Round 1

Reviewer 1 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

Accept in present form

Reviewer 2 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

That authors have addressed the issues raised during the first round of review and have now clearly indicated the intent, methodology and limitations of the study. I have no further comments.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Falasca et al., are reporting a retrospective study evaluating the effectiveness of fixed dose rosuvastatin/ezetimibe in lowering lipid parameters among virologically suppressed PLWH. Lipid profiles were assessed at baseline, 3, and 6 months, and the authors observed significant reductions in LDL and total cholesterol over time. An HIVnegative control cohort from a metabolic clinic was included for comparative analyses. The manuscript is generally clear and easy to follow, however, many grammatical and language errors are present and would benefit from thorough English language proofing. While the study question is straightforward and clinically relevant, I have several methodological concerns that I believe would be of vital importance and should be addressed.

- As acknowledged in the limitations, the HIV negative controls were not matched to PLWH for age/sex or cardiovascular risk factors. Since these greatly influence baseline LDL, statin pharmacokinetics, and time‑to‑response..etc, any observed differences in LDL‑lowering kinetics between groups could be attributed to demographic imbalance rather than HIV status. Analyses comparing PLWH to controls should therefore be interpreted as exploratory and adjusted for key covariates.

- The study assesses multiple lipid parameters at several time points without a clearly defined primary endpoint (perhaps percent change in LDL‑C at 6 months?). This exploratory approach increases the chance of false/positive, type I error can be inflated.

- As stated in the methodology section, the controls were recruited from a metabolic clinic and selected because they “required initiation” of rosuvastatin/ezetimibe. This can enrich the control group for more severe dyslipidemia or different comorbidity patterns compared with PLWH, creating indication bias. Baseline CVD risk scores, prior lipid‑lowering therapy, and comorbidities should be adjusted for.

- ART can alter rosuvastatin exposure and lipid response, and some INSTI‑based regimens increase rosuvastatin levels via transporter inhibition. 53% received INSTI‑based therapy but there is no information on the remaining regimens. It should be included as a covariate and stratified

- It is not clear to me what proportion of patients started rosuvastatin/ezetimibe de novo versus switched from other lipid‑lowering regimens.

- I thank the authors for the limitations section, as noted, the retrospective design limits external validity. Given the unmatched control group and potential confounding, the manuscript should avoid strong comparative claims between PLWH and controls and instead emphasize the (within‑PLWH) longitudinal LDL‑lowering effect, with adjusted models and clear confidence intervals. The “between‑group” analyses in my opinion should be framed as exploratory.

Minor comments:

- Line 31: “ This strategy facilitates achievement of stringent lipid targets outlined in current HIV guidelines, by delivering synergistic LDL reductions, superior to statin monotherapy”

Since statin monotherapy was not assessed in this study.

- Line 81: “We observed all patients who received “

- Line 83: “We included patients who met all of the following criteria”

- Line 85: Please clarify the inclusion criteria. The statement states that “effective (HIV RNA undetectable in last 2 blood samples) cART, LDL levels in last 2 blood samples>130 mg/dl or >100 mg/dl with CVD risk”, but what was the interval between the 1^st and second measurement?  1 year? 6 months?

-Line 96: “They were not individually matched to PLWH for age or sex, but were selected on the basis of a comparable cardiovascular risk profile requiring initiation of the same fixed-dose rosuvastatin/ezetimibe regimen 96 97” this is a potential problem, as it can introduce bias in the analysis as previously mentioned.

- Line 118: this should be included in a separate ethical statement

- The methodology is not clear to me. the first visit was when the rosuvastatin/ezetimibe was initiated? if not, then how much time has passed since initiation and initial visit?

- The manuscript would greatly benefit from a section on individual effects of ART and HIV itself on lipid levels and cardiovascular risk.

Author Response

Response to Reviewer 1

We appreciate the Reviewer’s recognition of the clinical relevance of the study and agree that the comparison between PLWH and HIV-negative controls must be interpreted cautiously. We have therefore repositioned this analysis as exploratory and made the within-PLWH longitudinal analysis the principal focus.

Comment 1

Reviewer comment. The HIV-negative controls were not matched to PLWH for age, sex, or cardiovascular risk factors. The observed differences in lipid-lowering kinetics could therefore reflect demographic imbalance rather than HIV status. The analyses should be interpreted as exploratory and adjusted for key covariates.

Response. We agree that the two cohorts were not individually matched for age or sex. However, the controls were consecutively recruited from the same institution and study period and had a comparable clinical indication for initiating the same fixed-dose rosuvastatin/ezetimibe regimen and a broadly comparable overall cardiovascular-risk burden. Baseline Framingham 10-year risk was similar in PLWH and controls (19.6 ± 18.7% versus 20.4 ± 14.3%, respectively). HIV infection was considered a relevant cardiovascular-risk modifier because persistent immune activation, chronic inflammation, antiretroviral exposure, and HIV-associated metabolic abnormalities may contribute to risk beyond conventional demographic factors. We acknowledge that age, sex, BMI, hypertension, diabetes, and other factors may influence lipid response and that conventional scores may not fully capture HIV-related inflammatory risk. Given the limited sample size, the between-group analysis was retained as an unadjusted exploratory analysis. A multivariable longitudinal model simultaneously adjusting for age, sex, BMI, diabetes, hypertension, baseline LDL-C, baseline cardiovascular risk, and ART regimen was not considered statistically stable because it would risk overfitting and imprecise estimates. Residual confounding therefore cannot be excluded.

Change in the revised manuscript. The Methods now describe the clinical indication and broadly comparable cardiovascular-risk burden of the controls, while explicitly stating that the cohorts were not individually matched. The Results report the baseline Framingham comparison. The Statistical Analysis and Limitations sections state that the between-group analysis was retained as unadjusted and exploratory because of the limited sample size. The Discussion and Limitations acknowledge residual confounding.

Comment 2

Reviewer comment. Several lipid parameters were evaluated at multiple time points without a clearly defined primary endpoint, increasing the risk of type I error.

Response. We agree that the primary endpoint should be stated explicitly. The primary endpoint is now defined as the percentage change in LDL-C from baseline to month 6 among PLWH. Secondary endpoints include absolute LDL-C change, target attainment, changes in total cholesterol, non-HDL cholesterol, HDL cholesterol, triglycerides, cardiovascular risk scores, and safety parameters. The analyses of individual lipid parameters and the control cohort are presented as secondary or exploratory analyses. We have also moderated the interpretation and clarified that no adjustment for multiplicity was prespecified in this retrospective study.

Change in the revised manuscript. The primary and secondary endpoints have been added to the Methods section, and the Results and Discussion have been reorganized accordingly. Where appropriate, p values are described as nominal and interpreted in the context of the exploratory design.

Comment 3

Reviewer comment. The control group was recruited from a metabolic clinic and selected because it required initiation of the same treatment. This may introduce indication bias. Baseline risk scores, previous lipid-lowering therapy, and comorbidities should be adjusted for.

Response. We agree that recruitment from a metabolic clinic and the requirement for treatment initiation may introduce indication bias. All 32 PLWH and all 43 HIV-negative controls initiated rosuvastatin/ezetimibe de novo; no participant had previously received a statin or another lipid-lowering agent. The groups differed in age, sex, BMI, and some comorbidities, so residual indication and selection bias cannot be excluded. Because of the limited sample size, a stable multivariable adjustment for the full set of covariates was not performed; the comparison is therefore reported as unadjusted and exploratory.

Change in the revised manuscript. The Methods and Results state that all 32 PLWH and all 43 HIV-negative controls initiated rosuvastatin/ezetimibe de novo; no participant was a switcher from previous lipid-lowering therapy. The Methods and Limitations describe the potential for indication and selection bias and the absence of a reliable covariate-adjusted model.

Comment 4

Reviewer comment. Antiretroviral therapy may alter rosuvastatin exposure and lipid response. Although 53% received an INSTI-based regimen, the other regimens are not described. ART should be included as a covariate and stratified.

Response. We thank the Reviewer for raising this important point. In our cohort, 53.1% of PLWH received an INSTI-based regimen combined with an NRTI backbone. The remaining 46.9% of PLWH received an unboosted dual-therapy regimen. No participant received ritonavir or cobicistat, which are pharmacokinetic boosters known to interact with statins and may affect statin exposure. Nevertheless, no treatment-limiting adverse events were observed during follow-up, and hepatic, renal, creatine-kinase, and viro-immunological parameters remained stable. Because of the limited sample size, no formal ART-stratified or covariate-adjusted analysis was performed because such analyses could be underpowered, unstable, and prone to overfitting.

Change in the revised manuscript. Section 2.3 reports the ART distribution, the NRTI backbone, and the absence of ritonavir- or cobicistat-containing regimens. The Discussion now states that these boosters are known to interact with statins, while also reporting the absence of treatment-limiting adverse events and the stability of viro-immunological parameters. No unsupported regimen-specific causal interpretation is made.

Comment 5

Reviewer comment. It is unclear what proportion of patients started rosuvastatin/ezetimibe de novo versus switched from another lipid-lowering regimen.

Response. All 32 PLWH and all 43 HIV-negative controls initiated rosuvastatin 5 mg plus ezetimibe 10 mg daily de novo. No participant had previously received a statin or another lipid-lowering agent, and no preceding statin-monotherapy trial was performed. The primary analysis therefore describes treatment initiation rather than a switch population.

Change in the revised manuscript. The Methods state that all PLWH and all controls initiated rosuvastatin/ezetimibe de novo. The Results confirm that all controls were lipid-lowering-treatment naïve, and the Discussion and Limitations clarify the absence of a statin-monotherapy comparator.

Comment 6

Reviewer comment. The retrospective design, unmatched controls, and potential confounding require avoidance of strong comparative claims. The manuscript should emphasize within-PLWH longitudinal LDL lowering, use adjusted models and confidence intervals, and frame between-group analyses as exploratory.

Response. We agree. The manuscript now emphasizes the within-PLWH longitudinal LDL-C response and frames the HIV-negative comparison as exploratory. Because the sample was small, the comparison was deliberately retained as unadjusted; a multivariable model including the principal covariates was not considered statistically stable and could have produced overfitted and imprecise estimates. Confidence intervals are now reported for the principal within-PLWH LDL-C changes and for the database-derived D:A:D correlations. No causal or definitive superiority claim is made. The Results now also report the exploratory mixed-design repeated-measures ANOVA, which showed significant group × time interactions for total cholesterol (F = 5.58, p = 0.005, partial η² = 0.071), LDL-C (F = 7.28, p = 0.001, partial η² = 0.091), and non-HDL cholesterol (F = 6.65, p = 0.002). These findings are explicitly presented as hypothesis-generating because the cohorts were not individually matched and the analysis was not adjusted for potential confounders. The Results and Table 2 now also report that both the 5-year and 10-year D:A:D scores decreased significantly over time, with overall Friedman-test p-values of 6.03 × 10⁻⁷ and 3.11 × 10⁻⁷, respectively. The significant longitudinal score reduction is distinguished from the exploratory LDL-C–D:A:D correlation analysis.

Change in the revised manuscript. The Abstract, Results, Discussion, Conclusions, Statistical Analysis, and Limitations have been revised to emphasize the within-PLWH response, the exploratory nature of the control comparison, the absence of adjusted between-group estimates, and the available 95% confidence intervals. Results and Table 2 now report the p-values for the significant longitudinal reductions in the 5-year and 10-year D:A:D scores. The Conclusions now emphasize the verified LDL-C reduction, target attainment, and significant D:A:D score reductions, and identify prospective studies to test treatment comparisons and the possible contribution of inflammation, ART exposure, and cardiometabolic factors to lipid kinetics. The Statistical Analysis section now specifies the Friedman test for the overall D:A:D time effect and Wilcoxon signed-rank tests for pairwise comparisons.

Minor comments

Comment 7a

Reviewer comment. The statement that the strategy provides LDL reductions superior to statin monotherapy is not supported because statin monotherapy was not assessed.

Response. We agree. The revised manuscript no longer presents the combination as superior to statin monotherapy. It states that all participants initiated the combination de novo in routine clinical practice because a substantial LDL-C reduction was expected to be necessary. The Discussion no longer contains the sentence asserting that the study does not establish superiority, while the relevant methodological limitation remains appropriately acknowledged elsewhere in the manuscript.

Change in the revised manuscript. The superiority wording was removed from the Discussion. The manuscript retains a cautious description of the treatment rationale and does not make a direct comparative efficacy claim.

Comment 7b

Reviewer comment. Please correct the wording “We observed all patients who received”.

Response. Corrected. The unclear wording has been removed and the Methods now describe the eligible participants who initiated rosuvastatin 5 mg plus ezetimibe 10 mg daily.

Change in the revised manuscript. Methods, Section 2.1.

Comment 7c

Reviewer comment. Please clarify the inclusion criteria and the interval between the two LDL and HIV-RNA measurements.

Response. We clarified that eligibility required LDL-C >130 mg/dL or LDL-C >100 mg/dL in the presence of cardiovascular risk factors on at least two separate qualifying measurements. The definitive analytical workbook did not contain the exact dates needed to calculate a fixed interval between measurements or treatment initiation. We therefore report no unverified interval and acknowledge this limitation.

Change in the revised manuscript. Methods, Sections 2.1 and 2.2.

Comment 7d

Reviewer comment. The control-group selection statement may introduce bias.

Response. We agree. The controls were recruited from the same institution and study period but were not individually matched. Their comparable treatment indication and broadly comparable baseline Framingham risk are described as clinical context, not as proof of exchangeability. The control comparison is explicitly exploratory and subject to selection bias and residual confounding.

Change in the revised manuscript. Methods and Limitations.

Comment 7e

Reviewer comment. The ethical information should be included in a separate ethical statement.

Response. We agree. Ethical approval, data protection, and consent information have been moved to a dedicated Ethics Approval and Consent to Participate section, separate from the laboratory methods.

Change in the revised manuscript. End matter and Methods.

Comment 7f

Reviewer comment. Clarify whether the first visit coincided with treatment initiation and how much time elapsed before the initial assessment.

Response. We clarified that the first clinical visit did not necessarily coincide with treatment initiation. T0 was defined as the laboratory assessment immediately preceding or closest to treatment initiation after eligibility had been confirmed. The exact dates needed to calculate a fixed interval were not available in the definitive workbook, so no interval is reported.

Change in the revised manuscript. Methods, Sections 2.1 and 2.2; Limitations.

Comment 7g

Reviewer comment. The manuscript would benefit from a section on the individual effects of ART and HIV on lipid levels and cardiovascular risk.

Response. The Introduction and Discussion explain that cardiovascular risk and dyslipidemia in PLWH may reflect chronic inflammation and immune activation, traditional risk factors, HIV-associated metabolic abnormalities, and exposure to specific antiretroviral agents. The manuscript also makes clear that this retrospective study cannot separate these mechanisms or attribute the observed kinetics specifically to HIV or ART.

Change in the revised manuscript. Introduction and Discussion.

Reviewer 2 Report

Comments and Suggestions for Authors

This is a retrospective, single-center study evaluating the real-world effectiveness of fixed-dose low-dose rosuvastatin 5 mg/ezetimibe 10 mg in 32 virologically stable people living with HIV (PLWH), with an exploratory comparison to 43 unmatched HIV-negative controls drawn from the same institution's metabolic disease clinic. Over 6 months, PLWH showed a significant LDL-C reduction (170.7→87.8 mg/dL, p<0.001), with 53.1% reaching guideline-recommended LDL targets, alongside improvements in cardiovascular risk scores. A mixed-design ANOVA suggested a different temporal pattern of lipid lowering between PLWH (gradual, sustained through month 6) and controls (rapid, plateauing by month 3).

The efficacy/safety data for PLWH are clinically useful and add to a still-limited real-world literature on statin/ezetimibe combinations in HIV. However, several methodological and reporting issues — detailed below — need to be addressed before the manuscript can be considered for publication

1. Possible circularity in the LDL–D:A:D correlation. The reported correlation between ΔLDL-C and ΔD:A:D score (r=0.74) is presented as evidence that "LDL cholesterol is a key driver of cardiovascular risk in this population." However, total/LDL cholesterol is a direct input variable in most D:A:D risk equations, so part of this correlation is mathematically built in rather than an independent clinical finding. Please clarify which D:A:D model version was used, whether lipid values are a direct input, and temper the causal interpretation accordingly.
2. Control group's own risk-score data are asserted but never shown. Table 2 reports D:A:D, Framingham, and SCORE2 trajectories only for PLWH. The statement that "SCORE2 values were numerically higher in controls" and the baseline Framingham comparison (20.4±14.3%) appear only in the text, with no accompanying table for controls' T0/T3/T6 risk scores. Please add a table (or expand Table 2) with the control group's risk-score data so the comparability claim can be verified by readers.
3. A key subgroup result appears for the first time in the Discussion. The Discussion states that only "20%" of very-high-risk patients reached LDL-C targets, but this stratified result (by CVD risk category) is never presented in the Results section, and the denominator/n is not given. All results discussed must first be reported in Results, with the number of patients per risk category and the proportion reaching target in each.
4. CRP was collected but never reported or used. Section 2.2 lists CRP among the measured biomarkers, yet no CRP values appear anywhere in Results. Given that the Discussion speculates that "HIV-related metabolic and inflammatory factors may contribute to the distinct lipid-lowering kinetics observed," CRP is directly relevant to testing that hypothesis and should be reported (baseline and over time, by group) rather than left unused.

 

5. Abstract: "Conclusions: :" — duplicate colon, needs correction.
6. Section 2.2: "After overnight fasting, and the following biomarkers were measured..." is a sentence fragment; needs rewording (e.g., "After overnight fasting, the following biomarkers were measured...").
7. Reference list formatting is inconsistent in a few entries (missing volume/page ranges); a copy-edit pass is recommended before typesetting.

Author Response

Response to Reviewer 2

We thank the Reviewer for the detailed assessment and for identifying several issues concerning the interpretation and completeness of the reporting. We have addressed each point below.

Comment 1

Reviewer comment. The correlation between ΔLDL-C and ΔD:A:D score may be circular because lipid values are direct inputs in most D:A:D equations. Please specify the model version, whether lipids are direct inputs, and temper the causal interpretation.

Response. We agree. The database was reanalyzed using the latest available full D:A:D cardiovascular risk model, corresponding to the 2016 update described by Friis-Møller et al. The model incorporates total cholesterol and HDL-C together with demographic, clinical, immunological, and antiretroviral-related variables; LDL-C is not entered as a separate direct variable. The previously reported correlation of 0.74 was not reproduced. The correlation between LDL-C change and 10-year D:A:D change was weak and not statistically significant (Spearman’s rho = 0.245, p = 0.177; n = 32), with an approximate 95% CI of −0.114 to 0.547. The corresponding 5-year correlation was also not statistically significant (rho = 0.255, p = 0.159), with an approximate 95% CI of −0.103 to 0.554. The association is now described as exploratory and potentially reflecting shared lipid-related information, not as an independent causal effect.

Change in the revised manuscript. Section 2.3 identifies the latest available full D:A:D model and its 2016 update. Results report the recalculated correlations, their 95% confidence intervals, and the non-causal interpretation. The significant longitudinal reductions in the 5-year and 10-year D:A:D scores are reported separately in the Results and Table 2 with their overall p-values. The 0.74 value and the previous causal wording were removed. The supporting D:A:D publication was added as Reference 19.

Comment 2

Reviewer comment. Control-group cardiovascular risk-score trajectories are not shown. Please add a table or expand Table 2 to allow verification of the comparability claim.

Response. We agree that complete longitudinal cardiovascular-risk data for the HIV-negative controls would have strengthened the comparison. However, the definitive database did not contain complete control risk-score trajectories. D:A:D scores were not calculated for controls because the model is HIV-specific, and no between-group risk-score trajectory claims are made. The manuscript reports only the available baseline Framingham comparison and explicitly identifies the incomplete control data as a limitation. For PLWH, the significant within-person reduction in D:A:D scores over time is now reported separately from the control comparison.

Change in the revised manuscript. Table 2 was expanded with a final p-value column reporting the overall longitudinal test: p = 6.03 × 10⁻⁷ for the 5-year D:A:D score and p = 3.11 × 10⁻⁷ for the 10-year score. The Results and Conclusions state that complete longitudinal risk-score data for controls were unavailable and that no between-group risk-score trajectories were claimed. The available baseline Framingham comparison remains reported in the baseline table. The Statistical Analysis section now specifies the Friedman test for the overall time effect and Wilcoxon signed-rank tests for pairwise comparisons, and the Table 2 footnote identifies the complete-case sample sizes.

Comment 3

Reviewer comment. The statement that only 20% of very-high-risk patients reached LDL-C targets appears for the first time in the Discussion. The denominator and number of patients are not provided.

Response. We agree. Because the denominator and category-specific target-attainment data could not be verified in the definitive analytical workbook, the statement that 20% of very-high-risk patients reached the LDL-C target was removed rather than retained or moved to the Results. Target attainment is now reported only as the overall proportion of PLWH reaching the recommended LDL-C target.

Change in the revised manuscript. The unverifiable 20% subgroup statement was removed from the manuscript and was not replaced by an unverified risk-category result. The Results now report only the verified overall target-attainment proportion.

Comment 4

Reviewer comment. CRP was collected but not reported or used, despite the Discussion speculating about inflammatory factors. Please report CRP at baseline and over time, by group, or remove the speculation.

Response. We agree. Complete CRP data suitable for a baseline and longitudinal by-group analysis were not available in the definitive database. CRP was therefore removed from the Methods and Results, and no mechanistic conclusion based on CRP is made. The remaining discussion of HIV-related inflammation is presented as biological context from the literature, not as an analysis of CRP in this cohort.

Change in the revised manuscript. CRP was removed from Section 2.2 and from the Results. No CRP-based claim is made. The limitations and conclusions call for prospective studies with complete inflammatory data.

Comment 5

Reviewer comment. The Abstract contains the duplicate wording “Conclusions: :”.

Response. Corrected. The duplicate colon has been removed. The Abstract and main text were also checked for consistency with the revised reporting of the significant lipid outcomes, the exploratory HIV-negative comparison, and the D:A:D analyses.

Change in the revised manuscript. Abstract and main text. The Abstract now emphasizes the significant LDL-C reduction, the significant reductions in total and non-HDL cholesterol, the significant month-6 triglyceride reduction, the significant longitudinal reductions in both D:A:D models, the 53.1% LDL-C target attainment, and the absence of treatment-limiting adverse events or viro-immunological deterioration. The Abstract no longer includes the non-significant LDL-C–D:A:D correlation or the limitation concerning unmatched controls. It now states that the exploratory comparison suggested a more rapid early lipid response among HIV-negative controls, with most of the reduction occurring by month 3. The detailed exploratory interpretation remains in the Results and Discussion.

Comment 6

Reviewer comment. The sentence “After overnight fasting, and the following biomarkers were measured” is a fragment.

Response. Corrected. The sentence now reads: “After overnight fasting, the following biomarkers were measured: …”.

Change in the revised manuscript. Methods, Section 2.2.

Comment 7

Reviewer comment. Reference formatting is inconsistent in some entries.

Response. We agree. The reference list was checked and the PDF footer contamination in the previous version of Reference 18 was removed. The D:A:D model source was added as Reference 19 with the complete citation and DOI. The remaining entries were reviewed for consistent journal, year, volume, article-number, and page-range formatting.

Change in the revised manuscript. Reference 18 was corrected, Reference 19 was identified as the D:A:D model source, and the reference list was checked for formatting consistency.

Additional language and consistency revisio

The manuscript was checked for grammar, terminology, inconsistent abbreviations, typographical errors, and consistency in the use of PLWH, HIV-negative controls, LDL-C, D:A:D, and fixed-dose combination therapy. The Conclusions were also revised to emphasize the significant LDL-C and D:A:D findings while framing future hypotheses concerning treatment comparisons, HIV-related inflammation, antiretroviral exposure, cardiometabolic factors, lipid kinetics, and clinical cardiovascular benefit.