Where Are We in Understanding the Natural History of Polycystic Ovary Syndrome?

A Systematic Review of Longitudinal Cohort Studies

Sylvia Kiconco; Chau Thien Tay; Kate Louise Rassie; Ricardo Azziz; Helena J. Teede; Anju E. Joham

Hum Reprod. 2022;37(6):1255-1273.

Abstract and Introduction

Abstract

Study Question: What is the natural history of reproductive, psychological and oncological features in women with polycystic ovary syndrome (PCOS) in comparison to those without PCOS across the life course?

Summary Answer: Existing longitudinal data on changes in reproductive, psychological and oncological features in PCOS are inadequate and conflicting, but the limited evidence suggests that total testosterone (T) and dehydroepiandrosterone sulphate (DHEAS) levels decline more significantly in women with PCOS than in those without PCOS, and the risk of gestational diabetes is higher in pregnant women with PCOS compared to their counterparts without PCOS.

What is Known Already: The progression of reproductive, psychological and oncological features in PCOS remains unclear, which limits prevention and early diagnosis strategies across the lifespan. Understanding the natural history of PCOS is one of the overarching priorities in PCOS research.

Study Design, Size, Duration: This is a systematic review of longitudinal cohort studies with a narrative presentation of findings. Databases MEDLINE, EMBASE, Ovid PsycInfo, CINAHL PLUS and EBM reviews were searched between 15 January 2020 and 11 February 2021 with no language restrictions. Only studies published from the year 1990 to February 2021 were included.

Participants/Materials, Setting, Methods: In line with current guidelines for the assessment and management of PCOS, we included studies where participants were females with PCOS diagnosed according to the 2003 Rotterdam or the 1990 National Institutes of Health (NIH) consensus criteria.

Main Results and the Role of Chance: A total of 21 longitudinal studies including 62 123 participants over four continents reported reproductive, psychological and/or oncological outcomes. Participants were females aged between 15 and 49 years at baseline, with follow-up periods ranging from 4 weeks to 32 years. Consistent evidence based on limited studies suggests that total T and DHEAS levels decline to a greater degree in women with PCOS compared to those without PCOS, and the risk gestational diabetes is higher in women with PCOS than in those without PCOS. Evidence reporting changes over time in the majority of the remaining outcomes was unclear due to conflicting and/or insufficient information.

Limitations, Reasons for Caution: There was extreme heterogeneity between studies in terms of study setting, population characteristics, follow-up period, effect measures used and laboratory testing approaches.

Wider Implications of the Findings: Understanding the natural history of PCOS and changes in diagnostic, reproductive, psychological and oncological features of PCOS across the lifespan is still a challenge and the existing literature is both limited and conflicting. It is important that future long-term prospective longitudinal studies are conducted in unselected and well-characterized populations.

Study Funding/Competing Interest(S): This specific study was not funded. S.K. is supported by scholarships from the Research Training Program of the Commonwealth of Australia and Monash University; H.J.T. is supported by an Australian National Health and Medical Research Council fellowship; and A.E.J. is supported by the Australian National Health and Medical Research Council's Centre for Research Excellence in Women's Health in Reproductive Life. R.A. was employed by the American Society for Reproductive Medicine and is a consultant to Spruce Biosciences and Fortress Biotech. The other authors have no conflicts of interest to declare.

Registration Number: Prospero registration number: CRD42020165546.

Introduction

Polycystic ovary syndrome (PCOS) is an endocrine-metabolic disorder diagnosed in adults based on the presence of at least two of three clinical features, including polycystic ovary morphology, oligo/amenorrhea and hyperandrogenism (clinical and/or biochemical). PCOS is a major public health issue that affects 5–15% women of reproductive age globally (Azziz et al., 2006; Diamanti-Kandarakis et al., 2006; March et al., 2010; Bozdag et al., 2016) but is complicated by diagnostic challenges including a lack of clear definitions for individual PCOS features. These contribute to misdiagnosis, delayed diagnosis and patient dissatisfaction (Dokras et al., 2017; Gibson-Helm et al., 2017), and up to 70% of women with the condition remain undiagnosed (March et al., 2010). The lack of information on changes in biochemical and clinical hyperandrogenism, cycle regularity and ovarian morphology, including the PCOS phenotype over the lifespan, also complicates diagnosis and warrants further investigation.

Women with PCOS are at increased risk of adverse reproductive, metabolic and psychological outcomes. Common reproductive features of the condition include biochemical hyperandrogenism, ovulatory and menstrual dysfunction, hirsutism, subfertility, endometrial hyperplasia and obstetrical complications (Teede et al., 2018a,b). Women with PCOS are also at a higher risk of infertility or reduced fertility than those without PCOS, which may be driven by changes in oocyte, endometrial and embryo function (Palomba, 2021). Metabolic features include increased risks for insulin resistance, dyslipidemia, impaired glucose tolerance, metabolic syndrome, gestational diabetes, type 2 diabetes and cardiovascular disease (Legro et al., 1999; Apridonidze et al., 2005). Psychological features include anxiety, depression, low self-esteem and poor body image (Moran et al., 2010; Teede et al., 2010; Deeks et al., 2011). Increased endometrial cancer risk has also been associated with PCOS (Charalampakis et al., 2016). These diverse PCOS features lead to a diminished quality of life in affected women (Dokras et al., 2018; Teede et al., 2018a,b). Overall, PCOS is associated with a substantial economic burden, conservatively estimated to exceed an annual total cost of $8 billion USD in the USA alone (2020 USD), including healthcare costs related to diagnosis, reproductive, metabolic, vascular and pregnancy-related morbidities (Riestenberg et al., 2022).

The recent international evidence-based guideline for the Diagnosis and Management of PCOS (Teede et al., 2018b) highlighted our limited understanding of the natural history of reproductive, psychological and oncological outcomes in PCOS and identified major gaps that currently limit the development of effective prevention strategies across the lifespan. Furthermore, understanding the natural history of PCOS emerged from the guideline process as one of the overarching priorities in PCOS research. Therefore, we now aim to explore the natural history of PCOS with a focus on reproductive and psychologic features as well as cancer risk, by conducting a systematic review of longitudinal cohort studies.

1 2 3 4 5

Next Section

Table I. Characteristics of included studies.
Study	Country	Design	Setting	PCOS group	Non-PCOS group	PCOS criteria	Follow-up duration	Outcomes measured	Risk of bias
Ahmad et al. (2018)	USA	Prospective cohort	Academic practice (PCOS clinic)	Age, 30.9 ± 6.46 years n = 31	Age, 36.06 ± 5.36 years n = 267	Rotterdam	PCOS (3.21 ± 1.62), controls 3.90 ± 0.79 years (2007 through 2013)	AMH	Moderate
Altinok et al. (2014)	Denmark	Retrospective	Academic practice (out-patient) and population (controls)	Age, 29 years n = 1124	Age, 29 n = 4213	Rotterdam	6.8 (PCOS) and 7.2 (controls) years (1997 to 2012)	Antidepressant prescription	Moderate
Brown et al. (2011)	Netherlands	Retrospective	Academic practice (out-patient clinic)	Age, 28.6 years, n = 254	Age, 29.9 years, n = 41	Rotterdam	7 years (1991 through 2009)	LH, FSH, LH/FSH, AMH, total T, DHEAS, FAI, SHBG	High
Carmina et al. (2012b)	Italy	Prospective	Academic practice	Age, 37 ± 1 years, n = 54	Age, 37 ± 1 years, n = 20	Rotterdam	5 years (period not specified)	LH, LH/FSH ratio, total T, DHEAS, AMH	Moderate
Carmina et al. (2012a)	Italy	Prospective	Academic practice (Endocrine Unit)	Age, 21.9 ± 2.1 years, n = 193	Age, 21.9 ± 2.1 years, n = 35 (controls were not followed up)	Rotterdam	20 years (baseline 1985/1990)	LH: FSH ratio, total T, DHEAS	High
Cheng-Che et al. (2015)	Taiwan	Retrospective	National health insurance registry	Age, 27 years, n = 3566	Age, 7 years, n = 14 264	NIH	7.15 years (2000/2004 to 2009)	Breast cancer, and uterine cancer	Low
Forslund et al. (2021)	Sweden	Prospective cohort	Academic practice	Age 49.4 ± 5.0 years, n = 21	Age 49.7 ± 5.6 years, n = 55	Rotterdam	32 years (1987 to 2019)	FSH, LH, DHEAS, SHBG, T, A₄, FAI	Moderate
Greenwood et al. (2019a)	USA	Prospective	Academic practice (PCOS clinic)	Age, 29 years n = 163	No controls	Rotterdam	5.5 years (2006–2017-consecutive)	BDI-FS score	Moderate
Greenwood et al. (2019b)	USA	Prospective	General population	Age 23 to 35 years, n = 83	Age 23 to 35 years, n = 1044	NIH	30 years at 5 year intervals (recruited 1985/1986)	CES-Depression symptom scores	Low
Harnod et al. (2020)	Taiwan	Retrospective cohort	Health insurance database	Age 27.74 ± 6.8 years, n = 7026	Age 27.74 ± 6.8 years, n = 28 104	NIH or Rotterdam	16 years (1996 to 2013)	Incident anxiety	Moderate
Huddleston et al. (2017)	USA	Prospective	Academic practice (PCOS clinic)	Age, 30.4 ± 5.6 years, n = 38	Age, 35.7 ± 5.5 years, n = 296	Rotterdam	3–4 years (2004–2014 cohort)	Total T	Low
Jarrett et al. (2020)	USA	Prospective cohort	General population	Age 26 ± 4 years, n = 26	Age 30 ± 6 years, n = 12	NIH	4 to 6 weeks (at every other day interval)	LH, FSH	High
Jakubowicz et al. (2002)	Venezuela	Retrospective nested in RCT	Hospital (Endocrinology Clinic)	Age 30.0 ± 3.2, years, n = 31	No controls	NIH	4.5 (1996–2000)	Free T, early pregnancy loss, pre-term births	High
Kerchner et al. (2009)	USA	Prospective	Academic practice (PCOS clinic)	Age 32 ± 6.3 years, n = 60	No controls	Rotterdam	22 ± 3.7 months (baseline 1997/1999)	Depression	High
Ng et al. (2019)	Hong Kong	Prospective cohort	General hospital and community (controls)	Age 30.6 ± 6.5 years, n = 199	Age 42.6 ± 7.0 years, n = 242	Rotterdam	10.6 ± 1.3 years (2003/2007 to 2016/2017)	FSH, LH, total T, FAI, AMH	Low
Palomba et al. (2014)	Italy	Prospective	Academic practice (Obstetrics and Gynecology department)	Age 27.8 ± 3.6 years, n = 150, Gestational age (5.4 weeks)	Age 27.4 ± 4.0 years, n = 150 Gestational age (5.4 weeks)	Rotterdam	27 gestational weeks (2003 through 2012)	Total T, A, DHEAS, SHBG, FAI, miscarriage, PIH, PE, GDM, B/W	Low
Palomba et al. (2007)	Italy	Prospective nested in RCT	Academic practice	Age 24.8 ± 2.7 years, n = 13	Age 25.6 ± 2.7 years, n = 10	NIH	24 months (recruited 2003/2004 to 2005/2006)	mFG score, total T, A₄, DHEAS, SHBG, FAI	Moderate
Reyes-Munoz et al. (2012)	Mexico	Prospective	Academic practice	Age 29.1 ± 3.9 years, n = 52	Age 29.0 ± 3.8 years, n = 52	Rotterdam	29.4 gestational weeks (2006 January through to December 2007	GDM, miscarriage, preterm birth, pre-eclampsia, stillbirth, congenital malformation, weight gain, newborn weight	High
Schmidt et al. (2011a)	Sweden	Prospective	Academic practice	Age 49.4 ± 4.9 years, n = 25	Age 49.7 ± 5.6 years, n = 68	Rotterdam	21 years (1987–2008)	Menopause age, FSH, LH, SHBG, total T, A₄, DHEAS, FAI	Moderate
Schmidt et al. (2011b)	Sweden	Prospective	Academic practice	Age 49.4 ± 4.9 years, n = 25	Age 49.7 ± 5.6 years, n = 68	Rotterdam	21 years (1987–2008)	Breast, and endometrial cancers	Moderate
Udesen et al. (2019)	Denmark	Prospective	Fertility clinic	Age 29.1 ± 4.1 years, n = 40	Age 30.0 ± 5.2 years, n = 8	Rotterdam	5.8 ± 0.8 years (recruited 2010/2012)	mFG score, total T, free T, DHEAS, A₄, SHBG, LH/FSH ratio	High

A₄, androstenedione; AMH, anti-Müllerian hormone; B/W, birth weight; BDI-FS, Beck Depression Inventory Fast Screen; CES-D, Center for Epidemiologic Studies-Depression; DHEAS, dehydroepiandrosterone sulphate; FAI, free androgen index; GDM, gestational diabetes mellitus; HTN, hypertension; mFG, modified Ferriman-Gallwey score; PCOS, polycystic ovary syndrome; PE, pre-eclampsia; PIH, pregnancy-induced hypertension; SHBG, sex hormone-binding globulin; T, testosterone.

Table II. Changes in reproductive outcomes over time.
Outcomes		Study author, year	Baseline age, sample size		Mean follow-up duration	Effect measures	Observed estimates
Outcomes		Study author, year	PCOS group	Non-PCOS group	Mean follow-up duration	Effect measures	Within PCOS group comparison	Within non-PCOS group comparison	PCOS group versus non-PCOS group
Clinical hyperandrogenism	Hirsutism (mFG score)	Palomba et al. (2007)	Age 24.8 years, n = 13	Age 25.6 years, n = 10	24 months	Mean comparison with baseline (24,18 months vs 6 months)	10.8 ± 1.8, 10.8 ± 2.2 vs 10.7 ± 1.8, P > 0.05	4.5 ± 1.4, 4.6 ± 1.4 vs 4.4 ± 1.6, P > 0.05	P < 0.05 at 18th and 24th visits (changes between groups not compared)
		Udesen et al. (2019)	Age 29.1 years, n = 40	Age 30.0 years, n = 8	5.8 ± 0.8 years	Mean comparison with baseline	5.0 (3.0–10.0) vs 6.0 (3.0–9.0), NS	2.5 (1.5–4.5) vs 1.5 (0–5.5), NS	–
	Acne	–	–	–	–	–	–	–	–
	Hair loss	–	–	–	–	–	–	–	–

Biochemical hyperandrogenism	Total T	Carmina et al. (2012a), Carmina et al. (2013)	Age 21.8 years, n = 193	Age 21.8 years, n = 35 (not followed up)	20 years	Mean comparisons (5th to 20th years) with baseline (ng/dl)	59 ± 28, 65 ± 25, 68 ± 22 vs 75 ± 26 (P < 0.05)	–	–
		Huddleston et al. (2017)	Age 30.4 years, n = 38	Age 35.7 years, n = 296	3–4 years	Change per year (nmol/l)	BMI >30: [−0.09 (95% CI −0.16 to −0.02)], P < 0.05 BMI ≤ 30: [−0.04 (95 % CI −0.11 to 0.03)], NS	Not reported	–
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Mean change from baseline (nmol/l)	−0.82 ± 0.88, P = 0.001	−0.36 ± 0.74, P = 0.001	P = 0.016
		Udesen et al. (2019)	Age 29.1 years, n = 40	Age 30.0 years, n = 8	5.8 years	Median comparison with baseline nmol/l	1.4 (1.0 to 2.0) vs 1.9 (1.4 to 2.5), P < 0.001	0.7 (0.4 to 0.9) vs 0.9 (0.7 to 1.6), P = 0.039	–
		Ng et al. (2019)	Age 30.6 years, n = 199	Age 42.6 years, n = 242	10.6 years	Change from baseline (nmol/l)	−0.5 ± 0.7 P < 0.001	–	–
		Forslund et al. (2021)	Age 49.4 years, n = 21	Age 49.7 6 years, n = 55	32 years	Change from baseline (nmol/l)	–1.2 ± 1.1, P < 0.01	–0.8 ± 0.7, P < 0.01	P = 0.48
		Palomba et al. (2007)	Age 24.8 years, n = 13	Age 25.6 years, n = 10	24 months	Mean comparison with baseline (24.18 vs 6 months)	1.5 ± 0.5, 1.6 ± 0.4 vs 1.7 ± 0.3 (ng/ml), P > 0.05	0.6 ± 0.2, 0.6 ± 0.2 vs 0.6 ± 0.2 (ng/ml), P > 0.05	P < 0.05 at 18th and 24th visits (changes between groups not compared)
		Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational weeks	27 gestational weeks	Mean comparison with baseline (32nd, 20th weeks vs pre study)	3.6 ± 2.4, 3.8 ± 1.5 vs 3.1 ± 2.3 (ng/dl), P < 0.05	1.0 ± 0.2, 1.0 ± 0.2 vs 0.9 ± 0.2 (ng/dl), NS	P < 0.01 at all visits (changes between groups not compared)
		Carmina et al. (2012b)	Age 37 years, n = 54	Age 37 years, n = 20	5 years	Mean comparison with baseline (nmol/l)	58 ± 19 vs 74 ± 22, P < 0.01	25 ± 16 vs 28 ± 11, NS	–
		Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year (ng/ml)	−2.3 (−2.9 to −1.4), P < 0.05	<0.0001 (−0.0001 to 0.0001), NS	P < 0.001
	Free T	Jakubowicz et al. (2002)	Age 30.0 years, n = 31	No controls	4.5 years	Mean comparisons baseline vs 6-week gestation, (ng/dl)	3.3 ± 0.2 vs 3.4 ± 0.3, NS	–	–
		Udesen et al. (2019)	Age 29.1 years, n = 40	Age 30.0 years, n = 8	5.8 ± 0.8 years	Mean comparison with baseline (nmol/l)	0.032 (0.019 to 0.050) vs 0.023 (0.014 to 0.036), P = 0.008	0.019 (0.012 to 0.023) vs 0.012 (0.007 to 0.014), NS
	FAI	Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year	<0.0001 (−0.0001 to 0.0001), NS	–	–
		Palomba et al. (2007)	Age 24.8 years, n = 13	Age 25.6 years, n = 10	24 months	Mean comparison with baseline (24.18 vs 6 months)	22.7 ± 5, 23.5 ± 4.3 vs 22.4 ± 5.7 (%), NS	4.3 ± 1.8, 4.2 ± 2.0 vs 4.5 ± 1.9 (%), NS	P < 0.05 at 18th and 24th visits (changes between groups not compared)
		Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 weeks gestational	Age 27.4 years, n = 150, 5.4 weeks of gestation	27 gestational weeks	Mean comparison with baseline (32nd, 20th weeks vs pre study)	11.3 ± 3.4, 10.5 ± 3.4 vs 13.0 ± 3.5 (%), P < 0.05	3.3 ± 2.5, 3.4 ± 2.1 vs 4.3 ± 2.6 (%), P < 0.05	P < 0.01 at all visits (changes between groups not compared)
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Mean change from baseline	−3.40 ± 4.45, P = 0.001	−1.63 ± 2.69, P = 0.001	P = 0.033
		Forslund et al. (2021)	Age 49.4 years, n = 21	Age 49.7 years, n = 55	32 years	Change from baseline	–4.5 ± 4.3, P < 0.01	–2.4 ± 2.7, P < 0.01	P = 0.08
		Ng et al. (2019)	Age 30.6 years, n = 199	Age 42.6 years, n = 242	10.6 years	Change from baseline	−3.0 ± 6.5, P < 0.001	–	–
	DHEAS	Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year (mg/ml)	−0.13 (−0.15 to −0.10), P < 0.001	<0.0001 (−0.0001 to 0.0001), NS	P < 0.001
		Carmina et al. (2012a)	Age 21.9 years, n = 193	Not followed up	20 years	Mean comparisons (5th–20th years) with baseline, (μg/ml)	2.00 ± 0.9, 2.1 ± 0.85, 2.2 ± 1.3 vs 2.7 ± 1.2, P < 0.01	–	–
		Carmina et al. (2012b)	Age 37 years, n = 54	Age 37 years, n = 20	5 years	Mean comparisons with baseline (μg/ml)	2.2 ± 1.2 vs 2.4 ± 1, P < 0.01	1.7 ± 1 vs 1.8 ± 1, NS	–
		Forslund et al. (2021)	Age 49.4 years, n = 21	Age 49.7 years, n = 55	32 years	Mean change from baseline, (μmol/l)	–2.9 ± 2.3, P < 0.01)	controls –1.4 ± 0.9, P < 0.01	P = 0.01
		Palomba et al. (2007)	Age 24.8 years, n = 13	Age 25.6 years, n = 10	24 months	Mean comparison with baseline, (ng/ml)	2579.2 ± 442.2, 2579.2 ± 368.5 vs 2616.1 ± 368.5, P > 0.05	1400.1 ± 405.3, 1326.5 ± 331.6 vs 1510.7 ± 368.5, P > 0.05	P < 0.05 at 18th and 24th visits (changes between groups not compared)
		Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational weeks	27 gestational weeks	Mean comparison with baseline (32nd, 20th weeks vs pre study), (ng/dl)	2762.1 ± 974.4, 2746.4 ± 977.3 vs 2684.0 ± 981.3, P < 0.05	1721.4 ± 713.3 1737.2 ± 722.4 vs 1709.2 ± 733.1, P > 0.005	P < 0.05 at all visits (changes between groups not compared)
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Change from baseline, (μmol/l)	−2.51 ± 1.58, P = 0.001	−1.71 ± 1.26, P = 0.001	P = 0.008
		Udesen et al. (2019)	Age 29.1 years, n = 40	Age 30.0 years, n = 8	5.8 years	Mean comparison with baseline, (Umol/l)	4857 ± 2190 vs 5676 ± 2764, P = 0.017	5181 ± 1962 vs 4437 ± 2550, NS	–
	A₄	Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year, (ng/dl)	−17.5 (−21.8, −12.9), P < 0.001	–	–
		Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational weeks	27 gestational weeks	mean Comparison with baseline (32nd, 20th weeks vs pre study), (ng/ml)	4.4 ± 3.6, 4.3 ± 3.7 vs 4.2 ± 3.5, P > 0.05	1.7 ± 1.3, 1.8 ± 1.0 vs 1.6 ± 1.1, P > 0.05	P < 0.05 at all visits (changes between groups not compared)
		Palomba et al. (2007)	Age 24.8 years, n = 13	Age 25.6 years, n = 10	24 months	Mean comparison with baseline (24.18 vs 6 months), (ng/ml)	1.8 ± 0.3, 1.9 ± 0.3 vs 1.9 ± 0.3, P > 0.05	0.6 ± 0.1, 0.7 ± 0.02, vs 0.6 ± 0.1, P > 0.05	P < 0.05 at all visits (changes between groups not compared)
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Change from baseline, (nmol/l)	−0.56 ± 3.58, P = 0.230	1.82 ± 2.91, P = 0.001	P = 0.001
		Udesen et al. (2019)	Age 29.1 years, n = 40	Age 30.0 years, n = 8	5.8 years	Median comparison with baseline, (nmol/l)	5.8 (4.0–8.5) vs 7.1 (4.7–9.0), P = 0.048	2.8 (1.8–3.4) vs 4.4 (3.0–5.7), P = 0.001	–
		Forslund et al. (2021)	Age 49.4 years, n = 21	Age 49.7 years, n = 55	32 years	Mean change from baseline, (nmol/l)	–1.5 ± 4.7 P = 0.23	0.1 ± 1.7, P = 1.00	P = 0.17
	SHBG	Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year, (mg/dl)	−0.03 (−0.05 to −0.001), P < 0.01	–	–
		Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational week	27 estational weeks	Mean comparison with baseline (32nd, 20th weeks vs pre study), (nmol/l)	25.0 ± 8.3, 24.7 ± 9.2 vs 18.0 ± 9.5, P > 0.05	48.2 ± 14.2, 49.0 ± 13.9 vs 42.7 ± 14.8, P > 0.05	P < 0.05 at all visits (changes between groups not compared)
		Palomba et al. (2007)	Age 24.8 years, n = 13	Age 25.6 years, n = 10	24 months	Mean comparison with baseline (24,18 vs 6 months), (nmol/l)	26.3 ± 4.1, 25.9 ± 4.3 vs 26.1 ± 3.5, P > 0.05	47.8 ± 7.2, 47.6 ± 6.8 vs 47.5 ± 6.8, P > 0.05	P < 0.05 at all visits (changes between groups not compared)
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Change from baseline, (nmol/l)	8.9 ± 29.4, P = 0.076	12.3 ± 32.5, P = 0.001	P = 0.290
		Udesen et al. (2019)	Age 29.1 years, n = 40	Age 30.0 years, n = 8	5.8 years	Median comparison with baseline, (nmol/l)	46.0 (36.5–83.0) vs 59.5 (40.0–83.0), P = 0.011	62.5 (45.0–73.5) vs 49.0 (30.0–72.5), NS	–
		Forslund et al. (2021)	Age 49.4 years, n = 21	Age 49.7 years, n = 55	32 years	Mean change from baseline, (nmol/l)	35 ± 20, P < 0.01	33 ± 56, P = 0.01	P = 0.75
Reproductive hormonal profile	AMH	Ahmad et al. (2018)	Age 30.9 years, n = 31	Age 36 years, n = 267	PCOS: 21 controls: 3.9 years	Rate of change/year and % change from baseline, (ng/m/l/year)	Rate/year, 2.96 ± 5.6 % change, −35.8% (−47.5% to −24.0%)	Rate/year 0.29 ± 0.56 % change, 8.0% (−12.0% to −3.9%)	Difference of −2.28 [−3.18 to −1.38]; P < 0.01 % change −35.8% vs. −8.0%; P < 0.01
		Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year, (ng/ml)	<0.0001 (95% CI −0.0001, 0.0001), NS	<0.0001 (95% CI −0.0001, 0.0001), NS	NS
		Carmina et al. (2012b)	Age 37 years, n = 54	Age 37 years, n = 20	5 years	Mean comparisons with baseline, (ng/ml)	3.9 ± 1.2 vs 6.7 ± 2.1, P < 0.01	1 ± 0.7 vs 1.7 ± 0.7, P < 0.01	Mean decrease: 40 ± 12% vs 41 ± 10% NS
		Ng et al. (2019)	Age 30.6 years, n = 199	Age 42.6 years, n = 242	10.6 years	Change from baseline, (pmol/l)	−13.5 ± 27.9, P < 0.001	–	–
	LH	Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year, (mIU/ml)	<0.0001 (−0.0001 to 0.0001), NS	−0.45 (−0.75 to −0.15), P < 0.05	P < 0.001
		Carmina et al. (2012b)	Age 37 years, n = 54	Age 37 years, n = 20	5 years	Mean comparisons with baseline, (mUI/ml)	8.8 ± 4 vs 10 ± 3.7, NS	6.5 ± 1.1 vs 6.4 ± 1.4, NS	–
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Change from baseline, (IU/l)	11.2 ± 10.0, P = 0.001	7.4 ± 14.1, P = 0.001	P = 0.153
		Forslund et al. (2021)	Age 49.4 years, n = 21	Age 49.7 years, n = 55	32 years	mean change from baseline, (IU/l)	13.6 ± 9.5, P < 0.01	8.0 ± 13.2, P < 0.01	P = 0.22
		Ng et al. (2019)	Age 30.6 years, n = 199	Age 42.6 years, n = 242	10.6 years	Change from baseline, (IU/l)	−0.1 ± 11.3, P = 0.89	–	–
		Jarrett et al. (2020)	Age 26 years, n = 26	Age 30 years, n = 12	4 to 6 weeks	Over follicular and luteal phases, (mIU/ml)	Follicular phase: 13.1 ± 3.8 Luteal phase: 6.2 ± 3.3	Follicular phase: 9.8 ± 2.5 Luteal phase: 6.1 ± 2.8	Follicular phase: P = 0.02 Luteal phase: P = 0.95, All days: P _(PCOS) = 0.01
	FSH	Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year, (mIU/ml)	0.18 (0.05, 0.32), P < 0.01	−0.59 (−0.98, −0.21), P < 0.05	P < 0.001
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Change from baseline, (U/l)	11.6 ± 38.0, P = 0.123	−13.9 ± 61.9, P = 0.12	P = 0.124
		Forslund et al. (2021)	Age 49.4 years, n = 21	Age 49.7 years, n = 55	32 years	Change from baseline, (IU/l)	24.9 ± 34.9, P = 0.03	2.7 ± 62, P = 0.73	P = 0.53
		Jarrett et al. (2020)	Age 26 years, n = 26	Age 30 years, n = 12	4 to 6 weeks	Over follicular and luteal phases, (mIU/ml)	Follicular phase: 6.0 ± 1.5 Luteal phase: 2.8 ± 1.2	Follicular phase: 7.8 ± 2.5 Luteal phase: 4.9 ± 2.0	Follicular phase: P = 0.05 Luteal phase: P < 0.01 All days: P _(PCOS) = 0.91
		Ng et al. (2019)	Age 30.6 years, n = 199	Age 42.6 years, n = 242	10.6 years	Change from baseline, (IU/l)	2.7 ± 8.0, P < 0.001	–	–
	LH:FSH ratio	Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	Median change per year	−0.06 (0.10 to −0.01), P < 0.05	<0.0001 (0.0001 to 0.0001), NS	P < 0.001
		Carmina et al. (2012b)	Age 37 years, n = 54	Age 37 years, n = 20	5 years	Mean comparisons with baseline	1.5 ± 0.7 vs 1.6 ± 0.7, NS	1 ± 0.3 vs 1.1 ± 0.2, NS	–
		Carmina et al. (2012a)	Age 21.9 years, n = 193	Not followed up	20 years	mean Comparisons (5th –20th years) with baseline	1.4 ± 0.5, 1.4 ± 0.6, 1.2 ± 0.4 vs 1.5 ± 0.6, NS	–	–
		Udesen et al. (2019)	Age 29.1 years, n = 40	Age 30.0 years, n = 8	5.8 years	Median comparison with baseline	1.7 (1.2–2.3) vs 1.6 (1.1–2.2), NS	0.7 (0.5–1.2) vs 0.8 (0.8–1.1), NS	–
Menstrual regularity		Brown et al. (2011)	Age 28.6 years, n = 254	Age 29.9 years, n = 41	7 years	% Regular cycles from baseline	4.6% vs 0%, P < 0.001	100% vs 100%, NS	P < 0.001
		Schmidt et al. (2011a)	Age 49.4 years, n = 25	Age 49.7 years, n = 68	21 years	Mean age of menopause (years)	50.1 ± 7.4	51.5 ± 4.8	P = 0.419
Chronic anovulation		–	–	–	–	–	–	–	–
Ovarian hyperstimulation syndrome		–	–	–	–	–	–	–	–
Pregnancy viability		–	–	–	–	–	–	–	–

A₄, androstenedione; AMH, anti-Müllerian hormone; DHEAS, dehydroepiandrosterone sulphate; FAI, free androgen index; mFG, modified Ferriman-Gallwey score; NS, not significant; PCOS, polycystic ovary syndrome; SHBG, sex hormone-binding globulin; T, testosterone.

Table III. Pregnancy-related outcomes.
Outcomes	Study author, year	Baseline age, sample size		Mean follow-up duration	Effect measures	Observed estimates
Outcomes	Study author, year	PCOS group	Non-PCOS group	Mean follow-up duration	Effect measures	Within PCOS group comparison	Within non-PCOS group comparison	PCOS group versus non-PCOS group
Live births	–	–	–	–	–	–	–	–
Miscarriage	Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational week	27 gestational weeks	Cumulative rates	16.0%	5.3%	P = 0.004
	Jakubowicz et al. (2002)	Age 30 years, n = 31	–	4.5 years	Cumulative proportion (%)	41.9%	–	–
	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Incidence proportion, RR	3.8%	1.9%	2.0 RR, 95% CI: 0.187–21.0
Stillbirths	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Incidence proportion, RR	1.9%	3.8%	0.5 RR, 95% CI 0.04–5.3
Neonatal mortality	–	–	–	–	–	–	–	–
Gestational weight gain	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Mean weight gain (kgs)	10.0 ± 5.4	10.8 ± 6.3	P = 0.513
Gestational diabetes	Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational weeks	27 gestational weeks	Cumulative rates	14.7%	5.3%	P = 0.011
	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Incidence proportion, RR	26.9%	9.6%	2.8 RR, 95% CI 1.08–7.2
Preterm birth	Jakubowicz et al. (2002)	Age 30 years, n = 31	No controls	4.5 years	Cumulative proportion (%)	33.3%	–	–
	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Incidence proportion, RR	11.5%	23.0%	0.5 RR, 95% CI 0.2–1.2
Hypertensive disease in pregnancy	Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational weeks	27 gestational weeks	Cumulative rates	12.7%	5.3%	P = 0.042
	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Incidence proportion, RR	9.6%	11.5%	0.5 RR, 95%CI 0.27–2.5
Baby birth weight	Palomba et al. (2014)	Age 27.8 years, n = 150, 5.4 gestational weeks	Age 27.4 years, n = 150 5.4 gestational weeks	27 gestational weeks	Mean	3105.3 ± 346.1(g)	3179.8 ± 300.4 (g)	P = 0.039
	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Mean	3,055 ± 552 (g)	2,976 ± 621 (g)	P = 0.513
Major congenital abnormalities	Reyes-Munoz et al. (2012)	Age 29 years, n = 52	Age 29 years, n = 52	29.4 gestational weeks	Incidence proportion	1.9%	1.9%	1.0 RR, 95% CI 0.06 to 15

kgs, kilograms; PCOS, polycystic ovary syndrome; RR, risk ratio.

Table IV. Psychologic and oncological outcomes.
Outcomes	Study author, year	Baseline age, sample size		Mean follow-up duration	Effect measures	Observed estimates
Outcomes	Study author, year	PCOS group	Non-PCOS group	Mean follow-up duration	Effect measures	Within PCOS group comparison	Within non-PCOS group comparison	PCOS group versus non-PCOS group
Psychologic outcomes
Depression	Altinok et al. (2014)	Age 29 years, n = 1124	Age 29.0 years, n = 4213	PCOS: 6.8 years, controls: 7.2 years	Incidence proportions/HR (antidepressant prescription)	20%	15%	Versus population control: 0.75 HR (95% CI 0.64 to 0.88), P < 0.001 vs HTN control; P < 0.020
	Greenwood et al. (2019a)	Age 29 years, n = 163	No controls	5.5 years	BDI-FS score: median change from baseline Depression risk antidepressant use: % change	BDI-FS score: 0 (–2 to 1) enduring depression: 63% recovery: 37%	–	–
	Greenwood et al. (2019b)	Age 23 to 35 years, n = 83	Age 23 to 35 years, n = 1044	30 years at 5-year intervals	CES-D score change over lifetime	Score range: 11.2 to 13.4	Score range: 9 to 11.5	Coefficient, 2.51, 95% CI 1.49 to 3.54, P < 0.001
	Kerchner et al. (2009)	Age 32 years, n = 60	No controls	22 ± 3.7 months	Depression incidence and persistence from first survey	19% (incidence) 21.6% (persistence)	–	–
Anxiety	Harnod et al. (2020)	Age 27.7 years, n = 7026	Age 27.7 years, n = 28 104	16 years	Incidence rate and HR	15.3 per 1000	12.8 per 1000	1.18 HR 95% CI 1.07–1.30
Eating disorders	–	–	–	–	–	–	–	–
Oncologic outcomes
Atypical hyperplasia	Cheng-Che et al. (2015)	Age 27 years, n = 3566	Age 7 years, n = 14 264	7.15 years	Incidence proportion/HR (breast cancer)	0.39%	0.21%	Cox: 1.98 HR (95% CI 1.03 to 3.80) Monte carlo: NS
	Schmidt et al. (2011b)	Age 49.4 ± 4.9 years, n = 25	Age 49.7 ± 5.6 years, n = 68	21 years	Incidence proportion (breast cancer)	9.4%	7.4%	NS
Endometrial cancer	Schmidt et al. (2011b)	Age 49.4 ± 4.9 years, n = 25	Age 49.7 ± 5.6 years, n = 68	21 years	Incidence proportion	0	0	–
	Cheng-Che et al. (2015)	Age 27 years, n = 3566	Age 7 years, n = 14 264	7.15 years	Incidence proportion/HR (uterine cancer)	0.14%	0.01%	8.4 HR 95%CI 1.6 to 43.9

BDI-FS, Beck Depression Inventory Fast Screen; CES-D, Center for Epidemiologic Studies-Depression; HR, hazard ratio; NS, not significant; PCOS, polycystic ovary syndrome.