Geographical Variations in Polycystic Ovarian Morphology: Comparison of India- and United States-Based Women with Polycystic Ovary Syndrome

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript addresses an important and clinically relevant question by comparing ovarian ultrasound morphology in women with PCOS from India and the United States, and by examining associations between morphologic markers and reproductive/metabolic characteristics. The topic is timely given ongoing debate about whether region-specific PCOM thresholds are needed. However, several methodological and analytical limitations reduce confidence in some inferences, and the current evidence is not fully sufficient to support the broader conclusions.

Major limitations/concerns:

• Differences in scan timing and clinical context between cohorts (e.g., natural follicular phase vs random timing, and withdrawal bleeding protocols) introduce a substantial risk of systematic bias in ovarian volume and follicle visibility, complicating cross-region comparisons.

• The analytic approach relies largely on unadjusted group comparisons and bivariate correlations, with limited control for key confounders (e.g., BMI, age, phenotype composition, menstrual status), making it difficult to disentangle geographic effects from differences in clinical severity and cohort composition.

• Laboratory and metabolic measurements were obtained using different platforms and thresholds across sites (including differing testosterone cutoffs and glucose/lipid measurement methods), which may materially affect both phenotype classification and reported metabolic differences between regions.

• Multiple outcomes and correlation tests are presented with emphasis on p-values and limited reporting of effect sizes and uncertainty, increasing the risk of spurious significance and making clinical relevance harder to judge.

• The cohorts originate from different sampling frames (clinical patient records vs research recruitment), raising concerns about selection bias and external validity; this likely influences the observed phenotype distribution and severity.

Minor issues:

• Apparent unit inconsistencies/possible typographical errors in hormone units (e.g., testosterone units) require careful verification.

• The manuscript’s presentation would benefit from clearer prioritization of key outcomes and more explicit linkage of observed differences to clinical significance, although the current framing can be difficult to interpret given the heterogeneity and limitations above.

Author Response

General comments: The manuscript addresses an important and clinically relevant question by comparing ovarian ultrasound morphology in women with PCOS from India and the United States, and by examining associations between morphologic markers and reproductive/metabolic characteristics. The topic is timely given ongoing debate about whether region-specific PCOM thresholds are needed. However, several methodological and analytical limitations reduce confidence in some inferences, and the current evidence is not fully sufficient to support the broader conclusions.

Response:

Major limitations/concerns:

Comments 1: Differences in scan timing and clinical context between cohorts (e.g., natural follicular phase vs random timing, and withdrawal bleeding protocols) introduce a substantial risk of systematic bias in ovarian volume and follicle visibility, complicating cross-region comparisons.

Response 1: We agree that ovarian morphology changes throughout the menstrual cycle. For this reason, analyses were standardized to the early follicular phase whenever possible. Given that the study population was oligomenorrheic, if exact menstrual timing was unknown or women reported long intervals between menstruation, images were collected in the absence of a dominant follicle or corpus luteum (lines 98-99). Standardization to these physiologic timepoints ensured consistent visualization of the ovarian markers of interest, including follicle populations, as demonstrated in our recent publication (Carter et al. 2026: doi: 10.1002/jum.70040). We are unaware of data showing differences in ovarian morphology in the follicular phase following natural versus pharmacologically-induced menstrual bleeding. We have acknowledged this as a limitation in lines 362-367.

Comments 2: The analytic approach relies largely on unadjusted group comparisons and bivariate correlations, with limited control for key confounders (e.g., BMI, age, phenotype composition, menstrual status), making it difficult to disentangle geographic effects from differences in clinical severity and cohort composition.

Response 2: Age and BMI did not differ by geographic location (Tables 1 and 2). As such, our statistician advised that that adjustment for age and BMI was not necessary. Additionally, introducing covariate adjustment given our non-parametric primary analysis would require a different modeling framework. The subgroup analysis involving only women with Phenotype A accounts for menstrual status (Tables 2, 4, and 6; Figure 2). We have moved these findings from the previous Supplemental Data to the manuscript full text.

Comments 3: Laboratory and metabolic measurements were obtained using different platforms and thresholds across sites (including differing testosterone cutoffs and glucose/lipid measurement methods), which may materially affect both phenotype classification and reported metabolic differences between regions.

Response 3: We agree this is an inherent limitation of multi-site studies, particularly those retrospective in nature. We agree that both populations represent PCOS cohorts as defined by the technology and clinical expertise available at each site. We have acknowledged this limitation in lines 357-361. However, in the case of our primary outcomes, all post-hoc measurements of ovarian morphology were made prospectively at a single site and represent a strength. Ultrasound images were collected using the same ultrasound system (lines 135-136) and standardized measurement techniques were applied post-hoc to images collected from both sites (lines 141-163). We have better clarified this point throughout the manuscript.

Comments 4: Multiple outcomes and correlation tests are presented with emphasis on p-values and limited reporting of effect sizes and uncertainty, increasing the risk of spurious significance and making clinical relevance harder to judge.

Response 4: We now include Cohen’s d effect sizes for our continuous variables (Tables 1-4). The bivariate analysis was intended to be exploratory (lines 343-345 and 361-362). We have tempered our language throughout the manuscript on the hypothesis generating nature of the results. We have also removed the bivariate analysis involving metabolic status markers to limit the number of exploratory analyses.

Comments 5: The cohorts originate from different sampling frames (clinical patient records vs research recruitment), raising concerns about selection bias and external validity; this likely influences the observed phenotype distribution and severity.

Response 5: We agree that this is a limitation of the study (lines 382-386). We have updated our methodology to reflect that sampling in both cohorts was consecutive, reducing some degree of selection bias (lines 80 and 93). We included a subgroup analysis of only women with Frank PCOS (Phenotype A) to account for differences in severity across clinical and research populations. We have moved the subgroup analysis to the main text to emphasize its relevance (Tables 2, 4, and 6; Figure 2). We are presenting all data available in a transparent manner. We believe this study provides a basis for future prospective studies that would more definitively address any need for regional specific criteria for polycystic ovaries (lines 373-375).

Minor issues:

Comments 6: Apparent unit inconsistencies/possible typographical errors in hormone units (e.g., testosterone units) require careful verification.

Response 6: Thank you for noting this. We verified all hormone units and corrected a typographical error in the Methods (line 113).

Comments 7: The manuscript’s presentation would benefit from clearer prioritization of key outcomes and more explicit linkage of observed differences to clinical significance, although the current framing can be difficult to interpret given the heterogeneity and limitations above.

Response 7: We appreciate the reviewer’s comment. We have updated the Introduction to clearly indicate that our primary outcome is ovarian morphology (line 145). Also, given the limitations of the study design, we have tempered our language to reflect the hypothesis generating nature of our results and the need for future prospective studies (lines 373-375).

Reviewer 2 Report

Comments and Suggestions for Authors

The purpose of this study was to directly compare ovarian morphology assessed by ultrasound in women with polycystic ovary syndrome (PCOS) residing in two geographical regions: India and the United States. In addition, the extent to which ovarian morphology reflected reproductive symptomatology across these regions was explored.

However, despite the large amount of data collected—including metabolic characteristics of both populations, as well as sonographic features and markers of ovarian morphology, this study suffers from a significant design bias.

Two fundamentally different populations were compared, as acknowledged by the authors themselves. The Indian cohort consisted of infertile women diagnosed with PCOS who were seeking evaluation for infertility, whereas the U.S. group comprised individuals who voluntarily enrolled in a research study on reproductive health and/or the pathophysiology of PCOS. It was not specified whether the North American population had previous offspring, nor in what proportion. Therefore, clinic-based populations in India are more likely to exhibit a more severe PCOS phenotype compared with volunteers from the general population, among whom milder and ovulatory forms of PCOS are relatively more common.

Consequently, the degree of variability observed is insufficient to justify regional definitions of ovarian morphology between these two populations.

Author Response

Comments 1: The purpose of this study was to directly compare ovarian morphology assessed by ultrasound in women with polycystic ovary syndrome (PCOS) residing in two geographical regions: India and the United States. In addition, the extent to which ovarian morphology reflected reproductive symptomatology across these regions was explored. However, despite the large amount of data collected—including metabolic characteristics of both populations, as well as sonographic features and markers of ovarian morphology, this study suffers from a significant design bias. Two fundamentally different populations were compared, as acknowledged by the authors themselves. The Indian cohort consisted of infertile women diagnosed with PCOS who were seeking evaluation for infertility, whereas the U.S. group comprised individuals who voluntarily enrolled in a research study on reproductive health and/or the pathophysiology of PCOS. It was not specified whether the North American population had previous offspring, nor in what proportion. Therefore, clinic-based populations in India are more likely to exhibit a more severe PCOS phenotype compared with volunteers from the general population, among whom milder and ovulatory forms of PCOS are relatively more common.

Response 1: We agree that these are inherent limitations of multi-site analyses that are retrospective in nature (lines 359-362). To address cohort differences, we ran a subgroup analysis restricted to Frank PCOS (Phenotype A) and excluded the milder and ovulatory forms of PCOS. The subgroup analysis has been moved from supplemental to the main text (Figure 2 and Tables 2, 4 and 6). We are using all available data in a transparent manner and hope this paper sets the stage for prospective data collection as now stated in lines 373-375.

Comments 2: Consequently, the degree of variability observed is insufficient to justify regional definitions of ovarian morphology between these two populations.

Response 2: We have tempered our language in the Discussion to reflect the limitations of the study design and ultimately agree that variations in ovarian morphology are not sufficient to warrant regional definitions at this time (lines 370-371). Prospective studies are needed to fully address whether regional definitions for polycystic ovarian morphology are needed (lines 373-375).

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

It is OK for all changes.

Reviewer 2 Report

Comments and Suggestions for Authors

Many thanks to the authors for their efforts in reviewing and correcting the manuscript, which has expanded the information and improved its accuracy. We believe that in this form it can be published, once the conclusions highlight the need for further prospective studies to confirm regional differences in PCO.