Polycystic ovary syndrome (PCOS) is a relatively common chronic heterogeneous endocrine and metabolic disorder of women of reproductive age that is characterized by a combination of reproductive, hormonal and metabolic abnormalities, which vary in women with PCOS.[1] It is estimated to have a large variation in prevalence depending on the criteria used for its diagnosis.[2] A recent large-scale systematic review and meta-analysis found prevalence of 12.1% (Rotterdam criteria), 7.9% (original National Institutes of Health criteria), and 12.7% (Androgen Excess-PCOS criteria) in women aged 18 years or older.[3] Importantly, the prevalence based on Rotterdam criteria was especially high in the Eastern Mediterranean region (15.1%) and South-East Asia (14.3%), showing the significant regional burden of the syndrome.[3]

PCOS is now known to be a condition that affects more than just reproductive health.[4] Its clinical features include polycystic ovarian morphology, insulin resistance and impaired glucose tolerance, chronic anovulation, clinical and biochemical hyperandrogenism, excessive growth of peripheral fat, increased long-term cardiometabolic risk, hypertension and dyslipidaemia, and menstrual irregularity.[5] The 2023 International Evidence-Based Guideline states that women with PCOS have a higher risk of metabolic and cardiovascular complications, and that the clinical picture is very variable, and should not be assumed to be a single, uniform disorder.[6]

According to the Rotterdam criteria, PCOS is categorized into four major phenotypes based on the presence or absence of hyperandrogenism (HA), ovulatory dysfunction (OD) and polycystic ovarian morphology (PCOM): phenotype A (HA+OD+PCOM), phenotype B (HA+OD), phenotype C (HA+PCOM), and phenotype D (OD+PCOM).[7] The hormonal and metabolic properties of these phenotypes can vary greatly. Recent studies have shown that women with phenotypes of hyperandrogenism and ovulatory dysfunction might be more prone to adiposity, insulin resistance, dyslipidemia, impaired glucose tolerance, and metabolic syndrome than women with the non-hyperandrogenic OD+PCOM phenotype.[8]

A total of 442 women with PCOS were included in a cross-sectional study in 2024, and there was a significant difference between BMI, WC, insulin levels, HOMA-IR, triglycerides, total cholesterol, and LDL cholesterol among phenotypes, with hyperandrogenic phenotypes showing a higher risk of insulin resistance, metabolic syndrome, impaired glucose tolerance, NAFLD and dyslipidemia.[9] Hormonal heterogeneity is also important.[10] Variations in circulating testosterone, free androgen index, LH, FSH, estradiol, and anti-Müllerian hormone (AMH), as well as insulin sensitivity, may play a role in the unique clinical presentations seen in each of the PCOS phenotypes.[11]

However, prior studies have not adequately evaluated metabolic and hormonal parameters in combination, with some studies differing in study population, diagnostic criteria, phenotype definition, and measured biomarkers. Therefore, it’s probable that some clinically significant phenotypical differences are overlooked when looking at PCOS as a diagnostic group. The heterogeneity of PCOS may justify the use of individual phenotypes that have different hormonal and metabolic characteristics as a way to better risk stratify and tailor clinical evaluation.

However, evidence of the phenotype is variable and may differ from those reported elsewhere in certain regional populations due to differences in metabolism and endocrinology. Thus, the aim of the present study was to compare the metabolic and hormonal profiles of the various Rotterdam phenotypes of PCOS and to see if there were any differences in anthropometric, glycemic, lipid, or reproductive hormonal parameters between them. The results could offer clinically relevant evidence for the clinical assessment of metabolic and endocrine risks in women with PCOS.

The study was an analytical cross-sectional study which included women with a diagnosis of polycystic ovary syndrome (PCOS) made based on the Rotterdam diagnostic criteria. The study was carried out in the Department of Gynecology, and Obstetrics, Abbasi Shaheed Hospital, Karachi during six months’ period from 1st November 2025 to 30th April 2026.

The sample size was derived using the OpenEpi online sample size calculator for body mass index (BMI) using data reported by Wen et al. (2024) who reported a mean BMI of 25.3±6.5 kg/m² for women with the ovulatory dysfunction–hyperandrogenism phenotype and 22.7±5.1 kg/m² for women with the ovulatory dysfunction–polycystic ovarian morphology phenotype. The difference between the two means is then 2.6 kg/m².[12] The OpenEpi calculation used a 95% confidence level, 80% statistical power, two-sided hypothesis testing, an equal allocation ratio (1:1), a standard deviation of 6.5 kg/m² in group 1 and 5.1 kg/m² in group 2, yielding a minimum sample size of 80 participants per group, which equals to a total of 176 participants.

A non-probability consecutive sampling technique was used.  Women between the ages of 18-40 who met the Rotterdam criteria for PCOS were included.[13] PCOS was diagnosed when at least two of the three criteria, i-e. ovulatory dysfunction, clinical or biochemical hyperandrogenism, polycystic ovarian morphology) were fulfilled after other disorders relevant to the diagnosis of PCOS had been excluded. The ability to perform all the necessary clinical, anthropometric, biochemical, hormonal, and ultrasonographic evaluations for phenotype classification was a requirement for all the participants. Participation was consented to in writing, and those who consented were included.

Pregnant or lactating women, women who had premature ovarian insufficiency, congenital adrenal hyperplasia, Cushing syndrome, androgen-secreting ovarian or adrenal tumors, hyperprolactinemia or clinically significant thyroid dysfunction were excluded. Other major endocrine conditions that could produce a hormone pattern similar to or significantly change the PCOS profile were also excluded. Hormonal therapy, COCP, anti-androgen, metformin, lipid-lowering, and other medical therapies reported to have a direct effect on reproductive or metabolic parameters within the prior three months were excluded from participation. Other factors included were women with known diabetes mellitus, severe hepatic or renal disease, and other significant systemic illness. These exclusions were made to reduce potential for confounding of metabolic and hormonal measurements.

Data were collected using a structured data collection proforma after ethical clearance from the relevant institutional ethics committee and informed consent from participants. Sociodemographic and reproductive data such as age, menstrual pattern, duration of symptoms, parity, history of infertility and medical history were collected. Anthropometric data were collected with standard methods and included weight, height, BMI, waist circumference, hip circumference, and WHR. Blood pressure readings were also taken. Clinical hyperandrogenism was determined by the modified Ferriman–Gallwey score and menstrual history by the questionnaire method. Pelvic ultrasonography was performed, and the polycystic ovarian morphology was classified based on the diagnostic criteria.

Venous blood samples were taken after a 12-hour fasting period for the determination of metabolic and hormonal parameters. Fasting blood glucose, fasting insulin, HbA1c, total cholesterol, triglycerides, HDL-C, LDL-C, and insulin resistance by the homeostatic model assessment of insulin resistance (HOMA-IR) were measured as part of metabolic investigations. Hormonal assessments performed included total testosterone, free testosterone/free androgen index, LH, FSH, estradiol, prolactin, TSH, and anti-Müllerian hormone (AMH), which were available at the study center. Standard laboratory procedures were employed in the laboratory investigations. Data were then gathered for the classification of the participants into Rotterdam phenotypes A, B, C, and D and for the comparison of their metabolic and hormonal profiles.

The data collected were entered, coded, and analyzed with SPSS version 26.0. The distribution of continuous variables was determined by the Shapiro–Wilk test and was reported as mean ± standard deviation (SD) or median with interquartile range (IQR) in the case of non-normally distributed variables. The frequencies and percentages of categorical variables were presented. One-way analysis of variance (ANOVA) was used to compare continuous variables across the four PCOS phenotypes and a post hoc test for pairwise comparisons was used for normally distributed variables; Kruskal–Wallis test was used for non-normally distributed variables. Chi-square test and Fisher's exact test were used to compare categorical variables. Spearman's correlation coefficient was used to determine the correlations between the selected hormonal and metabolic parameters. A multivariable linear or logistic regression analysis was conducted to assess if the association of PCOS phenotype persisted after potential confounders (age, BMI) were controlled. A p-value ≤0.05 was considered statistically significant.

A total of 176 women with PCOS were included, with 44 women per Rotterdam phenotype. There was no difference between the four groups with regard to age, duration of symptoms, parity, or infertility history. In line with the phenotype definitions, the clinical hyperandrogenism and menstrual irregularity were significantly more common in phenotype A, B, and D, with a lower frequency in phenotype C. The hyperandrogenic phenotypes also had significantly higher Ferriman–Gallwey scores, while polycystic ovarian morphology was noted in all phenotypes except B (Table 1)

Table 1. Sociodemographic and reproductive characteristics of participants according to PCOS phenotype (n=176)

The anthropometric assessment revealed differences between the phenotypes in weight, BMI, waist and hip circumference, and WHR. Phenotypes A and B tended to have higher adiposity and central adiposity scores, while phenotype D had the lowest scores. There were no significant differences in height, systolic blood pressure, or diastolic blood pressure between the groups. (Table 2)

Table 2. Anthropometric and blood pressure parameters according to PCOS phenotype

A significant difference was seen between the phenotype and glycemic, insulin-resistance and lipid parameters. Similarly, fasting glucose, 2-hour glucose, HbA1c, fasting insulin, and HOMA-IR were significantly elevated and total cholesterol, triglycerides, and LDL-C were elevated in phenotypes A and B, but were lower in phenotype D, while HDL-C was lower in phenotypes A and B. (Table 3)

Table 3. Glycemic, insulin-resistance and lipid parameters according to PCOS phenotype

There were also large differences between the phenotypes in the hormonal profile. Hyperandrogenic phenotypes (especially phenotypes A–C) had significantly elevated total testosterone, free androgen index, LH, LH/FSH ratio, and AMH levels, and significantly decreased SHBG levels. There was no significant difference in the levels of FSH, estradiol, prolactin, or TSH between the four groups. (Table 4)

Table 4. Hormonal profile according to PCOS phenotype

Metabolic complications were significantly more common in phenotypes A and B than in phenotype D, and showed significant differences between phenotypes, with phenotype D having the lowest frequency in each case. There was no significant difference between the groups in terms of the frequency of type 2 diabetes mellitus. (Table 5)

Table 5. Frequency of metabolic abnormalities according to PCOS phenotype

After adjustment for age, BMI and waist circumference, phenotypes A, B, and C were independently associated with insulin resistance compared to phenotype D in the multivariable analysis. Waist circumference and BMI were also independently linked with insulin resistance. Testosterone, free androgen index, and BMI were all positively correlated with HOMA-IR, and SHBG was negatively associated with HOMA-IR, whereas AMH was only weakly but statistically significantly positively correlated with HOMA-IR by correlation analysis. (Table 6)

Table 6. Multivariable analysis and correlations of hormonal and metabolic parameters

In the current study, significant differences in metabolic and hormonal characteristics were found between women with PCOS based on Rotterdam phenotype. Generally, both phenotypes A and B (hyperandrogenism and ovulatory dysfunction) presented with higher levels of adiposity, insulin resistance, dyslipidaemia, and adverse glycemic parameters, while phenotype D (non-hyperandrogenism) presented with the lowest levels of these parameters. Testosterone, free androgen index, LH/FSH ratio and AMH were also higher in hyperandrogenic phenotypes, accompanied by lower SHBG. These finding reinforce the view that PCOS is not a homogeneous condition and is a spectrum of conditions.

The anthropometric results of our study revealed significantly higher weight, BMI, WC, HC, and WHR in phenotypes A and B than in phenotype D. This likewise follows the prospective study by Mehra et al. (2023) in which hyperandrogenic phenotypes were associated with a higher frequency of glucose and lipid abnormalities and phenotype A was associated with a strong correlation between BMI and metabolic parameters. They were also able to confirm that the least metabolic disturbance was in phenotype D.[14]  Gürsu et al. also found that BMI was significantly higher in phenotype A than in the other Rotterdam phenotypes, which further confirmed a link between the complete phenotype and increased adiposity.[15]

Insulin resistance was a significant finding we made in our research. In phenotypes B and A there was the highest level of HOMA-IR and fasting insulin, and in almost 75% of the participants of these phenotypes insulin resistance was present, whereas in phenotype D, the lowest level of HOMA-IR and fasting insulin was found. This is in line with Gürsu et al., who found a significant increase in fasting insulin and HOMA-IR in phenotype A compared to the other three phenotypes. Wen et al. (2024) performed a study on 442 women with PCOS and compared the three phenotypes. They discovered that the insulin levels and

HOMA-IR were higher in the OD-HA and HA-PCOM groups than in the OD-PCOM group, with significant differences in the prevalence of insulin resistance and metabolic syndrome. These observations  reflect our ability to identify women with a more metabolically adverse profile from the presence of hyperandrogenism, particularly in combination with ovulatory dysfunction.[12, 15]

This correlation is further confirmed by the lipid results of our study. Mehra et al. also noted that phenotypes A and B had significantly higher total cholesterol, triglycerides, and LDL-C, while having significantly lower HDL-C than phenotype D.[14] Wen et al. found similar results, with significantly higher total cholesterol, triglycerides, and LDL-C occurring in the OD-HA and HA-PCOM groups compared with the OD-PCOM group.[12] Our results are similar to those obtained by Mitra et al. (2024), who analyzed 144 women stratified by all four Rotterdam phenotypes and observed no significant differences in IR, lipid profile and MS among the phenotypes despite the differences in anthropometry and AMH. This difference could be due to differences in population size, distribution of phenotypes, ethnic distribution, BMI distribution, or statistical power in identifying between-phenotype metabolic differences.[16]

In our study, there was also a correlation between phenotype and glucose metabolism. Phenotypes A and B had significantly higher fasting glucose, 2-hour glucose, and HbA1c levels, and impaired glucose tolerance was more common in these groups. Our data is consistent with that of Wen et al., who found a much greater prevalence of impaired glucose tolerance, insulin resistance and metabolic syndrome in the hyperandrogenic phenotypes than in OD-PCOM.[17] In a much larger study conducted across the country in India, Ganie et al. (2024) also showed a significant prevalence of dyslipidemia, NAFLD, impaired glucose tolerance, diabetes, and metabolic syndrome in women diagnosed with PCOS, but their distribution of phenotypes was quite different, with phenotype C being the most common.[18] This difference in the distribution of phenotypes between the two studies (hospital-based vs population-based) highlights the impact of population characteristics and recruitment site on phenotype proportions.

It is also biologically plausible that increased NAFLD and metabolic syndrome prevalence among phenotypes A and B in our study could be due to the association of these two phenotypes with greater central adiposity and insulin resistance, as hyperandrogenism is associated with both features. Wen et al. found that the hyperandrogenic phenotypes were significantly more likely than the OD-PCOM phenotype to have NAFLD and metabolic syndrome.[17]  In contrast, Mitra et al. found no differences in insulin resistance or metabolic syndrome between Rotterdam phenotypes.[16] Based on the available literature, a relationship between Rotterdam phenotype and metabolic risk is reproducible in some populations but not universal, indicating the need to take BMI, WC, and other metabolic determinants into account, in addition to the Rotterdam phenotype classification.

Hyperandrogenic and non-hyperandrogenic phenotypes were well distinguished by the hormonal findings in our study. Phenotypes A, B, and C had significantly higher total T and free androgen index, and significantly lower SHBG than phenotype D, while significantly higher total T, free T, and lower SHBG than phenotypes A and B. These results corroborate those of Mehra et al. who reported higher biochemical and metabolic abnormalities in hyperandrogenic phenotypes.[16] A more recent study by Savran Üçok et al. involving 226 women confirmed that phenotype A had the most severe hyperandrogenic and metabolic changes, whereas phenotype D had the least, and FAI was highly correlated with hirsutism, while SHBG was inversely correlated.[19] These findings are similar to our results, and emphasize the importance of androgen excess in identifying clinically significant subgroups of PCOS.

The LH findings in our study also demonstrated higher LH concentrations and LH/FSH ratios in the hyperandrogenic phenotypes, particularly phenotypes A and B. Gürsu et al. reported significantly higher LH and LH/FSH ratios in phenotype A, whereas FSH, estradiol, prolactin, and TSH did not differ significantly between phenotypes. This is generally in agreement with the other observation that FSH, estradiol, prolactin and TSH were similar among groups.[15] A cross-sectional study of clinical phenotypes of PCOS in 2023 also found distinct hormonal patterns for LH, LH/FSH ratio, testosterone and FAI, but with a different set of phenotypes than the traditional Rotterdam classification.[20]

In our study, AMH levels varied significantly across the phenotypes, with the highest level being in phenotype C and levels above in the other hyperandrogenic phenotypes than in phenotype D. This discovery needs to be understood in the context of PCOM, as AMH represents the stock of small antral follicles. The AMH varied significantly between the four Rotterdam phenotypes, with the highest level seen in phenotype D and the lowest level in phenotype B, showing that AMH does not necessarily follow the same trend between each population. Likewise, the study by Mani and John in 2026 specifically examined AMH in PCOS phenotypes and controls, continuing to show interest in AMH as a phenotyped biomarker.[21]  A recently published systematic review and meta-analysis of AMH (2026) also highlighted significant differences in AMH between phenotypes and issues with standardization of assays and phenotype-specific interpretation.[22]

Our study also supports an interaction between androgen excess, adiposity, and insulin resistance through correlation analysis. A positive correlation was observed between Testosterone, free androgen index, BMI and HOMA-IR, while SHBG was inversely correlated with HOMA-IR. These relationships are similar to those reported by Mehra et al. who observed significant correlations between BMI and biochemical abnormalities, especially in phenotype A.[14]  The results are also similar to those obtained in the study conducted by Savran Üçok et al. in 2026, where FAI, testosterone, and SHBG were closely related to the clinical presentation of hyperandrogenism and metabolic indices.[19] Importantly, the correlation between AMH and HOMA-IR in our study was weaker compared to that of BMI and the various androgen parameters, indicating that AMH might be a direct reflection of the ovary's morphology or follicular excess rather than systemic metabolic abnormality.

In this study, which involved over 2,500 women with PCOS, van der Ham et al. (2024) also found that the concept of conventional Rotterdam phenotypes may only capture a part of the biological heterogeneity of PCOS, with the metabolically characterized PCOS subgroup having several metabolically and reproductively different subtypes, particularly the metabolically characterized phenotype A.[23] This finding resonates with our study findings as it indicates a potential clinical value of Rotterdam phenotypes as markers but not as full metabolic risk markers. Likewise, the large study in India in 2024 showed significant differences in the distribution of phenotypes in comparison to our balanced study population, highlighting the population heterogeneity.[18]

These metabolic differences are clinically relevant as evidenced by other phenotype studies. In a large IVF/ICSI cohort, Wang et al. (2022) observed that Rotterdam phenotypes A and D had higher rates of adverse pregnancy outcomes than other phenotypes and reported differences in BMI, fasting glucose, LH, testosterone, and AMH among the phenotypes. Although their main purpose was to identify reproductive rather than metabolic groups, their results support the idea that phenotypically distinct groups can be identified with clinically distinct groups.[24] In contrast, the present study specifically shows that differences in the phenotype are present in several domains including anthropometric, glycemic, lipid, insulin-resistance, and hormonal domains.

In conclusion, our results corroborate the recent literature, indicating that hyperandrogenic phenotypes (A and B) of PCOS have an overall more metabolically adverse profile than phenotype D, and that phenotype C generally do not show hyperandrogenism or insulin resistance. The greatest strength of the present study is its combination of assessment of anthropometric, glycemic, lipid, insulin-resistance, and hormonal parameters with the use of multivariable analysis. The results reinforce the use of a phenotype-based approach when diagnosing PCOS, but do not preclude the use of other measures of metabolic and hormonal status, such as BMI, waist circumference, and other individual metabolic and hormonal variables.

LIMITATIONS

There were a number of limitations in this study. The analytical cross-sectional design of the study prevented the determination of temporal and causal relationships between PCOS phenotype and metabolic or hormonal abnormalities. The study involved only one tertiary care hospital and was a consecutive non-probability sampling, which might restrict the generalizability of the findings to other groups. Although well-balanced for analytical purposes, it is important to note that the distribution of phenotypes in the broader population of women with PCOS may be different. Other metabolic and hormonal parameters may also be affected by lifestyle, dietary, genetic and socio-economic factors that have not been measured. Furthermore, insulin resistance was assessed by HOMA-IR instead of actual insulin sensitivity. More extensive assessment of potential confounders and larger multicenter prospective studies are warranted that follow metabolic outcomes longitudinally.

Metabolic and hormonal characteristics were found to differ between women with different Rotterdam phenotypes of PCOS. Higher adiposity, insulin resistance, dyslipidaemia, adverse glycemic parameters, and androgen excess were linked to hyperandrogenic phenotypes A and B, while the metabolic disturbances were least severe for phenotype D. Even after adjusting for age, BMI, and waist circumference, phenotype was correlated with insulin resistance. The results of this work corroborate the phenotype-based approach to the clinical evaluation of PCOS and highlight the need for personalized assessment of metabolic and hormonal risk.

1. joham, A.E., et al., Polycystic ovary syndrome. The lancet Diabetes & endocrinology, 2022. 10(9): p. 668-680.
2. Chiaffarino, F., et al., Prevalence of polycystic ovary syndrome in European countries and USA: a systematic review and meta-analysis. European Journal of Obstetrics & Gynecology and Reproductive Biology, 2022. 279: p. 159-170.
3. Neven, A.C.H., et al., Prevalence of polycystic ovary syndrome: a global and regional systematic review and meta-analysis. Hum Reprod Update, 2026. 32(3): p. 277-312.
4. Parua, S., et al., Exploring female infertility: A comprehensive review of polycystic ovary syndrome (PCOS) and its impact on reproductive health. Obesity Medicine, 2025. 55: p. 100619.
5. Pililis, S., et al., The cardiometabolic risk in women with polycystic ovarian syndrome (PCOS): From pathophysiology to diagnosis and treatment. Medicina, 2024. 60(10): p. 1656.
6. Jamshaid, M., et al., Bridging the Practice Gap: Polycystic Ovary Syndrome as an Under-Recognised Risk State for Ischaemic Stroke and Cardiovascular Disease. Journal of Clinical Medicine, 2026. 15(10): p. 3577.
7. Tatarchuk, T., et al., Distribution and anthropometric characteristics of Rotterdam criteria-based phenotypic forms of Polycystic ovaries syndrome in Ukraine. European Journal of Obstetrics & Gynecology and Reproductive Biology, 2024. 295: p. 104-110.
8. Wang, X. et al., Divergent pathophysiological drivers of polycystic ovary syndrome: insulin resistance independently fuels the hyperandrogenic phenotype whilst neuroendocrine factors dominate non-hyperandrogenic presentations. Frontiers in Endocrinology, 2026. 17: p. 1758861.
9. Wen, X., L. Wang, and E. Bai, Metabolic characteristics of different phenotypes in reproductive-aged women with polycystic ovary syndrome. Frontiers in Endocrinology, 2024. Volume 15 - 2024.
10. Emanuel, R.H., et al., A review of the hormones involved in the endocrine dysfunctions of polycystic ovary syndrome and their interactions. Frontiers in Endocrinology, 2022. 13: p. 1017468.
11. Malhotra, N., et al., Serum anti-mullerian hormone (AMH) levels among different PCOS phenotypes and its correlation with clinical, endocrine, and metabolic markers of PCOS. Reproductive Sciences, 2023. 30(8): p. 2554-2562.
12. Wen, X., L. Wang, and E. Bai, Metabolic characteristics of different phenotypes in reproductive-aged women with polycystic ovary syndrome. Frontiers in Endocrinology, 2024. 15: p. 1370578.
13. Kim, J., et al. Position and characteristics of adolescents diagnosed as PCOS under the original Rotterdam criteria but excluded under the 2018 updated guideline. in Endocrine Abstracts. 2023. Bioscientifica.
14. Mehra, T., et al., Correlation of Body Mass Index with Anthropometric and Biochemical Parameters Among Polycystic Ovary Syndrome Phenotypes. Indian J Clin Biochem, 2023. 38(2): p. 231-241.
15. Gürsu, T., A. Eraslan, and B. Angun, Comparison of body mass index, anti-müllerian hormone and insulin resistance parameters among different phenotypes of polycystic ovary syndrome. Gynecology and Obstetrics Clinical Medicine, 2022. 2(4).
16. Mitra, S., G.K. Saharia, and S.K. Jena, Cardio-metabolic risk in Rotterdam clinical phenotypes of PCOS. Ann Endocrinol (Paris), 2024. 85(1): p. 44-47.
17. Wen, X., L. Wang, and E. Bai, Metabolic characteristics of different phenotypes in reproductive-aged women with polycystic ovary syndrome. Front Endocrinol (Lausanne), 2024. 15: p. 1370578.
18. Ganie, M.A., et al., Prevalence, phenotypes, and comorbidities of polycystic ovary syndrome among Indian women. JAMA Network Open, 2024. 7(10): p. e2440583.
19. Savran Üçok, B., et al., Comparative analysis of hormonal and metabolic indices in phenotypic subgroups of polycystic ovary syndrome. J Turk Ger Gynecol Assoc, 2026. 27(1): p. 29-35.
20. Elsayed, A.M., et al., Clinical Phenotypes of PCOS: a Cross-Sectional Study. Reproductive Sciences, 2023. 30(11): p. 3261-3272.
21. Mani, M.M. and S. John, Diagnostic utility of serum anti-Müllerian hormone in different phenotypes of polycystic ovary syndrome: a case-control study from a tertiary care centre. International Journal of Reproduction, Contraception, Obstetrics and Gynecology, 2026. 15(7): p. 2560-2565.
22. Schwenck-Carvalho, P.J., et al., Anti-Müllerian Hormone Levels Across Phenotypes in Polycystic Ovary Syndrome: A Systematic Review and Meta-analysis. Journal of the Endocrine Society, 2026. 10(3).
23. van der Ham, K., et al., Clustering Identifies Subtypes With Different Phenotypic Characteristics in Women With Polycystic Ovary Syndrome. J Clin Endocrinol Metab, 2024. 109(12): p. 3096-3107.
24. Wang, Q., et al., Association of Polycystic Ovary Syndrome Phenotypes With Adverse Pregnancy Outcomes After In Vitro Fertilization/Intracytoplasmic Sperm Injection. Front Endocrinol (Lausanne), 2022. 13: p. 889029.