AATS 2026 关注

LB4. Patterns and Predictors of Recurrence Following Neoadjuvant Chemoimmunotherapy and Surgery in Non-Small Cell Lung Cancer

• Abstract Presenter:Nanruoyi Zhou, Memorial Sloan Kettering Cancer Center
  

Objective: Neoadjuvant chemoimmunotherapy (chemoIO) is standard of care for locally advanced, resectable NSCLC without targetable EGFR/ALK alterations. However, little is known regarding predictors of recurrence and recurrence patterns after surgery. To address this knowledge gap, we investigated clinicopathologic risk factors associated with recurrence following neoadjuvant chemoIO and surgery in this patient population.

Methods: This is a multicenter, international, retrospective cohort study of patients with clinical stage IA3-III NSCLC without known targetable EGFR/ALK alterations who underwent neoadjuvant chemoIO and surgical resection from 1/2017-3/2024. Exclusion criteria were prior history of lung cancer, treatment with investigational immunotherapies or other neoadjuvant treatments. The primary endpoint was recurrence-free survival (RFS), which was measured from date of surgery and estimated using the Kaplan-Meier approach with Cox proportional hazards models for univariable and multivariable analysis.

Results: 291 patients with a median follow-up of 29.3 months (95% CI 27.6-31.5 months) were included. Patients were treated with regimens containing anti-PD-(L)1 inhibitors in a neoadjuvant-only (247, 85%) or perioperative (44, 15%) approach. Clinical N2 disease was present in 119 (41%) patients. R0 resection was achieved in 281 (97%) cases and 182 (63%) cases were completed minimally-invasively. A major pathologic response (MPR) was seen in 141 (48%) cases and a pathologic complete response (pCR) was seen in 84 (29%) cases. Median residual viable tumor (RVT) was 10% (IQR 0-60%). Following surgery, 72 (25%) patients received adjuvant therapy. Of these, 14 (19%) received post-operative radiation (PORT) alone, 45 (63%) received systemic therapy alone, and 13 (18%) received a combination of systemic therapy and PORT. Systemic therapy regimens included chemotherapy alone (13/72, 18%), immunotherapy alone (37/72, 51%), and chemoIO (7/72, 10%) for a median of 4.5 cycles (range 1-14). A total of 90 recurrences were observed during the study period, of which 35 (39%) were locoregional, 36 (40%) were distant, and 19 (21%) were both locoregional and distant. The most common site of locoregional recurrence was lymph nodes (33/90, 37%). The most common site of distant recurrence was the brain (19/90, 21%). In patients with a pCR, 9/84 (11%) experienced recurrences, with 5/9 (56%) locoregional recurrences in lymph nodes. Seven patients with 90-day postoperative mortalities and 10 patients with R1 resections were excluded from RFS analyses. The 2-year RFS rate was 71.6% (95% CI 66.2-77.4%). On univariable analysis, never smoking status, cN2 disease , adenocarcinoma histology, increasing primary tumor invasive size, increasing ypN stage, increasing RVT%, absence of MPR, absence of pCR, lymphovascular invasion, visceral pleural invasion, spread through airspaces, and PD-L1<1% were associated with worse RFS (p<0.05). On multivariable analysis, never smoking status, absence of MPR and residual nodal disease independently predicted worse RFS (Table 1).

Conclusions: Despite high rates of MPR and pCR, recurrence occurred in nearly one-third of patients by two years after surgery. Never smoking status, absence of MPR, and residual nodal disease were independent predictors of recurrence. These findings support prospective evaluation of risk-adapted postoperative treatment strategies, including surveillance, in this patient population.

RF76. R+ Resection for Clinical Stage III Non-Small Cell Lung Cancer: Should We Have Just Given Chemoradiation?

• Abstract Presenter:Grace Ha, Montefiore Medical Center
  

Objective:
The optimal treatment standard for patients with clinical stage III (cIII) non-small cell lung cancer (NSCLC) is debated, with resectability playing a critical role. While surgeons must try to avoid R+ resections, chemoradiation is also known to frequently leave patients with residual local disease. The objective of this study is to compare outcomes in patients with cIII NSCLC treated with chemoradiation (CRT) versus surgery with positive margins.

Methods:
From the National Cancer Database, we identified patients with cIII NSCLC (excluding N3) treated with either CRT or surgery with an R1 outcome (SR1) (2004-2022). Logistic regression evaluated predictors of treatment modality. Survival in matched groups was analyzed by Kaplan-Meier and Cox regression.

Results:
Of 72,908 patients, 96.5% (n=70,324) received CRT and 3.5% (n=2,584) underwent SR1. The SR1 group accounted for 6.2% of those patients who underwent surgery for cIII disease (n=41,873) and was associated with left-sided (aOR 1.13, 95% CI 1.02-1.25), middle lobe (aOR 1.60, 95% CI 1.29-1.98), and lower lobe (aOR 1.33, 95% CI 1.19-1.48) disease. CRT use was associated with increasing age (aOR 1.01, 95% CI 1.01-1.02), cT4 disease (aOR 1.69, 95% CI 1.47-1.95), and node-positivity (aOR 4.24, 95% CI 3.62-4.95). A total of 541 (20.9%) SR1 patients had neoadjuvant therapy, including 166 (n=6.2%) with immunotherapy. The majority of resections were open (57.8%, n=909) and the most common operation was a lobectomy (66.6%, n=1722). Of the SR1 cohort, 9.4% (n=244) had nodal upstaging and 51.9% (n=1339) had identical clinical and pathologic nodal stage. Adjuvant therapy rates in the SR1 group were 51.4% (n=1329) for systemic therapy and 43.3% (n=1118) for radiation. In a matched cohort comparing CRT and SR1 groups, SR1 was initially associated with worse survival (71.5% vs. 75.2%) at one year but improved survival after 2.5 years (46.9% vs. 45.2%) (HR 0.92, 95% CI 0.85-0.99, p=0.04; Figure 1).

Conclusions:
Survival after incomplete resection appears comparable to that seen in patients undergoing nonsurgical therapy, suggesting that surgery could be reasonably considered over CRT even in marginally resectable patients. With higher pathologic response rates identified in the era of neoadjuvant chemoimmunotherapy, it is anticipated that patients with cIII NSCLC will increasingly be offered surgical management.

RF77. The Prognosis of Single-site Oligometastatic Non-Small Cell Lung Cancer Treated with Local Therapies in the Era of Immunotherapy

• Abstract Presenter:Gabriela Esnaola, Yale School of Medicine
  

Objective
Systemic immunotherapy (IO) has been shown to improve survival in patients with non-small cell lung cancer (NSCLC). Single metastatic sites have a more favorable prognosis than multiple sites, and patients with single-site metastatic NSCLC have historically been considered candidates for local therapy (LT) in the form of radiation or surgery. However, most studies demonstrating a possible benefit to local therapy originated prior to widespread IO adoption, and a recent randomized trial has shown no advantage for LT in patients who received IO with less than three metastatic sites. National survival outcomes for patients receiving LT and systemic therapy (ST) for lung cancer with single metastatic sites in the era of immunotherapy remain poorly described. We used the National Cancer Database (NCDB) to assess how site of metastasis and the use of LT in combination with systemic IO is associated with prognosis of patients with single-site metastatic NSCLC in the United States (US).
Methods
The NCDB (2018–2023) was used to retrospectively evaluate patients with single-site synchronous oligometastatic NSCLC (clinical stage IVA) treated with IO. Patients were stratified based on metastatic site and treatment type: systemic IO with or without LT (IO+LT or IO). All patients could have additionally received chemotherapy (CTX). Prognosis was evaluated using Kaplan-Meier survival analysis landmarked at the median time from diagnosis to LT initiation.
Results
Among included patients (N=4,149), the most common sites of metastasis were the lung (N=990; 23.9%), brain (N=984; 23.7%), bone (N=717; 17.3%), distant lymph node (N=272; 6.6%), liver (N=152; 3.7%), and other sites (N=1,034; 24.9%). 992 patients (23.9%) were treated with IO and 3,157 (76.1%) with IO+LT (79.4% of all patients received CTX, 62.9% radiation, and 14.2% surgery). Poor OS was associated with bone (median survival 22.6 months, 25.5% 5-year survival) and liver (median survival 16.7 months, 12.0% 5-year survival) metastases compared to brain, lung, or nodal metastases (median survival 28.4 months, 31.5% 5-year survival). Significantly longer survival was associated with patients selected for LT (Figure 1).
Conclusions
The addition of local therapy to a systemic regimen containing immunotherapy was common for patients with single metastatic sites in the US. Patients selected for local therapy had a favorable prognosis, which varied by metastatic location.

RF78. Feasibility of Panitumumab-IRDye800 molecular imaging for intraoperative tumor localization of lung cancer

• Abstract Presenter:Carolyn C. Chang, Stanford University Medical Center
  

Objective: Panitumumab-IRDye800 (pan800) is a molecular imaging agent composed of panitumumab, an epidermal growth factor receptor (EGFR) antibody, conjugated to IRDye800, a near-infrared fluorescent dye. We aimed to evaluate EGFR expression in non-small cell lung cancer (NCSLC) and determine the feasibility of pan800 molecular imaging for intraoperative tumor localization. Methods: We conducted a preclinical study to assess EGFR expression in NSCLC compared to normal lung in surgical specimens. Subsequently, patients with NCSLC scheduled for surgical resection were enrolled in a single-center, phase 1, open-label, dose-escalation study (NCT03582124). Patients received pan800 infusion at a range of doses and days before surgery. Surgeons used white light, then pan800 molecular imaging, for tumor localization; surgical resection was then performed per standard of care. The primary objective was to determine the optimal dose and timing for intraoperative tumor localization, with secondary endpoints including quantification of ex vivo tumor fluorescence. Results: In our preclinical study, we found significantly higher EGFR expression in tumor versus normal lung in 12 adenocarcinoma (51.8% vs. 14.6%; p<0.0001) and 14 squamous cell carcinoma (69.5% vs. 9.6%; p<0.0001) specimens (Figure 1A). Subsequently, 11 patients with adenocarcinoma were enrolled and received a pan800 infusion (25, 50, or 75 mg) 1–5 days before surgery (Figure 1B-G). Four (36.4%) patients had successful intraoperative tumor localization using pan800 imaging; the median in vivo tumor-to-background ratio (TBR) was 2.70 [IQR 2.27–3.03], ex vivo TBR 3.34 [IQR 2.92–3.46], tumor size 2.3 cm [IQR 1.3–3.1], and tumor-to-pleura distance 0.1 cm [IQR 0.0–0.2]. Seven (63.6%) patients had no intraoperative fluorescence; median ex vivo TBR 2.21 [1.67-2.7], tumor size 2.2 cm [IQR 1.0–2.6] and tumor-to-pleura distance 1.9 cm [IQR 0.95–3.0], the latter significantly higher (p=0.01). The optimal dose was determined to be 50 or 75 mg, and timing of 2-4 days before surgery; 4/7 (57.1%) patients who received pan800 infusion in this range had successful intraoperative tumor localization. Conclusions: Our study demonstrates that NSCLC overexpresses EGFR suggesting pan800 is a promising imaging agent for intraoperative tumor localization. Preliminary findings from our clinical trial suggest an optimal dose of 50-75 mg and timing at 2-4 days before surgery, and enrollment is ongoing to refine these parameters

RF79. Radiographic Tumor Regression as a Predictor for Pathologic Response after Neoadjuvant Therapy for Non-Small Cell Lung Cancer

• Abstract Presenter:Justin Bader, Yale New Haven Hospital
  

Objective: Evaluate whether radiographic changes in non-small cell lung cancer (NSCLC) correspond to pathologic response in patients who undergo neoadjuvant therapy.

Methods: Patients with stage IB-IV NSCLC who received neoadjuvant therapy and surgery between 2015-2025 in a large health system were included. Salvage surgery cases or surgery for oligoprogression were excluded. Chart review collected demographic, pathologic, treatment, and radiographic data. Per RECIST 1.1 CT scan criteria, patients were stratified by radiographic tumor size decrease ≥30% from baseline (objective radiographic response, ORR) vs tumor size decrease <30% (radiographic stable disease). Tumor size was defined by largest diameter on CT, and radiographic response was defined by difference between tumor size on CT prior to neoadjuvant therapy initiation and CT prior to surgical resection. Pathologic response to neoadjuvant therapy was measured by expert pathologist review using major pathologic response (MPR) ≤10% viable tumor and complete pathologic response (pCR) as 0% viable tumor.

Results: Among patients with neoadjuvant therapy, 62% (53/86) had adenocarcinoma, 31% (27/86) had squamous cell carcinoma, and 7% (6/86) had other histology. Neoadjuvant therapy included 39% (35/90) chemotherapy only, 9% (8/90) immunotherapy only, and 52% (47/90) chemotherapy and immunotherapy. ORR occurred in 43% (39/90). Radiographic progression prior to surgery was seen in 11% (10/90). Tumors with ORR were significantly more likely than radiographic stable disease to have lower percentage of viable tumor (21 vs 32, p=0.05) and higher rates of pCR (33%(10/30) vs 11%(5/47), p=0.02) after treatment (Table I). Logistic regression showed tumors with ORR were 3.9x more likely to achieve pCR compared to tumors with radiographic stable disease (Odds Ratio: 3.9, p=0.03). However, when analyzing tumors with the highest radiographic response (>60% response), 20% (1/5) had residual viable tumor, 20% (1/5) had pCR, and an additional 60% (3/5) had MPR.

Conclusions: NSCLC patients with radiographic tumor response ≥30% per RECIST criteria after neoadjuvant therapy were significantly more likely to achieve complete pathologic response. However, among patients with significant radiographic response, there remain many patients without pCR. Further studies, radiomic markers, and long-term recurrence data are needed to explore if radiographic response can serve as a viable predictor of pathologic response.

RF81. Uninvolved vs Metastatic Tumor-Draining Lymph Nodes in NSCLC: Distinct Immune Ecologies Inform Selective Dissection and Immunotherapy

• Abstract Presenter:Yang Zhang, Fudan University Shanghai Cancer Center
  

Objective:
To define immunologic differences between uninvolved tumor-draining lymph nodes (uiLNs) and metastatic lymph nodes (metLNs) in non-small cell lung cancer (NSCLC) and assess their implications for selective mediastinal lymph node dissection and neoadjuvant PD-1 blockade.
Methods:
A translational cohort comprising 106 matched specimens (primary tumors, uiLNs, metLNs, and blood) from treatment-naïve patients and patients receiving neoadjuvant PD-1 blockade underwent single-cell RNA sequencing with TCR/BCR profiling and spatial transcriptomics. Immune composition, functional states, clonal sharing along the LN–tumor axis, and spatial neighborhoods were analyzed; key findings were functionally tested with adoptive transfer.
Results:
uiLNs and metLNs exhibited fundamentally different immune ecologies. uiLNs were enriched for activated CD8⁺ T cells and dendritic cells with higher cytotoxic/effector signatures, greater TCR clonality, and progenitor phenotypes, whereas metLNs showed increased Tregs, exhaustion programs, and fibroblast/epithelial enrichment indicating tumor infiltration. After PD-1 blockade, a CD200⁺ progenitor-exhausted CD8 subset (Tpex) expanded markedly in uiLNs (≈2.2-fold), but not equivalently in metLNs. Clonal sharing of Tex/Tpex between uiLNs and tumors was significantly higher in pathologic complete responders, linking uiLN-derived clones to intratumoral effector replenishment, while metLNs contributed fewer shared, less functional clones. Spatial modeling identified tumor-resident C1QA⁺ dendritic cells that secrete CXCL9/10 and colocalize with Tex to form immunostimulatory niches favoring uiLN-to-tumor recruitment. In contrast, metLNs contained fibroblast-enriched neighborhoods (TpCN5) with extracellular-matrix barriers that diminished CXCL9 diffusion and physically restricted Tpex ingress. In vivo, transferred CD200⁺ Tpex differentiated into effectors and synergized with PD-1 blockade to suppress tumor growth.
Conclusions:
uiLNs function as immunologically competent reservoirs that generate and export tumor-reactive T-cell clones under PD-1 blockade, whereas metLNs are immunosuppressive, barrier-forming sinks. These divergent roles support a selective dissection paradigm-resect metLNs while preserving uiLNs-to minimize morbidity and maintain the immune machinery that augments neoadjuvant immunotherapy in operable NSCLC.

147. Quality of Life Beyond Five Years After Lung Cancer Resection: An Analysis of the Boston Lung Cancer Study

• Commentator:Cherie Erkmen, Temple University Health Systems
• Abstract Presenter:Alexandra Potter, Massachusetts General Hospital
  

OBJECTIVE
Patient-reported quality of life (QOL) beyond five years after lung cancer resection is largely unknown. The objective of this study was to evaluate QOL among long-term lung cancer survivors who underwent lung cancer resection in the Boston Lung Cancer Study (BLCS).

METHODS
The BLCS is a cancer epidemiology cohort study that recruited patients newly diagnosed with lung cancer in Boston, MA from 1993-2025. Beginning in 2024, the BLCS administered the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C301 to all surviving patients in the cohort. For the present study, we identified BLCS participants who were sent the EORTC QLQ-C30 questionnaire >5 years after they underwent lung resection for stage I lung cancer. We evaluated QOL scores on each of the different scales in the study cohort and compared these QOL scores to the average QOL scores from a representative sample of the general U.S. population aged ≥70 years; this age group was selected to most closely match the age of patients in our cohort.

RESULTS
Among 564 patients who met inclusion criteria and were sent the EORTC QLQ-C30 survey at least five years after lung cancer surgery, 200 (35.4%) responded. Among participants who responded, most were female (67.0%). The median age at lung cancer resection was 65.8 (IQR: 58.1, 71.5) years, and the median time from surgery to survey completion was 8.6 (IQR: 6.0-11.0) years.

In the overall cohort, the mean (SD) global health score was 69.2 (19.5); this was numerically higher than that among individuals of a comparable age in the general U.S. population (mean: 64.3, SD: 21.9). The overall cohort had either similar or better QOL when compared individuals of a comparable age in the general U.S. population across the other 13 scales (Figure).

In a propensity score-matched analysis of 46 patients who underwent lobectomy and 46 who underwent sublobar resection, there were no statistically significant differences in QOL life scores on the global health score or the other 13 scales.

CONCLUSIONS
In this analysis of long-term survivors after lung cancer resection in the Boston Lung Cancer Study, patients who underwent lung cancer resection reported similar or slightly better QOL compared to comparably aged individuals in the general U.S. population. Long-term patient-reported QOL was similar between patients who underwent sublobar resection versus lobectomy.

148. Impact of Neoadjuvant Chemo-Immunotherapy on Lung Cancer Surgery Peri-Operative Outcomes in a Statewide Quality Collaborative

• Abstract Presenter:Andrew Popoff, Henry Ford Health
• Commentator:Varun Puri, MD, MSCI, Barnes Jewish Hospital
  

Objective: Recent neoadjuvant and perioperative chemo immunotherapy (CI) regimens for non-small cell lung cancer (NSCLC) have dramatically improved oncologic outcomes for patients. Studies regarding the impact of these regimens on surgical outcomes are limited. We hypothesized that neoadjuvant CI would not have an impact on surgical outcomes at a statewide level.

Methods: A retrospective review of prospectively collected data was conducted for all Stage IB and greater patients undergoing curative intent resection for NSCLC from July 1, 2023 to June 30, 2025 at 21 centers in a statewide quality collaborative. Outcomes included operative time, surgical approach and conversion rate, post operative morbidity, length of stay (LOS), ICU admission, and 30-day mortality. Fisher's exact test was used to compare cohorts.

Results: A total of 570 patients received surgery, of which 122 (21.3%) received neoadjuvant CI. In the neoadjuvant group, 43% (52) were female. The mean age was 67 in both cohorts. The most common resection performed was lobectomy, 113 (92.62%), followed by pneumonectomy 8 (6.6%), with the remainder undergoing wedge resection. Those who received CI were more likely to undergo an open operation, 32.79% (40) vs 17.1% (77) (p=0.0002), had a higher conversion to open rate 13.9% (17) vs 7.3% (33) (p < 0.02), and had longer mean operative times, 244 mins vs 202 mins (p <0.0001), Table. Transfusion, ICU and reoperation rates were similar between cohorts. Rates of major post operative morbidity were similar between groups, 8.2% (10) vs 7.4% (33) (p = 0.14) including prolonged air leak 15.6% (19) vs 12.7% (57) (p = 0.08) and atrial arrhythmia 13.9% (17) vs 8.2% (37) (p=0.06). While readmission rates were similar 10.9% (13) vs 9.2% (41) (p=0.11), the neoadjuvant CI cohort had a longer mean LOS 5.5 days vs 4.8 (p = 0.002), and higher 30-day mortality, 4.9% (6) vs 0.89% (4) (p = 0.007).

Conclusion: Neoadjuvant CI in resectable NSCLC is associated with more open surgery, more conversions, and longer operative times. While major post operative morbidity is similar to those not receiving neoadjuvant CI, LOS is slightly longer and perioperative mortality is worse.

150. Septum-Guided Segmentectomy for Deep Early-Stage Non-Small Cell Lung Cancer: Equivalent Outcomes to Lobectomy Without Margin Dependence

• Abstract Presenter:Qing Wang, Shanghai Chest Hospital Affiliated to Shanghai Jiaotong University
• Commentator:Eric Grogan, Vanderbilt University Medical Center
  

Objective
Segmentectomy has emerged as a lung-preserving alternative to lobectomy for peripheral early-stage non-small cell lung cancer (NSCLC). However, its role in tumors located in the inner two-thirds of the lung parenchyma remains controversial under margin-dependent paradigms. This study evaluated the oncological safety and feasibility of uniportal video-assisted thoracoscopic surgery (VATS) septum-guided segmentectomy, a novel approach utilizing intersegmental septa and membranous venous sheaths as natural anatomical boundaries.
Methods
We retrospectively analyzed 492 patients with cT1abN0M0 non-small cell lung cancer in the inner two-thirds of the lung parenchyma who underwent uniportal VATS septum-guided segmentectomy or lobectomy between January 2019 and March 2025. For septum-guided segmentectomy, the intersegmental plane was delineated with meticulous dissection of intersegmental veins and membranous sheaths serving as natural boundaries. Segmental bronchus and artery were divided at the hilum, and parenchymal dissection was performed along venous planes using energy devices centrally and staplers peripherally to preserve lung tissue. After applying inclusion criteria and propensity score matching (PSM), 344 patients (before PSM) and 260 matched patients (after PSM) were included. Perioperative outcomes, recurrence-free survival (RFS), and overall survival (OS) were assessed with a median follow-up of 24 months.
Results
Before PSM, 7 recurrences or deaths occurred in the lobectomy group, whereas none were observed in the segmentectomy group. Survival analysis showed no significant difference in RFS or OS between groups (p = 0.11 and 0.20 before PSM; p = 0.24 and 0.41 after PSM). Segmentectomy required longer operative time (131 vs. 98 minutes; p < 0.001), but all cases achieved R0 resection. In the segmentectomy group, 92 margin distances were recorded, 32 of which did not meet NCCN criteria (>20 mm or tumor diameter); distances ranged from 2–40 mm (median 15 mm). Perioperative complications and hospital stay were comparable between groups (all p > 0.05).
Conclusions
Uniportal VATS septum-guided segmentectomy provides oncological outcomes equivalent to lobectomy for deeply located early-stage NSCLC. By leveraging intersegmental veins and membranous septa as anatomical boundaries, this approach eliminates reliance on fixed margin distances and establishes a precise, tissue-preserving paradigm for complex parenchymal resections.