Review Reports - Occult Malignancy Presenting as Pericardial Effusion: The Diagnostic Contribution of Pericardial Fluid Cytology

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This retrospective study addresses a clinically relevant question regarding the diagnostic contribution of pericardial fluid cytology in patients presenting with significant pericardial disease. The identification of eight patients in whom cytology provided the first pathological evidence of a previously unrecognized malignancy is particularly noteworthy. The manuscript is potentially suitable for publication; however, several methodological and reporting issues require substantial clarification before the conclusions can be fully supported.

The manuscript frequently uses the terms “pericardial fluid cytology,” “pathological assessment,” and “neoplastic involvement” in ways that are not always clearly distinguishable. The main analysis includes 95 episodes with pathological assessment, whereas cytological and histopathological findings were reportedly collected separately. Please provide a clear breakdown of cases evaluated by cytology only, histology only, both modalities, or neither, together with the corresponding malignant and non-malignant findings. The definition of the composite term “neoplastic pericardial involvement” should also be explicitly stated.
A dedicated Statistical Analysis subsection is needed. The manuscript reports Fisher’s exact tests, Mann–Whitney U tests, Kruskal–Wallis tests, Fisher–Freeman–Halton exact testing with Monte Carlo estimation, odds ratios with confidence intervals, and Benjamini–Hochberg false discovery rate correction, but these methods are not adequately described in the Methods section. Please specify the statistical software and version, assumptions and criteria for test selection, significance threshold, calculation of confidence intervals, handling of missing data, and the approach to multiple comparisons.
The unit of analysis is the clinical episode rather than the individual patient. Since 23 patients contributed more than one eligible admission, observations are not completely independent. This issue should be addressed analytically rather than solely acknowledged as a limitation. At minimum, a sensitivity analysis restricted to the first eligible episode per patient is recommended. Alternatively, a patient-clustered analytical approach may be considered.
Pathological assessment was available in only 95 of 234 episodes. Because specimen acquisition was determined by clinical and procedural circumstances, this creates a potentially important selection bias. Please clarify how many patients undergoing invasive drainage had cytology performed, how many had histopathological examination, how many had both, and why pathological assessment was unavailable in the remaining cases. A comparison of the baseline characteristics of pathology-evaluable versus non-evaluable episodes would help readers assess the extent of potential selection bias.
The cytopathological methodology requires further description. Please clarify specimen processing, whether cell blocks and immunocytochemistry were used, how malignant, suspicious, atypical, and negative findings were classified, and whether standardized reporting criteria were applied. This is especially important because six of the eight occult malignancy cases were considered to be of pulmonary origin. The authors should clarify whether the presumed primary origin was based on cytomorphology alone, immunocytochemical findings, subsequent imaging, or tissue confirmation.
The eight patients in whom pericardial cytology provided the first pathological evidence of malignancy represent the most clinically informative subgroup. A supplementary patient-level table summarizing age, sex, effusion characteristics, procedural approach, cytological diagnosis, immunocytochemical findings, subsequent diagnostic work-up, and final oncological diagnosis would substantially strengthen the manuscript.
Figure 3B appears to contain a color/legend inconsistency. The text states that 13 of 21 neoplastic findings occurred in patients with previously known malignancy and eight occurred in patients without a previous malignancy. However, the colors of the corresponding pie-chart segments appear to be reversed relative to the legend. Please verify and correct the figure.
Several secondary analyses are based on small subgroups and multiple comparisons. Although false discovery rate correction was appropriately used for the laboratory analysis according to cardiac tamponade, several other comparisons are presented with nominal p-values. These analyses should be clearly designated as exploratory, and borderline findings should be interpreted cautiously.
Cohort identification was based on hospitalizations coded with a primary or secondary diagnosis of pericarditis, although the study focuses primarily on pericardial effusion and tamponade. Please report the diagnostic codes used for case identification and clarify whether cases coded primarily as pericardial effusion or cardiac tamponade could have been missed. Operational definitions of pericardial effusion severity, hemodynamic compromise, and cardiac tamponade should also be provided.
The STROBE checklist should be revised after the methodological section is updated. Several checklist entries currently indicate that statistical methods and missing-data handling are described in the Methods section, although these details are not sufficiently provided in the submitted manuscript. The supplementary file also contains an unrelated title at the beginning and should be corrected.

The manuscript would also benefit from language editing and shortening of repetitive sections in the Introduction and Discussion. Overall, the study has a clinically relevant message, but substantial methodological clarification and re-analysis are required before the findings can be reliably interpreted.

Author Response

We appreciate the reviewer’s detailed evaluation of our manuscript. The comments have prompted several revisions aimed at improving methodological transparency, the presentation of the results, and the interpretation of our findings.

The main revisions include:

Pathological assessment: We clarified the distinction between cytological and histopathological examinations, defined neoplastic pericardial involvement more precisely, and acknowledged the limitations of the retrospective records in distinguishing the contributions of each diagnostic modality.
Statistical methodology: We expanded the Materials and Methods section to describe the statistical procedures, criteria for test selection, handling of missing data, and approach to multiple comparisons.
Repeated hospitalizations: We performed an additional sensitivity analysis restricted to the first eligible hospitalization per patient. The findings were consistent with the primary episode-based analysis, supporting the stability of the principal results.
Selection bias: We clarified the circumstances affecting the availability of pathological assessment and emphasized that the proportion of neoplastic findings applies only to the pathologically evaluable subgroup, rather than the entire cohort.
Pathological interpretation: We clarified how pathological diagnoses were obtained and documented, including the basis for the presumed pulmonary origin in several cases. We also acknowledged the absence of consistently available information on specimen processing and subsequent investigations performed at other institutions.
Occult malignancy cases: We added a patient-level table to the main manuscript, presenting the clinical characteristics, procedural approaches, pathological findings, and available immunocytochemical information for the eight patients in whom malignancy was first identified through pericardial assessment.
Figures: We corrected the color and legend inconsistency in Figure 3B without altering the underlying data.
Exploratory analyses: We clarified the exploratory nature of secondary comparisons and strengthened the discussion of statistical uncertainty, particularly for small subgroups and nominally significant findings.
Cohort identification: We specified that patient identification was restricted to the Department of Cardiovascular Surgery and clarified how cardiac tamponade was identified from the clinical documentation. The potential for incomplete case ascertainment was also acknowledged.
Reporting and presentation: We revised the STROBE checklist to reflect the updated methodology, corrected the supplementary file, and edited the Introduction and Discussion to improve readability and reduce repetition.

These changes provide a clearer account of the study design, the available evidence, and the limitations inherent to its retrospective nature. The principal findings and clinical focus of the study remain unchanged.

All modifications addressing Reviewer 1’s comments have been highlighted in yellow in the revised manuscript to facilitate their identification and verification.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

First of all, congratulations for your interesting work. I hope that my hints will help you in the next steps of improvement and the final manuscript will be really valuable for the readers. I would consider it clinically relevant and potentially publishable, but its main contribution is currently more descriptive/diagnostic than mechanistic or precision-medicine oriented, therefore several elements would require corrections.

The finding that cytology can detect malignancy at the first presentation is also not surprising, but it is clinically useful. Recent multi-institutional evidence actually supports the importance of cytology, showing higher sensitivity of pericardial-fluid cytology than biopsy in malignant effusions. Therefore, the paper is not challenging established knowledge, but rather adds a useful real-world, cardiovascular-surgical cohort demonstrating how that knowledge operates in patients presenting with clinically significant effusion.

The main issue is that the manuscript sometimes presents this as if it were establishing the diagnostic accuracy of cytology. It does not. The authors correctly acknowledge this: only 40.6% of episodes had pathological assessment, and cytology and tissue histology were not systematically paired. Only 95/234 episodes underwent pathological assessment. Importantly, sampling was determined by clinical/procedural circumstances rather than by a predefined study protocol. Therefore, the reported 22.1% rate of neoplastic involvement cannot be interpreted as the prevalence of malignant pericardial effusion in this cohort. This is probably the most important limitation because the manuscript's central message concerns the diagnostic contribution of cytology. The subgroup that underwent sampling may have been systematically different; eg. larger effusions, tamponade, known cancer, suspicious imaging, recurrent disease, etc. The cohort itself is also strongly enriched for severe disease: 91% had large effusions and 74.4% had haemodynamic compromise.

Secondly, the study does not really establish the diagnostic performance of cytology. The title and conclusions emphasize the "diagnostic contribution" of cytology, but the design cannot provide sensitivity, specificity, negative predictive value, positive predictive value, or diagnostic accuracy. There was no systematic cytology–histology pairing, and there was no longitudinal verification of patients with negative cytology. The authors acknowledge that the absence of follow-up prevents estimation of false-negative rates and negative predictive value. Pericardial This distinction is important because current literature already contains comparative diagnostic studies of cytology and biopsy. Recent multi-institutional data, for example, directly evaluated their relative sensitivity. The present manuscript therefore has a different and potentially valuable question, but it should make that distinction much sharper. I would change the framing from "diagnostic performance/value of cytology" toward "clinical diagnostic contribution of pathological assessment in patients with pericardial effusion." The authors could then emphasize that their novel contribution is not proving that cytology is sensitive, but demonstrating its clinical pathway value, particularly in occult malignancy.

Moreover, the paper is not yet sufficiently aligned with modern personalised oncology and multi-omics This is, in my opinion, the most important opportunity for strengthening the manuscript. The study stops essentially at morphological diagnosis and broad tumour origin. For example, six of eight occult malignancies were considered pulmonary in origin, but the study does not address molecular subtype, actionable alterations, PD-L1 status, immune phenotype, tumour mutational burden, or other information increasingly relevant to treatment selection. This is particularly relevant because pericardial fluid is not only a diagnostic specimen; it can potentially be a source of tumour-derived material for molecular characterization. Modern oncology increasingly asks not simply "Is this malignant?" but "What molecularly defined malignancy is this, and what treatment is appropriate for this individual patient?" I would therefore suggest that the Discussion be expanded to introduce a precision-diagnostics concept: cytology could be the first step, followed: when malignant cells are identified and material is sufficient, of course by molecular testing. For lung cancer, for example, this could potentially include targeted genomic profiling and PD-L1 assessment, depending on specimen adequacy and clinical indication.

Finally, I would like to thank you for the excellent tables you have prepared for the document, they enhance the value of your work and facilitate the understanding process. Would it be possible to add some figures and graphs too? It will definitely increase the value of your already well-written manuscript. There are several mechanisms and many molecules described in the paper, it may not be easy to digest for younger readers, and good pictures showing these pathways and molecules, also aberrant places, will definitely be of great value.

Author Response

We sincerely thank the reviewer for the thorough and constructive evaluation of our manuscript and for recognizing its clinical relevance. The comments have helped us refine the scientific focus of the study, clarify its methodological limitations, and better define its contribution within the existing literature.

In response to the reviewer's recommendations, we have made the following revisions:

Clarification of the study population and potential selection bias. We have emphasized that pathological assessment was available in only 95 of 234 episodes (40.6%) and that the reported 22.1% proportion of neoplastic findings applies exclusively to this subgroup. We have also clarified that specimen availability depended on clinical and procedural circumstances rather than a predefined sampling protocol. The implications of this selection process for the interpretation and generalizability of our findings are now explicitly discussed.
Distinction between diagnostic performance and clinical diagnostic contribution. We have refined the Introduction to clarify that our study does not evaluate the sensitivity, specificity, predictive values, or diagnostic accuracy of pericardial fluid cytology. Instead, its contribution lies in describing the clinical role of pathological assessment in a real-world cardiovascular surgical cohort, particularly in patients whose pericardial disease represented the first manifestation of an otherwise unrecognized malignancy. The relevant methodological limitations are also addressed in the Discussion.
Integration of a precision-oncology perspective. We have expanded the Discussion to acknowledge the potential role of pericardial fluid in molecular characterization and personalized oncological management. At the same time, we have emphasized that improving access to routine cytological assessment and standardizing specimen collection and processing remain important priorities, particularly given the limited availability of pathological evaluation in our cohort. Molecular testing is presented as a potential subsequent step in appropriately selected patients, not as an investigation performed in the present study.
Addition of a graphical representation. We have introduced Figure 6, entitled “Proposed clinical pathway for suspected malignant pericardial involvement,” to illustrate the integration of cytological findings with further clinical, imaging, and oncological investigations. The figure emphasizes that negative cytology does not exclude malignancy and highlights the potential role of additional investigations, including molecular characterization, when clinically indicated. We have explicitly identified this pathway as conceptual and not prospectively validated.

We have also reviewed the Introduction and Discussion to improve clarity, reduce repetition, and maintain a more focused scientific narrative.

To facilitate the review process, newly introduced revisions addressing Reviewer 2 have been highlighted or underlined in green in the revised manuscript. Modifications previously introduced in response to Reviewer 1 remain highlighted in yellow.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript addresses a clinically relevant question and the limitations are acknowledged. One substantive issue and several minor points require attention.

The principal concern is terminology. The Methods state that cytological and histopathological results could not be separated in the database, and “pathological assessment” is therefore used. However, the title, abstract, Table 1 and Conclusions attribute all eight occult-malignancy diagnoses to pericardial fluid cytology. Five of these patients underwent surgical drainage, and the finding in case 6 (“tumour emboli from breast carcinoma”) suggests a tissue-based diagnosis. The authors should review the eight pathology reports and indicate for each case whether the diagnosis was established on fluid, tissue, or both. If this cannot be determined, the title and conclusions should refer to pathological assessment rather than cytology.

Immunocytochemistry was not recorded in seven of eight cases. The basis for attributing pulmonary origin should be stated. If imaging findings contributed, this should be made explicit and the conclusion regarding tumour-origin attribution revised accordingly.

In case 6, only 90 mL was drained, which is inconsistent with the statement that all eight patients presented with large effusions and haemodynamic compromise. Please clarify.

The analysis is episode-based and 23 patients contributed multiple admissions. The authors should confirm that no patient appears in both oncological-status groups across admissions, and report the odds ratio from the first-episode sensitivity analysis.

Reference 5 is cited in support of statements derived from the 2015 ESC Guidelines on pericardial diseases and from the ESC/ACVC consensus document. The appropriate sources should be cited.

The Introduction should be substantially shortened. The Discussion repeats numerical results. Section 3.7 is peripheral to the study aim and could be moved to the Supplementary Material.

Author Response

Comments

In case 6, only 90 mL was drained, which is inconsistent with the statement that all eight patients presented with large effusions and haemodynamic compromise. Please clarify.

Reference 5 is cited in support of statements derived from the 2015 ESC Guidelines on pericardial diseases and from the ESC/ACVC consensus document. The appropriate sources should be cited.

The Introduction should be substantially shortened. The Discussion repeats numerical results. Section 3.7 is peripheral to the study aim and could be moved to the Supplementary Material.

Response

We sincerely thank the reviewer for the careful evaluation of our manuscript and the constructive suggestions. We have carefully considered the observations and revised the manuscript to improve the accuracy and clarity of our findings while preserving its clinical focus. The main modifications have been highlighted in yellow to facilitate their identification in the revised version.

To address the distinction between pericardial fluid cytology and tissue histopathology, we individually reviewed the original pathology reports of all eight patients presenting with previously unrecognized malignancy. This verification confirmed that malignant cells were identified in pericardial fluid cytology in every case. In Case 6, cytology provided the first pathological evidence of malignancy, preceding tissue histopathology, which subsequently demonstrated intravascular tumor emboli and neoplastic infiltration, with strong CK7 and GATA3 positivity and negative TTF-1 staining. We have revised the Materials and Methods and Results sections and updated Table 1 to distinguish cytological, histopathological, and immunostaining findings, including their respective specimen types. As positive pericardial fluid cytology was confirmed in all eight patients, we have retained the original manuscript title.

We have also refined the interpretation of the presumed primary tumor origin. In Case 4, TTF-1 positivity was documented in the pericardial fluid cell sediment, while in Case 8, tissue immunohistochemistry demonstrated variable positivity for TTF-1 and Napsin A, supporting a pulmonary origin. In the remaining cases, the presumed pulmonary origin was based on the original pathology reports, without additional immunostaining results documented in the available records. Section 3.4 and the Abstract have been revised to distinguish confirmed malignancy from presumed tumor origin, and to acknowledge that subsequent investigations performed at external institutions were unavailable. Accordingly, the six cases are now described as having findings suggestive of a pulmonary origin.

Regarding Case 6, we have clarified that cardiac tamponade is determined by hemodynamic consequences rather than by a specific volume of pericardial fluid. Echocardiographic findings were consistent with cardiac tamponade despite the relatively small drained volume of 90 mL. Section 3.4 has been revised to distinguish effusion size from hemodynamic compromise and to clarify the clinical findings in this patient.

To account for repeated hospitalizations, we performed a sensitivity analysis restricted to the first eligible hospitalization per patient. Among 209 individuals, pathological assessment was available in 85, with neoplastic findings identified in 19. The association between previously known malignancy and neoplastic pathological findings remained statistically significant (OR, 3.16; 95% CI, 1.11–9.05), supporting the consistency of the primary episode-based findings. These results have been incorporated into Section 3.1.

We have also reviewed and updated the relevant guideline citations. The 2022 ESC Guidelines on Cardio-Oncology (Lyon et al.) have been retained for statements concerning neoplastic pericardial involvement, while the 2025 ESC Guidelines for the management of myocarditis and pericarditis and the ESC/ACVC consensus document (Gevaert et al.) have been cited where appropriate to support recommendations concerning pericardial effusion, cardiac tamponade, and acute pericardial disease in patients with cancer.

Finally, we have made minor editorial adjustments to the Introduction, removing redundant statements while preserving the scientific rationale of the study. We have retained the principal numerical findings in the Discussion where they support interpretation and comparison with previous studies. We have also carefully considered the suggestion to move Section 3.7 to the Supplementary Materials. Given the clinical focus of our investigation, we believe that retaining this section in the main manuscript provides important context for understanding the relationship between clinical presentation, laboratory findings, and pathological results, and preserves the continuity of the study.

We are grateful for the reviewer's thoughtful observations, which have helped us improve the methodological transparency and clinical interpretation of our findings.