Surgical and Pathological Outcomes of Lung Cancer Patients Who Did Not Radiologically Respond to Neoadjuvant Chemoimmunotherapy

1. Introduction

Perioperative and neoadjuvant chemo-immunotherapy has emerged as a promising treatment strategy for patients with resectable non-small cell lung cancer (NSCLC) at clinical stages II–III [1,2,3,4,5]. Induction chemo-immunotherapy has been shown to prolong event-free and/or overall survival in clinical trials, potentially by targeting micro-metastatic lesions in the earliest phase of the treatment. Furthermore, an additional benefit of induction chemo-immunotherapy is tumor shrinkage, which can increase the likelihood of achieving an R0 resection with adequate surgical margins and, in some patients, this will facilitate the use of less invasive surgical approaches.

Clinical trials of these perioperative and neoadjuvant treatments reported objective response rates of 53.6% (95% confidence interval: 46.0–61.1; CheckMate816 study) [6] and 56.3% (51.0–61.4; AEGEAN study) [2]. These results indicate that nearly half of patients do not achieve a radiological response after induction chemo-immunotherapy. Although perioperative and neoadjuvant chemo-immunotherapy trials have demonstrated favorable surgical outcomes, the association between radiological response and surgical outcomes remains unclear.

Conversely, it has sometimes been noted that some patients who do not demonstrate a radiological response to induction chemo-immunotherapy nonetheless exhibit a pathological response [7,8]. This discrepancy is partly attributable to the difficulty of distinguishing a viable tumor from residual fibrotic tissue, inflammation, or treatment-induced scarring on computed tomography (CT). Clinically, it would be valuable to predict pathological responses among patients who do not exhibit a radiological response to treatment.

The current study aimed to clarify these issues in patients who did not demonstrate a radiological response to induction chemo-immunotherapy, using a real-world registry database comprised of patients who received induction chemo-immunotherapy at 29 institutions in Japan (CReGYT-04 Neo-Venus study).

2. Materials and Methods

2.1. Patients

Data of patients with resectable NSCLC who received induction nivolumab plus chemotherapy between March 2023 and July 2024 at 29 centers in Japan (N = 131) were collected using an electrical data collection system reported previously (the CReGYT-04 Neo-Venus study) [9]. Clinical and pathological stage data were determined using the 8th edition of the TNM classification. This study was approved by the National Cancer Center Institutional Review Board (approval number: 2024-091 / approval date: July 11, 2024). The requirement for written informed consent from patients was waived because of the retrospective nature of this study. Anonymized patient data were electronically submitted to the registration database from each institution. Data were managed independently by biostatisticians at the Yamaguchi University Graduate School of Medicine.

2.2. Data Collection and Statistical Analysis

The radiological response was evaluated using the Response Evaluation Criteria in Solid Tumors (version 1.1) [10] based on CT examinations performed before and after induction chemo-immunotherapy. Surgery-related adverse events (AEs) of grade ≥ 3 within postoperative 30 days were recorded, and any grades of immune-related AEs (irAEs) observed postoperatively were collected from the medical records. A pathological complete response (pCR) was defined as 0% viable tumor in the resected specimen, and a major pathological response (MPR) was defined as ≤10% viable tumor in the primary tumor bed [11].

Fisher’s exact test was used to statistically evaluate associations between two categorical variables. For the continuous variables of blood tests, we employed cutoff values used in clinical practice to divide the cohort into two groups. Immune-nutritional indices that are calculated by combining blood data were classified into two groups based on cutoff values selected from within the range of various previously reported thresholds, approximately at the midpoint [12,13,14]. Trends in proportions across ordered groups were evaluated using the Cochran–Armitage trend test. Statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R (version 4.3.1 and R Commander version 2.9-1, The R Foundation for Statistical Computing, Vienna, Austria). EZR is a modified version of R commander designed to add statistical functions frequently used in biostatistics [15].

3. Results

3.1. Patient Characteristics

A flow diagram of the patients included in the analyses is summarized in Figure 1. Among 124 patients who underwent post-neoadjuvant evaluation, 84 were radiologically diagnosed with complete or partial responses (CR/PR), 35 with stable disease (SD), and 5 with progressive disease (PD). Of the 84 patients with CR/PR, 2 patients were unable to undergo surgical resection because of irAEs. Among the 35 patients with SD, 2 refused surgical resection. All five patients with PD were unable to undergo surgical resection.

3.2. Surgical Outcomes

3.2.1. Surgical Procedures

Surgical outcomes were compared between 33 patients with SD and 82 patients with CR/PR. Surgical procedures were compared between these two groups after excluding one SD patient who underwent exploratory thoracotomy (Table 1). Patient characteristics, such as age, sex, and smoking status, were similar between the SD and CR/PR groups. The rates of thoracotomy rather than less invasive approaches were also similar between the groups (53% in SD vs 52% in CR/PR). The application of bronchial and/or pulmonary artery plasty was also comparable (19% in SD vs 16% in CR/PR), whereas the rates of bi-lobectomy or pneumonectomy were numerically higher in the SD group (19%) compared with the CR/PR group (10%), although this difference was not statistically significant (p=0.21). There were no significant differences in operation time or blood loss between the groups.

3.2.2. Residual Tumor (R) Classification

In the analysis of residual tumor (R) classification, after excluding 1 SD patient without available pathological data, 88% (28/32 patients) of SD patients and 98% (80/82 patients) of CR/PR patients achieved R0 resection. In the SD group, one patient (3%) had an R1 resection and three patients (9%) had R2 resections, whereas two patients (2%) in the CR/PR group were diagnosed with R1 resection. The distribution was shown to differ significantly when using the Cochran–Armitage trend test (Figure 2A, p=0.0084).

3.2.3. Adverse Events

An analysis of the postoperative clinical course indicated that 5 patients (16%) in the SD group and 15 patients (18%) in the CR/PR group experienced a surgery-related AE (grade ≥ 3) within 30 days after surgery (Table 1). Although the overall frequencies of surgery-related AEs were similar between the groups, we observed some differences in the types of AEs (Figure 3). The frequencies of empyema (2.4% vs 0%), pneumonia (8.5% vs 3.1%), chylothorax (2.4% vs 0%), and prolonged air leak (9.8% vs 3.1%) were numerically higher in the CR/PR group, whereas the frequency of atelectasis (2.4% vs 9.4%) was higher in the SD group. Despite some numerical differences between the groups, we did not perform statistical tests because of the low number of events. We also found that the frequencies of postoperatively occurring irAEs (any grade) were similar between the groups (13% in both groups, Table 1).

3.3. Pathological Evaluation Within SD Group

Pathological data were available for 31 SD patients after excluding 1 with no submitted pathological data and 1 who underwent exploratory thoracotomy. In comparison with 82 CR/PR patients, we observed a significant difference between the groups (Figure 2B, p<0.001). Although 74% of SD patients showed a non-MPR, it is noteworthy that 26% had an MPR, including 10% with pCR.

Next, we explored which clinical factors were associated with MPR versus non-MPR among patients who had radiological SD after neoadjuvant chemo-immunotherapy. There were no significant differences in age, sex, body mass index, clinical T and N factors, or histology between the MPR and non-MPR groups, whereas the pre-treatment PD-L1 status was significantly associated with a pathological response (Table 2). None of the radiological non-responders with < 1% PD-L1 achieved MPR, and 4 of 5 patients with high PD-L1 (≥ 50%) achieved an MPR despite the absence of a radiological response. Smoking data were not included because most patients were smokers (Table 1). The magnitude of the radiological response (tumor shrinkage rate ≤ 10% vs 10%–29%) did not correlate with the pathological response (p=0.69), whereas a decrease in the maximum standardized uptake value (SUVmax) of the primary tumor (≥ 50% vs < 50%) after neoadjuvant chemo-immunotherapy was significantly associated with a pathological response (p=0.03), although only about half of the patients received positron emission tomography (PET)-CT after neoadjuvant chemo-immunotherapy. We observed that 5 out of 9 patients (56%) with a ≥ 50% decrease in the SUVmax had an MPR (including pCR), whereas none of the patients with a < 50% decrease did so (Table 2).

3.4. Immune-Nutritional Indices

We also explored immune-nutritional indices between patients with an MPR and non-MPR individuals among SD patients (Table 3). We observed that none of the examined indices, including the neutrophil-to-lymphocyte ratio, pan-immune-inflammation value, or prognostic nutrition index, was significantly associated with a pathological response.

4. Discussion

This multicenter retrospective study evaluated the surgical and pathological outcomes after neoadjuvant chemo-immunotherapy, with a particular focus on patients who did not achieve a radiological response. First, we observed that 4% of the evaluable patients did not receive surgery because of PD. This underscores the importance of informing patients about the potential risk of surgery cancellation related to PD, although this risk in real-world settings is as low as that observed in clinical trials [1,16,17].

Most radiological SD patients received surgery; however, the R0 resection rate was substantially lower (88%) compared with the CR/PR group (98%). We observed that other surgical outcomes—including surgical procedures, applications of bronchial and/or PA plasty, operation time, and blood loss—were similar between the groups, suggesting that a radiological response alone may not predict surgical complexity or technical feasibility. The factors associated with the risk of conversion to open thoracotomy, such as the presence of pre-treatment nodal disease in a previous study [18], are clinically important; however, we did not analyze this because there were only four patients with conversion—two in the SD group and two in the CR/PR group. Surgery-related AEs and irAEs did not differ significantly between the SD and CR/PR groups, although some severe complications were numerically frequent in the CR/PR group, indicating the need for extensive postoperative management, regardless of radiological response.

Although SD is generally considered a suboptimal, or absence of, response to the treatment, it is of note that 26% of SD patients achieved an MPR, including pCR in 10% of SD patients, at pathological evaluation, highlighting that radiological assessment may underestimate the true biological efficacy of neoadjuvant chemo-immunotherapy. Because the pre-surgical identification of patients with an MPR among the SD group would be clinically beneficial, we explored the clinical predictors of an MPR in patients who showed radiological SD after neoadjuvant chemo-immunotherapy. Although the magnitude of radiological shrinkage (< 10% vs 10%–29%) or immuno-nutritional indices were not associated with a pathological response, we found that the pre-treatment PD-L1 status and a decrease in the SUVmax of the primary tumor (≥ 50% vs < 50%) were significantly associated with a pathological response. These observations align with better pathological responses in patients with higher PD-L1 status in clinical trials [2,6] as well as emerging literature supporting the use of metabolic imaging to assess immunotherapy responses [19,20,21]. Therefore, we consider that pre-treatment PD-L1 evaluation is useful for predicting pathological responses in cases of radiological SD. In addition, we consider that post-neoadjuvant PET-CT is also beneficial in these patients, not only to detect unexpected distant metastasis, but also to predict pathological responses.

Our study had several limitations. The retrospective design and modest cohort size, particularly for patients with radiological SD, may have limited statistical power. Although the exploratory analysis suggested trends in immune-nutritional indices among MPR versus non-MPR patients (Table 3), these did not reach statistical significance; therefore, larger studies are needed to validate these potential biomarkers.

5. Conclusions

In conclusion, although approximately 18% of radiological non-responders did not proceed to surgery, the surgical outcomes were comparable between the SD and CR/PR groups. Among those who received surgical resection, 26% of SD patients achieved MPR, and a ≥ 50% decrease in the SUVmax predicted MPR, supporting the role of metabolic imaging for pre-surgical evaluation.