Consistency analysis of programmed cell death ligand-1/thyroid transcription factor-1 dual staining in malignant pleural effusion and lung adenocarcinoma histology specimens

Abstract

Background

Lung cancer remains the leading cause of cancer-related deaths worldwide, with non-small cell lung cancer (NSCLC) constituting the majority of cases. Immune checkpoint inhibitors have significantly improved outcomes in advanced NSCLC, particularly for patients with programmed cell death ligand-1 (PD-L1) expression. However, obtaining tissue biopsies for PD-L1 assessment in advanced disease is often challenging. Malignant pleural effusion (MPE), frequently associated with advanced lung adenocarcinoma, offers an accessible alternative source for PD-L1 testing. Nevertheless, the reliability of PD-L1 evaluation in cytological specimens from MPE, especially regarding staining methodologies, requires further validation. This study aimed to assess the impact of PD-L1 single staining versus PD-L1/thyroid transcription factor-1 (TTF-1) dual staining on the concordance of PD-L1 tumor proportion scores between paired histologic and cytologic MPE samples.

Methods

This retrospective study involved 112 lung adenocarcinoma patients, using paired histological and MPE cytological samples for PD-L1 evaluation via immunohistochemical staining. Tumor proportion score measured PD-L1 levels, withTTF-1 marking lung adenocarcinoma cells. The study compared PD-L1 staining across samples and techniques.

Results

This study aimed to evaluate the impact of staining methods on PD-L1 scoring in histological and MPE cytological specimens. The results indicate that the selection of PD-L1 single-stained or PD-L1/TTF-1 dual-stained methodologies within the same sample type did not lead to significant differences in the distribution of PD-L1 tumor proportion scores. However, when evaluating concordance with paired histological specimens, PD-L1/TTF-1 dual staining in MPE cytology samples achieved a significantly higher concordance rate (82.14%) compared to PD-L1 single staining (69.64%), with kappa values increasing from 0.59 to 0.75. Notably, in MPE cytology specimens with low tumor density, PD-L1/TTF-1 dual staining demonstrated a marked advantage over PD-L1 single staining, with the concordance rate of tumor proportion score with histology specimens increasing from 56.86% to 84.31%, with kappa values increasing from 0.47 to 0.78.

Conclusions

The application of PD-L1/TTF-1 dual staining in MPE cytological specimens significantly enhances the consistency of PD-L1 expression within tissue specimens, particularly in advanced lung adenocarcinoma patients. This method provides a reliable, evidence-based medical reference that complements histopathological analysis.

Highlight box.

Key findings

• Programmed cell death ligand-1 (PD-L1)/thyroid transcription factor-1 (TTF-1) dual staining in malignant pleural effusion (MPE) cytology achieved 82.1% concordance with histology (vs. 69.6% in PD-L1 single staining).

• The study demonstrated a marked 27.5-percentage-point improvement in concordance with histology specimens (from 56.9% to 84.3%) in low tumor-density MPE cytology using dual staining.

• No significant differences in PD-L1 tumor proportion score distribution were observed between single and dual staining within the same sample type.

What is known and what is new?

• PD-L1 testing guides non-small cell lung cancer immunotherapy.

• MPE is a PD-L1 test alternative but faces tumor-density challenges.

• Dual staining significantly improves PD-L1 concordance in MPE-histology comparisons.

• This study provides evidence for dual staining on overcoming PD-L1 limitations in low-cellularity MPE.

What is the implication, and what should change now?

• This study validates MPE as reliable PD-L1 test source for advanced lung adenocarcinoma.

• Dual staining reduces false-negative risk in cytology specimens.

• Update PD-L1 testing guidelines to recommend TTF-1 co-staining for MPE.

• Implement dual staining in cytopathology laboratories.

Introduction

Lung cancer is the leading cause of cancer-related mortality globally. Non-small cell lung cancer (NSCLC), which predominantly includes adenocarcinoma and squamous cell carcinoma, constitutes 85% of all lung cancers (1). Although the increase in public awareness and the implementation of lung cancer screening programs have raised the detection rate of early-stage NSCLC, a substantial majority of NSCLC patients present with locally advanced or metastatic disease at the time of diagnosis (2,3). In such cases, obtaining invasive surgical resections or tissue biopsies can be challenging, while malignant pleural effusion (MPE) is frequently observed in NSCLC patients, particularly those with lung adenocarcinoma (LUAC), often serving as the only available diagnostic specimen in patients with advanced NSCLC. Hence, there has been a marked interest in recent years in developing less invasive detection methods for MPE (4,5).

Traditionally, treatment options have been limited for patients with locally advanced or metastatic NSCLC lacking a targetable genetic alteration in the tumor. However, the advent of immune checkpoint inhibitors has revolutionized the standard of care for advanced NSCLC (6-8). Programmed cell death ligand-1 (PD-L1) is an immune checkpoint protein expressed on tumor cells and tumor-infiltrating immune cells. PD-L1 binds to the PD-1 receptor allowing tumor cells to evade surveillance by T cells. Given that 24–60% of NSCLC cases express PD-L1, immune checkpoint inhibitors that block the PD-1/PD-L1 interaction, such as pembrolizumab and nivolumab, have shown active and enduring clinical efficacy in the treatment of advanced NSCLC (9,10). However, these immune checkpoint inhibitors may potentially induce a range of immune-related adverse events, including pneumonitis and endocrinopathy, and their high cost of treatment makes it essential to accurately identify patients who are likely to respond to PD-1/PD-L1 inhibitor therapy (11-13). In 2017, the National Comprehensive Cancer Network (NCCN) guidelines recommended assessing PD-L1 expression in tumor cells of advanced NSCLC patients (14). The utility of immunohistochemical (IHC) staining for detecting PD-L1 expression in formalin-fixed, paraffin-embedded (FFPE) specimens has been clinically validated (14). Although surgical resection or biopsy specimens are preferred for PD-L1 assessment, research indicates that cytopathological specimens from MPE are easily sampled and show reliable correlation with surgical specimens in detecting PD-L1 expression. Sufficient tumor cells in MPE cell blocks are vital for accurate PD-L1 testing, but the sparse distribution of tumor cells among non-tumor cells, such as mesothelial and immune cells, complicates the assessment (15-17).

Accurate identification of PD-L1 in MPE-derived cell blocks is crucial. In NSCLC, thyroid transcription factor-1 (TTF-1) is a frequently utilized biomarker with high sensitivity and specificity for LUAC, aiding in tumor differentiation and metastasis regulation (18-20). The PD-L1/TTF-1 double staining method offers significant advantages over single staining by enhancing differentiation between tumor and non-tumor cells, particularly in cytology samples, and reducing interobserver variability, thus improving consistency and accuracy. This method minimizes false positives and negatives, providing a more reliable PD-L1 assessment. Despite these benefits, the clinical significance of dual PD-L1/TTF-1 staining in MPE cytopathology is not well understood. This study aims to evaluate the clinical implications of PD-L1/TTF-1 dual staining through a retrospective analysis of paired histological and MPE cytopathological specimens from LUAC patients. We present this article in accordance with the STARD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1226/rc).

Methods

Patient selection

This retrospective cohort study analyzed pathological data from 1,700 patients with LUAC in Beijing Chao-Yang Hospital, Capital Medical University, between January 2021 and November 2023. Enrolled patients had histological samples confirming the diagnosis of LUAC and matched cytopathological samples from MPE. Inclusion criteria comprised of specimens obtained within the past 3 years, no history of prior radiotherapy, chemotherapy, targeted therapy, or immunotherapy, each patient providing tissue specimens along with three or more corresponding MPE cytology specimens. Specifically, six patients each provided four MPE cytology specimens, and the remaining 123 patients each provided three MPE cytology specimens. Both specimen types required TTF-1 positive tumor cells with a minimum of 100 viable tumor cells (VTC). A total of 112 consecutively eligible cases meeting all criteria were included, with the patient selection flowchart depicted in Figure 1. This cohort represents all available qualifying patients within the study period. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study received approval from the Ethics Committee of Beijing Chao-Yang Hospital (No. 2023/603), with waived patient informed consent.

Figure 1.

The flowchart of patient selection. H&E, hematoxylin and eosin; IHC, immunohistochemical; LUAC, lung adenocarcinoma; MPE, malignant pleural effusion; PD-L1, programmed death-ligand 1; TTF-1, thyroid transcription factor-1; VTC, viable tumor cells.

Histology and MPE cell blocks processing

Histopathology specimens were immediately immersed in 10% neutral buffered formalin (NBF) upon resection and left for a duration of 6 to 72 hours before being processed into tissue blocks. Pleural effusion specimens, collected in different batches during the same hospitalization period, each ranging from 200 to 800 mL, were allowed to stand at room temperature for 30 minutes. Following centrifugation at 2,000 rpm for 5 minutes using a BY-320C centrifuge (Beijing Baiyang Medical Instruments Co., Ltd, Beijing, China), at least 100 mL of the sediment layer was collected. The cell sediment was fixed in 10% neutral buffered formalin (NBF) for 2 hours. Subsequently, the pelleted cell clumps underwent centrifugation at 2,500 rpm for 5 minutes, after which they were encapsulated in Bio-Wrap and dehydrated alongside the histopathology specimens in a fully automated tissue dehydration system (VIP6 AI-J2, Sakura Co., Ltd., Japan). The dehydrated samples were then embedded in paraffin (Leica Biosystems, Buffalo Grove, IL, USA), sectioned into 4-micrometer-thick slices, and stained with hematoxylin and eosin (H&E) using the H&E600 kit (Beijing Barry Medical Devices Co., Ltd., Beijing, China) and the H&E600 Fully Automated Stainer (Roche, Co., Ltd., USA) for morphological observation.

IHC staining and scoring of PD-L1 expression

PD-L1 (clone SP263, rabbit monoclonal antibody, working fluid; Ventana Medical Systems) and TTF-1 (clone LBP1-TTF1, mouse monoclonal antibody, working fluid; Guangzhou LBP Medical Technology Co., Ltd., Guangzhou, China) were utilized to assess PD-L1 expression levels and TTF-1 positivity in both histological and cytopathological specimens. IHC staining was performed using the Ventana Benchmark Ultra automatic staining system from Roche Diagnostics, Rotkreuz, Switzerland, following the manufacturer’s protocols, with human placental tissue samples as positive controls. The secondary antibodies included single/dual staining reagent kits, in conjunction with the OptiView DAB chromogenic system for signal enhancement. The tumor proportion score (TPS) was determined by pathologists trained and certified in assessing PD-L1 expression. Positive staining was defined as tumor cell membrane staining intensity exceeding that of the background, excluding cytoplasmic PD-L1 staining in tumor cells and staining of tumor-associated immune cells (e.g., lymphocytes, monocytes, macrophages, plasma cells, and neutrophils), as well as adjacent stromal cells, necrotic cells, and carbon pigment particles. This exclusion criterion specifically addresses the risk of false positives from non-specific PD-L1 expression in non-tumor cells—a challenge proactively resolved by PD-L1/TTF-1 dual staining through TTF-1-guided tumor cell identification. PD-L1 expression levels were categorized into cutoffs based on TPS: <1% (negative), 1–49% (low expression), and ≥50% (high expression). TTF-1 staining was used to confirm the presence of LUAC cells. Both PD-L1 single-staining and PD-L1/TTF-1 double-staining were independently reviewed by two experienced pathologists in a blinded manner, and discrepancies were resolved through consensus.

Statistical analysis

The consistency between histological and paired cytological specimens was evaluated using weighted Cohen’s kappa (κ) statistic, where a value of κ<0.4 indicates poor agreement, 0.4 to 0.74 indicates fair to moderate agreement, and ≥0.75 indicates good agreement. P values less than 0.05 were considered statistically significant. Statistical comparisons between staining methods (single vs. dual) within the same sample type were conducted using Wilcoxon signed-rank tests for continuous TPS values. The correlation between PD-L1/TTF-1 expression in pleural effusion cytology and patient age was assessed using the Mann-Whitney U test. Associations with patient gender, smoking status, specimen type, origin, and genetic mutation status were analyzed using Pearson’s chi-squared (χ²) test or Fisher’s exact test as appropriate. Statistical significance was considered at a two-tailed P value of <0.05. Statistical analyses were conducted using SPSS version 25.0 (IBM Corporation, Armonk, NY, USA).

Results

Clinicopathological features of patients and specimens

Our study included 112 paired histological and cytological specimens from patients, aged 34 to 84 years, with a median of 67 years. The histological samples consisted of 31 surgical resection specimens (resected lung tissue or lymph nodes) and 81 small biopsy specimens obtained through tracheoscopy or puncture. These samples were derived from primary lung lesions (61 cases) and metastatic lesions (66 cases), which sourced from lymph nodes (23 cases) and pleura (43 cases), as detailed in Table 1.

Table 1. Clinical and pathologic details of patients and specimens (n=112).

Variable	Data
Age (years)
Median	67
Range	34–84
Sex
Male	58 (51.8)
Female	54 (48.2)
Smoking history
Yes	50 (44.6)
No	62 (55.4)
Sample type
Resection	31 (27.7)
Small biopsy	81 (72.3)
Sample origin
Lung	46 (41.1)
Lymph node	23 (20.5)
Pleura	43 (38.4)
Tumor type
Primary	46 (41.1)
Metastasis	66 (58.9)
Genetics
EGFR mutations	59 (52.7)
No mutations	53 (47.3)
Tumor fraction
<0.5	44 (39.3)
≥0.5	68 (60.7)

Data are presented as n (%) unless otherwise stated. Tumor fraction refers to the ratio of tumor cells to non-tumor cells in MPE cytopathology slides, where non-tumor cells include histiocytes, mesothelial cells, and inflammatory cells, etc. EGFR, epidermal growth factor receptor; MPE, malignant pleural effusion.

Comparative analysis of PD-L1 expression in MPE cytopathology and corresponding histopathology specimens

To evaluate the consistency between PD-L1 single-stained and PD-L1/TTF-1 dual-stained samples, we performed distinct analyses on both histology and MPE cytology samples. Our findings indicate that the selection of PD-L1 single-stained or PD-L1/TTF-1 dual-stained methodologies did not influence the PD-L1 TPS in either sample (with P values of 0.12 and 0.89, respectively) (Table S1). Table S2 provides a detailed comparison of the variances in TPS between PD-L1 single-stained and PD-L1/TTF-1 dual-stained PD-L1 in histology and MPE cytology samples. Histopathology samples, which are PD-L1 single-stained, are the standard for evaluating PD-L1 expression. Previous studies have shown a strong correlation between PD-L1 expression in these histopathological samples and corresponding cytological specimens from MPE (15-17). To evaluate the effectiveness of PD-L1/TTF-1 dual-staining in aligning cytological with histopathological specimens, a comprehensive analysis was conducted. Table 2 illustrates that in cytopathological PD-L1 single-stained samples, 69.64% (78/112) showed concordant results with histopathological specimens, indicating moderate agreement (κ=0.59, P<0.001). Notably, discrepancies were observed in 25 false positive cases and 9 instances of inconsistent PD-L1 grading (3 negative, 3 low expression, and 3 high expression). For cytopathological PD-L1/TTF-1 dual-staining, 82.14% (92/112) of specimens aligned with the PD-L1 single-stained histopathological specimens, showing strong agreement (κ=0.75, P<0.001). However, there were 12 false positive cases and 8 cases with discordant PD-L1 grading (4 negative, 2 low expression, 2 high expression). These results suggest that PD-L1/TTF-1 dual-staining in cytopathological specimens significantly improves alignment with PD-L1 TPS in histopathological specimens, indicating a more reliable method for assessing PD-L1 expression.

Table 2. Concordance of PD-L1 expression between MPE cytopathology specimens and matched histopathology specimens across three categories (n=112).

Histopathology single-stained	Cytopathology single-stained						Cytopathology dual-stained
	<1%	1–49%	≥50%	κ	P		<1%	1–49%	≥50%	κ	P
<1%	42	24	1	0.59	<0.001		55	11	1	0.75	<0.001
1–49%	3	23	3				4	23	2
≥50%	0	3	13				0	2	14

MPE, malignant pleural effusion; PD-L1, programmed death-ligand 1.

In a parallel evaluation of different PD-L1 staining methods in MPE cytology and corresponding histopathology specimens, we categorized PD-L1 expression as negative (<1%) and positive (1% to 49% and >50%). Seventy-five percent (84/112) of single-stained MPE cytopathology samples and 85.71% (96/112) of dual-stained MPE cytopathology samples were consistent with corresponding histopathological specimens, with the κ value improving from 0.52 to 0.71 (Table 3).

Table 3. Concordance of PD-L1 expression between MPE cytopathology specimens and matched histopathology specimens with two categories (n=112).

Histopathology single-stained	Cytopathology single-stained					Cytopathology dual-stained
	Negative	Positive	κ	P		Negative	Positive	κ	P
Negative	42	25	0.5	<0.001		55	12	0.7	<0.001
Positive	3	42				4	41

MPE, malignant pleural effusion; PD-L1, programmed death-ligand 1.

To address the potential impact of non-specific PD-L1 staining on sample interpretation, we implemented a classification strategy to identify non-tumor cells, such as histiocytes, mesothelial cells, and inflammatory cells, which serve as stromal components. This distinction was aimed at separating non-tumor cells (including histiocytes, mesothelial cells, and inflammatory cells, etc.) from tumor cells, thereby enhancing the precision of PD-L1 expression analysis. The density of tumor cells was quantified using the tumor fraction, defined as the ratio of tumor cells to total non-tumor cells. Adjustments were made based on relevant literature to enhance the methodological robustness (21). Specifically, we defined tumor fraction <0.5 and ≥0.5 as indicators of low and high tumor cell densities, respectively. In our comparative analysis of PD-L1 detection in MPE cytology samples using single-stained versus dual-stained methods, we observed significant outcomes. Under low tumor cell density conditions, the concordance rate between PD-L1 single-stained MPE cytology and histological PD-L1 TPS was 56.86% (29/51), with a moderate agreement (κ=0.47, P<0.001) (Figure 2A and Table S3). Conversely, the concordance rate for PD-L1/TTF-1 dual-stained was 84.31% (43/51) and the agreement was better (κ=0.78, P<0.001) in the same scenario (see Figure 2B and Table S4). In contrast, in high tumor cell density cases, the agreement between PD-L1 single-stained and PD-L1/TTF-1 dual-stained MPE cytology with PD-L1 TPS in histology was consistently 80%, with both methods showing a κ value of 0.72 (P<0.001) (Figure 2C,2D). These results indicated that MPE cytology PD-L1/TTF-1 dual staining was more sensitive in detecting PD-L1 expression in tumor cells under conditions of low tumor cell density, where non-specific PD-L1 staining may be prevalent. Specific consistency analyses were detailed in Tables S3,S4.

Figure 2.

Confusion matrix analysis of histopathology and MPE cytology PD-L1 single-stained and PD-L1/TTF-1 dual-stained concordance on PD-L1 TPS under tumor fraction <0.5 (A,B) and tumor fraction ≥0.5 (C,D). MPE, malignant pleural effusion; PD-L1, programmed death-ligand 1; TPS, tumor proportion score; TTF-1, thyroid transcription factor-1.

Subsequent analyses delved deeper into the agreement between PD-L1/TTF-1 dual-stained MPE cytopathology specimens and histopathology specimens sourced from various sample types, as illustrated in Figure 3. Within MPE cytopathology, the concordance between PD-L1/TTF-1 dual-stained MPE specimens and surgical resection samples stood at 90.32% (28/31), indicating a substantial level of agreement (κ=0.87, P<0.001) in contrast to small biopsy or puncture specimens, which was only 79.01% (64/81), resulting in a κ value of 0.70 (P<0.001). This discrepancy may stem from the heightened spatial heterogeneity present in small biopsy or puncture specimens compared to surgical resection samples. Moreover, 73.91% (34/46) of MPE cytopathology specimens demonstrated concordance in PD-L1/TTF-1 dual-stained results with PD-L1 TPS of primary lung lesions, indicating moderate agreement (κ=0.6, P<0.001). In contrast, 87.9% (58/66) of MPE cytopathology specimens displayed concordance with PD-L1 TPS of metastatic lesions, showcasing stronger agreement in κ test results compared to primary lung lesions (κ=0.8, P<0.001). Detailed consistency analyses are provided in Table S5.

Figure 3.

Confusion matrix assessing concordance between cytopathological and histopathological classifications of PD-L1/TTF-1 dual-stained specimens, stratified by sample type (A: resection; B: biopsy) and tumor origin (C: primary; D: metastatic). PD-L1, programmed death-ligand 1; TTF-1, thyroid transcription factor-1.

Additionally, a comprehensive analysis was conducted on the correlation between PD-L1/TTF-1 dual-stained specimens in MPE cytological pathology and patients’ clinical characteristics. This relationship is outlined in Table S6, revealing that PD-L1 expression in MPE cytological pathology specimens did not exhibit significant correlations with patients’ gender, age, smoking history, sample type, and source. Notably, a significant association was identified with the presence of EGFR gene mutations (P<0.001, Table S6). These findings emphasize the need for further exploration into the interplay between PD-L1 expression and EGFR mutations in MPE cytological pathology specimens.

Comparability of PD-L1 single-stained with PD-L1/TTF-1 in matched cases

The PD-L1 TPS in cytopathological specimens from MPE showed good concordance with histopathology. PD-L1/TTF-1 dual staining in cytopathology facilitates easier analysis of PD-L1 IHC quantification (Figure 4). It was worth noting that the presence of non-tumor cell components (stroma) surrounding tumor cells in cytopathology specimens of MPE could result in some samples showing higher cytopathological PD-L1 single-stained interpretation in some cases due to non-specific PD-L1 staining of stromal cells (Figure 5A,5B), impacting 16 of 112 cases (14.3%) with higher PD-L1 TPS (Table S2). When there is no non-specific PD-L1 staining in non-tumor components (stroma), neither single nor dual staining leads to bias in PD-L1 TPS scoring (Figure 5C,5D). However, even for samples with tumor fraction ≥0.5, non-specific stromal PD-L1 IHC staining would occur in both single-staining and double-staining images, affecting the PD-L1 TPS score (Figure 5E,5F). Additionally, 2 of 112 cases (1.8%) of pleural effusion cytopathology samples exhibited lower PD-L1 TPS in PD-L1 single-stained interpretations compared to PD-L1/TTF-1 double-stained interpretations (Table S2), attributed to overlooked shrunken, smaller tumor cells during PD-L1 single staining (Figure 5G,5H).

Figure 4.

Consistency of PD-L1 TPS in MPE cytopathology samples and matched histopathology samples from LUAC patients with TPS > 90% (magnification ×200). Representative images illustrate H&E staining of various specimens, including a lung bronchial mucosal tumor tissue (A), pleural metastatic tumor tissue (F), lymph node metastatic tumor tissue (K), and PD-L1 IHC single staining of corresponding histopathology samples (B,G,L) and the H&E staining of their matched MPE cytopathology samples (C,H,M). PD-L1 IHC single-stained cytopathology specimens (D,I,N) and PD-L1/TTF-1 IHC dual-stained cytopathology specimens (E,J,O). H&E, hematoxylin and eosin; IHC, immunohistochemical; LUAC, lung adenocarcinoma; MPE, malignant pleural effusion; PD-L1, programmed death-ligand 1; TTF-1, thyroid transcription factor-1.

Figure 5.

Comparative analysis of PD-L1 IHC single vs. dual staining in MPE cytopathology (magnification ×200). For tumor fraction <0.5, PD-L1 IHC single-stained images (A) demonstrate non-specific stromal expression while dual-stained samples from the same case show consistent stromal PD-L1 IHC staining (B). Samples without stromal PD-L1 IHC staining are shown in single-stained (C) and dual-stained (D) formats. For tumor fraction ≥0.5, non-specific stromal PD-L1 IHC staining appears in single-stained (E) and dual-stained (F) images. Critically, shrunken tumor cells with PD-L1 IHC expression—which may be misclassified as non-specific staining in single-stained preparations (G)—are confirmed as tumor-derived via TTF-1 nuclear labeling in dual-stained images (H), reducing under-scoring risk. IHC, immunohistochemical; MPE, malignant pleural effusion; PD-L1, programmed death-ligand 1; TTF-1, thyroid transcription factor-1.

Discussion

Studies have indicated that cytopathology specimens, including fine needle aspiration cytology samples, reliably reflect the overall PD-L1 expression status in tumor cells of advanced-stage LUAC patients (22,23). Multiple investigations have shown a strong correlation and consistency between PD-L1 expression in cytopathology and histopathology, establishing them as dependable resources for evaluating immune therapy response and effectiveness (14-16). However, current cytopathology studies predominantly involve fine needle aspiration cytology specimens, such as transbronchial needle aspiration or percutaneous computed tomography-guided biopsies (15,24,25), with limited focus on using MPE cell block samples exclusively. This study highlights that cytopathology derived from MPE demonstrates a notably high level of agreement with tissue samples when assessing PD-L1 expression in advanced LUAC patients. Moreover, the utilization of PD-L1/TTF-1 dual staining can further improve the precision of cytopathology in evaluating PD-L1 expression, especially in cases where cytopathology samples contain a larger proportion of non-tumor cells or exhibit morphological alterations in tumor cells.

TTF-1 was selected as a marker for LUAC tumor cells, given that only approximately 10% of AC patients in the Asian population exhibit complete TTF-1 negativity (26). Consistent with this, our data revealed that a mere 7.8% (10/129) of LUAC patients within our study cohort exhibited no TTF-1 expression in their tumor cells. While TTF-1-negative cases were excluded from this study to ensure data reliability, such specimens predominantly represent moderately/poorly differentiated LUAC subtypes with distinct cytomorphological atypia, enabling identification through careful morphological assessment. Additionally, established diagnostic guidelines recommend employing alternative markers such as Napsin A—either as a secondary marker or in combination with TTF-1—to maintain adenocarcinoma identification accuracy in clinical practice (27). To ensure PD-L1 assessment accuracy and mitigate biases from pleural fluid processing, our study exclusively enrolled cases with at least three MPE cytology submissions. Additionally, samples with VTC ≤100 were excluded to eliminate confounding effects from low tumor cellularity. The cohort comprised 112 cases with multiple MPE submissions, stratified as follows: among 6 patients undergoing four submissions, 33.3% (2/6) had VTC >100 in two submissions and 66.7% (4/6) in three submissions; concurrently, among 106 patients with three submissions, 39.6% (42/106) achieved VTC >100 in two submissions while 60.4% (64/106) maintained this threshold across all three submissions (Table S7). This finding highlights the heterogeneity among MPE samples, indicating that selecting cases with three or more MPE submissions is necessary to ensure sufficient VTC in cytology specimens for reliable PD-L1 detection. This approach enhanced the robustness and reliability of PD-L1 expression data in MPE within this study.

Given the high concordance in PD-L1 expression assessment between the 22C3 and SP263 clones, we classified PD-L1 expression levels into three cutoffs (<1%, 1–49%, ≥50%) based on the reference 22C3 study protocol (28). Using IHC on histopathology as the control group, we observed that PD-L1/TTF-1 dual staining improved the concordance in PD-L1 TPS compared to single staining, from 69.64% to 82.14%. Dual-stained cytology samples showed a higher concordance rate and an increased κ value from 0.59 to 0.75 (Table 2). Additionally, using a 1% threshold for PD-L1 TPS binary classification, dual staining further enhanced the agreement between MPE cytopathology and histopathology (Table 3). Our results are consistent with previous studies focused on MPE cytology. For instance, Grosu et al. compared 82 MPE cytology cases with histological specimens and found a concordance rate of 52.4% with a κ value of 0.8 using the 22C3 PD-L1 antibody (14). Similarly, Sun et al. reported a 71.1% concordance rate with a κ value of 0.5 in their study of 90 MPE cell block pairs from NSCLC patients with matched primary lung cancer tissue using SP263 (29). Zou et al. reported an 85.1% concordance rate with a κ value of 0.8 using the SP263 antibody in 124 MPE cell blocks and paired histology samples. However, they also observed that PD-L1/TTF-1 dual staining reduced sensitivity in detecting PD-L1 expression compared to single staining (17). In this study, we used an enhanced DAB detection kit and adjusted the concentration of the TTF-1 reagent, among other measures, to prevent a decrease in the sensitivity of PD-L1/TTF-1 dual staining.

The high heterogeneity of pleural effusions, characterized by a large proportion of non-tumor cells such as mesothelial cells, immune cells, and histiocytes, etc., which frequently exhibit non-specific PD-L1 expression, presents a significant challenge for PD-L1 detection in advanced NSCLC (15). This challenge stems primarily from non-specific PD-L1 expression in non-tumor cells (e.g., mesothelial cells)—precisely the false-positive risk targeted by the IHC exclusion criterion for non-tumor cells (section “IHC staining and scoring of PD-L1 expression”). Crucially, PD-L1/TTF-1 dual staining resolves this limitation through TTF-1-guided tumor cell identification, enabling accurate differentiation of LUAC cells from PD-L1+ stromal backgrounds. Due to the critical difficulty in distinguishing LUAC tumor cells from morphologically similar mesothelial cells, compounded by limited tumor cells and non-tumor cells PD-L1 expression, the accuracy of PD-L1 scoring is compromised (17). The integration of TTF-1—a highly specific nuclear marker for LUAC—in dual staining directly overcomes this challenge by enabling unambiguous identification of tumor cells amidst mesothelial backgrounds (20). To address this issue, we classified tumor cell density by tumor fraction and found that in low tumor density samples (tumor fraction <0.5), PD-L1/TTF-1 dual staining significantly improved concordance and κ value versus single staining (84.31% vs. 56.86%; κ=0.78 vs. 0.47). For high tumor density (tumor fraction ≥0.5), both methods showed comparable performance (80% concordance; κ=0.72) (Figure 2). These results underscore that the dual-staining approach—by leveraging TTF-1 to unambiguously identify tumor cells—is fundamental to accurate PD-L1 assessment in cytologically complex effusions.

In advanced-stage LUAC patients, PD-L1 testing is typically performed on small biopsy specimens (30). However, due to tumor heterogeneity and spatial heterogeneity, small biopsy specimens may not fully reflect the overall status of the tumor. Our study further analyzed the concordance of PD-L1 TPS between MPE cytopathology and different types and sources histopathology specimen. The results indicate that the agreement of PD-L1/TTF-1 dual-stained in MPE cytopathology with surgical resection specimens is higher than that with small biopsy or puncture specimens, with a consistency (90.3% vs. 79.0%) and κ value (0.9 vs. 0.70) decreased. And MPE cytopathology PD-L1/TTF-1 dual-stained specimens showed a higher concordance (87.9% vs. 73.9%) and κ value (0.8 vs. 0.6) with metastatic lesions compared to primary lung lesions. This difference may be due to the fact that MPE cytopathology specimens are also representative of metastatic tissue. Additionally, we analyzed the correlation between PD-L1 TPS from MPE cytology PD-L1/TTF-1 dual-stained samples and the clinical pathological features of patients. The results indicated that PD-L1 TPS from dual staining was correlated with the presence of EGFR gene mutations. This suggests that EGFR gene mutation status may be an important factor influencing PD-L1 expression levels in pleural effusion cytology. These findings are consistent with previous research showing that in advanced NSCLC, EGFR mutations can affect PD-L1 expression. This correlation underscores the complex interplay between genetic mutations and immune checkpoint expression, with potential implications for personalized therapeutic strategies and prognosis in NSCLC patients (31-33).

Limitations of this study include its retrospective design, variations in sample handling methods, and the use of different PD-L1 and TTF-1 antibodies and staining systems between institutions, which may affect the interpretation of PD-L1 expression during dual staining and require validation by more research centers and laboratories. Also, LUAC patients lacking TTF-1 expression may not be amenable to this double staining approach. In our study, there were 9 cases (9/112, 8.0%) where the PD-L1 TPS in histological samples was less than 1%, but dual staining in MPE cytology showed a TPS of greater than 1%. The lack of follow-up data on the clinical immunotherapy of these patients limits our ability to further elucidate the clinical utility of PD-L1 testing in MPE cell blocks. Additionally, the absence of relevant clinical management information, such as immunotherapy response rates and prognosis, necessitates further clarification of the clinical utility of PD-L1 detection in MPE cell blocks.

Conclusions

In summary, our study demonstrates the reliability of PD-L1 expression assessment in MPE cell blocks, and the use of dual PD-L1/TTF-1 staining can improve the consistency of PD-L1 TPS between cytopathology and histopathology samples, especially in the presence of a significant number of non-tumor cells in pleural effusion. In the future, artificial intelligence can be used to further automate the evaluation of the degree of PD-L1 expression by identifying PD-L1-positive tumor cells in scanned section images, reducing time and labor costs while improving the accuracy of the evaluation.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study received approval from the Ethics Committee of Beijing Chao-Yang Hospital (No. 2023/603), with waived patient informed consent.

Data Sharing Statement

Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1226/dss

jtd-18-02-100-dss.pdf ^{(166.8KB, pdf)}

DOI: 10.21037/jtd-2025-1226