Introduction

Immunotherapy such as that using immune checkpoint inhibitors (ICIs) is generally administered to patients with advanced non-small cell lung cancer (NSCLC). Immunotherapy consists of programmed death-1 (PD-1) blockade or anti-PD-1 antibody plus cytotoxic T-lymphocyte antigen-4 (CTLA4) antibody determined by the expression of programmed death-ligand-1 (PD-L1) in cancer cells. Currently, there is no optimal biomarker for predicting the efficacy of anti-PD-1 (nivolumab) plus anti-CTLA4 (ipilimumab; Nivo-Ipi) antibody-based immunotherapy for the treatment of NSCLC.

Recently, several new agents targeting tumor specific molecules have been developed to establish late-line standard of care for patients with advanced NSCLC. Of these, trophoblast surface antigen 2 (TROP2) is a transmembrane glycoprotein expressed in NSCLC, with its high expression closely associated with a poor outcome, making it a promising therapeutic target1. Recent prospective studies demonstrated promising antitumor activity of datopotamab deruxtecan (Dato-DXd), a novel TROP2-directed antibody–drug conjugate (ADC), in pretreated patients with advanced NSCLC2. Dato-DXd significantly improved progression-free survival (PFS) compared to docetaxel in a phase III study3. However, it is unclear whether TROP2 expression can be used as a biomarker of therapeutic efficacy and outcome in patients with advanced NSCLC.

Previous studies have reported that TROP2 overexpression in tumor cells is a predictor of worse prognosis in patients with NSCLC harboring early-stage disease4,5. Further, Jiang et al. showed that TROP2 overexpression was closely correlated with poor prognosis in patients with advanced NSCLC who received platinum-based chemotherapy or chemo-radiotherapy6. Currently, a randomized phase III study comparing Dato-DXd with pembrolizumab to pembrolizumab alone is ongoing worldwide in patients with advanced NSCLC7. A recent study observed a significant association of TROP2 overexpression with a poor outcome on PD-1 blockade based on transcriptomic data from 891 patients with NSCLC in two large, randomized trials8. A preclinical study also suggested that TROP2 expression could affect the functions of cytotoxic T cells by leading them towards apoptosis9. However, it remains unclear whether the TROP2 expression can be used as a biomarker for the efficacy of treatment with CTLA4 antibody in addition to PD-1 blockade in patients with advanced NSCLC.

Here, we conducted a retrospective study to evaluate the clinicopathological relevance of TROP2 expression in patients with advanced NSCLC receiving Nivo-Ipi therapy.

Results

Patient demographics

Patient characteristics are summarized in Table 1. The median age was 72 years, ranging from 45 to 86 years. The detailed information was as follows: 89 patients were male and 21 were female; 37 patients had a PS of 0 and 50 had a PS of 1; 56 patients had adenocarcinoma (AD), 41 had squamous cell carcinoma (SCC), and 13 had tumors of other histological types; 47 patients had PD-L1 < 1%, 51 had 1–49%, 10 had ≥ 50%, and 2 patients had an unknown status. The objective response rate (ORR) of 110 patients was 37.3% [1 complete response (CR), 40 partial response (PR), 33 stable disease (SD), 28 progressive disease (PD), and 8 not evaluated (NE)].

Table 1 Patient characteristics.
Full size table

Nivo-Ipi therapy was mainly chosen for patients with PS 0–1 and low PD-L1 expression (particularly ≤ 10%). It was occasionally used in patients with PS 2–3, particularly in younger patients or those with low PD-L1 expression. A few patients with PD-L1 ≥ 50% also received this therapy, usually when chemotherapy was considered inappropriate or when a durable immune response was anticipated. Driver gene mutation status was available in 98 patients. Among them, 89 had no mutations, while nine patients harbored driver mutations (EGFR, n = 1; KRAS, n = 7; others, n = 2). No ALK rearrangements were detected. Mutation status was unavailable in 12 patients.

Immunohistochemical evaluation

TROP2 protein expression was immunohistochemically detected in 110 tumors. Of them, 86 were biopsy samples and 24 were surgically resected. Figure 1 displays representative IHC images. TROP2 immunostaining was detected in the cytoplasm and the plasma membrane. Overexpression of TROP2 was observed in 46 (41.8%) patients (AD: 20, SCC: 20, Others: 6). Cytoplasmic staining of TROP2 with an intensity score of 3 was observed in 64 (58.1%) patients (AD: 29, SCC: 26, Others: 9). TROP2 staining with a proportion score of 4 was observed in 68 (61.8%) patients (AD: 36, SCC: 25, Others: 7). The patient characteristics according to TROP2 expression are shown in Table 2. Overexpression of TROP2 was significantly associated with poor response.

Fig. 1
figure 1

Representative immunohistochemical staining of TROP2 expression in lung cancer. The assessment of TROP2 expression was performed based on intensity score of 0 to 3 (A) and proportion score of 1 to 4 (B).

Full size image
Table 2 Patient’s characteristics according to TROP2 expression.
Full size table

TROP2 expression with an intensity score of 3 was significantly related to lower incidence of brain metastasis and low leukocyte count. TROP2 expression with a proportion score of 4 was also significantly related to the frequency of immune-related Adverse Events (irAEs). Patient characteristics according to PD-L1, histology, and TROP2 expression are shown in Table A1 (Supplementary Information).

Survival analysis

The median PFS and OS were 210 and 413 days, respectively. Eighty-one patients experienced tumor recurrence, and 59 died due to disease progression. The Kaplan–Meier curves for PFS and OS according to TROP2 expression are depicted in Fig. 2. Overexpression of TROP2 was associated with a shorter PFS and OS in all patients, and worse prognosis in PFS and OS was observed in patients with TROP2 overexpression even after stratification by PD-L1 expression rates (PD-L1 < 1%; Fig. 2C,D, PD-L1 ≥ 1%; Fig. 2E,F) and histological type (AD; Fig. 2G,H, non-AD; Fig. 2I,J). Based on these subgroups, patients with TROP2 overexpression had significantly shorter PFS and OS in PD-L1 < 1% or non-AD histology groups.

Fig. 2
figure 2

Kaplan–Meier survival curve according to TROP2 expression in all patients [PFS (A) and OS (B)], subgroup of PD-L1 < 1% [PFS (C) and OS (D)], those of PD-L1 ≥ 1% [PFS (E) and OS (F)], those of AD [PFS (G) and OS (H)], and those of non-AD [PFS (I) and OS (J)]. Median PFS and OS values, hazard ratios (HR) with 95% confidence intervals (CI), and log-rank p-values were as follows: all patients: median PFS 273 vs 140 days (HR = 1.30, 95% CI 1.05–1.63, p = 0.017) [A]; median OS 784 vs 272 days (HR = 1.46, 95% CI 1.13–1.89, p = 0.003) [B]. PD-L1 < 1%: median PFS 336 vs 113 days (HR = 1.41, 95% CI 1.01–1.98, p = 0.041) [C]; median OS 714 vs 269 days (HR = 1.54, 95% CI 1.06–2.25, p = 0.021) [D]. PD-L1 ≥ 1%: median PFS 262 vs 145 days (HR = 1.33, 95% CI 0.99–1.79, p = 0.056) [E]; median OS NR vs 273 days (HR = 1.50, 95% CI 1.04–2.15, p = 0.023) [F]. AD: median PFS 298 vs 179 days (HR = 1.13, 95% CI 0.82–1.56, p = 0.474) [G]; median OS NR vs 314 days (HR = 1.41, 95% CI 0.98–2.03, p = 0.061) [H]. non-AD: median PFS 262 vs 127 days (HR = 1.45, 95% CI 1.05–1.99, p = 0.019) [I]; median OS 784 vs 251 days (HR = 1.52, 95% CI 1.04–2.24, p = 0.028) [J].

Full size image

Next, we conducted a survival analysis based on the intensity and proportional expression of TROP2, including subgroups defined by PD-L1 expression status (PD-L1 < 1% and PD-L1 ≥ 1%) and histology (AD and non-AD) (Fig. 3). High intensity of TROP2 expression was associated with worse PFS and OS in all patients, whereas high proportional expression of TROP2 showed no prognostic value. Stratified by PD-L1 expression status, high intensity of TROP2 expression was significantly associated with worse OS, while no significant association with OS was observed with high proportional TROP2 expression. In patients with AD, high intensity of TROP2 expression was closely associated with worse PFS. Conversely, in patients with non-AD histology, high proportional expression of TROP2 was significantly associated with worse PFS.

Fig. 3
figure 3figure 3

Kaplan–Meier survival curves for PFS and OS in all patients based on TROP2 intensity [PFS (A), OS (B)] and proportion [PFS (C), OS (D)]. Subgroup analyses by PD-L1 expression (PD-L1 < 1%: PFS (E), OS (F); PD-L1 ≥ 1%: PFS (G), OS (H)) and histology (AD: PFS (I), OS (J); non-AD: PFS (K), OS (L)) for TROP2 intensity. Those of PD-L1 expression (PD-L1 < 1%: PFS (M), OS (N); PD-L1 ≥ 1%: PFS (O), OS (P)) and histology (AD: PFS (Q), OS (R); non-AD: PFS (S), OS (T)) for TROP2 proportion. AD, adenocarcinoma. Median PFS and OS values, hazard ratios (HR) with 95% confidence intervals (CI), and log-rank p-values were as follows: all patients (Intensity): median PFS 298 vs 151 days (HR = 1.26, 95% CI 1.01–1.59, p = 0.041) [A]; median OS NR vs 356 days (HR = 1.52, 95% CI 1.14–2.02, p = 0.003) [B]. all patients (Proportion): median PFS 285 vs 189 days (HR = 1.19, 95% CI 0.94–1.49, p = 0.138) [C]; median OS 784 vs 443 days (HR = 1.22, 95% CI 0.93–1.60, p = 0.150) [D]. PD-L1 < 1% (Intensity): PFS 336 vs 139 days (HR = 1.25, 95% CI 0.88–1.76, p = 0.207) [E]; OS 784 vs 372 days (HR = 1.50, 95% CI 0.99–2.28, p = 0.048) [F]. PD-L1 ≥ 1% (Intensity): PFS 276 vs 151 days (HR = 1.38, 95% CI 1.01–1.88, p = 0.038) [G]; OS NR vs 353 days (HR = 1.65, 95% CI 1.10–2.49, p = 0.012) [H]. AD (Intensity): PFS 336 vs 178 days (HR = 1.24, 95% CI 0.91–1.71, p = 0.169) [I]; OS NR vs 365 days (HR = 1.60, 95% CI 1.08–2.37, p = 0.016) [J]. non-AD (Intensity): PFS 255 vs 139 days (HR = 1.23, 95% CI 0.88–1.72, p = 0.211) [K]; OS 784 vs 353 days (HR = 1.40, 95% CI 0.93–2.11, p = 0.099) [L]. PD-L1 < 1% (Proportion): PFS 325 vs 139 days (HR = 1.21, 95% CI 0.86–1.70, p = 0.266) [M]; OS 714 vs 383 days (HR = 1.30, 95% CI 0.88–1.93, p = 0.179) [N]. PD-L1 ≥ 1% (Proportion): PFS 270 vs 189 days (HR = 1.28, 95% CI 0.92–1.77, p = 0.139) [O]; OS NR vs 545 days (HR = 1.27, 95% CI 0.86–1.88, p = 0.232) [P]. AD (Proportion): PFS 298 vs 228 days (HR = 1.03, 95% CI 0.75–1.42, p = 0.855) [Q]; OS NR vs 591 days (HR = 1.16, 95% CI 0.78–1.71, p = 0.470) [R]. non-AD (Proportion): PFS 285 vs 134 days (HR = 1.39, 95% CI 1.00–1.93, p = 0.046) [S]; OS 714 vs 291 days (HR = 1.34, 95% CI 0.91–1.97, p = 0.134) [T].

Full size image

The results of univariate and multivariate analyses for PFS and OS in all patients are summarized in Table 3. Univariate analysis of all patients identified PS, liver metastases, bone metastases, NLR, TROP2, PLR, SII, PNI, and ALI as significant predictors of PFS and OS. The univariate log-rank test enabled screening for variables with p < 0.05 for subsequent multivariate analysis. The results of the multivariate analysis identified liver metastases, bone metastases, TROP2, and SII as independent predictors of PFS. Liver metastases, bone metastases, NLR, TROP2, SII, and PNI were also identified as independent predictors of OS.

Table 3 Univariate and multivariate survival analysis in all patients.
Full size table

Discussion

Here, we conducted a clinicopathological study to evaluate the prognostic significance of TROP2 expression in patients with advanced NSCLC who received Nivo-Ipi therapy. We found that overexpression of TROP2 was a significant biomarker for predicting worse outcome after Nivo-Ipi, regardless of PD-L1 expression. Apart from TROP2 expression, liver and bone metastases and inflammatory markers such as NLR and SII were also identified as significant predictors. Our findings suggest that the expression of TROP2 was not associated with liver and bone metastases, and inflammatory and nutritional induces. Therefore, TROP2 may be a unique predictor for the therapeutic efficacy of Nivo-Ipi treatment in patients with NSCLC. In particular, TROP2 expression could be a helpful predictor for patients with PD-L1 < 1% and non-AD histology.

The expression of TROP2 was immunohistochemically specific for tumor cells, and correlated with poor outcomes in patients with NSCLC4,5,6,8. TROP2 promotes tumor cell growth, proliferation, and metastasis by regulating the calcium signaling pathway and cyclin expression10. TROP2 overexpression has been reported in patients with breast, pancreatic, prostate, and ovarian cancers, as well as NSCLC10. Kobayashi K et al. reported that TROP2 overexpression was observed in 87 out of 130 lung adenocarcinoma cases (66%)5. TROP2 overexpression was identified in 41.8% of cases in the present study, consistent with the results of previous reports indicating TROP2 overexpression in NSCLC4,5,6. In this study, high TROP2 expression was significantly associated with lower incidence of brain metastasis. Previous studies have suggested that SCC shows a lower frequency of brain metastasis and numerically higher rates of TROP2 overexpression compared with AD11,12. In our cohort, SCC accounted for 41 out of 110 cases (37.3%), and TROP2 overexpression was numerically more frequent in SCC (48.7%) compared with AD (35.7%). According to a population-based study using the SEER–NPCR database, SCC constituted 29.4% of NSCLC cases in the United States13, indicating that the proportion of SCC in our study was relatively higher than in previous reports. Another possible explanation is that patients with high TROP2 expression may experience rapid disease progression, leading to death before developing brain metastases. Taken together, the histological composition of our cohort and the potential for rapid disease progression in TROP2-high tumors may underlie the lower incidence of brain metastases observed in these patients.

Recent studies revealed a close association between TROP2 expression and the tumor immune microenvironment, and that TROP2 overexpression resulted in reduced levels of T-cell infiltration in patients with NSCLC8,14. Bessede et al. described that TROP2 overexpression was closely related to therapeutic resistance to ICIs but not to chemotherapy in patients with advanced NSCLC8. We did not identify a significant correlation between TROP2 and PD-L1 expression, which has not been investigated in detail in previous studies. The absence of a correlation between TROP2 and PD-L1, despite TROP2’s involvement in the immune microenvironment such as in tumor-infiltrating lymphocytes (TILs), indicates that TROP2 may have a distinct and independent role in tumor immunity. Notably, the intracellular, but not the membrane, expression of TROP2 was significantly correlated with a worse outcome after PD-1 blockade monotherapy8. The functional protein domain of TROP2 is released from the membrane and accumulates within the nucleus of tumor cells8. We observed that intense TROP2 expression was closely associated with a poor outcome, whereas the proportional of TROP2 expression was not a significant biomarker for the prediction of Nivo-Ipi efficacy. As our study defined the intensity level as significant staining of intracellular and membrane lesions, an intensity score of 3 may be due to the intracellular expression of TROP2. Therefore, not only the membrane, but also intracellular expression of TROP2 could predict the resistance to Nivo-Ipi therapy in patients with advanced NSCLC.

TROP2 was associated with decreased OS in some solid tumors, affecting the tumor immune microenvironment14. The expression of TROP2 in cervical cancer was associated with increased intratumoral CD3 + /CD8 + TILs and PD-L1 expression15. Conversely, the NSCLC tumors with higher TROP2 expression had lower gene expression of immune cell markers and were linked to a negative response to ICIs8,14. Moreover, patients with advanced NSCLC exhibited reduced levels of T gamma delta, Th1 cells, and macrophages in high TROP2-expressing tumors8. Thus, overexpression of TROP2 may be associated with reduced T-cell infiltration in the tumor microenvironment in patients with NSCLC, which could contribute to ICI resistance. Our study suggests that TROP2 overexpression was a potential biomarker of therapeutic resistance to not only PD-1 blockade but also to anti-CTLA4 antibody.

ADCs targeting TROP2 are undergoing clinical trials for various tumors including lung cancer10. For example, sacituzumab govitecan has been approved for the treatment of triple-negative breast cancer, advanced urothelial carcinoma, and HR + /HER2-breast cancer10. A prospective study of TROPION-Lung02 examined the combination of Dato-DXd and pembrolizumab in 47 patients with advanced NSCLC, and identified an objective response rate of 38% and median PFS of 10.8 months16. The efficiency of TROP2-targeted ADCs is predicted to increase with TROP2 overexpression. Previous studies in breast cancer demonstrated an association between TROP2 expression and therapeutic efficiency, although a similar correlation was also demonstrated in lung cancer17. Considering the high frequency of TROP2 overexpression observed in our study, the potential effectiveness of ADCs in patients receiving Nivo-Ipi therapy is expected. Further studies are required to elucidate clinical benefits from a combination of ICIs and anti-TROP2 agents.

There are several limitations in our study. First, our sample size was limited, which may cause some bias in the results. Second, we did not examine whether the predictive significance of TROP2 expression could be different between patients receiving ICIs and those receiving chemotherapy. A previous study showed that TROP2 overexpression was a significant predictor for PD-1 blockade monotherapy efficacy8. It remains unclear whether the expression of TROP2 varies according to different therapeutic regimens. Finally, the quantitative assessment of TROP2 expression within tumor cells remains unclear. The criteria used in our study was determined based on previous studies4,5,6. However, the calculation of tumor proportion and intensity may be necessary for the accurate evaluation of TROP2 expression. Further studies are warranted to elucidate the optimal assessment of TROP2 expression using IHC. In addition, differences in treatment duration among patients may have influenced the observed incidence of irAEs, and landmark analysis was not performed in this retrospective study.

In conclusion, TROP2 expression was identified as a significant predictor for Nivo-Ipi treatment outcome in patients with advanced NSCLC. The prognostic role of TROP2 was uniform regardless of PD-L1 expression, however, TROP2 may be a unique predictor related to therapeutic efficacy in patients with PD-L1 < 1% and non-adenocarcinoma histology. Further studies, including a prospective study combining Nivo-Ipi with ADCs targeting TROP2, are required in patients with advanced NSCLC.

Methods

Patients

Between December 2020 and December 2022, 166 patients with NSCLC were treated with Nivo-Ipi therapy as a first-line treatment at our institution. Of these, tumor specimens at baseline were unavailable for 48 patients and immunostaining assessment on tissue slides was difficult for eight. Thus, 110 patients were eligible for this study, some of whom have been reported previously18. This study was approved by the Institutional Review Board (IRB) of Saitama Medical University, International Medical Center (approval number: 20–125). The requirement for written informed consent was waived by the IRB due to the retrospective study design. In accordance with the IRB-approved protocol, an opt-out method was employed, and patients were given the opportunity to decline participation after public disclosure of study details. All procedures involving human participants were conducted in accordance with the ethical standards of the institutional review board and the 1964 Declaration of Helsinki and its later amendments.

Treatment and evaluation

All patients were intravenously administered nivolumab (240 mg/day) once every 3 weeks and 1 mg/kg ipilimumab once every 6 weeks, as previously described19. The chief physicians evaluated the test results (physical examinations, complete blood counts, and biochemical tests) and graded the adverse events (AEs). Moreover, white blood cells, neutrophils, lymphocytes, and albumin were extracted within one week of the first cycle of Nivo-Ipi. Toxicity was graded based on the Common Terminology Criteria for AEs, version 4.0. Tumor response was assessed according to the Response Evaluation Criteria in Solid Tumors (RECIST), version 1.120. IrAEs were assessed from the initiation of Nivo-Ipi therapy until discontinuation of therapy due to disease progression, death, or other reasons.

Assessment of inflammatory and nutritional indices

Clinical and biological data (total protein, albumin, and C-reactive protein [CRP] levels; white blood cell, neutrophil, platelet, and lymphocyte counts; and height and weight) were extracted from medical records for analysis. Six indices reflecting systemic inflammatory and nutritional status based on previous studies21 were calculated at baseline within one week of the first cycle of each treatment. The inflammatory indices used were neutrophil to lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII) (platelet count × neutrophil count/lymphocyte count)18,22,23. The nutritional indices included prognostic nutritional index (PNI) = 10 × albumin (g/dL) + 0.005 × lymphocyte count14; advanced lung cancer inflammation index (ALI) = body mass index + albumin level (g/dL)/NLR22,23, and Glasgow prognostic score (GPS). The optimal cut-off value for the blood markers was the median value, and values greater than this cut-off were defined as high. The GPS was tabulated as follows: 0, no abnormal values (good); 1, abnormal value (intermediate); and 2, two abnormal values (poor)24. Abnormal values were defined as CRP level > 10 mg/mL and albumin < 3.5 g/dL. A GPS of 0 was defined as low, while a GPS of 1 or 2 was classified as high.

Immunohistochemistry (IHC)

The assessment and procedure for immunohistochemical staining were performed as previously described4,5,6. TROP2 expression was immunohistochemically examined by incubating tumor specimens with a rabbit monoclonal antibody against TROP2 (ab214488, clone EPR20043, high pH retrieval, Abcam, Waltham, MA, USA) at a dilution of 1:2000 in antibody diluent (Dako Glostrup, Denmark; S2022), overnight at 4 °C, followed by incubation at room temperature for 30 min. TROP2 expression was considered positive when its staining was observed in the plasma membrane and cytoplasm. The expression of TROP2 was evaluated by the intensity and proportion scores as described previously4,5,6. The intensity score was defined as score 0 (negative), score 1 (weak), score 2 (moderate), or score 3 (strong), and the proportion score presented the estimated fraction of positively stained tumor cells (1 =  < 10%, 2 = 10–50%, 3 = 50–80%, 4 = 80–100%) (Fig. 1). The immunostaining score of TROP2 was calculated as the product of a proportion score and intensity score. Therefore, the total score ranged from 0–12. Overexpression of TROP2 was defined as a total score of 12. Mouse monoclonal Ki67 (M7240 clone MIB-1,1:100, pH6 retrieval DAKO, Glostrup, Denmark) was quantified as previously described25. The cutoff value of Ki-67 expression was the median value, and high and low expression levels were defined according to this cutoff.

The sections were evaluated by at least two researchers (ST and KK) using a light microscope (× 200 and × 400 magnification) in a blinded manner. In case of discrepancies, both investigators simultaneously evaluated the slides until a final consensus was reached. The investigators were blinded to the patient outcomes.

Statistical analysis

Continuous variables were dichotomized as described above, and categorical variables were compared using the χ2 test. The statistical significance level was set at p < 0.05. PFS was defined as the time from initial treatment to disease progression or death. Overall survival (OS) was defined as the time from initial treatment to death from any cause. PD-L1 (22C3) expression in tumors was counted as tumor proportional score classified into three groups: < 1%, 1–49%, and ≥ 50%. The Kaplan–Meier method was used to estimate survival over time, with survival differences analyzed using the log-rank test. Variables that showed significant differences in the Kaplan–Meier analyses were simultaneously included in the multivariate Cox proportional hazards model. All statistical analyses were performed using GraphPad Prism (v.10.0; GraphPad Software, San Diego, CA, USA) and JMP 16.0 (SAS Institute Inc., Cary, North Carolina, USA).