors such as smoking and alcohol use when studying women in the LRS. Interestingly, type of alcohol consumed, e.g. red wine vs liquor, was associated differently with onset of perimenopause (33). Smoking was associated with lower AMH levels which were also influenced by age (34).

Reports identified in the search for LRS yielded few studies of symptoms, perhaps reflecting the assumption that this period is one in which women are not bothered by symptoms associated with menopause such as hot flashes, mood, and sleep disruption. To the contrary, the few studies of symptoms revealed that women who were presumably in the LRS reported some, specifically poor sleep (35,36,44) and depressive symptoms (37,44). In addition, a validated measure of menopause-related symptoms documented the presence of psychological, vasomotor, and somatic symptoms among women prior to the MT stage (i.e. during the LRS). These increased in frequency as women progressed through the MT. Moreover, each symptom group demonstrated a different pattern across the premenopause (i.e. LRS), early and late MT and postmenopause (38). Investigators studying symptom clusters have also identified clusters of mood and cognitive symptoms (tension, panic, nervous, depressed, difficulty concentrating) occurring during LRS and persisting into the MT. In addition, a cluster of breast pain, sleep disruption, hot flashes and vaginal dryness symptoms was characteristic of the late reproductive stage but not during later stages (10). Investigators also explored a variety of gene variants in LRA women associated with symptoms such as hot flashes and depressed mood.

Most research identified in searches of LRA is challenging to interpret with respect to the LRS as defined by STRAW +10. Although some investigators (Hale, Robertson, and Freeman) incorporated or approximated the STRAW criteria in their studies, many did not, leaving readers uncertain about the meaning of LRA with respect to changes occurring during LRS (stage −3).

Of interest is that the largest longitudinal study of the menopausal transition, SWAN, has not incorporated the STRAW criteria in their published results. Although SWAN investigators subsequently added use of monthly menstrual calendars along with a Daily Hormone Study to characterize the menopausal transition in greater detail (65), investigators have not yet attempted to harmonize their classification approaches with the STRAW criteria. Therefore, determining LRS symptoms in their study is not possible. The exception is Gracia’s efforts to harmonize the STRAW, SWAN and POAS staging systems (47).

Implementation of the STRAW criteria has underscored the need for a classification system that women, clinicians, and researchers could and would use. Balancing ease of classification based on self-report of menstrual cycle characteristics with precision of the classification achieved is challenging. Increased usage of menstrual tracking devices may help refine features of LRS cycle patterns. Current menstrual tracking apps allow operationalization of at least two dimensions of menstrual cycle changes: cycle length can be calculated automatically and changes over several months noted more easily. In addition, many apps allow women to note the amount of flow daily. These data can be viewed as trends and could allow for operationalizing a meaningful metric of change. One consideration in recording these subtle changes is allowing women to anchor their ratings to their own menstrual cycle experiences, e.g. amount flow relative to her own heaviest flow. A recent evaluation of menstrual tracking applications indicate they have insufficient support for facilitating menstrual literacy, especially for perimenopausal users, prompting consideration of enriching the content associated with the tracking devices (66).

With the emergence of more sensitive assays, such as those for AMH, it may be possible to identify the LRS more precisely with endocrine markers in addition to menstrual cycle characteristics. Finally, genetic markers indicating the likely timing of this stage may contribute to more clear delineation of its onset.

Multiple approaches to staging reproductive aging by investigators such as Mitchell, Freeman, Gracia, Hale, Robertson, and Miro, has enhanced understanding of the hormonal patterns and menstrual characteristics of cycles that precede the MT. Simultaneously, this has demonstrated a need to harmonize these systems to the extent possible in order to integrate results of studies, reduce the likelihood of misclassification bias, and minimize women’s and clinicians’ confusion about which of women’s experiences in the late reproductive years can be attributed to approaching menopause.

Reflection on the contributions of studies of LRS and LRA also bears consideration of investigators’ aims. Although researchers studying LRS were motivated to understand the transition to menopause, including cyclic and hormonal changes and women’s experiences, researchers studying LRA were more often motivated by interest in fertility and reproductive endocrinology. Hale, Gracia, and Mitchell attempted to integrate the understanding of this stage as one in which women’s fertility was changing and the transition to menopause was beginning. Women, themselves, were and are motivated to understand their personal experience, including cycle changes and symptoms. They are interested in anticipatory guidance, symptom management, as well as fertility changes.

This scoping review was limited by inclusion of research reports not published in English. In addition, tracking research using terms such as LRS and LRA may have been limited by investigators’ use of these terms in their publications and by our ability to find research that employed other terminology to indicate this stage of reproductive aging.

Conclusions and Recommendation

A critical next step in enhancing the utility of the STRAW framework in both scientific research and clinical care is a better characterization of LRS with respect to symptoms, menstrual attributes, and hormonal profiles as well as improved criteria to identify its inception. The literature presented in this review indicates that the LRS warrants more attention as a distinct stage of reproductive aging beyond its role as an uninteresting comparator in studies of the menopausal transition.

Research described in this scoping review indicates that LRS can be distinguished from the MT based on variability in reproductive endocrine patterns, changing menstrual cycle patterns and related physiologic changes. In addition, women report a variety of symptoms during the LRS, some of which seem related to reproductive aging.

Harmonization of data from prior studies of women of LRS and LRA is required so that the LRS can be differentiated from the middle reproductive stage and be more completely understood. Efforts to enhance women’s ability to describe their LRS experiences along with clinicians’ recognition of the LRS present an opportunity for anticipatory guidance and health promotion during this important stage of women’s lives.

Further efforts to increase women’s, clinician’s and researcher’s awareness of the stages of reproductive aging, including LRS, will depend on a more robust and nuanced understanding of the relevant physiology, integrated with careful study of women’s experiences.