Value of RASSF1A and SHOX2 methylation detection in intraoperative pleural lavage fluid for potential microdissemination and prognosis evaluation in non-small cell lung cancer

Abstract

Objective

To evaluate the potential of RASSF1A and SHOX2 methylation status in intraoperative pleural lavage (IPL) fluid for assessing invasiveness and predicting postoperative recurrence in non-small cell lung cancer (NSCLC).

Methods

Forty-three NSCLC patients who underwent lung tumor resection and IPL between November 2021 and June 2022 were prospectively enrolled. The patients included in the analysis did not receive adjuvant chemotherapy after surgery. Methylation of RASSF1A and SHOX2 in the IPL fluid was determined. Correlation of methylation status with clinicopathological or imaging features was assessed by Spearman correlation. Progression-free survival (PFS) and overall survival (OS) were compared by Kaplan-Meier curves and log-rank tests. Receiver operating characteristic curves evaluated the predictive performance of RASSF1A, SHOX2, and their combination for PFS.

Results

Positive methylation rates were 25.58% for RASSF1A, 27.91% for SHOX2, and 18.86% for combined detection. RASSF1A methylation positively correlated with neuron-specific enolase, CYFRA21-1, and imaging features (irregular tumor boundaries, lymph node enlargement, and tumor infiltration) (P < 0.01), and negatively with tumor density (solid) (P = 0.016). SHOX2 methylation correlated positively with CYFRA21-1 and imaging features (all P < 0.01). Kaplan-Meier analysis showed that the PFS of RASSF1A, SHOX2, and their combination positive patients was significantly shorter than that of the negative group (all P < 0.001). OS analysis also showed a similar trend, but due to the limited number of deaths, the relevant results need to be interpreted with caution. ROC analysis showed that the AUC of RASSF1A, SHOX2, and combined detection for predicting PFS were 0.817, 0.923, and 0.962, respectively, with the combined detection performing the best.

Conclusion

RASSF1A and SHOX2 methylation in IPL fluid closely correlates with aggressive pathological features and unfavorable NSCLC prognosis. In an exploratory cohort without postoperative adjuvant therapy, combined testing showed potential value in stratifying the risk of postoperative recurrence and is more suitable as a supplementary molecular reference for postoperative adjuvant therapy decision-making and follow-up management. However, it still needs to be further validated in multi-center populations receiving standard treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12957-026-04408-1.

Introduction

Lung cancer (LC) remains the most frequently diagnosed and lethal malignant tumor worldwide, accounting for approximately 2.5 million new cases and 1.8 million deaths globally in 2022 [1], which is nearly one-fifth of all cancer-related fatalities. Non-small cell lung cancer (NSCLC), the most common histological subtype, represents about 85% of all cases [2]. Despite improved resection rates in early-stage NSCLC through advancements in imaging and surgical techniques, postoperative recurrence and occult micrometastasis still compromise long-term survival [3]. A cumulative risk of second primary or recurrence in NSCLC patients 10 years after surgery reaches as high as 33.3%, with a recurrence rate after 5 years close to 50% [4]. Routine clinical examinations currently lack the sensitivity to accurately detect pleural micrometastasis. Computed tomography (CT) only visualizes millimeter-sized lesions. Histopathology frequently fails to identify minimal residual disease or potential pleural dissemination. Thus, there is an urgent need for a detection method that can more sensitively reflect the potential tumor residue or microdissemination status in the pleural cavity during the perioperative period, in order to compensate for the limitations of imaging and traditional pathological evaluation.

Intraoperative pleural lavage cytology (PLC) is sensitive for detecting subclinical tumor cell dissemination and can indicate shedding of tumor cells via pleural or lymphatic routes [5]. Recuero et al. [6]. observed a 2.2% (15/684) PLC positivity rate in NSCLC; the two-year recurrence rate in PLC-positive patients was 73.3% (11/15) versus 27.2% (182/669) in PLC-negative patients; multivariate and propensity-score analyses demonstrate that PLC positivity significantly increased early recurrence risk (OR = 3.46, P < 0.001). However, traditional cytological methods have limited sensitivity and insufficient ability to identify very small numbers of tumor cells, potentially underestimating the risk of pleural cavity microdissemination. Higher-sensitivity molecular assays are required to detect minimal tumor signals in the pleural cavity. DNA methylation, a key epigenetic mechanism regulating gene expression and maintaining cellular identity, critically participates in the development and progression of multiple cancers [7–9]. DNA methylation markers can be applied to cancer screening, auxiliary diagnosis, and prognostic monitoring [10]. Early LC diagnosis could be achieved by detecting methylation alterations of genes, such as SHOX2 and RASSF1A, in tissue, blood, or bronchoalveolar lavage fluid [11–13]. Jin et al. [14]. reported comparable predictive performance of joint SHOX2 and RASSF1A methylation detection in plasma and bronchoalveolar lavage fluid (BALF) (area under the curve [AUC] 0.806 vs. 0.781). Thus, plasma could serve as an alternative to BALF for sensitive and specific early LC diagnosis. Moreover, joint SHOX2 and RASSF1A methylation detection in pleural effusion outperformed conventional cytology [15], supporting the feasibility and potential of pleural-fluid-based methylation assays. Intraoperative pleural lavage (IPL) fluid can provide a unique sample source for tumor-derived cells or free DNA detection in the local pleural cavity environment, thus providing the possibility for perioperative microdissemination risk assessment and postoperative risk stratification.

We hypothesized that RASSF1A and SHOX2 methylation in IPL fluid could serve as sensitive molecular markers for detecting the potential risk of pleural cavity microdissemination and predicting postoperative recurrence. Using NSCLC patients undergoing curative surgery as the study population, we systematically evaluated the relationships of RASSF1A and SHOX2 methylation in IPL fluid with tumor pathological features, imaging manifestations, and postoperative outcomes, aiming to explore its potential application value as a molecular marker of micro-dissemination in the assessment of LC invasiveness and prognosis prediction, providing new molecular basis for individualized risk stratification and precision treatment.

Methods

Study population

The single-center prospective study enrolled 43 NSCLC patients who underwent lung resection and IPL at our institution between November 2021 and June 2022. Inclusion criteria: (1) age ≥ 18 years; (2) pre-operative chest X-ray, CT, and B-ultrasound tests confirming absence of pleural effusion within one week before surgery; (3) with liver B-ultrasound, cranial CT, and bone scan tests confirming absence of distant metastasis; (4) without any previous traumatic examination (excluding fiberoptic bronchoscopy examination) of the ipsilateral pleural cavity except fiber-optic bronchoscopy; (5) without evidence of malignancy in other sites or organs; (6) with willingness to undergo 3-year follow-up and to cooperate with relevant imaging or tumor-marker examinations. Exclusion criteria: (1) pregnant or lactating women; (2) with benign lung tumors, extensive pleural adhesions, or definite malignant pleural effusion before surgery; (3) with an inability to understand or sign informed consent. During the follow-up period of the study, all patients included in the analysis did not receive adjuvant chemotherapy after surgery. For patients in stage III and above, the absence of postoperative adjuvant chemotherapy is mainly related to real-world clinical factors, including patient refusal, comorbidities, inappropriate physical condition, and limited treatment compliance, rather than the standard treatment exclusion set by the study protocol. The postoperative treatment decision is independent of the methylation test results in this study.

All patients provided written informed consent. The study protocol was approved by the hospital’s ethics committee (Approval No.: 2021KY56J01) and was conducted in accordance with the Declaration of Helsinki. The final diagnosis for all patients was confirmed by postoperative pathological results.

Data collection

Clinical, imaging, and laboratory data were collected to evaluate the relationship of RASSF1A and SHOX2 methylation in IPL fluid with tumor invasiveness and NSCLC prognosis. Clinical data included demographic characteristics (sex, age), medical history (hypertension, diabetes, etc.), smoking history, and surgical approach (wedge resection, segmentectomy, or lobectomy). Pathological results included lymph-node metastasis status, pathological stage, histological subtype, maximum tumor diameter and Ki-67 index. Hematological data comprised white blood cell count (WBC), red blood cell count (RBC), platelet count (PLT), and hemoglobin (Hb); serum tumor markers including carcinoembryonic antigen (CEA), cytokeratin 19 fragment antigen 21 − 1 (CYFRA21-1), neuron-specific enolase (NSE), squamous cell carcinoma antigen (SCCA), and pro-gastrin-releasing peptide (proGRP). Imaging variables included tumor location (central or peripheral) from pre-operative chest CT findings, tumor border characteristics, mediastinal and hilar lymph-node enlargement, tumor infiltration, and tumor density.

Sample collection

IPL fluid was collected (1) after fluid addition in the pleural cavity and before lung resection (2), after tumor resection and completion of systematic lymph-node dissection, and (3) before chest closure. In our exploratory experiments, the (2) post-resection lavage sample showed the highest methylation positivity, more sufficiently reflecting intraoperative tumor shedding, and was therefore used for further analysis. Briefly, 500 mL sterile normal saline was instilled to thoroughly irrigate the pleural cavity, and ≥ 30 mL of IPL fluid was retrieved.

Sample processing and methylation detection

As described in the literature [16], genomic DNA was extracted from 10 mL of IPL fluid centrifuged at 10,000 rpm for 5 min and subjected to bisulfite treatment: unmethylated cytosines were converted to uracil, whereas methylated cytosines remained unchanged. Real-time qPCR was run on the Applied Biosystems 7500 system (Thermo Fisher Scientific, China). The Human SHOX2 and RASSF1A gene methylation detection kit (Tellgen Co., Ltd., Shanghai, China) was used to quantify methylation in promoter CpG islands. Methylated SHOX2 and RASSF1A DNA plasmids served as positive controls [17].

qPCR results were interpreted. RASSF1A methylation was considered positive if the amplification curve of FAM fluorescence signal showed a smooth “S” shape and the threshold cycle (Ct) was less than 35; negative when Ct ≥ 35. SHOX2 methylation was positive if the VIC amplification fluorescence signal showed a smooth “S” and Ct < 32; negative when Ct ≥ 32 [12].

Follow-up

Patients were followed up via electronic medical records, outpatient visits, and telephone contact until June 2025. Progression-free survival (PFS) was defined as the time from the date of the patient’s first surgery to disease progression or tumor-related death. Those who had not experienced any incidents were excluded as of the last valid follow-up. OS was defined as the time from the date of surgery to death from any cause. Survivors were excluded as of the last valid follow-up [18].

Statistical analysis

Data analyses were run in SPSS 26.0 and GraphPad Prism 9.0. Normally distributed continuous variables were presented as mean ± SD (x̄ ± s) and intergroup differences were examined by t-tests. Skewed data were expressed as median (interquartile range), and intergroup comparison was conducted by the Mann-Whitney U test. Categorical variables were reported as n (%) and compared by χ² or Fisher’s exact test. Associations between methylation status and tumor markers or imaging features were analyzed using Spearman correlation with 95% confidence intervals (CI) derived from 1,000 bootstrap replicates. PFS and OS were estimated using the Kaplan-Meier method, with between-group differences compared with the log-rank test. Factors associated with PFS were analyzed with Cox proportional hazards regression. Predictive performance was evaluated using ROC curve analysis, with predictive ability expressed as the AUC. Due to limited sample sizes and event counts in some subgroups after stratification, the stability of HR and 95% CI estimated by the Cox model was insufficient; therefore, stratified survival analyses were primarily presented using Kaplan-Meier curves and log-rank tests. Two-sided P < 0.05 was considered statistically significant.

Results

Baseline characteristics of patients

Forty-three surgically treated NSCLC patients were enrolled: 25 males (58.14%) and 18 females (41.86%), median age 63.00 years. Lobectomy was performed in 39 cases (90.70%). In terms of pathological stage distribution, stage I accounted for 53.49% (23), stage III for 37.21% (16), and stages II and IV each for 4.65% (2). Median maximum tumor diameter was 2.00 cm; 60.47% (26) were 1–3 cm. Adenocarcinoma accounted for 72.09% and squamous cell carcinoma for 23.26%. Patient’s pathological results, laboratory indicators, and imaging characteristics are listed in Table 1. Preoperative CT findings indicated peripheral-type tumors accounting for 69.77% (30), central-type for 30.23% (13), mediastinal/hilar lymph-node enlargement for 23.26% (10), and solid lesions for 72.09% (31).

Table 1.

Baseline clinical pathological characteristics, laboratory indicators, and imaging features of patients

Characteristics	Median (quartile) /n (%)
General information
Gender
Male	25 (58.14)
Female	18 (41.86)
Age	63.00 (57.00, 70.00)
Hypertension history	15 (34.88)
Diabetes history	7 (16.28)
Smoking history	15 (34.88)
Surgical approach
Wedge resection	2 (4.65)
Segmentectomy	2 (4.65)
Lobectomy	39 (90.70)
pathological results
lymph node metastasis
Yes	14 (32.56)
No	29 (67.44)
pathological staging
Ⅰ	23 (53.49)
Ⅱ	2 (4.65)
Ⅲ	16 (37.21)
Ⅳ	2 (4.65)
pathological type
adenocarcinoma	31 (72.09)
squamous cell carcinoma	10 (23.26)
adenosquamous carcinoma	2 (4.65)
maximum tumor diameter	2.00 (1.50, 3.00)
< 1 cm	6 (13.95)
1–3 cm	26 (60.47)
≥ 3 cm	11 (25.58)
Ki-67 (%)	40.00 (10.00, 60.00)
Laboratory indicators
WBC (×10⁹/L)	5.91 (4.80, 7.16)
RBC (×10¹²/L)	4.30 (4.03, 4.58)
PLT (×10⁹/L)	243.00 (189.00, 285.00)
Hb (g/L)	131.00 (121.00, 142.00)
CYFRA21-1 (ng/mL)	2.35 (1.89, 3.10)
CEA (ng/mL)	2.87 (1.85, 6.09)
NSE (ng/mL)	8.81 (6.50, 10.65)
SCCA (ng/mL)	0.50 (0.35, 0.66)
proGRP (pg/mL)	33.20 (26.10, 42.50)
Imaging features
Tumor Category
peripheral tumor	30 (69.77)
Center tumor	13 (30.23)
Tumor Margin Feature
smooth	13 (30.24)
burred	15 (34.88)
pleural retraction/vascular penetration	15 (34.88)
Lymphadenopathy
Yes	10 (23.26)
No	33 (76.64)
Tumor Invasion
Yes	3 (6.98)
No	40 (93.02)
Tumor Density
Solid	31 (72.09)
Non-solid	12 (27.91)

No patient received adjuvant chemotherapy after surgery

WBC White Blood Cell count, RBC Red Blood Cell count, PLT Platelet count, Hb Hemoglobin, Ki-67 Ki-67 proliferation index, CYFRA21-1 Cytokeratin 19 fragment antigen 21 − 1, CEA Carcinoembryonic Antigen, NSE Neuron-Specific Enolase, SCCA Squamous Cell Carcinoma Antigen, proGRP Pro-Gastrin-Releasing Peptide

The association of RASSF1A and SHOX2 methylation with lymph node metastasis and tumor size

The methylation positivity of RASSF1A and SHOX2 were 25.58% (11) and 27.91% (12), respectively, and the joint positivity was 18.86% (8). In the presence of lymph-node metastasis, positivity rates for RASSF1A and SHOX2 were both 57.14% (8), and joint positivity was 42.86% (6), all significantly higher than in patients without nodal involvement (Fig. 1A, P < 0.01). Positivity rates were numerically higher in larger tumors but did not reach statistical significance (Fig. 1B).

Fig. 1.

Methylation-positive rates according to lymph-node metastasis status (A) and maximum tumor diameter (B)

The association of RASSF1A and SHOX2 methylation with serum tumor markers and imaging features

Spearman correlation results are shown in Table 2. Among serum tumor markers, RASSF1A methylation was positively correlated with NSE (r = 0.430, P = 0.004) and CYFRA21-1 (r = 0.387, P = 0.010); no significant correlations were observed with CEA, SCCA or pro-GRP. SHOX2 methylation was positively correlated with CYFRA21-1 only (r = 0.389, P = 0.010). Regarding imaging features, RASSF1A methylation was positively correlated with tumor border characteristics (r = 0.428, P = 0.004), lymph-node enlargement (r = 0.687, P < 0.001) and tumor infiltration (r = 0.467, P = 0.002), and negatively correlated with solid tumor density (r=-0.365, P = 0.016). SHOX2 methylation was also positively correlated with tumor border characteristics (r = 0.417, P = 0.005), lymph-node enlargement (r = 0.517, P < 0.001) and tumor infiltration (r = 0.440, P = 0.003). Joint RASSF1A and SHOX2 methylation was positively correlated with central tumor location (r = 0.336, P = 0.028); all other correlations with serum markers and imaging features were consistent with single-gene analyses.

Table 2.

Correlation of RASSF1A and SHOX2 methylation status with serum tumor markers and imaging features

Feature	RASSF1A			SHOX2			Joint
	r	95% CI	P-value	r	95% CI	P-value	r	95% CI	P-value
Tumor marker
CYFRA21-1	0.387	0.034–0.661	0.010	0.389	0.065–0.647	0.010	0.400	0.018–0.658	0.008
CEA	0.150	-0.147-0.420	0.336	0.175	-0.118-0.449	0.260	0.063	-0.254-0.373	0.690
NSE	0.430	0.162–0.640	0.004	0.278	-0.040-0.542	0.071	0.397	0.112–0.610	0.008
SCCA	0.103	-0.209-0.377	0.510	0.203	-0.136-0.484	0.192	0.161	-0.123-0.410	0.301
pro GRP	0.090	-0.207-0.412	0.565	0.059	-0.258-0.336	0.709	0.063	-0.252-0.349	0.690
Imaging feature
Tumor Category	0.194	-0.143-0.517	0.212	0.268	-0.070-0.572	0.083	0.336	0.003–0.635	0.028
Tumor Margin Feature	0.428	0.196–0.632	0.004	0.417	0.145–0.637	0.005	0.496	0.263–0.682	0.001
Lymphadenopathy	0.687	0.394–0.894	< 0.001	0.517	0.190–0.801	< 0.001	0.587	0.258–0.842	< 0.001
Tumor Invasion	0.467	0.248–0.709	0.002	0.440	0.234–0.687	0.003	0.573	0.323–0.842	< 0.001
Tumor Density	-0.365	-0.509–0.227	0.016	-0.272	-0.457–0.008	0.078	-0.297	-0.432–0.162	0.053

The Spearman correlation coefficient was calculated using Spearman rank correlation analysis, and the P-value was obtained based on Spearman rank correlation test. The 95% CI was estimated through Bootstrap resampling (1000 times). Due to the different statistical inference methods used, there may be slight inconsistencies between the P-value and the 95% CI boundary under small sample conditions. The relevant results should be interpreted cautiously in conjunction with the effect size and overall trend

CI Confidence Interval, CYFRA21-1 Cytokeratin 19 fragment antigen 21 − 1, CEA Carcinoembryonic Antigen, NSE Neuron-Specific Enolase, SCCA Squamous Cell Carcinoma Antigen, pro-GRP Pro-Gastrin-Releasing Peptide

Differences in PFS and OS according to methylation status

By the cutoff date, 11 patients had experienced recurrence and 5 had died. K-M analyses (Fig. 2) showed that PFS was significantly shorter in patients with RASSF1A methylation positivity than in those with negativity (log-rank χ²=25.47, P < 0.001). Similarly, PFS was markedly shortened in patients with SHOX2 methylation positivity and in those with joint positivity (both P < 0.001). Median PFS was 19.00 months in the RASSF1A-positive group, 20.00 months in the SHOX2-positive group, and 18.50 months in the joint-positive group; no negative groups reached median PFS (Fig. 2A-C). In OS analysis, the RASSF1A, SHOX2, and combined methylation positive groups showed a generally poor survival trend (all P < 0.001). However, due to the limited total number of deaths, the relevant results should be interpreted with caution. The median OS of the positive group in the joint detection was 32.00 months, while the other groups did not reach the median survival time (Fig. 2D-F).

Fig. 2.

Progression-free survival (A-C) and overall survival (D-F) for RASSF1A methylation, SHOX2 methylation, and their combined methylation

Furthermore, considering the significant impact of pathological stage on prognosis, we performed stratified Kaplan-Meier analyses by pathological stage (I-II vs. III-IV). After stratification, the direction of the association between each methylation marker and poorer prognosis was generally consistent with the overall analysis. In the PFS analysis, among both stage I-II and stage III-IV patients, those positive for RASSF1A, SHOX2, and joint methylation exhibited significantly shorter PFS compared to the negative groups (all P < 0.001 for stage I-II; all P < 0.01 for stage III-IV, Figure S1). In the OS analysis, due to the limited total number of death events (n = 5), the Kaplan Meier analysis of OS after staging and stratification was only presented for exploratory purposes (Figure S2). Overall, the positive groups of various methylation indicators showed a poor OS trend, but the relevant results need to be interpreted with caution.

Predictive value of joint RASSF1A and SHOX2 methylation for recurrence

To evaluate the predictive efficacy of RASSF1A and SHOX2 gene methylation and their combined detection for PFS, this study conducted an exploratory ROC analysis with the occurrence of progression during follow-up as the binary outcome. The AUC was 0.817 for RASSF1A methylation, 0.923 for SHOX2 methylation, and 0.962 for the joint model, indicating superior predictive efficacy of the joint approach (Fig. 3).

Fig. 3.

ROC curves of RASSF1A, SHOX2, and joint methylation detection for PFS. A Univariate ROC curves comparing the predictive performance of RASSF1A, SHOX2, and joint detection for progression-free survival; (B) Multivariate ROC curves after adjusting for the confounding factor (lymph node metastasis), showing the prognostic discrimination of each model after adjustment

Discussion

We assessed the methylation of RASSF1A and SHOX2 in IPL fluid of NSCLC patients and evaluated their associations with tumor invasiveness, imaging features, and postoperative prognosis. RASSF1A and SHOX2 methylation were detected in 25.58% and 27.91% of cases, respectively, and were significantly increased in patients with lymph node metastasis. In addition, the combined methylation status is positively correlated with central-type tumors. Kaplan Meier analysis further showed that methylation positivity was significantly associated with shorter PFS, and OS also showed a similar trend. However, due to the limited number of deaths, the relevant results should be interpreted with caution. The joint detection showed better recurrence discrimination ability than a single indicator in this exploratory cohort, indicating its potential clinical application value.

RASSF1A, a classic tumor suppressor gene, critically maintains cellular homeostasis by regulating the cell cycle, apoptosis, and microtubule stability [19–21]. SHOX2 methylation is considered a sensitive molecular marker for early diagnosis and potential microdissemination monitoring in liquid biopsy of lung cancer [22, 23]. Both RASSF1A and SHOX2 methylation in IPL fluid were positively correlated with imaging features indicative of tumor invasiveness (e.g., indistinct borders, lymph node enlargement, and tumor infiltration). Thus, they may reflect molecular signals of tumor microdissemination. This aligns with Nonaka et al.’s findings that patients with no pleural signs on preoperative thin-section CT and a consolidation/tumor ratio ≤ 0.5 rarely showed PLC positivity or pleural dissemination recurrence [24]. Although the detection methods differ, our molecular findings corroborate those imaging observations. Compared to traditional methods, methylation signal detection may be more sensitive in revealing potential micro-dissemination tendencies.

RASSF1A methylation was significantly positively correlated with serum tumor markers such as NSE and CYFRA21-1, while SHOX2 methylation was positively correlated with CYFRA21-1. Methylation signals in pleural fluid reflect locally shed tumor cells or cell-free DNA and may correspond to a subset of tumor cells with high invasive potential or specific differentiation features, such as neuroendocrine-differentiated cells [25, 26]. We speculate that RASSF1A methylation positivity may lead to a more aggressive phenotype by reducing tumor suppressive activity and altering pathways related to adhesion and migration, which may partly explain its association with adverse imaging features and recurrence risk [27–30]. Such cells may acquire enhanced motility and anti-apoptotic capabilities at the epigenetic level, gaining the potential to breach the basement membrane, disseminate via the pleura, or metastasize through circulation, providing early molecular signals for postoperative recurrence. From a translational perspective, the coordinated changes in methylation signals and serological markers offer new insights for constructing multi-modal risk prediction models.

Methylation of RASSF1A and SHOX2, individually and in combination, was associated with shorter PFS and OS. The joint assay outperformed single-gene testing, indicating complementary value in reflecting residual disease or intraoperative microdissemination. Consistently, Tomizawa et al. [31]. found that high RASSF1A methylation in tumor tissue from 110 stage I lung adenocarcinoma patients was significantly linked to poor survival and strongly correlated with poor differentiation, vascular invasion and pleural invasion, suggesting early involvement in tumor progression. Similarly, Klacz et al. [32]. demonstrated that RASSF1A promoter hypermethylation and downregulation of its mRNA/protein were significantly associated with tumor progression, higher grade and adverse survival outcomes in clear-cell renal cell carcinoma, reinforcing the cross-cancer prognostic utility of RASSF1A methylation. Deng et al. [33]. showed that RASSF1A and SHOX2 methylation in bronchoalveolar lavage fluid were significantly related to shorter PFS in NSCLC patients; unmethylated RASSF1A predicted a better response to pemetrexed-based therapy, indicating prognostic value and potential as a chemotherapy-sensitivity marker. Notably, the prognostic significance of RASSF1A methylation is not uniform across tumor types. A recent meta-analysis of 27 studies in ovarian cancer found no significant association between RASSF1A, BRCA1 or APC promoter methylation and OS or PFS [34], highlighting tissue specificity of RASSF1A methylation biology. Using IPL fluid as the detection matrix, the present assay more directly captures tumor cells and cell-free DNA shed from the pleural surface and indicates signals related to potential micro-dissemination within the pleural cavity, thus enhancing the sensitivity and clinical utility of prognostic evaluation.

From a clinical translation perspective, methylation signals in IPL fluid are more suitable as supplementary molecular markers for postoperative risk stratification. Its core advantage lies in the fact that the second pleural lavage fluid used in this study was collected after tumor resection and systematic lymph node dissection, and its clinical significance is more inclined to reflect residual tumor-derived molecular signals in the local pleural cavity environment. Furthermore, the local enrichment effect of pleural lavage fluid enhances the sensitivity for detecting methylation signals, facilitating the capture of minimal tumor dissemination [35]. Therefore, the more practical application scenario of this detection is to identify patients with a higher risk of recurrence after surgery, providing a reference for closer follow-up monitoring, discussion of adjuvant therapy, and multidisciplinary evaluation.

However, this study has certain limitations. First, the sample size is relatively limited and derived from a single center, which may introduce selection bias and necessitate further validation of the generalizability of the results. Second, the concentration of tumor-derived DNA in IPL fluid may be influenced by factors such as surgical technique, lavage volume, and local blood contamination, potentially affecting the sensitivity of methylation detection. Third, this study did not include a concurrent head-to-head comparison between traditional cytological examination of IPL fluid and methylation detection; therefore, a direct evaluation of their differential sensitivity and specificity for detecting pleural cavity dissemination could not be performed. Four, given the small sample size, limited events, and insufficient consideration of time to event information in ROC analysis, the above discriminant results should mainly be interpreted as exploratory findings. Finally, although no adjuvant chemotherapy after surgery was not predetermined in the study protocol, the fact that stage III patients did not receive treatment due to real-world factors such as patient refusal, comorbidities, or compliance may still enrich the population with a higher risk of recurrence in this cohort, leading to an overestimation of the observed expected after effects. Therefore, the conclusion of this study is mainly applicable to the exploratory population without postoperative adjuvant therapy, and its applicability in patients receiving standard adjuvant therapy still needs further verification. Future multicenter, large-sample prospective studies are needed to standardize the collection and detection protocols for IPL fluid and to conduct head-to-head comparisons between traditional cytology and methylation detection, in order to further validate their clinical value in identifying pleural dissemination and assessing prognosis.

Conclusion

In summary, the methylation detection of RASSF1A and SHOX2 in IPL fluid is associated with the invasive characteristics and potential micro-dissemination status of NSCLC. In an exploratory cohort without postoperative adjuvant therapy, combined detection is associated with a higher risk of postoperative recurrence, suggesting that it can serve as a potential complementary molecular marker for postoperative risk stratification and adjuvant therapy decision-making discussions. However, the above findings still need to be further validated in larger samples, multiple centers, and populations receiving standard treatment.

Authors’ contributions

Conceptualization: Jianbin Hou, Junmeng Xiao. Data curation: Bo Yang. Formal Analysis: Jianbin Hou, Junmeng Xiao, Ran Yang. Investigation: Weijie Wang. Methodology: Jianbin Hou, Junmeng Xiao. Project administration: Ziliang Shi. Resources: Xiaoyu Niu. Software: Yang Zhao. Supervision: Xu Guo. Validation: Jianwei Cao. Visualization: Bing Xu. Writing – original draft: Jianbin Hou, Junmeng Xiao, Bo Yang, Ran Yang, Weijie Wang. Writing – review & editing: Ziliang Shi, Xiaoyu Niu, Yang Zhao, Xu Guo, Jianwei Cao, Bing Xu.

Funding

This work was supported by Major Science and Technology Project of Anyang City (2023A02SF001).

Data availability

The data and materials in the current study are available from the corresponding author.

Declarations

Ethics approval and consent to participate

The study protocol was conducted in accordance with the Declaration of Helsinki and approved by Ethics Committee of Anyang Tumor Hospital(2025WZ27K01). Informed consent was waived by the committee because of the retrospective nature of the study.

Competing interests

The authors declare no competing interests.