CEACAM5 as a biomarker of semi-solid, lepidic lung adenocarcinoma

Introduction

Semi-solid, sub-solid, and ground-glass opacities (GGO) seen on chest computed tomography (CT) scan often represent a slow growing form of lung adenocarcinoma with lepidic histology (1-3). Many subsolid GGO have indolent growth and can be safely observed, however some lesions have malignant potential and ultimately require treatment (4,5). Distinguishing between benign and malignant pulmonary lesions is crucial for optimizing therapeutic decision-making while also avoiding unnecessary surgical morbidity. Patients frequently present with multiple nodules in different lobes, making resection of all semi-solid nodules often not feasible (6,7); therefore, the highest-risk nodules for progression to invasive cancer must be selectively identified for surgery. Additionally, the development of second lung cancers is common. An ideal treatment approach to subsolid nodules should target the most aggressive nodules while avoiding unnecessary interventions for nodules that are low risk of invasive cancer (8,9). There is a clinical need for improved biomarkers of lung nodules, both to distinguish between benign and malignant small nodules found incidentally on chest CT scan, but also to guide therapeutic decisions in semi-solid lepidic lung adenocarcinoma that present radiographically as GGOs.

Carcinoembryonic antigen-related cell adhesion molecule 5 (CEACAM5) is a transmembrane glycoprotein with roles in cell adhesion, migration, and survival. CEACAM5 is commonly upregulated in several epithelial malignancies, including colorectal, prostate, gastric, and non-small cell lung cancer (NSCLC) (10-12). In colorectal cancer, anti-CEACAM5 antibody drug conjugates are under active investigation in clinical trials as a treatment for select patients (13). Similarly, CEACAM5 is being studied as a potential prognostic biomarker in gastric cancer and for treatment-susceptibility in prostate cancer (14-16). In lung cancer, CEACAM5 is overexpressed in about 20% of lung adenocarcinoma and is not expressed in normal lung parenchyma. Importantly, high CEACAM5 expression in lung adenocarcinoma has been associated with worse overall survival (17-19).

We have previously shown that RNA expression of CEACAM5 is increased among semi-solid lung nodules between the non-solid ground-glass and the more invasive solid components within the same nodule (20). However, it is unknown if these changes in gene expression translate to elevated protein expression within evolving GGOs and subsolid lepidic lung adenocarcinoma. In this study, we examine the potential role of CEACAM5 as a diagnostic biomarker by comparing protein expression of CEACAM5 within subsolid lepidic lung adenocarcinoma to solid NSCLC. Subsolid lepidic lung adenocarcinoma is characterized by slow growth kinetics and excellent survival. It is unknown if what underlies the slow growth is tumor biology driven or regulated by the immune system. To better characterize CEACAM5’s role in tumor biology and progression of NSCLC, we also investigate correlations of CEACAM5 expression with common lung cancer driver mutations and the immune infiltration of CD8⁺ T-cells and CD20⁺ B-cells in the tumor microenvironment (TME). We present this article in accordance with the REMARK reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-1-0102/rc).

Methods

Lepidic adenocarcinoma GGO specimen selection

Patients with radiographic findings of GGO or subsolid lung nodule (SSN) were identified retrospectively, and those that underwent surgical resection of their GGO or SSN between 2019 and 2022 at Yale New Haven Hospital were studied. GGO were defined radiographically as focal pulmonary opacities of increased attenuation that preserved underlying bronchial and vascular markings on chest CT scan. SSN included both pure ground-glass nodules as well as part-solid nodules which contained both ground-glass and solid components on chest CT. Patients who underwent prior systemic therapy or radiation to GGO were excluded. Hematoxylin and eosin (H&E) stains were reviewed for all patients by an expert thoracic pathologist, who was blinded to clinical data and tumor subtype. Following this pathologic evaluation, 42 specimens were confirmed to be lepidic-predominant lung adenocarcinoma consistent with the radiographic GGO finding. Specimens with mucinous or non-lepidic predominant histologies were excluded. Manual chart review of the electronic medical record was performed to collect demographics, medical history, radiographic findings, and post-surgery follow-up information. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Institutional Review Board of Yale New Haven Hospital (No. 2000032694, date: September 27^th, 2022). Individual consent for this retrospective analysis was waived because the research involved minimal risk and used retrospective data analysis without patient-identifying information.

Lung cancer tissue microarray (TMA) creation

For comparison to a large cohort of solid NSCLC specimens, a TMA of 98 different solid NSCLC tumors was used. H&E section of formalin-fixed, paraffin-embedded (FFPE) surgical specimens were reviewed by a thoracic pathologist and a TMA was created using 0.6-mm diameter cores obtained from the corresponding FFPE blocks using standard TMA preparation methods (21). H&E of the resulting TMA was reviewed by a thoracic pathologist and cores with majority stromal compartments or poor tissue quality were excluded from analysis. The final TMA was comprised of 98 patients with NSCLC including 57 patients with adenocarcinoma, 20 patients with squamous cell carcinoma, 10 patients with large cell carcinoma, and 9 patients with adenosquamous carcinoma.

CEACAM5 immunohistochemistry (IHC) of lung cancer TMA and GGO cohort

IHC with mouse monoclonal antibodies against CEACAM5 was performed in a CLIA setting on 5 µm sections cut from the TMA and GGO FFPE blocks validated by pathologists. Staining was performed using a CEACAM5 primary antibody (Santa Cruz Biotechnology; #sc-23928; clone CI-P83-1; 1:100 dilution) on the Leica Bond III platform with ER2 antigen retrieval for 20 minutes followed by a 10-minute antibody incubation. Staining was done in batches with validated positive and negative controls used as quality assurance with each run. The CEACAM5-specific staining was evaluated by a thoracic pathologist using the Histoscore (H-score) classification system which incorporates both the proportion of positively stained cells and the staining intensity (0= no staining; 1+= faint, partially stained tumor cell; 2+= weakly to moderately stained tumor cell; and 3+= completely stained tumor cell). This system was used to be more comprehensive and reflective of true protein expression across a sample. Staining quality for each specimen was assessed through thoracic pathologist review and GGO specimens with poor immunohistochemistry staining including significant overstaining were excluded (n=6). The H-score for CEACAM5 expression was calculated for each TMA core and GGO specimen and used for further analysis.

Multiplex immunofluorescence of GGO cohort

GGO specimens were cut from the corresponding FFPE blocks, mounted onto slides, and placed in a 60 ℃ oven overnight. The slides underwent an established multiplex immunofluorescence staining protocol as shown by Schalper et al. (22). In brief, the slides were deparaffinized and rehydrated followed by antigen retrieval performed for 1 hour at 96 ℃ in 1mM EDTA-H₂O solution (pH 8; Millipore-Sigma #03690). Slides were then incubated with dual-endogenous enzyme blocking agent (Agilent Technologies; #S200389), rinsed, and then blocked with 0.3% BSA-TBST.

Primary CD8 antibodies (Agilent; #M710301-2; 1:250 dilution) and CD20 (Agilent; #M075501-2; 1:150) were mixed in a solution with 0.3% BSA-TBST. Slides were incubated with the primary antibodies at 4 ℃ overnight. Pairs of horseradish peroxidase (HRP)-conjugated secondary antibodies and Cf dye tyramide conjugates (Biotium, Freemont, CA, USA) were then added to the slides for 1 hour and 10 minutes, respectively, including Rat anti-mouse IgG1 (Invitrogen; #18401582; 1:100) and Cf488a Cf dye tyramide (Biotium; #92171, 1:100); and goat anti-mouse IgG2a (Abcam; #ab97245; 1:200) and Cf680r Cf dye tyramide (Biotium; #92196, 1:150). Secondary antibodies were diluted in 0.3% BSA-TBST, and tyramides were diluted in Tyramide Amplification Buffer (Biotium; #22029-T). After incubation with each tyramide, horseradish peroxidase (HRP) activity was blocked with a benzhydrazide-H₂O₂ solution (Millipore Sigma; #B13071). Slides were then incubated with an Alexa Fluor 532-conjugated pan-cytokeratin primary antibody (Novus Biologicals; #NBP2-33200AF532, 1:100) in 0.3% BSA-TBST. Finally, slides were stained with 4',6-diamidino-2-phenylindole (DAPI) and mounted with Vectashield Vibrance Antifade Mounting Medium (Vector Laboratories; #H-1700-10).

The stained slides were scanned using the Stellaris 8 Falcon confocal microscope imaging system (Leica Biosystems) under fluorescent conditions at 10× magnification with resolution of 1,024×1,024 pixels and laser excitation at 405, 488, 532, 647, and 680 nm. Detector gain and laser intensity were optimized for each fluorophore to maximize signal while minimizing background. Identical acquisition settings were maintained across all samples and sequential scanning was used to minimize spectral overlap between fluorophores. Each slide was analyzed using QuPath digital image analysis software version 0.6.0 and cell segmentation was performed using QuPath’s built-in cell detection algorithm. Fluorescence intensity thresholds for marker positivity were visually validated to distinguish positive and negative cells and identical thresholds were applied across all specimens for each marker. Cell quantification and analysis were performed across the entire analyzed tissue rather than by hotspot selection to minimize bias. Fluorescent intensities of each cell type were quantified (CD8⁺ as a surface marker of T cells, CD20⁺ as a surface marker of B cells, and PanCK⁺ as a surface marker to identify epithelial tumor cells), and optimal intensity thresholds were set for accurate individual cell identification. The final data were expressed as the number of cells of each type based on fluorescent signal above the predetermined fluorescent intensity threshold. Cells with fluorescent signal for multiple markers were excluded. Cells with DAPI nuclei fluorescence were included for determining the total cells in the tissue microenvironment which could include, for example, stromal fibroblasts, endothelial cells, alveolar or parenchymal cells, or macrophages. Cell-type percentages were calculated as the number of marker-positive cells divided by the total number of DAPI-positive nucleated cells within the entire tissue area.

Next-generation sequencing (NGS)

A targeted NGS was performed using an institutionally developed and prospectively validated genomic profiling assay studying cancer-related genes on the Ion Torrent platform. In brief, DNA is extracted from the FFPE specimens using the commercially available kit (Qiagen Inc., Germantown, MD, USA) and quantitated on a Qubit 2.0 fluorimeter. The isolated DNA was amplified using the Ion AmpliSeq Cancer Hotspot Panel v2 multiplex PCR primer set. These primers amplify 207 amplicons covering exonic regions of 50 cancer-related genes in which mutations have been reported in various types of cancer. In aggregate, ~30 kilobases of DNA sequence were amplified for subsequent nucleotide sequence analysis, which was performed on an Ion Torrent S5XL next generation sequencer. After sequencing, the data were processed using a customized bioinformatics analysis pipeline to detect various genomic alterations.

Statistical analysis

Patients with lepidic adenocarcinoma GGO were stratified into high and low CEACAM5 tumor expression based on H-score ≥100 or <100, and bivariate analysis was performed using Chi-squared test for categorical variables and unpaired t-test for continuous variables with P value <0.05 for significance. This included comparison of patient demographics, medical history, radiographic findings, follow-up information, pathologic features, tumor genetic profiling, and TME cell composition. The CEACAM5 H-score for the GGO cohort was also compared to the TME cell composition for CD8⁺ cells, CD20⁺ cells, and PanCK⁺ cells. Patient characteristics, pathologic staging, and CEACAM5 H-score were compared between the lepidic adenocarcinoma GGO cohort and solid NSCLC. This analysis was also repeated on subsets of the NSCLC cohort including by histologic subtype of NSCLC. CEACAM5 H-scores were compared among all patients in both the GGO cohort and each cancer type from the lung cancer TMA.

Results

Lepidic adenocarcinoma ground glass opacities vs. solid NSCLC patient demographics

The cohort of lepidic adenocarcinoma GGO patients had a mean age at diagnosis of 68 years with majority of patients in the cohort being female (n=28, 78%) and white (n=31, 94%). Most patients were current or former smokers (n=28, 78%) and had a mean smoking pack-years of 28 [standard deviation (SD) =15]. Only 2.8% (n=1) of patients had a history of prior lung cancer and 25% (n=9) had a history of other non-lung cancer (Table 1). Patients with lepidic adenocarcinoma GGO were more likely to be female (78% vs. 53%, P=0.01) and less likely to be a current or prior smoker (78% vs. 95%, P=0.006) compared to patients with solid NSCLC (Table 2). The mean total nodule size on preoperative CT scan was 22 mm with mean solid component size of 8 mm. Of the patients who had non-solid nodules on first GGO diagnosis, 61% of GGO (n=11) transformed into a part-solid nodule during interval CT scans. In the cohort, 53% (n=19) of patients had multiple GGO present on their preoperative CT scan, and 36% (n=4) of patients who started with a single GGO at first diagnosis would develop a 2^nd new primary GGO on interval CT scans.

CEACAM5 expression in lepidic adenocarcinoma GGO vs. solid NSCLC

IHC of CEACAM5 expression was performed for lepidic adenocarcinoma GGO versus solid NSCLC specimens encompassing all histologic adenocarcinoma subtypes, and protein expression read and H-scores were calculated by expert pathologist review (Figure 1). Levels of CEACAM5 protein expression, as measured by IHC H-scores, were compared between the lepidic adenocarcinoma GGOs and solid NSCLC tumors (Table 2). Overall, lepidic adenocarcinoma GGOs had significantly higher levels of CEACAM5 expression measured by H-scores (53 vs. 27, P<0.001) compared to the solid NSCLC group.

Figure 1 GGO with CEACAM5 immunohistochemistry. Surgically resected GGO specimens had sectioning of tissue onto microscope slides followed by CEACAM5 immunohistochemistry and H-score quantification by a thoracic pathologist. The specimens (n=36) were compared to evaluate trends in the CEACAM5 expression compared to solid NSCLC specimens. Regions of GGO were classified with Histoscore for intensity of CEACAM5 immunohistochemistry staining with 0 for no staining (A), 1+ for weak staining (B), 2+ for moderate staining (C), and 3+ for strong staining (D). CEACAM5, carcinoembryonic antigen-related cell adhesion molecule 5; GGO, ground-glass opacity; NSCLC, non-small cell lung cancer.

Lepidic adenocarcinoma GGOs were also compared to the subgroup of solid NSCLC patients with adenocarcinoma. Patients with lepidic adenocarcinoma GGO were more likely to be female (78% vs. 37%, P<0.001) and had tumors with significantly higher CEACAM5 expression compared to solid lung adenocarcinoma (H-score 53 vs. 35, P=0.009) (Table 2). When comparing CEACAM5 expression among other solid lung cancer histology, lepidic adenocarcinoma GGO also had significantly higher CEACAM5 expression than patients with squamous cell carcinoma (H-score 53 vs. 4.6, P=0.003) and large cell carcinoma (H-score 53 vs. 12, P=0.02) (Figure 2).

Figure 2 CEACAM5 H-Score distributions among GGO cohort and TMA lung cancer types. CEACAM5 expression was compared using H-score for patients with lepidic adenocarcinoma GGO versus different subgroups of patients from the NSCLC cohort. These subgroups within the NSCLC cohort included lung adenocarcinoma, squamous cell carcinoma, and large cell carcinoma. The distribution of H-scores, including the mean for each cohort (colored bar), is shown with statistical significance represented by * for P<0.05 and ** for P<0.01. Adeno, adenocarcinoma; CEACAM5, carcinoembryonic antigen-related cell adhesion molecule 5; GGO, ground-glass opacity; NSCLC, non-small cell lung cancer; SCC, squamous cell carcinoma; TMA, tissue microarray.

Patient demographics, radiographic growth patterns, and tumor characteristics were compared between tumors with high and low levels of CEACAM5 expression. Less smoking exposure history was found in patients with lower CEACAM5 expression compared to those with higher CEACAM5 expression (18 vs. 33 pack-years, P=0.01). Although this may reflect distinct biologic differences in smoking pack-years, this finding should be interpreted cautiously given the limited sample size. Radiographically prior to surgical resection, all lepidic adenocarcinoma GGO specimens with high CEACAM5 expression (n=5) started as radiographic non-solid GGO and then actively developed a new radiographic solid component compared to only 46% of lepidic adenocarcinoma GGO with lower CEACAM5 expression, although this association did not reach statistical significance (100% vs. 46%, P=0.10) (Table 1).

GGO CEACAM5 immunohistochemistry and TME

Immune cell infiltration of CD8⁺ T-cells and CD20 B-cells was observed among surgically resected lepidic adenocarcinoma subsolid nodules with a mean of 1.06% CD20 positive cells (SD =1.49), 3.30% CD8⁺ T-cells (SD =2.80), and 3.0% PanCK positive cells (SD =4.19) (Figures 3,4). Patients with lepidic adenocarcinoma GGO were stratified based on high and low CEACAM5 protein expression as measured by IHC H-score <100 vs. ≥100 with examples of IHC scoring in Figure 1. Lepidic adenocarcinoma GGO patients with high CEACAM5 expression had higher proportion of PanCK positive tumor cells in their tissue microenvironment (5.5% vs. 2.2%, P=0.02) than patients with lower CEACAM5 expressing tumors (Table 1). CD8⁺ and CD20⁺ immune cell infiltration into lepidic adenocarcinoma tumors were not significantly correlated with CEACAM5 expression (Figure 5).

Figure 3 Immunofluorescence images of GGO tumor immune microenvironment. GGO tissue imaged by multiplex immunofluorescence demonstrates distribution of all cell types (A) including DAPI (dark blue) for nucleic identification of all cell types, pan-cytokeratin (green) for malignant epithelial cells, CD20⁺ (cyan) for B-cells, and CD8⁺ (yellow) for cytotoxic T-cells. 10× magnification of the tissue area (red box in A) is shown in (B-E) for the distribution of all cell types (B), pan-cytokeratin positive cells (C), CD20 positive cells (D) and CD8 positive cells (E). DAPI, 4',6-diamidino-2-phenylindole; GGO, ground-glass opacity.

Figure 4 Distribution of cell types among patients with GGO. Surgically resected GGO specimens were sectioned onto microscope slides and underwent antibody staining, immunofluorescence microscopy, and cell quantification. The distribution of cell types within the tissue microenvironment are shown as violin plots across all GGO specimens (n=36) for CD8⁺ T cells, CD20⁺ B cells, and PanCK⁺ tumor epithelial cells with the thick dashed lines as the means, and thin dashed lines as upper and lower quartiles for each subgroup. GGO, ground-glass opacity.

Figure 5 CEACAM5 H-score of GGO for CD8, CD20, and PanCK composition of tumor microenvironment. Distribution of the percentages of cell types detected by immunofluorescence in the TME of GGO specimens including CD8⁺ T cells (A), CD20⁺ B cells (B), and PanCK⁺ tumor cells (C). Surgically resected GGO specimens had sequential sectioning of tissue onto microscope slides with one slide undergoing CEACAM5 immunohistochemistry with H-score quantification and the other slide undergoing antibody staining, immunofluorescence microscopy, and cell quantification. The specimens (n=36) were compared to evaluate trends in the CEACAM5 expression and the cell type composition within the tissue microenvironment. CEACAM5, carcinoembryonic antigen-related cell adhesion molecule 5; GGO, ground-glass opacity; TME, tumor microenvironment.

Next generation sequencing was performed on the resected lepidic adenocarcinoma specimens (n=36) and found 38% (n=9) with driver mutations in EGFR, 38% (n=9) with KRAS driver mutations, 13% (n=3) with a BRAF mutation, 25% (n=6) with TP53 co-mutation, and 4.2% (n=1) with co-mutations in STK11. We observed no differences in the distribution of mutations among lepidic adenocarcinoma GGOs based on CEACAM5 expression, though sample size was small (Table 2).

Since the lepidic adenocarcinoma GGO cohort utilized IHC and IF of the full resected tissue section, additional analysis evaluated the spatial distribution of H-score (0, 1+, 2+, and 3+) components across the tissue area. Among specimens with H-score 50–100, the percentage of tissue with any CEACAM5 staining (regardless of staining intensity) was 30.0% compared to 53.3% and 93.3% for specimens with H-scores of 100–200 and 200–300, respectively. Furthermore, among specimens with the strongest CEACAM5 staining intensity (3+), the mean tissue area demonstrating 3+ staining increased with higher H-score category, measuring 7.5% for H-scores 50–100, 22.5% for H-scores 100–200, and 56.7% for H-scores 200–300.

Discussion

In this study, the potential role of CEACAM5 as a biomarker for subsolid lepidic adenocarcinoma was explored. Subsolid, lepidic adenocarcinoma had greater CEACAM5 expression on immunohistochemistry than other types of solid NSCLC, even when specifically compared to lung adenocarcinoma specimens. These findings align with previous work showing elevated levels of CEACAM5 gene expression within GGO and elevated levels of CEA in blood among patients with GGO (20,23). Recent studies from the United Kingdom Proteomics Project integrated analyses of lung cancer-associated genetic variations with proteomic profiling of circulating biomarkers, and they identified CEACAM5 as one of the top disease-associated proteins which may be detectable in plasma before lung cancer diagnosis (24). These findings reinforce CEACAM5 as a biomarker of high research interest in early lung cancer and lung cancer screening.

CEACAM5 is actively being investigated for diagnostic and therapeutic targeting in clinical trials in lung cancer. The phase 1b/2 multicenter PROCEADE PanTumor trial is currently evaluating a CEACAM5 antibody-drug conjugate for patients with locally-advanced or metastatic CEACAM5 expressing tumors including gastric, NSCLC, and pancreatic adenocarcinoma (25). Another phase 1 trial (NCT06768151) is recruiting patients with CEACAM5-positive, advanced stage NSCLC to evaluate treatment with chimeric antigen receptor T lymphocytes (CAR-T) therapy (26). Interestingly, the CARMEN trial, which investigated CEACAM5 as a therapeutic target in NSCLC, initially failed to show significant improvement in progression-free survival for patients treated with the CEACAM5 antibody-drug conjugate, tusamitamab ravtansine, compared to standard chemotherapy (27,28). However, a subsequent subgroup analysis of the data showed that significantly better objective response rates were observed among patients with the highest levels of CEACAM5 staining as well as patients with detectable circulating levels of carcinoembryonic antigen (CEA) (29).

In addition to CEACAM5-targeted therapies for NSCLC, CEACAM5 may also have potential to localize tumors during surgical resection. Anti-CEACAM5 fluorochromes can be administered systemically for patients with CEACAM-5 positive tumors and illuminate tumor targets during surgical resection. This is being investigated as a target for intraoperative molecular imaging with studies utilizing anti-CEACAM5 fluorochromes to localize CEACAM5-positive tumors during pulmonary resection ensure negative margins (30). This approach is being studied for colon cancer metastases to the lung, and there is preclinical data to support the feasibility of this approach in primary lung tumors (31,32). Furthermore, a recent large study of IHC evaluation of CEACAM5 in solid NSCLC tumors similarly found that CEACAM5 was more expressed among lung adenocarcinoma than squamous cell carcinoma which is in concordance with our data. However, no association was seen between CEACAM5 IHC expression and overall survival, tumor stage, PD-L1 expression, tumor mutational burden, and EGFR or KRAS mutations (33).

Other studies have shown high CEACAM5 expression in advanced stage lung adenocarcinoma compared to lower/intermediate stages (17,18). This trend has also been observed in other solid tumors, including pancreatic and colon cancer (11,12). Interestingly, studies in gastric cancer have shown a potential bimodal distribution with high CEACAM5 expression in premalignant gastric lesions and also advanced gastric cancer (34,35). Taken with the prior findings of high CEACAM5 expression in advanced NSCLC (36,37), it is possible that our findings support NSCLC following a similar bimodal stage distribution pattern as gastric cancer. These findings should not be interpreted as suggesting a simple monotonic increase in CEACAM5 expression from lepidic to fully solid adenocarcinoma across all tumors. Rather, CEACAM5 in NSCLC may simply be a biomarker for a unique subtype of lung adenocarcinoma, arising with specific patterns of growth. Not all solid NSCLC express CEACAM5, and CEACAM5-positive tumors can exist at all stages. Our data show that lepidic adenocarcinoma GGO has higher levels of CEACAM5 than other forms of adenocarcinoma, and CEACAM5 may therefore be a marker of this subtype. There were trends within our data to suggest that within this subgroup, the highest-expressing lesions may exhibit greater radiographic evolution, although this observation is limited by sample size, did not reach statistical significance, and should be interpreted with caution. Importantly, the biological mechanisms underlying the increased CEACAM5 expression in lepidic adenocarcinoma are not fully defined but may relate to its established roles in epithelial adhesion. As a transmembrane glycoprotein involved in cell-cell adhesion, CEACAM5 is thought to contribute to the epithelial organization in lung adenocarcinoma (38,39). In the context of lepidic adenocarcinoma presenting as radiographic GGO, elevated CEACAM5 expression may reflect the lepidic growth pattern which relies heavily on cell-cell adhesion molecules for proliferation along intact alveolar structures. However, the mechanisms of CEACAM5 in lung cancer are not fully understood, and future studies will continue to investigate this.

This study has several important limitations. First, the sample size of the lepidic adenocarcinoma GGO cohort is modest with only a small subset of tumors demonstrating high CEACAM5 expression. While the primary comparison between GGO and solid NSCLC revealed statistically significant differences in CEACAM5 expression, the limited sample size reduces the statistical power of subgroup analyses. Future studies with larger sample size are needed to validate and better clarify these relationships. Another limitation of our study comes from the selection of patients who all underwent surgical resection. These patients may represent a subgroup with greater clinical concern for malignancy and may not fully reflect the broader population of patients who are under lung nodule surveillance. Therefore, future studies would evaluate CEACAM5 expression over time in GGOs undergoing radiographic surveillance.

Conclusions

This study demonstrates that CEACAM5 is significantly overexpressed in lepidic adenocarcinoma GGO compared to other forms of solid NSCLC. This study provides a biologic rational to support the potential utility of CEACAM5 as a diagnostic biomarker for this subtype of NSCLC. Larger, prospective studies incorporating longitudinal imaging and molecular profiling could help clarify whether CEACAM5 tissue expression or serum measurement of circulating CEA predicts radiographic progression or histologic transformation of GGO lesions. Future research should also aim to integrate CEACAM5 with other emerging biomarkers, including genomic alterations and radiomic features, to develop multi-dimensional risk stratification models. This could improve clinical decision-making in the management of subsolid and ground glass pulmonary nodules, particularly in distinguishing lesions that warrant intervention from those that can be observed.

Acknowledgments

We would like to thank the Braude Foundation for their support of this work and Yale Pathology Tissue Services for sectioning of tissue for analysis and immunohistochemistry support. We also would like to thank the West Campus Imaging core facilities for assistance with setting up the Stellaris microscope used for immunofluorescence. An abstract for this study was presented at the 2025 IASLC World Conference on Lung Cancer as an electronic poster. Reprinted from Journal of Thoracic Oncology, Bader JM et al., CEACAM5 as a Biomarker for Differentiating Ground-Glass Opacities from Other Lung Cancers, Volume 20, Issue 10, Supplement 1, S706-S707, October 2025, with permission from Elsevier.

Footnote

Reporting Checklist: The authors have completed the REMARK reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-1-0102/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-1-0102/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-1-0102/prf

Funding: This study was supported by the Braude Foundation and a young investigator award from the IASLC International Lung Cancer Foundation.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-1-0102/coif). L.C. is the founder and a member of the scientific advisory board of Normunity Biotechnology. He holds company and financial interests in Tayu Biotech Group, NextCure, CTTQ Pharmaceuticals Co., Tpioneer, and ONC4C, and serves on advisory boards for selected entities. S.D. reports advisory board participation and serves as a consultant for AstraZeneca and advisory board participation for Genentech. G.A.W. reports advisory board participation for AstraZeneca and Bristol Myers Squibb. The other authors have no conflicts of interest to declare.

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. The study was approved by the Institutional Review Board of Yale New Haven Hospital (No. 2000032694, date: September 27^th, 2022). Individual consent for this retrospective analysis was waived because the research involved minimal risk and used retrospective data analysis without patient-identifying information.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Hammer MM, Hatabu H. Subsolid pulmonary nodules: Controversy and perspective. Eur J Radiol Open 2020;7:100267. [Crossref] [PubMed]
2. Infante M, Berghmans T, Heuvelmans MA, et al. Slow-growing lung cancer as an emerging entity: from screening to clinical management. Eur Respir J 2013;42:1706-22. [Crossref] [PubMed]
3. Korb ML, Burt BM. The elusive ground glass opacity, revealed. J Thorac Dis 2018;10:S3828-30. [Crossref] [PubMed]
4. Loverdos K, Fotiadis A, Kontogianni C, et al. Lung nodules: A comprehensive review on current approach and management. Ann Thorac Med 2019;14:226-38. [Crossref] [PubMed]
5. Hodnett PA, Ko JP. Evaluation and management of indeterminate pulmonary nodules. Radiol Clin North Am 2012;50:895-914. [Crossref] [PubMed]
6. Akçiçek E, Durhan G, Düzgün SA, et al. Non-metastatic causes of multiple pulmonary nodules. Insights Imaging 2024;15:288. [Crossref] [PubMed]
7. Shrager JB. Approach to the patient with multiple lung nodules. Thorac Surg Clin 2013;23:257-66. [Crossref] [PubMed]
8. Chen K, Liu A, Wang C, et al. Multidisciplinary expert consensus on diagnosis and treatment of multiple lung cancers. Med 2025;6:100643. [Crossref] [PubMed]
9. Tian H, Bai G, Yang Z, et al. Multiple primary lung cancer: Updates of clinical management and genomic features. Front Oncol 2023;13:1034752. [Crossref] [PubMed]
10. Zhou J, Fan X, Chen N, et al. Identification of CEACAM5 as a Biomarker for Prewarning and Prognosis in Gastric Cancer. J Histochem Cytochem 2015;63:922-30. [Crossref] [PubMed]
11. Blumenthal RD, Leon E, Hansen HJ, et al. Expression patterns of CEACAM5 and CEACAM6 in primary and metastatic cancers. BMC Cancer 2007;7:2. [Crossref] [PubMed]
12. Thomas J, Klebanov A, John S, et al. CEACAMS 1, 5, and 6 in disease and cancer: interactions with pathogens. Genes Cancer 2023;14:12-29. [Crossref] [PubMed]
13. Kopetz S, Boni V, Kato K, et al. Precemtabart tocentecan, an anti-CEACAM5 antibody-drug conjugate, in metastatic colorectal cancer: a phase 1 trial. Nat Med 2025;31:3504-13. [Crossref] [PubMed]
14. Imberti C, De Gregorio R, Korsen JA, et al. CEACAM5-Targeted Immuno-PET in Androgen Receptor-Negative Prostate Cancer. J Nucl Med 2024;65:1043-50. [Crossref] [PubMed]
15. Martinelli S, Peri S, Anceschi C, et al. Targeting CEACAM5: Biomarker Characterization and Fluorescent Probe Labeling for Image-Guided Gastric Cancer Surgery. Biomedicines 2025;13:1812. [Crossref] [PubMed]
16. Zhang L, Zhang C, Liu N. CEACAM5 targeted by miR-498 promotes cell proliferation, migration and epithelial to mesenchymal transition in gastric cancer. Transl Oncol 2022;24:101491. [Crossref] [PubMed]
17. Hu R, Huffman KE, Chu M, et al. Quantitative Secretomic Analysis Identifies Extracellular Protein Factors That Modulate the Metastatic Phenotype of Non-Small Cell Lung Cancer. J Proteome Res 2016;15:477-86. [Crossref] [PubMed]
18. Papadaki MA, Messaritakis I, Fiste O, et al. Assessment of the Efficacy and Clinical Utility of Different Circulating Tumor Cell (CTC) Detection Assays in Patients with Chemotherapy-Naïve Advanced or Metastatic Non-Small Cell Lung Cancer (NSCLC). Int J Mol Sci 2021;22:925. [Crossref] [PubMed]
19. Tabernero J, Bedard PL, Bang YJ, et al. Tusamitamab Ravtansine in Patients with Advanced Solid Tumors: Phase I Study of Safety, Pharmacokinetics, and Antitumor Activity Using Alternative Dosing Regimens. Cancer Res Commun 2023;3:1662-71. [Crossref] [PubMed]
20. Woodard GA, Ding V, Cho C, et al. Comparative genomics between matched solid and lepidic portions of semi-solid lung adenocarcinomas. Lung Cancer 2023;180:107211. [Crossref] [PubMed]
21. Camp RL, Chung GG, Rimm DL. Automated subcellular localization and quantification of protein expression in tissue microarrays. Nat Med 2002;8:1323-7. [Crossref] [PubMed]
22. Carvajal-Hausdorf DE, Schalper KA, Neumeister VM, et al. Quantitative measurement of cancer tissue biomarkers in the lab and in the clinic. Lab Invest 2015;95:385-96. [Crossref] [PubMed]
23. Tomita M, Ayabe T, Chosa E, et al. Correlation between Serum Carcinoembryonic Antigen Level and Histologic Subtype in Resected Lung Adenocarcinoma. Asian Pac J Cancer Prev 2015;16:3857-60. [Crossref] [PubMed]
24. Johnson MA, Nieves-Rodriguez S, Hou L, et al. Machine learning-based proteogenomic data modeling identifies circulating plasma biomarkers for early detection of lung cancer. Commun Med (Lond) 2026; Epub ahead of print. [Crossref]
25. EMD Serono Research & Development Institute, Inc. PROCEADE PanTumor: A Phase 1b/2, Multicenter, Open-Label Study of Anti-CEACAM5 Antibody-Drug Conjugate M9140 in Participants With Advanced Solid Tumors (Master Protocol). clinicaltrials.gov; 2025. Accessed December 10, 2025. Available online: https://clinicaltrials.gov/study/NCT06710132
26. Chongqing Precision Biotech Co., Ltd. Clinical Study of CEA Targeting Chimeric Antigen Receptor T Lymphocytes (CAR-T) for CEA Positive Advanced Lung Cancer. clinicaltrials.gov; 2025. Accessed December 10, 2025. Available online: https://clinicaltrials.gov/study/NCT06768151
27. CARMEN-LC03: Tusamitamab Ravtansine vs Docetaxel in Previously Treated Advanced Nonsquamous NSCLC - The ASCO Post. Accessed August 4, 2025. Available online: https://ascopost.com/news/october-2024/carmen-lc03-tusamitamab-ravtansine-vs-docetaxel-in-previously-treated-advanced-nonsquamous-nsclc/
28. Besse B, Russo GL, Lena H, et al. OA08.05 Tusamitamab Ravtansine vs Docetaxel in Previously Treated Advanced Nonsquamous NSCLC: Results from Phase 3 CARMEN-LC03 Trial. J Thorac Oncol 2024;19:S25-S26.
29. Gazzah A, Ternès N, Lee JS, et al. Biomarker analysis from a Phase 1/1b study of tusamitamab ravtansine in patients with advanced non-small cell lung cancer. Transl Oncol 2026;63:102615. [Crossref] [PubMed]
30. Gutowski M, Framery B, Boonstra MC, et al. SGM-101: An innovative near-infrared dye-antibody conjugate that targets CEA for fluorescence-guided surgery. Surg Oncol 2017;26:153-62. [Crossref] [PubMed]
31. Azari F, Kennedy GT, Chang A, et al. Glycoprotein Receptor CEACAM5-Targeted Intraoperative Molecular Imaging Tracer in Non-Small Cell Lung Cancer. Ann Thorac Surg 2023;116:631-41. [Crossref] [PubMed]
32. Azari F, Meijer RPJ, Kennedy GT, et al. Carcinoembryonic Antigen-Related Cell Adhesion Molecule Type 5 Receptor-Targeted Fluorescent Intraoperative Molecular Imaging Tracer for Lung Cancer: A Nonrandomized Controlled Trial. JAMA Netw Open 2023;6:e2252885. [Crossref] [PubMed]
33. Hsu YR, Almutrafi A, Hueniken K, et al. The Landscape of CEACAM5 Expression by Immunohistochemistry in NSCLC. JTO Clin Res Rep 2026;7:100943. [Crossref] [PubMed]
34. Jang B, Lee SH, Dovirak I, et al. CEACAM5 and TROP2 define metaplastic and dysplastic transitions in human antral gastric precancerous lesions and tumors. Gastric Cancer 2024;27:263-74. [Crossref] [PubMed]
35. Liu JN, Wang HB, Zhou CC, et al. CEACAM5 has different expression patterns in gastric non-neoplastic and neoplastic lesions and cytoplasmic staining is a marker for evaluation of tumor progression in gastric adenocarcinoma. Pathol Res Pract 2014;210:686-93. [Crossref] [PubMed]
36. Zhang X, Han X, Zuo P, et al. CEACAM5 stimulates the progression of non-small-cell lung cancer by promoting cell proliferation and migration. J Int Med Res 2020;48:300060520959478. [Crossref] [PubMed]
37. Lefebvre AM, Adam J, Nicolazzi C, et al. The search for therapeutic targets in lung cancer: Preclinical and human studies of carcinoembryonic antigen-related cell adhesion molecule 5 expression and its associated molecular landscape. Lung Cancer 2023;184:107356. [Crossref] [PubMed]
38. Matsumura M, Mitsui H, Woo T, et al. Identification of key molecules in micropapillary progression of lung adenocarcinoma: A comprehensive gene expression analysis study using the spatial gene expression solution methodology. Oncol Lett 2025;30:533. [Crossref] [PubMed]
39. Wu Z, Zeng X, Wang H, et al. LncRNA ARAP1-AS1 contributes to lung adenocarcinoma development by targeting miR-8068 to upregulate CEACAM5. Cancer Biomark 2023;38:177-89. [Crossref] [PubMed]