Clinical Determinants and Bone Metabolic Correlates of 24-h Urinary PGE2 and PGEM Excretion in Chinese Adults: A Multicenter Cross-Sectional Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript „Clinical Determinants and Bone Metabolism Markers Correlates of 24-Hour Urinary PGE2 and PGEM Excretion in Healthy Adults: A Multicenter Cross-Sectional Study” is very well written and clearly structured. Overall, a manuscript is scientifically sound, well referenced, and effectively guides the reader toward the research question.

The introduction is well written and have a clear logical structure. The methodology is also well constructed and inclusion of a large multicenter cohort across nine tertiary hospitals enhances the representativeness of the sample and reduces potential center-related bias. In addition, the exclusion criteria are clearly defined and appropriately applied, as the authors carefully eliminated conditions that could influence both PGE2 levels and bone metabolism, which is essential for the validity of the study objectives.

Another strength of this study is the 24-hour urine collection protocol,  with exclusion of periods potentially affecting urinary biomarkers (e.g., menstruation and infection).

The authors have done a good job adjusting for a wide range of relevant confounders. However, physical activity was not assessed, even though it could influence PGE2 levels and bone metabolism markers. This should be mentioned more clearly as a limitation.

There seems to be a small inconsistency that would benefit from clarification: diabetes is listed among the exclusion criteria, yet it also appears as a covariate in the adjusted models. It would be helpful if the authors briefly explained how diabetes was handled in the analysis and resolved this apparent discrepancy.

The results are presented clearly, with a balance between descriptive statistics and multivariable analyses.  It would help to clarify whether any sex-stratified analyses or interaction tests between age and sex were performed, especially in relation to postmenopausal status, considering its known effects on bone and inflammatory pathways.

The discussion is also well structured and provides reasonable biological explanations for the main findings. The authors also do a good job of placing their results in the context of existing literature and highlighting the complex nature of PGE2 signaling. It would be helpful for the authors to briefly acknowledge and discuss additional unaddressed limitations, particularly regarding menopausal status and physical activity. Reference section is good.

The English language is appropriate and understandable

A manuscript can be accepted after minor revision.

Author Response

Comments 1: The authors have done a good job adjusting for a wide range of relevant confounders. However, physical activity was not assessed, even though it could influence PGE2 levels and bone metabolism markers. This should be mentioned more clearly as a limitation.

Response 1: We thank the reviewer for raising this important point. We fully agree that physical activity is a relevant confounding factor that could affect both PGE2 levels and bone metabolism markers, and its absence in our study is indeed a clear limitation. We have taken this suggestion seriously and have now revised the Discussion section to state this limitation more explicitly and transparently.

Revised Discussion Section in Line 406:

“We did not assess physical activity, hormone replacement therapy (HRT) use or other inflammatory markers such as other prostaglandin metabolites and cytokines, which could have influenced the PGE2-bone metabolism associations or provided a broader inflammatory context.”

Comments 2: There seems to be a small inconsistency that would benefit from clarification: diabetes is listed among the exclusion criteria, yet it also appears as a covariate in the adjusted models. It would be helpful if the authors briefly explained how diabetes was handled in the analysis and resolved this apparent discrepancy.

Response 2: We sincerely thank you for your careful reading and for pointing out this inconsistency. We apologize for this oversight, and it was entirely our writing error in the exclusion criteria. In our actual study, we did not exclude diabetic participants; instead, we included them in the analysis and adjusted for diabetes as a covariate in all multivariable models, because diabetes may influence PGE2 and bone metabolism through inflammatory and renal pathways. To resolve this discrepancy, we have revised the exclusion criteria in the Methods section.

Revised Methods Section in Line 112:

 “Subjects with the following conditions were excluded: (1) serious diseases affecting the pulmonary, cardiovascular, gastrointestinal, hematopoietic, renal or nervous systems; (2) conditions known to affect bone metabolism, such as osteogenesis imperfecta, Paget’s disease of bone, primary hyperparathyroidism, rheumatoid arthritis or malignant tumors; (3) concurrent use of medications known to influence urinary PGE2 levels or bone metabolism (e.g., cyclooxygenase inhibitors, diuretics, synthetic steroid hormones, epinephrine or anticonvulsants); or (4) pregnancy or lactation.”

Comments 3: The results are presented clearly, with a balance between descriptive statistics and multivariable analyses.  It would help to clarify whether any sex-stratified analyses or interaction tests between age and sex were performed, especially in relation to postmenopausal status, considering its known effects on bone and inflammatory pathways.

Response 3: We sincerely thank the reviewer for this insightful comment. We conducted additional subgroup analyses stratified by sex and age group. We also performed menopause-stratified analyses among women, comparing the associations of 24-hour U-PGE2/U-PGEM with bone metabolism markers between premenopausal and postmenopausal participants. In addition, we formally tested the interaction between age and sex by including an age×sex interaction term in the multivariable regression models.

The overall association were broadly consistent across sex-stratified and age-stratified analyses. Among women, the direction and magnitude of the main associations were generally similar between premenopausal and postmenopausal participants. The age×sex interaction was not statistically significant between 24-hour U-PGE2/U-PGEM with bone metabolism-related markers (P>0.05).

The results were provided in the Supplementary Materials. We have also revised the Methods and Results sections accordingly. These additional analyses strengthen the robustness of our findings while acknowledging that residual confounding related to reproductive hormone profiles and other unmeasured factors cannot be completely excluded.

Revised Methods Section in Line 198:

“Additional subgroup analyses were performed after stratification by sex and age group. Among women, analyses were further stratified according to menopausal status. ”

Revised Result Section in Line 313:

“In sex-stratified and age-stratified analyses, the overall direction of the associations between urinary PGE2/PGEM excretion and the main bone metabolism-related markers was broadly consistent with that observed in the total population. Among women, similar patterns were observed in premenopausal and postmenopausal subgroups.”

Comments 4: The discussion is also well structured and provides reasonable biological explanations for the main findings. The authors also do a good job of placing their results in the context of existing literature and highlighting the complex nature of PGE2 signaling. It would be helpful for the authors to briefly acknowledge and discuss additional unaddressed limitations, particularly regarding menopausal status and physical activity.

Response 4: We sincerely thank the reviewer for this constructive suggestion. We completely agree that both menopausal status and physical activity are important factors that could influence PGE2 levels and bone metabolism markers. First, we re-examined our data and successfully obtained and incorporated information on menopausal status for all female participants. We also performed menopause-stratified analyses among women, comparing the associations of 24-hour U-PGE2/U-PGEM with bone metabolism markers between premenopausal and postmenopausal participants. The results were provided in the Supplementary Materials. We have also revised the Methods and Results sections accordingly. Second, regarding physical activity, we acknowledge that we did not collect data on this variable. As the reviewer rightly pointed out, physical activity may affect PGE2 secretion and bone turnover through mechanical loading and inflammatory pathways. We have now explicitly discussed this as a notable limitation in the revised Discussion section.

Revised Methods Section in Line 199:

“Among women, analyses were further stratified according to menopausal status. ”

Revised Result Section in Line 316:

“Among women, similar patterns were observed in premenopausal and postmenopausal subgroups.”

Revised Discussion Section in Line 406:

“We did not assess physical activity, hormone replacement therapy (HRT) use or other inflammatory markers such as other prostaglandin metabolites and cytokines, which could have influenced the PGE2-bone metabolism associations or provided a broader inflammatory context.”

Reviewer 2 Report

Comments and Suggestions for Authors

Manuscript Title: Clinical Determinants and Bone Metabolism Marker Correlates of 24-Hour Urinary PGE2 and PGEM Excretion in Healthy Adults: A Multicenter Cross-Sectional Study

The manuscript addresses an interesting and clinically relevant topic and is generally well written. However, several issues should be addressed to improve the clarity, methodological rigor, and overall quality of the manuscript. The following comments and suggestions are provided for the authors’ consideration:

1) Please consider including a schematic diagram or flowchart illustrating the study design, participant recruitment, sample collection, and analytical procedures to improve the clarity and readability of the study methodology.

2) Tables 1 and 2 should be reformatted into a horizontal layout, if feasible, to improve readability and facilitate comparison of the presented data.

3) The meaning of the statistical significance indicator (**) in Figure 2 should be clearly defined in the figure legend and explained in the main text where appropriate.

4) Information regarding menopausal status should be collected and presented in the Results section, as estrogen deficiency during menopause is known to accelerate bone resorption and influence bone metabolism. Additionally, the use of hormone replacement therapy (HRT) should be considered as an inclusion/exclusion criterion or, alternatively, discussed as a potential confounding factor.

5) The accreditation or certification status of the central laboratory responsible for biochemical analyses should be provided in Section 2.3 (Laboratory Assays). In addition, a brief description of the sample preparation procedures for urinary PGE2 and U-PGEM measurements should be included to improve methodological transparency and reproducibility.

6) The reported median BMI in Section 3.1 (22.92 kg/m²) appears inconsistent with the value presented in Table 1 (23.29 ± 4.23 kg/m²). Please verify these values and provide clarification where necessary.

7) The units used for 24-hour urinary PGE2 and U-PGEM should be presented consistently throughout the manuscript. Specifically, the text reports values in pg/mmol creatinine, whereas Figure 1 presents values in pg/mg Cr. Please verify and standardize the units.

8) The number of participants included in each group shown in Figure 2 should be reported either in the main text, figure legend, or supplementary materials to facilitate interpretation of the results.

9) The resolution and overall quality of Figures 3 and 4 should be improved to enhance readability and allow clearer visualization of the presented data.

Author Response

Comments 1: Please consider including a schematic diagram or flowchart illustrating the study design, participant recruitment, sample collection, and analytical procedures to improve the clarity and readability of the study methodology.

Response 1: We sincerely thank the reviewer for this thoughtful and constructive suggestion. We completely agree that a visual overview of the study workflow would greatly enhance the clarity and readability of the methodology section, especially given the multi‑step nature of our procedures.

In response, we have now designed a comprehensive flowchart that systematically depicts the entire study process, including participant recruitment, eligibility screening, sample collection, laboratory measurements, and the main analytical steps. This flowchart has been included as Figure 1 in the revised manuscript Line 120. We believe that this schematic diagram will help readers quickly grasp the overall study design and procedural sequence, complementing the detailed textual description.

Revised Methods Section in Line 106:

“A comprehensive flowchart illustrating the study design, participant recruitment, sample collection, and analytical procedures was showed in Figure 1. ”

Comments 2: Tables 1 and 2 should be reformatted into a horizontal layout, if feasible, to improve readability and facilitate comparison of the presented data.

Response 2: We sincerely thank the reviewer for this helpful suggestion. Accordingly, we reformatted both tables into a horizontal (landscape) layout in the revised manuscript Line 208 and 251.

Comments 3: The meaning of the statistical significance indicator (**) in Figure 2 should be clearly defined in the figure legend and explained in the main text where appropriate.

Response 3: We sincerely thank the reviewer for this helpful comment. In the revised manuscript, we have added an explicit explanation of the significance indicators in the main text and Figure 3 legend (due to the addition of Figure 1 Flowchart for the study process, Figure 2 has been changed to Figure 3).

Revised Figure 3 Legend in Line 230:

"Figure 3. Distribution of 24-hour U-PGE2 and U-PGEM by age group. (A) 24-hour U-PGE2; (B) 24-hour U-PGEM. The numbers of participants in the 18–30, 31–45, 46–60, and ≥61 years groups were 187, 156, 174, and 220, respectively. Overall differences across age groups were assessed using the Kruskal–Wallis test, followed by Bonferroni-adjusted post hoc pairwise comparisons. *P < 0.05; **P < 0.01. U-PGE2, urinary prostaglandin E2; U-PGEM, urinary prostaglandin E metabolite."

Comments 4: Information regarding menopausal status should be collected and presented in the Results section, as estrogen deficiency during menopause is known to accelerate bone resorption and influence bone metabolism. Additionally, the use of hormone replacement therapy (HRT) should be considered as an inclusion/exclusion criterion or, alternatively, discussed as a potential confounding factor.

Response 4: We sincerely thank the reviewer for this important and insightful comment. We fully agree that menopausal status and HRT use are critical factors that can profoundly affect bone metabolism, and we appreciate the opportunity to address them more thoroughly.

In response to your first point, we have now incorporated information on menopausal status for all female participants, and performed subgroup analyses stratified by menopausal status to examine whether the observed associations differ between the two groups. The results of these stratified analyses are provided in the Supplementary Materials. We believe this addition substantially strengthens our study by accounting for the known effects of estrogen deficiency on bone resorption and metabolism.

Regarding your second point on HRT use, we acknowledge that we did not collect detailed data on HRT in our original study, nor did we use it as an inclusion or exclusion criterion. We fully recognise that HRT may influence PGE2 levels and bone metabolism markers, and therefore its absence is a potential confounder. In the revised manuscript, we have now explicitly addressed this in the Discussion section.

Revised Methods Section in Line 193:

“Among women, analyses were further stratified according to menopausal status. ”

Revised Result Section in Line 300:

“Among women, similar patterns were observed in premenopausal and postmenopausal subgroups.”

Revised Discussion Section in Line 388:

“We did not assess physical activity, hormone replacement therapy (HRT) use or other inflammatory markers such as other prostaglandin metabolites and cytokines, which could have influenced the PGE2-bone metabolism associations or provided a broader inflammatory context.”

Comments 5: The accreditation or certification status of the central laboratory responsible for biochemical analyses should be provided in Section 2.3 (Laboratory Assays). In addition, a brief description of the sample preparation procedures for urinary PGE2 and U-PGEM measurements should be included to improve methodological transparency and reproducibility.

Response 5: We sincerely thank the reviewer for this valuable suggestion. We fully agree that providing the laboratory’s accreditation status and detailed sample preparation procedures is essential for ensuring methodological transparency and reproducibility.

In response, we first have now explicitly stated the accreditation status of the central laboratory where all biochemical analyses were performed in Methods Section. The laboratory is accredited under CNAS MT0048 (15189), and we have included the accreditation number and certifying body to ensure full transparency. Second, we have added a detailed description of the sample preparation procedures for urinary PGE2 and U-PGEM measurements in Methods Section. 

Revised Methods Section in Line 140 and 152:

“All blood specimens were sent to the central laboratory for biochemical evaluation, and the laboratory is accredited under CNAS MT0048 (15189). ”

“The 24-hour U-PGE2 and U-PGEM levels were detected using competitive enzyme-linked immunosorbent assays (ELISA; Cayman Chemicals, Ann Arbor, MI, USA; item 500141 for PGE2 and item 514531 for PGEM) according to the manufacturer’s instructions as we described in previous study (19). Prior to analysis, urine samples were centrifuged at 4 °C for 10 minutes, and the supernatants were then extracted and diluted to three different concentrations using the diluent provided in the ELISA kit. ” 

Comments 6: The reported median BMI in Section 3.1 (22.92 kg/m²) appears inconsistent with the value presented in Table 1 (23.29 ± 4.23 kg/m²). Please verify these values and provide clarification where necessary.

Response 6: We sincerely apologize for our oversight. We have carefully verified and revised the values accordingly.

Revised Result Section in Line 200:

“All the participants (men: 304; women: 433) had a mean age of 48.0 years and a mean BMI of 23.29 kg/m2.”

Comments 7: The units used for 24-hour urinary PGE2 and U-PGEM should be presented consistently throughout the manuscript. Specifically, the text reports values in pg/mmol creatinine, whereas Figure 1 presents values in pg/mg Cr. Please verify and standardize the units.

Response 7: We sincerely apologize for our oversight. We have since revised Figure 2 accordingly in Line 217 (due to the addition of Figure 1 Flowchart for the study process, Figure 1 has been changed to Figure 2).

Comments 8: The number of participants included in each group shown in Figure 2 should be reported either in the main text, figure legend, or supplementary materials to facilitate interpretation of the results.

Response 8: We sincerely thank the reviewer for this helpful suggestion. We have added the number of participants in each age group to the revised Figure 3 legend (due to the addition of Figure 1 Flowchart for the study process, Figure 2 has been changed to Figure 3).

Revised Figure 3 Legend in Line 230:

"Figure 3. Distribution of 24-hour U-PGE2 and U-PGEM by age group. (A) 24-hour U-PGE2; (B) 24-hour U-PGEM. The numbers of participants in the 18–30, 31–45, 46–60, and ≥61 years groups were 187, 156, 174, and 220, respectively. Overall differences across age groups were assessed using the Kruskal–Wallis test, followed by Bonferroni-adjusted post hoc pairwise comparisons. *P < 0.05; **P < 0.01. U-PGE2, urinary prostaglandin E2; U-PGEM, urinary prostaglandin E metabolite."

Comments 9: The resolution and overall quality of Figures 3 and 4 should be improved to enhance readability and allow clearer visualization of the presented data.

Response 9: We sincerely thank the reviewer for this helpful suggestion. We have re-uploaded the images, all of which have a DPI of over 400.

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript addresses an interesting and underexplored topic by establishing reference values for 24-hour urinary PGE2 and PGEM excretion in healthy adults and exploring their associations with markers of bone metabolism. The multicenter design and relatively large sample size are strengths.

Comments

1. The Introduction is well written and provides sufficient background on the biological functions of PGE2 and its involvement in several pathological conditions. The study aim is clearly stated. However, the rationale for investigating healthy individuals rather than patients with bone disorders should be more clearly articulated. Establishing physiological determinants and reference relationships is a valid objective; nevertheless, the Introduction should better explain how these findings may support the future application of U-PGE2 and U-PGEM as biomarkers in skeletal diseases. While the authors extensively discuss the association between PGE2 and various diseases, the transition to the investigation of U-PGE2 and U-PGEM in healthy adults and their associations with calcium–phosphorus homeostasis and bone turnover markers is somewhat abrupt and would benefit from a stronger biological and clinical justification.

 

1. Methods

The exclusion criteria indicate that subjects with diabetes mellitus were excluded; however, diabetes was included as a covariate in the adjusted regression models. The authors should clarify this apparent inconsistency.

1. Results

Statistically well executed and consistent with the objective.

4. The Discussion is generally well argued and places the findings in the context of the existing literature. I appreciate that the authors have addressed the main limitations of the study; however, further consideration of potential selection bias and residual confounding factors would strengthen the manuscript.
   1. residual confounding factors: Although the authors adjusted for several important covariates, the potential influence of residual confounding factors should be acknowledged. Variables such as menopausal status in women may affect both urinary PGE2/PGEM levels and bone metabolism markers and could therefore have influenced the observed associations.
   2. potential selection bias: The study population was recruited from tertiary care hospitals rather than from a community-based sample. Although extensive exclusion criteria were applied, this recruitment strategy may have introduced selection bias, as participants attending hospital-based health examinations may differ from the general healthy population in terms of health awareness, comorbidities, and lifestyle characteristics. The potential impact of this limitation on the generalizability of the proposed reference data should be discussed.

Author Response

Comments 1: However, the rationale for investigating healthy individuals rather than patients with bone disorders should be more clearly articulated. Establishing physiological determinants and reference relationships is a valid objective; nevertheless, the Introduction should better explain how these findings may support the future application of U-PGE2 and U-PGEM as biomarkers in skeletal diseases. While the authors extensively discuss the association between PGE2 and various diseases, the transition to the investigation of U-PGE2 and U-PGEM in healthy adults and their associations with calcium–phosphorus homeostasis and bone turnover markers is somewhat abrupt and would benefit from a stronger biological and clinical justification.

Response 1: We sincerely thank the reviewer for this thoughtful and constructive comment. We fully agree that the rationale for studying healthy individuals, rather than focusing directly on patient populations, was not sufficiently articulated in our original Introduction, and that the transition from disease‐oriented discussions to the healthy‐population investigation appeared abrupt.

In response, we have substantially revised the Introduction to address these points. Specifically, we have:

Strengthened the rationale by emphasising that establishing physiological reference ranges and identifying key determinants of U‑PGE2 and U‑PGEM in healthy adults are essential prerequisites for interpreting abnormal values in disease states. Without such normative data, it is difficult to distinguish pathological elevations from normal biological variation.

Smoothened the transition by explicitly linking the urinary biomarkers to bone metabolism in the healthy population. We now highlight that calcium–phosphorus homeostasis and bone turnover markers are well‑characterised physiological axes, and investigating their relationships with PGE2 metabolites in a healthy cohort can provide a baseline framework for understanding how these biomarkers behave in the absence of overt pathology. This, in turn, will inform future studies in skeletal disorders such as PHO, osteoporosis, and osteoarthritis.

Articulated future clinical applications more explicitly, including using U‑PGE2 and U‑PGEM to monitor disease activity, evaluate treatment responses (e.g., to COX‑2 inhibitors or other NSAIDs), and potentially serve as early warning markers for accelerated bone turnover or metabolic bone diseases.

Revised Introduction Section in Line 84, 93 and 99:

“Nevertheless, establishing normative reference data in general population constitutes an essential prerequisite for the reliable clinical application of urinary PGE₂ and PGEM as biomarkers. In the absence of well‑characterised physiological ranges and key influencing factors, the interpretation of elevated levels in disease state remains inherently ambiguous.”

“Clarifying these links could establish a baseline for future skeletal disease research, and aid early detection and therapy evaluation, especially for PGE₂‑targeted interventions.”

“The ultimate aim is to establish a robust normative framework that will facilitate the future application of these urinary biomarkers in the diagnosis, monitoring, and management of skeletal and other PGE₂‑related diseases.”

Comments 2: The exclusion criteria indicate that subjects with diabetes mellitus were excluded; however, diabetes was included as a covariate in the adjusted regression models. The authors should clarify this apparent inconsistency.

Response 2: We sincerely thank you for your careful reading and for pointing out this inconsistency. We apologize for this oversight, and it was entirely our writing error in the exclusion criteria. In our actual study, we did not exclude diabetic participants; instead, we included them in the analysis and adjusted for diabetes as a covariate in all multivariable models, because diabetes may influence PGE2 and bone metabolism through inflammatory and renal pathways. To resolve this discrepancy, we have revised the exclusion criteria in the Methods section.

Revised Methods Section in Line 112:

 “Subjects with the following conditions were excluded: (1) serious diseases affecting the pulmonary, cardiovascular, gastrointestinal, hematopoietic, renal or nervous systems; (2) conditions known to affect bone metabolism, such as osteogenesis imperfecta, Paget’s disease of bone, primary hyperparathyroidism, rheumatoid arthritis or malignant tumors; (3) concurrent use of medications known to influence urinary PGE2 levels or bone metabolism (e.g., cyclooxygenase inhibitors, diuretics, synthetic steroid hormones, epinephrine or anticonvulsants); or (4) pregnancy or lactation.”

Comments 3: The Discussion is generally well argued and places the findings in the context of the existing literature. I appreciate that the authors have addressed the main limitations of the study; however, further consideration of potential selection bias and residual confounding factors would strengthen the manuscript.

residual confounding factors: Although the authors adjusted for several important covariates, the potential influence of residual confounding factors should be acknowledged. Variables such as menopausal status in women may affect both urinary PGE2/PGEM levels and bone metabolism markers and could therefore have influenced the observed associations.

potential selection bias: The study population was recruited from tertiary care hospitals rather than from a community-based sample. Although extensive exclusion criteria were applied, this recruitment strategy may have introduced selection bias, as participants attending hospital-based health examinations may differ from the general healthy population in terms of health awareness, comorbidities, and lifestyle characteristics. The potential impact of this limitation on the generalizability of the proposed reference data should be discussed.

Response 3: We sincerely thank the reviewer for this insightful comment. To further assess the robustness of the primary findings and to explore potential heterogeneity across clinically relevant subgroups, we conducted additional subgroup analyses stratified by sex, age group and menopausal status among women. The associations between 24-hour urinary PGE2/PGEM excretion and the major bone metabolism-related markers were generally consistent with those observed in the overall population. In particular, the direction of the associations between urinary PGE2/PGEM and serum calcium, P1NP, and PTH remained broadly similar across the sex- and age-stratified analyses. Among women, the overall pattern of associations was also generally comparable between premenopausal and postmenopausal participants. The detailed results are presented in Supplementary Materials.

These additional analyses support the overall robustness of our findings and suggest that the observed associations were not solely driven by differences in sex, age, or menopausal status. Nevertheless, we acknowledge that subgroup analyses cannot completely eliminate residual confounding. Other unmeasured or incompletely measured factors, including reproductive hormonal profiles, physical activity, dietary sodium intake, low-grade inflammatory status, and additional lifestyle characteristics, may still influence both urinary PGE2/PGEM levels and bone metabolism markers. We have therefore revised the Discussion to explicitly acknowledge this limitation and to avoid causal interpretation of the observed associations.

We also agree that recruitment from tertiary care hospitals may have introduced selection bias. Although the multicenter design, standardized protocol, and strict exclusion criteria helped reduce the influence of overt diseases and medications affecting PGE2 metabolism or bone metabolism, participants recruited through hospital-based health examination settings may differ from the general community population in health awareness, healthcare-seeking behavior, comorbidity burden, and lifestyle characteristics. Therefore, the present findings should be interpreted as hospital-based estimates derived from carefully screened Chinese adults, rather than as definitive population-based reference intervals.

Accordingly, we have revised the Discussion to clarify the potential impact of residual confounding and hospital-based selection bias on the generalizability of the findings. Future community-based, population-representative studies with comprehensive assessment of reproductive status, sex hormones, lifestyle factors, inflammatory biomarkers, and longitudinal follow-up are needed to validate these findings.

Revised Discussion Section in Line 385 and 395:

“The cross-sectional design precludes causal inference, and the modest sample size together with the exclusively Chinese cohort from tertiary care hospitals limits statistical precision and generalizability.”

“Although the subgroup analyses showed broadly consistent patterns, other unmeasured lifestyle factors cannot be excluded. Thus, the lack of independent associations in adjusted models should not be interpreted as evidence against biological relevance. Future community-based representative studies with comprehensive assessment of reproductive status, sex hormones, lifestyle, and inflammatory biomarkers are warranted.”

Reviewer 4 Report

Comments and Suggestions for Authors

This multicenter cross-sectional study examines clinical determinants of 24-hour urinary PGE2 (U-PGE2) and its metabolite (U-PGEM) in 737 healthy Chinese adults, and tests their associations with calcium–phosphorus homeostasis and bone turnover markers using multivariable linear regression and restricted cubic splines (RCS). The reported findings are that both biomarkers rise with age and in summer/autumn, are inversely associated with serum calcium, show a threshold-type association with P1NP, and that U-PGEM (but not U-PGE2) is inversely associated with PTH.

The study has real strengths: a properly timed 24-hour collection rather than spot urine, a sizable multicenter sample, well-characterized biochemistry, and a sensible analytic toolkit. Population-level normative data on 24-hour U-PGE2/U-PGEM are genuinely scarce, so the descriptive contribution is worthwhile. My concerns are mostly about the inferential claims and a few internal inconsistencies that need to be resolved before the associations can be trusted at face value. One contradiction (diabetes as both an exclusion criterion and a reported, adjusted-for covariate) must be fixed. My detailed comments follow.

Major comments

1. Diabetes is both excluded and present in the cohort. The exclusion criteria (Section 2.1, item 2) list diabetes mellitus among the conditions “known to affect bone metabolism” that disqualified participants, yet Table 1 reports 43 diabetic participants (5.8%), and Model 2 explicitly adjusts for diabetes. These statements cannot both be correct. Please clarify whether diabetics were enrolled (in which case the exclusion text and the “healthy adults” framing need revision) or excluded (in which case Table 2 should not contain a diabetes term). The same point applies more softly to hypertension (16.0%): a cohort with this prevalence of hypertension and possible diabetes is arguably a community sample rather than “healthy adults,” and the title/abstract wording should reflect the actual inclusion.
2. Multiplicity is not addressed in the regression or spline analyses. Table 2 tests on the order of 17 covariates × 2 outcomes × 2 models, and the RCS section adds roughly 14 outcome–marker associations, all judged at an uncorrected two-tailed P < 0.05. Bonferroni correction was applied to the Kruskal–Wallis post hoc tests, but apparently nowhere else. With this many comparisons, the robust signals (serum calcium, P ≈ 0.006–0.008; age, P < 0.001; season, P < 0.001) will survive, but the borderline findings—BMI (P = 0.020), ALT (P = 0.041), the P1NP nonlinear term (P = 0.006–0.024), and the U-PGEM–PTH association (P = 0.024)—are exactly the ones at risk of being false positives. I would ask the authors either to apply a multiplicity correction (FDR is reasonable here) or to explicitly frame the weaker associations as hypothesis-generating, and to be careful that the Abstract and Conclusion do not present borderline results with the same confidence as the strong ones.
3. Creatinine normalization of a timed collection may confound the headline age association. Normalizing to urinary creatinine is standard for spot urine, but this is a full 24-hour collection, for which total 24-hour excretion is the more natural denominator. This matters because urinary creatinine excretion falls with age as muscle mass declines; dividing PGE2/PGEM by a denominator that itself decreases with age can inflate an apparent positive age association. Since age is one of the central findings, please report a sensitivity analysis using absolute 24-hour excretion (and/or per-kg or per-body-surface) and confirm the age relationship holds. At a minimum, this limitation should be discussed.
4. Inconsistent units for the primary outcome. The text, abstract, and tables report U-PGE2 and U-PGEM in “pg/mmol creatinine,” but the axes of Figure 1 read “pg/mg Cr.” mmol and mg are not interchangeable, so as written, it is impossible to know which unit the reported values (133.87, 246.76, etc.) are in. Please reconcile throughout; this affects every quantitative statement in the paper and any future comparison to other cohorts.
5. The P1NP “threshold” finding is over-interpreted. The claim of a “novel threshold-type non-linear association” with a break at 57 ng/mL (Section 3.4, Figure 4C–D) rests on RCS curves with wide confidence bands, and the 57 ng/mL cut appears to be read off the fitted curve post hoc rather than pre-specified. The Discussion then interprets the inverse relationship causally—“compensatory elevation of PGE2 when bone formation is low”—which a cross-sectional design cannot support, and which could equally run the other way. Please soften this to an observed association, avoid presenting the threshold as a validated cut-point, and consider reporting the knot placement and a test of whether the nonlinear term adds to a simpler linear model.
6. Season handling is unusual and imbalanced. Collapsing season into “Spring + Winter” versus “Summer + Autumn” combines non-adjacent seasons and is not self-explanatory—one would expect either four seasons or a warm/cold contrast that does not pair spring with winter. The groups are also heavily imbalanced (158 vs 579, i.e., 79% of samples in one stratum), which raises the possibility that the seasonal “effect” partly reflects collection logistics or differential recruitment rather than biology. Please justify the grouping, ideally show the four-season analysis, and discuss potential collection-period confounding.
7. Model specification, collinearity, and center effects. Two related points. (a) Serum calcium, phosphorus, PTH, and 25(OH)D are physiologically interdependent; if they were entered together (or alongside one another across models), collinearity and over-adjustment could distort the estimates. Please state exactly which variables were in which model and report collinearity diagnostics (e.g., VIF). (b) This is a 9-hospital study with an inter-assay CV for U-PGEM of 8.3%, yet there is no adjustment for site and no assessment of between-center variability. A center random effect (or at least a sensitivity analysis by site) would guard against site-level batch effects driving the associations.

Minor comments

1. Serum total calcium was used without mention of albumin correction or ionized calcium. Because the calcium association is a headline result, please confirm whether albumin-adjusted calcium changes the finding, or note this as a limitation.
2. The median dietary calcium intake (≈232 mg/day; IQR 78–486) is implausibly low for an adult population and well below typical Chinese intake estimates. This suggests the FFQ may capture only part of dietary calcium. Please comment on the instrument’s validity and whether this affects the (null, in Model 2) diet-calcium associations.
3. Median 25(OH)D is 16 ng/mL, i.e., the cohort is broadly vitamin D insufficient. This is worth noting both for the “healthy” characterization and when interpreting bone-marker relationships.
4. Competitive ELISA for urinary PGE2 is known to be vulnerable to PGE2 instability and cross-reactivity, with PGEM generally regarded as the more reliable systemic readout. The authors partly acknowledge this (“better systemic representation of U-PGEM”), but a brief explicit measurement caveat for U-PGE2 would strengthen the interpretation of the U-PGE2-specific null results.
5. The sample is described as “relatively modest” in the Limitations but “large” in the Conclusion. Please make these consistent.
6. Figure formatting needs attention. Panel labels overlap the axis titles in Figures 3 and 4 (e.g., “Calcium (mmol/L)” running into “(a)”); the Figure 2 violin-plot significance markers render as stray symbols; and Figure 1 appears to plot raw, highly skewed values with a linear fit, even though regressions used log10-transformed data—consider plotting on the log scale or clarifying. Figure 1 also has heavy overplotting near the origin.
7. The Figure 4 caption reads “between 24-hour U-PGEM and U-PGEM,” which is a duplication—presumably one should be U-PGE2 or the marker name.
8. Language and typographical pass needed. Examples: “Written informed [consent] was obtained” (Section 2.1, missing word); “moderate correlated” → “moderately correlated”; “statistical significant” → “statistically significant”; “1,α hydroxylase” → “1α-hydroxylase”; the stray colon after “no conflicts of interest.”; full-width parentheses/commas in several table cells; and the title phrase “Markers Correlates” reads awkwardly. The abbreviation list also defines U-CaE as “urinary calcium,” while the text uses “urinary calcium excretion,” and P1NP is listed without “total.”

Author Response

Comments 1: Diabetes is both excluded and present in the cohort. The exclusion criteria (Section 2.1, item 2) list diabetes mellitus among the conditions “known to affect bone metabolism” that disqualified participants, yet Table 1 reports 43 diabetic participants (5.8%), and Model 2 explicitly adjusts for diabetes. These statements cannot both be correc Please clarify whether diabetics were enrolled (in which case the exclusion text and the “healthy adults” framing need revision) or excluded (in which case Table 2 should not contain a diabetes term). The same point applies more softly to hypertension (16.0%): a cohort with this prevalence of hypertension and possible diabetes is arguably a community sample rather than “healthy adults,” and the title/abstract wording should reflect the actual inclusion.

Response 1: We sincerely thank you for your careful reading and for pointing out this inconsistency. We apologize for this oversight, and it was entirely our writing error in the exclusion criteria. In our actual study, we did not exclude participants with diabetes and hypertension; instead, we included them in the analysis and adjusted for diabetes as a covariate in all multivariable models, because diabetes may influence PGE2 and bone metabolism through inflammatory and renal pathways. To resolve this discrepancy, we have revised the exclusion criteria in the Methods section. Additionally, we fully agree that our study population is not accurately described as “healthy adults”,  and this was an imprecise wording choice on our part. In response, we have revised the title and abstract to reflect the actual samples more accurately.

Revised Methods Section in Line 112:

 “Subjects with the following conditions were excluded: (1) serious diseases affecting the pulmonary, cardiovascular, gastrointestinal, hematopoietic, renal or nervous systems; (2) conditions known to affect bone metabolism, such as osteogenesis imperfecta, Paget’s disease of bone, primary hyperparathyroidism, rheumatoid arthritis or malignant tumors; (3) concurrent use of medications known to influence urinary PGE2 levels or bone metabolism (e.g., cyclooxygenase inhibitors, diuretics, synthetic steroid hormones, epinephrine or anticonvulsants); or (4) pregnancy or lactation.”

Revised Title in Line 2:

“Clinical Determinants and Bone Metabolic Correlates of 24-Hour Urinary PGE2 and PGEM Excretion in Chinese Adults: A Multicenter Cross-Sectional Study.”

Revised Abstract in Line 19:

“Methods: In this multicenter, cross-sectional study, 737 Chinese adults underwent standardized 24-hour urine collection.”

Comments 2: Multiplicity is not addressed in the regression or spline analyses. Table 2 tests on the order of 17 covariates × 2 outcomes × 2 models, and the RCS section adds roughly 14 outcome–marker associations, all judged at an uncorrected two-tailed P < 0.05. Bonferroni correction was applied to the Kruskal–Wallis post hoc tests, but apparently nowhere else. With this many comparisons, the robust signals (serum calcium, P ≈ 0.006–0.008; age, P < 0.001; season, P < 0.001) will survive, but the borderline findings—BMI (P = 0.020), ALT (P = 0.041), the P1NP nonlinear term (P = 0.006–0.024), and the U-PGEM–PTH association (P = 0.024)—are exactly the ones at risk of being false positives. I would ask the authors either to apply a multiplicity correction (FDR is reasonable here) or to explicitly frame the weaker associations as hypothesis-generating, and to be careful that the Abstract and Conclusion do not present borderline results with the same confidence as the strong ones.

Response 2: We sincerely thank the reviewer for this important and methodologically rigorous comment. In the revised manuscript, we applied the Benjamini–Hochberg false discovery rate (FDR) procedure to control for multiple testing. For the RCS analyses, we applied FDR correction conservatively across all spline-related tests, including both the overall association tests. Raw P values and FDR-adjusted P values have been added to Figure 4 and Figure 5 (due to the addition of Figure 1 Flowchart for the study process, Figure 3 has been changed to Figure 4, Figure 4 has been changed to Figure 5). The results of this study were based on the exploratory analysis, we have revised the Methods and Results sections accordingly.

Revised Methods Section in Line 191:

“To account for multiplicity in the restricted cubic spline analyses, P values were adjusted using the Benjamini–Hochberg false discovery rate procedure.”

Revised Results Section in Line 262-271，285-296:

“To explore potential dose-response relationships, RCS were employed to model the associations of serum calcium, phosphorus, U-CaE levels with 24-hour U-PGE₂ and U-PGEM (Figure 4). The results revealed a significant inverse linear association between serum calcium levels and both 24-hour U-PGE₂ (Overall P = 0.010, FDR-adjusted P=0.030; Figure 4A) and U-PGEM (Overall P = 0.004, FDR-adjusted P=0.024; Figure 4B). Furthermore, the tests for non-linearity were not statistically significant (Nonlinear P = 0.168 and P = 0.791, respectively), confirming that these dose-response relationships are predominantly linear. However, neither 24-hour U-PGE₂ nor U-PGEM exhibited any statistically significant linear or non-linear associations with serum phosphorus or U-CaE levels (all Overall P>0.05; Figure 4C-F).”

“The analyses revealed a significant non-linear association between P1NP and both 24-hour U-PGE2 (Overall P = 0.021, Nonlinear P = 0.006, FDR-adjusted P=0.048; Figure 5C) and U-PGEM (Overall P = 0.017, Nonlinear P = 0.024,FDR-adjusted P=0.048; Figure 5D). The fitted curves showed lower estimated U-PGE2 and U-PGEM levels across increasing P1NP concentrations in the lower-to-middle range, with a flatter pattern at higher P1NP concentrations.”

“24-hour U-PGEM demonstrated a significant inverse linear association with PTH levels (Overall P = 0.024, Nonlinear P = 0.413, FDR-adjusted P=0.048; Figure 5F), whereas the association for U-PGE2 was not statistically significant (Overall P = 0.404, FDR-adjusted P=0.485; Figure 5E). Furthermore, neither 24-hour U-PGE2 nor U-PGEM exhibited any statistically significant linear or non-linear associations with the β-CTX-I (Figure 5A, 5B) or 25(OH)D levels (all FDR-adjusted P > 0.05; Figure 5G, 5H).”

Comments 3: Creatinine normalization of a timed collection may confound the headline age association. Normalizing to urinary creatinine is standard for spot urine, but this is a full 24-hour collection, for which total 24-hour excretion is the more natural denominator. This matters because urinary creatinine excretion falls with age as muscle mass declines; dividing PGE2/PGEM by a denominator that itself decreases with age can inflate an apparent positive age association. Since age is one of the central findings, please report a sensitivity analysis using absolute 24-hour excretion (and/or per-kg or per-body-surface) and confirm the age relationship holds. At a minimum, this limitation should be discussed.

Response 3: 

We sincerely thank the reviewer for this important and methodologically insightful comment.

In the original analysis, urinary PGE2 and PGEM were expressed relative to 24-hour urinary creatinine to standardize inter-individual differences in urinary excretion and to provide an internal quality-control measure for the completeness of the 24-hour urine collection. In response to the reviewer’s suggestion, we additionally analyzed absolute 24-hour PGE2 and PGEM excretion, calculated as urinary analyte concentration multiplied by total 24-hour urine volume. In these sensitivity analyses,  The results revealed a significant positive linear association between age and both 24-hour U-PGE2 (Overall P=0.001; Supplementary Figure 1A) and U-PGEM (Overall P<0.001; Supplementary Figure 1B)

We did not use body-weight-normalized or body-surface-area-normalized PGE2/PGEM excretion as the principal alternative outcome. Unlike urinary creatinine, body weight and body surface area are not established physiological denominators for urinary PGE2/PGEM excretion, and there is no evidence that PGE2/PGEM excretion scales proportionally with total body weight or body surface area. In addition, body weight reflects both lean and adipose compartments and is strongly related to age, sex, and BMI. Instead, in analyses of absolute 24-hour excretion, we adjusted BMI as the covariates in the regression model.

Accordingly, we have revised the Abstract, Results, Discussion, and Conclusions to avoid overinterpretation of the age-related findings. We have also added a limitation acknowledging that creatinine normalization may be affected by age-related variation in muscle mass and urinary creatinine excretion.

Revised Methods Section in Line 180:

“As a sensitivity analysis, we applied RCS to analysis the assiciation between age and  absolute 24-hour U-PGE2 and U-PGEM excretion, calculated as urinary analyte concentration multiplied by the corresponding 24-hour urine volume. ”

Revised Results Section in Line 224:

“In sensitivity analyses using absolute 24-hour excretion rather than creatinine-normalized concentrations, age remained positively associated with both 24-hour PGE2 and PGEM excretion (overall P = 0.001 and P < 0.001, respectively; Supplementary Figure 1A-B). ”

Comments 4: Inconsistent units for the primary outcome. The text, abstract, and tables report U-PGE2 and U-PGEM in “pg/mmol creatinine,” but the axes of Figure 1 read “pg/mg Cr.” mmol and mg are not interchangeable, so as written, it is impossible to know which unit the reported values (133.87, 246.76, etc.) are in. Please reconcile throughout; this affects every quantitative statement in the paper and any future comparison to other cohorts.

Response 4: We sincerely apologize for our mistakes. We have since revised Figure 2 in Line 217 accordingly ( due to the addition of Figure 1 Flowchart for the study process, Figure 1 has been changed to Figure 2).

Comments 5: The P1NP “threshold” finding is over-interpreted. The claim of a “novel threshold-type non-linear association” with a break at 57 ng/mL (Section 3.4, Figure 4C–D) rests on RCS curves with wide confidence bands, and the 57 ng/mL cut appears to be read off the fitted curve post hoc rather than pre-specified. The Discussion then interprets the inverse relationship causally—“compensatory elevation of PGE2 when bone formation is low”—which a cross-sectional design cannot support, and which could equally run the other way. Please soften this to an observed association, avoid presenting the threshold as a validated cut-point, and consider reporting the knot placement and a test of whether the nonlinear term adds to a simpler linear model.

Response 5: We sincerely thank the reviewer for this careful and constructive comment. First, we acknowledge that the apparent change in the slope of the fitted curves around a P1NP value of approximately 57 ng/mL was identified visually from the fitted RCS curves and was not a prespecified clinical, biological, or statistically validated cut-point. We have therefore removed the terms “novel threshold-type non-linear association,” “threshold effect,” and the specific value of 57 ng/mL throughout the Results, Discussion, and Conclusions. The findings are now described more cautiously as exploratory non-linear associations between P1NP and 24-hour urinary PGE2/PGEM excretion.

Second, we have clarified the RCS modelling strategy in the Methods. The RCS models were fitted using three knots placed at the 10^th, 50^th and 90^th of the P1NP distribution. The P value for non-linearity was obtained by testing the joint contribution of the nonlinear spline terms, corresponding to a comparison of the full RCS model with the nested linear model. In the revised analysis, there remained statistical evidence of deviation from linearity for the associations of P1NP with both 24-hour U-PGE2 (P for non-linearity = 0.006) and U-PGEM (P for non-linearity = 0.024).

Finally, we have removed the causal interpretation that low bone formation induces a compensatory elevation of PGE2. We now describe these findings as hypothesis-generating and state that longitudinal and mechanistic studies are required to clarify temporality and underlying biological mechanisms.

Revised Methods Section in Line 188:

“Restricted cubic spline models were fitted using three knots placed at the 10^th, 50^th and 90^th of the P1NP distribution. Overall association and non-linearity were evaluated using joint tests of the spline terms.”

Revised Results Section in Line 285:

“The fitted curves showed lower estimated U-PGE₂ and U-PGEM levels across increasing P1NP concentrations in the lower-to-middle range, with a flatter pattern at higher P1NP concentrations.”

Revised Discussion Section in Line 366:

“Although the fitted curves suggested an inverse pattern that became less pronounced at higher P1NP concentrations, the cross-sectional design precludes conclusions regarding temporality or causality. These findings may reflect bidirectional biological relationships, reverse causation, residual confounding, or other unmeasured mechanisms. Therefore, the observed pattern should be considered hypothesis-generating rather than evidence of a compensatory PGE₂ response to low bone formation [3,40,41].”

Comments 6: Season handling is unusual and imbalanced. Collapsing season into “Spring + Winter” versus “Summer + Autumn” combines non-adjacent seasons and is not self-explanatory—one would expect either four seasons or a warm/cold contrast that does not pair spring with winter. The groups are also heavily imbalanced (158 vs 579, i.e., 79% of samples in one stratum), which raises the possibility that the seasonal “effect” partly reflects collection logistics or differential recruitment rather than biology. Please justify the grouping, ideally show the four-season analysis, and discuss potential collection-period confounding.

Response 6: We sincerely thank the reviewer for this important and constructive comment. The original grouping was applied pragmatically because of the small number of winter collections; however, we acknowledge that combining non-adjacent seasons is not an optimal analytic approach. In response, we have removed the dichotomous seasonal classification from the revised primary analyses. The numbers of participants collected in each season were as follows: spring, n = 176 (23.9%); summer, n = 382 (51.8%); autumn, n = 173 (23.5%); and winter, n = 6 (0.8%). These distributions have been added to the revised Table 1. We have also revised the Table 2.

We have also expanded the Discussion to acknowledge that the observed seasonal associations may not represent a purely biological effect. The timing of recruitment was not randomized, and the unequal distribution of collections across seasons may have been influenced by recruitment schedules, site-specific enrollment patterns, participant characteristics, unmeasured environmental exposures, or other collection-period factors. Therefore, the observed seasonal differences should be interpreted as exploratory associations rather than evidence of a causal seasonal effect on urinary PGE2 or PGEM excretion. Future studies with prospectively balanced sampling across seasons and repeated measurements within individuals will be required to clarify whether these patterns reflect true biological seasonality.

Revised Discussion Section in Line 343:

“Although seasonal differences in urinary PGE2 and PGEM were noted, the unbalanced sampling, especially few winter collections, limits distinguishing true seasonality from confounding. These exploratory findings thus require validation in studies with balanced seasonal recruitment.”

Comments 7: Model specification, collinearity, and center effects. Two related points. (a) Serum calcium, phosphorus, PTH, and 25(OH)D are physiologically interdependent; if they were entered together (or alongside one another across models), collinearity and over-adjustment could distort the estimates. Please state exactly which variables were in which model and report collinearity diagnostics (e.g., VIF). (b) This is a 9-hospital study with an inter-assay CV for U-PGEM of 8.3%, yet there is no adjustment for site and no assessment of between-center variability. A center random effect (or at least a sensitivity analysis by site) would guard against site-level batch effects driving the associations.

Response 7: We sincerely thank the reviewer for this important and methodologically valuable comment. We have clarified the analytical structure of the regression and restricted cubic spline analyses. The fully adjusted models were not constructed by simultaneously entering serum calcium, phosphorus, PTH, 25(OH)D, and bone turnover markers into the same model. Rather, each calcium-phosphorus or bone metabolism marker was evaluated separately as the primary exposure variable in an individual model, with log10-transformed urinary PGE2 or PGEM as the outcome. Each model was adjusted for a common prespecified set of covariates: age, sex, BMI, and season of sample collection. We assessed multicollinearity among all independent variables using the variance inflation factor (VIF) (Table 1). The results showed that all VIF values were below 5, indicating no severe multicollinearity.

To address potential center-level clustering and site-related variation, we re-estimated the principal adjusted models using linear mixed-effects models with hospital site specified as a random intercept (Table 2). The calcium associations were attenuated after accounting for hospital-level clustering and therefore require cautious interpretation.

Table 1. The result of Multicollinearity test

Outcome	Exposure	VIF for covariates
		Exposure	BMI	age	Sex	season
24h U-PGE2	Calcium	1.071	1.088	1.071	1.033	1.09
	Phosphorus	1.091	1.087	1.084	1.079	1.053
	U-CaE	1.033	1.099	1.063	1.048	1.027
	β-CTX-I	1.028	1.087	1.061	1.043	1.047
	P1NP	1.013	1.088	1.069	1.033	1.028
	25(OH)D	1.209	1.087	1.112	1.111	1.107
	PTH	1.042	1.091	1.074	1.033	1.044
24h U-PGEM	Calcium	1.060	1.087	1.066	1.036	1.09
	Phosphorus	1.080	1.087	1.078	1.076	1.06
	U-CaE	1.035	1.096	1.058	1.055	1.037
	β-CTX-I	1.032	1.087	1.059	1.048	1.055
	P1NP	1.016	1.088	1.066	1.036	1.038
	25(OH)D	1.248	1.087	1.127	1.121	1.14
	PTH	1.037	1.089	1.066	1.036	1.053

 

 

 Table 2. The results of mixed-effects models for associations between bone metabolism markers and 24h U-PGE2 and U-PGEM.

	24h U-PGE2		24h U-PGEM
	β (95%CI)	P	β (95%CI)	P
Ca	-0.015(-0.106, 0.076)	0.749	-0.107(-0.215, 0.001)	0.052
P	0.049(-0.065, 0.163)	0.401	-0.016(-0.153, 0.122)	0.824
UCA	-0.01(-0.023, 0.003)	0.118	-0.01(-0.025, 0.005)	0.202
CTX	-0.06(-0.178, 0.057)	0.311	-0.105(-0.248, 0.037)	0.148
P1NP	0.001(-0.001, 0.001)	0.953	-0.001(-0.002, 0.001)	0.100
PTH	-0.001(-0.002, 0.001)	0.301	-0.002(-0.004, 0.001)	0.020
OHD	0.004(0.001, 0.008)	0.068	0(-0.005, 0.005)	0.955

Comments 8: Serum total calcium was used without mention of albumin correction or ionized calcium. Because the calcium association is a headline result, please confirm whether albumin-adjusted calcium changes the finding, or note this as a limitation.

Response 8: We sincerely apologize for not providing sufficient clarification regarding the assessment of serum calcium. In this study, serum albumin and ionized calcium were not measured. Therefore, we were unable to calculate albumin-adjusted calcium concentrations or evaluate whether the observed associations between urinary PGE2/PGEM and serum calcium would remain unchanged when using albumin-corrected or ionized calcium values.

We have expanded the Discussion to explicitly acknowledge that total serum calcium may be influenced by albumin concentration and other protein-related factors. Accordingly, the observed inverse associations between 24-hour urinary PGE2/PGEM excretion and serum calcium should be interpreted cautiously as associations with unadjusted total serum calcium, rather than as evidence of an association with biologically active ionized calcium or calcium homeostasis.

In addition, we have revised the relevant wording in the Discussion, and Conclusions to avoid over-interpretation of this finding. We now emphasize that future studies incorporating serum albumin, albumin-adjusted calcium, and ionized calcium measurements are needed to validate and better characterize the observed association.

Revised Discussion Limitation Section in Line 391:

“Regarding calcium-related measures, serum albumin and ionized calcium were unavailable, preventing albumin-adjusted calcium calculations; additionally, dietary calcium was estimated via a 1-week self-reported FFQ, which is subject to recall error and potential underestimation, and may have attenuated the adjusted associations.”

Comments 9: The median dietary calcium intake (≈232 mg/day; IQR 78–486) is implausibly low for an adult population and well below typical Chinese intake estimates. This suggests the FFQ may capture only part of dietary calcium. Please comment on the instrument’s validity and whether this affects the (null, in Model 2) diet-calcium associations.

Response 9: We sincerely thank the reviewer’s suggestion. Dietary calcium intake was assessed using a standardized semi-quantitative food frequency questionnaire (FFQ), and calcium intake was calculated according to the reported food type, portion size, and frequency of consumption during the preceding week. Although a standardized FFQ was used, this approach relied on participants’ recall of their recent dietary intake. Therefore, incomplete reporting of foods, beverages, portion sizes, eating occasions, or calcium-containing items may have occurred. Accordingly, we acknowledge that the FFQ-derived calcium estimate should be interpreted primarily as a self-reported, short-term estimate of dietary calcium intake rather than a definitive measure of habitual total calcium intake.

We agree that measurement error in dietary calcium assessment may have influenced the null association observed after multi-variable adjustment. Incomplete or imprecise dietary reporting could attenuate an underlying association toward the null and may partly explain why dietary calcium intake was not independently associated with urinary PGE2 or PGEM in Model 2. Therefore, we have revised the manuscript to avoid interpreting the adjusted null findings as evidence that dietary calcium has no relationship with urinary PGE2/PGEM excretion. We have added this to the limitation section and clarified that future studies should incorporate more comprehensive and validated dietary assessment methods, such as repeated dietary records, multiple 24-hour dietary recalls, or biomarker-supported dietary assessment, to better characterize the potential association between calcium intake and urinary PGE2/PGEM levels.

Revised Discussion Limitation Section in 391:

“Regarding calcium-related measures, serum albumin and ionized calcium were unavailable, preventing albumin-adjusted calcium calculations; additionally, dietary calcium was estimated via a 1-week self-reported FFQ, which is subject to recall error and potential underestimation, and may have attenuated the adjusted associations.”

Comments 10: Median 25(OH)D is 16 ng/mL, i.e., the cohort is broadly vitamin D insufficient. This is worth noting both for the “healthy” characterization and when interpreting bone-marker relationships.\

Response 10: We thank you for this insightful observation. We fully agree that a median 25(OH)D level of 16 ng/mL indicates widespread vitamin D insufficiency in our cohort, which is an important consideration both for characterizing our study population and for interpreting the bone‑marker relationships.

In response, we have revised the title and abstract to remove the term “healthy”; we now consistently describe the participants as “Chinese adults” to reflect the actual inclusion criteria (which did not exclude individuals with hypertension, diabetes, or low vitamin D status).

Furthermore, we have explicitly noted the high prevalence of vitamin D insufficiency in our cohort, acknowledged that this may influence bone turnover markers and PGE2/PGEM associations, and discussed that residual confounding related to vitamin D status cannot be completely ruled out.

Revised Title in Line 2:

“Clinical Determinants and Bone Metabolic Correlates of 24-Hour Urinary PGE2 and PGEM Excretion in Chinese Adults: A Multicenter Cross-Sectional Study.”

Revised Discussion Section in Line 381:

“The high prevalence of vitamin D insufficiency in our cohort may also influence bone turnover markers and PGE2/PGEM associations, so the residual confounding related to vitamin D status cannot be completely ruled out.”

Comments 11: Competitive ELISA for urinary PGE2 is known to be vulnerable to PGE2 instability and cross-reactivity, with PGEM generally regarded as the more reliable systemic readout. The authors partly acknowledge this (“better systemic representation of U-PGEM”), but a brief explicit measurement caveat for U-PGE2 would strengthen the interpretation of the U-PGE2-specific null results.

Response 11: We appreciate your methodological insight. We agree that competitive ELISA for urinary PGE2 has inherent limitations, and we thank you for the suggestion to make this caveat more explicit.

Revised Discussion Section in Line 377:

“Given U-PGE2 ELISA’s susceptibility to degradation and cross-reactivity, unlike the more stable PGEM, the U-PGE2 null results should be interpreted cautiously due to potential measurement error.”

Comments 12: The sample is described as “relatively modest” in the Limitations but “large” in the Conclusion. Please make these consistent.

Response 12: We thank the reviewer for catching this inconsistency. We agree that the same sample should not be described differently in two sections of the manuscript. To address this, we have revised the Conclusion section to align with the Limitations, and we have now consistently used “relatively modest” throughout the text.

Comments 13: Figure formatting needs attention. Panel labels overlap the axis titles in Figures 3 and 4 (e.g., “Calcium (mmol/L)” running into “(a)”); the Figure 2 violin-plot significance markers render as stray symbols; and Figure 1 appears to plot raw, highly skewed values with a linear fit, even though regressions used log10-transformed data—consider plotting on the log scale or clarifying. Figure 1 also has heavy overplotting near the origin.

Response 13: We sincerely thank the reviewer’s suggestion. We have revised Figures 1~5 as suggested.

Comments 14: The Figure 4 caption reads “between 24-hour U-PGEM and U-PGEM,” which is a duplication—presumably one should be U-PGE2 or the marker name.

Response 14: We thank the reviewer for carefully identifying this typo. We have corrected the duplication in the Figure 5 caption from “between 24-hour U-PGEM and U-PGEM” to “between 24-hour U-PGEM which correctly describes the variables presented in the figure (due to the addition of Figure 1 Flowchart for the study process, Figure 4 has been changed to Figure 5).

Comments 15: Language and typographical pass needed. Examples: “Written informed [consent] was obtained” (Section 2.1, missing word); “moderate correlated” → “moderately correlated”; “statistical significant” → “statistically significant”; “1,α hydroxylase” → “1α-hydroxylase”; the stray colon after “no conflicts of interest.”; full-width parentheses/commas in several table cells; and the title phrase “Markers Correlates” reads awkwardly. The abbreviation list also defines U-CaE as “urinary calcium,” while the text uses “urinary calcium excretion,” and P1NP is listed without “total.”

Response 15: We sincerely thank you for your careful reading and for pointing out the language and typographical issues. We have thoroughly revised the manuscript to address each point. Below is our point‑by‑point response.

1. We have added the missing word, and the sentence now reads “Written informed consent was obtained from all participants.” in Line 108.

2. We have corrected “moderate correlated”to “moderately correlated” in the Results section in Line 201.

3. We have changed “statistical significant” to “statistically significant” throughout the manuscript.

4. We have corrected “1,α hydroxylase” to “1α‑hydroxylase” as per standard nomenclature.

5. We have removed the redundant colon. The statement now ends with a period only “The authors declare no conflicts of interest.” 

6. We have carefully reviewed all tables and replaced any full‑width punctuation with their standard half‑width equivalents. All table cells now use proper English punctuation.

7. We agree with the your suggestion that Title phrase “Markers Correlates” reads awkwardly. We have revised the title to “Clinical Determinants and Bone Metabolic Correlates of 24-Hour Urinary PGE2 and PGEM Excretion in Chinese Adults: A Multicenter Cross-Sectional Study”. 

8. We have updated the abbreviation definition to match the text. It now reads “U‑CaE: urinary calcium excretion” and “total procollagen type 1 N‑terminal propeptide”. in Abbreviations List

Round 2

Reviewer 4 Report

Comments and Suggestions for Authors

The authors have done an excellent job addressing the reviewer's technical concerns. It's ready to be accepted.