Dropout after Neoadjuvant Chemoimmunotherapy for Non-Small Cell Lung Cancer in the Veterans Health Administration

Abstract

Objectives:

Multiple studies have demonstrated excellent survival for patients with advanced non-small cell lung cancer who undergo induction chemoimmunotherapy followed by surgery. However, up to 22% of patients in these studies did not undergo surgical resection. National Veterans Health Administration data outlining the uptake, utilization, and dropout in this cohort are limited. We sought to describe utilization, non-progression to surgery, and outcomes in this cohort.

Methods:

Using the United States Veterans Health Administration database, we identified patients with clinical stage (cStage) IB-III-N2 non-small cell lung cancer who underwent chemoimmunotherapy (2021–2024). All patients received at least 2 cycles CIO. Exclusion criteria included radiation during chemoimmunotherapy, clinical N3 disease, >4 cycles chemoimmunotherapy, or cStage IV disease. Clinicopathological characteristics and treatment patterns were described. Overall survival (OS) was estimated using the Kaplan-Meier method.

Results:

In total, 86 patients met final inclusion criteria: 60 (69.8%) underwent surgery and 26 (30.2%) did not. The groups had similar cStage, age, and comorbidities. Among patients who did not undergo surgery, 46.2% developed distant metastatic disease during induction therapy. The 2-year OS was 80.2% for the entire cohort (86.7% surgical; 65.8% non-surgical).

Conclusions:

In our nationally representative cohort, patients undergoing chemoimmunotherapy with intent for surgical resection had favorable intermediate-term survival. however, nearly one-third of patients did not undergo surgery, most commonly due to disease progression. These findings highlight the need to better identify patients who do not progress to resection and define management strategies for these patients.

Introduction

Lung cancer is the most common type of cancer, accounting for the most deaths amongst all cancers, with an approximately 227,000 diagnoses and 125,000 deaths in the United States in 2025.¹ Despite significant improvements in therapy, survival remains poor, with recent data demonstrating a 5-year overall survival of 69% for stage IB disease and as low as 31% for stage IIIB non-small cell lung cancer (NSCLC).²

More recently, multiple randomized controlled trials have demonstrated that induction chemoimmunotherapy followed by surgery for stage IB-III-N2 NSCLC results in dramatic improvements in survival versus chemotherapy alone.^3–6 These studies demonstrated similar 2-year OS of around 80% among patients who underwent neoadjuvant chemoimmunotherapy versus around 65% for chemotherapy alone.^3–6 Subsequently, utilization of neoadjuvant therapy prior to surgical resection for advanced NSCLC has become a new standard.

One potential issue with offering neoadjuvant therapy is that patients may not undergo surgical resection. Across trials evaluating induction chemoimmunotherapy for stage IB-III NSCLC, 18–23% of patients did not proceed to surgery for various reasons.^3–6 As the uptake of chemoimmunotherapy as an induction method becomes increasingly popular, it is important to understand how these trial results translate into clinical practice. Using a large, nationally representative cohort within the Veterans Health Administration (VHA) system, we aim to evaluate the implementation of neoadjuvant chemoimmunotherapy within the VHA, characterize the patient population selected for this approach, describe patients who do not progress to surgical resection, and evaluate outcomes of patients who undergo surgery and those who do not. This study is intended as a descriptive hypothesis-generating analysis of utilization patterns and outcomes within the VHA rather than a comparative or causal evaluation of treatment strategies.

Methods

We performed a retrospective study of Veterans within the VHA, the largest integrated healthcare system in the United States. Relevant patient-, treatment-, and tumor-related covariates were abstracted from the Veterans Affairs Corporate Data Warehouse (CDW) which captures clinical and administrative data across VHA facilities nationwide. Clinical and administrative data elements were queried from the CDW and the Veterans Affairs Surgical Quality Improvement Program (VASQIP) database and subsequently organized using the VHA Informatics and Computing Infrastructure (VINCI) system as performed in multiple previous publications by our group.^7–11 We obtained death data from the Death Ascertainment file. Natural language processing, manual data extraction, and cross-reference of multiple datasets were performed to validate the extracted data.

The St. Louis Veteran’s Health Administration Research and Development Committee (#1214632, August 2, 2019) and Institutional Review Board approved the study protocol. Consent was waived given the deidentified nature of the analyses. The study was conducted in accordance with the Declaration of Helsinki (revised 2013).

Veterans with clinical stage IB-III-N2 NSCLC (as defined by ICD-O-3 histology and site codes) were identified withing the CDW using structured oncology and administrative data between 2021–2024. All VHA Medical Centers throughout the United States were included in our initial query for potential patients. Patients were included if they had been evaluated by a surgeon and received at least two cycles of platinum-based chemotherapy in conjunction with immunotherapy, consistent with neoadjuvant or perioperative checkpoint blockade strategies.^4–6 We required at least two cycles of therapy to ensure meaningful exposure to induction chemoimmunotherapy and to reduce potential misclassification of patients who may have received only a single dose for metastatic disease but stopped due to early toxicity, treatment interruption, or documentation variability within administrative datasets. Patients who were not assessed for potential resection by a surgeon were not included in this cohort. We excluded patients who received neoadjuvant radiation, who had >4 cycles of chemoIO prior to surgery (a regimen not used in the clinical trials), or those with clinical N3 disease to align with previous phase III randomized trials.^4–6

Extracted patient variables included patient demographics (age, sex, and race), clinical characteristics, tumor features, and treatment-related variables. Medical comorbidities were determined using ICD-9 and ICD-10 codes and used to calculate a composite Charlson Comorbidity Index (CCI). Tumor variables included tumor location, size, and histopathology. Treatment variables included timing of therapy and interval between treatments. For patients that underwent surgery, operative variables including surgical approach, type of resection, margin status. Perioperative outcomes were obtained from the VASQIP and CDW Surgery files. Oncology treatment data, including type of chemotherapy or immunotherapy and number of cycles received, were extracted from using the CDW Oncology Raw domain.

The primary aim of this study was to describe the characteristics and outcomes of Veterans undergoing induction chemoimmunotherapy with intent for surgical resection. We described outcomes among patients who did not undergo resection (overall survival, reasons for non-progression to surgery, and subsequent therapies received) and well as those who underwent resection (pathological response, perioperative outcomes, overall survival). Overall survival (OS) was defined as time from initiation of chemoimmunotherapy to death or last follow-up (Aug 15, 2025). OS was estimated using the Kaplan-Meier method. Descriptive comparisons between groups were performed. Continuous variables were summarized using means (standard deviations) and compared using Kruskal-Wallis Test, while categorical variables were summarized as frequencies (proportions) were compared using χ² tests. All statistical tests were two-tailed, and p-values < 0.05 were considered statistically significant. All analyses were performed in SAS version 9.3.

Results

In total, 86 patients met the final inclusion criteria, and 42 different VHA medical centers contributed at least 1 patient to the dataset (Figure 1). Of these 86 patients, 26 patients (30.2%) did not undergo surgery, while 60 patients (69.8%) underwent surgery. In the entire cohort, the majority of patients were stage III (N=47, 54.7%). However, some had clinical stage IB (N=4, 4%) and stage II disease (N=35, 40.7%). Overall, 31.4% had clinical N2 disease, while the majority had cN0 disease (N=37, 43.0%). The majority of patients had adenocarcinoma (N=47, 54.7%) and were male (N=80, 93%) and White (N=64, 74.4%) (Table 1). Median follow-up was 2.51 years (Interquartile range, 2.17–3.09).

Figure 1. Flow Diagram of Cohort Selection for Veterans with NSCLC Undergoing Chemoimmunotherapy.

Flow diagram illustrating cohort selection from a national Veterans Health Administration dataset. Patients with clinical stage IB–III(N2) non–small cell lung cancer (NSCLC) were identified and stratified by receipt of surgical resection. Patients receiving ≥2 cycles of chemoimmunotherapy were included and underwent chart review to confirm eligibility, treatment classification, and surgical candidacy. Patients initially classified as non-surgical who subsequently underwent resection were reclassified to the surgical cohort. The final analytic cohort consisted of 86 patients.

Table 1.

Baseline Demographic, Clinical, and Treatment Characteristics Stratified by Progression to Surgery

Variables	No surgery (N=26)	Surgery (N=60)	p-value
Age, mean (SD), y	71.3 (6.6)	68.5 (8.2)	0.142
CCI, median (SD)	2 (1,4)	1.5 (1,3)	0.160
Sex, n (%)			1.000
Female	2 (6.7%)	4 (7.1%)
Male	28 (93.3%)	52 (92.9%)
Race, n (%)			0.30
White	17 (65.4%)	47 (78.3%)
Black	6 (23.1%)	11 (18.3%)
Other	1 (3.9%)	1 (1.7%)
Unknown	2 (7.7%)	1 (1.7%)
Primary insurance, n (%)			0.294
Non-VA	3 (11.5%)	10 (16.7%)
VA	22 (84.6%)	50 (83.3%)
Unknown	1 (3.9%)	0
Care Setting (treatment locations), n (%)			0.390
VA only	18 (69.2%)	44 (74.6%)
Any community care	8 (30.8%)	15 (25.4%)
Histology, n (%)			0.874
Adenocarcinoma	15 (57.7%)	32 (53.3%)
Squamous	10 (38.5%)	23 (38.3%)
*Other	1 (3.9%)	5 (8.3%)
Tumor Location, n (%)			0.118
RUL	11 (42.3%)	31 (51.7%)
LUL	4 (15.4%)	14 (23.3%)
RML	0	5 (8.3%)
RLL	6 (23.1%)	7 (11.7%)
LLL	3 (11.5%)	2 (3.3%)
Unknown	2 (7.7%)	1 (1.7%)
Clinical Tumor Stage, n (%)			0.344
T1	7 (26.9%)	19 (31.7%)
T2	6 (23.1%)	13 (21.7%)
T3	6 (23.1%)	18 (30.0%)
T4	5 (19.2%)	10 (16.7%)
Tx	2 (7.7%)	0
Clinical Nodal Stage, n (%)			0.978
N0	11 (42.3%)	26 (43.3%)
N1	6 (23.1%)	14 (23.3%)
N2	9 (34.6%)	18 (30.0%)
Nx	0	2 (3.3%)
Overall Stage, n (%)			0.620
I	2 (7.7%)	2 (3.3%)
II	9 (34.6%)	26 (43.3%)
III	15 (57.7%)	32 (53.3%)
Chemo drug, n (%)			0.010
Carboplatin	25 (83.3%)	31 (55.4%)
Cisplatin	5 (16.7%)	25 (44.6%)
Immunotherapy Agent, n (%)			0.0006
Durvalumab	0	1 (1.8%)
Nivolumab	20 (66.7%)	53 (94.6%)
Pembrolizumab	10 (33.3%)	2 (3.6%)
Pre-Operative Workup, n (%)
PET Scan	25 (96.2%)	60 (100%)	0.127
Invasive Mediastinal Staging	18 (69.2%)	52 (88.1%)	0.195
Brain MRI	20 (70.9%)	45 (76.3%)	0.191
PET Scan, Invasive Mediastinal Staging, and Brain MRI	14 (58.85%)	44 (74.58%)	0.094
Treatment only at VA, n (%)	18 (69.2%)	44 (74.6%)	0.390

Baseline demographic, clinical, and treatment characteristics of patients stratified by progression to surgery following induction therapy. Values are presented as number (percentage), mean (SD), or median (IQR), as appropriate. P-values reflect comparisons between groups.

^*Other histology includes less common NSCLC subtypes defined by ICD-O-3 coding, including large cell carcinoma, adenosquamous carcinoma, sarcomatoid carcinoma variants, and NSCLC not otherwise specified.

Patients who underwent surgery had similar clinicopathological features to those who did not in terms of clinical stage, pathology, age, and comorbidities (Table 1). A subset of patients received portions of their care outside of the VA system, with 30.8% of patients in the no-surgery group and 25.4% in the surgery group receiving community-based care. Pre-operative staging techniques were also similar between groups (Table 1). Patients who did not undergo surgery were more likely to receive carboplatin over cisplatin (83.3%), which may reflect oncologist choice due to functional status. In addition, the distribution of immunotherapy agents differed between groups (Table 1).

In the surgical cohort 57% of patients underwent a minimally invasive resection and 81.7% underwent a lobectomy. Of 51 patients with known pathology data, 21 (41.1%) had a complete pathological response. The perioperative outcomes, surgical resection data, and pathological staging information are noted in Table 2. In the no-surgery group, 13 patients (50%) went on to undergo radiation therapy, of whom 5 (38.5%) received concurrent chemoradiation.

Table 2.

Clinicopathological Characteristics and Perioperative Outcomes of the Surgical Cohort

Variable	Number (%) or Median (IQR)
Approach
Thoracotomy	17 (36.2%)
Minimally Invasive	30 (63.8%)
Resection Extent
Lobectomy	43 (91.5%)
Pneumonectomy	1 (2.1%)
Wedge	3 (6.4%)
Perioperative outcomes
30-day mortality	2 (3.33%)
90-day mortality	3 (5.00%)
Any major complication	13 (21.6%)
Complete Pathological Response (N=52)	21 (41.2%)
Clinical to Pathological Tumor Stage (N=52)	21 (41.2%)
Downstage	31 (60.1%)
Unchanged	16 (31.4%)
Upstage	4 (7.8%)
Clinical to Pathologic Nodal Stage (N=52)
Downstage	22 (43.1%)
Unchanged	25 (49.0%)
Upstage	4 (7.8%)

Baseline clinicopathologic characteristics, operative approach, perioperative outcomes, and pathologic response among patients who underwent surgical resection following induction therapy. Values are presented as number (percentage)

Next, we evaluated the survival of the cohort. Median follow-up was 2.51 (2.17, 3.09) years. The 2-year OS for the cohort was 80.23%. Patients who underwent surgery had a 2-year OS of 86.67%, and those who did not have a 2-year OS of 65.38% (Figure 2). Survival in patients with stage I and II and patients with stage III disease are demonstrated in Figure 3A & B.

Figure 2. Overall Survival Stratified by Progression to Surgery.

Kaplan–Meier overall survival stratified by progression to surgery following induction therapy. Two-year overall survival was 86.7% for surgery and 65.4% for no surgery.

Figure 3. Overall Survival Stratified by Progression to Surgery, by Clinical Stage.

Kaplan–Meier overall survival stratified by progression to surgery following induction therapy, shown separately for (A) clinical stage I–II and (B) clinical stage III disease. In stage I–II disease, 2-year overall survival was 85.7% for surgery and 72.7% for no surgery. In stage III disease, 2-year overall survival was 87.5% for surgery and 60.0% for no surgery.

We evaluated reasons for non-progression to surgery. We found that the majority of patients developed distant disease during induction (N=12, 46.2%). Only 2 patients were deferred due to concern for inability to achieve a negative margin (N=2, 7.7%). One patient had a complete clinical response and chose not to undergo an operation (Table 3).

Table 3:

Reasons for Non-Progression to Surgery Following Induction Therapy

Reason for No surgery	Number (%)
Distant Progression of Disease	12 (46.2%)
Unknown	4 (15.4%)
Patient Decision	3 (11.5%)
Complete Clinical Response	1 (3.8%)
Social Reason	2 (7.7%)
Unable to Tolerate Surgery	3 (11.5%)
Risk of Positive Margin	2 (7.7%)

Finally, we evaluated outcomes for patients who did not progress to surgery. In the non-surgery group, patients who underwent radiation therapy (either alone or in combination with chemotherapy) had a 2-year OS of 76.9% (N=13). Patients that received chemotherapy alone had a 2-year OS of 53.9%.

Discussion

This study describes the utilization and outcomes of induction chemoimmunotherapy within the VHA, highlighting outcomes in patients who undergo surgery and the substantial proportion of patients who do not proceed to surgical resection. Interestingly, only 86 patients from 2021–2024 underwent attempted neoadjuvant chemoimmunotherapy in the VHA. However, we observed that patients who undergo induction chemoimmunotherapy with intent of surgical resection have excellent 2-year OS of approximately 80%. In the patients who undergo a resection, the pathological complete response rate was 41%. While randomized controlled trials reported dropout rates of 18–22%, we found that about 30% of patients did not undergo surgical resection. The most common reason for failure to proceed to surgery was development of distant metastatic disease during induction therapy. There were no major differences between patients who did and did not undergo surgery in age, comorbidity, or cT, cN or overall cStage, or pre-operative staging techniques. Finally, patients who did not undergo surgery and were eligible and went on to receive radiotherapy had a 2-year OS of 76.9%. Patients who failed to proceed to an operation due to disease progression likely had poor tumor biology which ultimately led to an inability to undergo surgical resection. Collectively, our study highlights the good intermediate-term survival for patients who undergo neoadjuvant chemoimmunotherapy with or without resection but also highlights a higher failure rate to undergo resection and potential for lower utilization of this approach in the VHA setting.

One of the major findings in our study was a favorable intermediate-term OS of 80.2%. This OS rate appears similar to the 2-year OS reported in trials (80–83%).^4,6 There are important differences between our population and the trial cohorts. Patients in this study were less likely to have cN2 disease (N=27/86, 31.4%) compared with an approximately 40% clinical N2 rate in randomized trials.^4,6 In addition, in our study, 50% of the patients had clinical stage III disease compared with trials that had a 63–70% cStage III population.^4,6 While our patients may have had less advanced disease, they were also older, with an average age of 70 versus 63–66 in the trials.^4,6 Overall, these data suggest that utilization of induction chemoimmunotherapy within the VHA may preferentially occur in older patients with larger, node-negative tumors, rather than patients with clinical N2 disease. However, despite these differences, OS remains consistent with trials.

An additional observation from this study is the relatively small number of Veterans receiving neoadjuvant chemoimmunotherapy within the VHA during the study period. Despite evaluating patients across all VHA centers between 2021 and 2024, only 86 patients met inclusion criteria. Given the increasing adoption of chemoimmunotherapy in randomized trials and academic centers, this may suggest that neoadjuvant chemoimmunotherapy remains underutilized within the VHA. This may reflect differences in referral patterns for surgical evaluation, institutional treatment preferences, increased disease burden or local invasion for this particular patient population, or increased comorbidity and frailty of the Veteran population limiting tolerance of multimodality therapy with surgery.

We also found a higher rate of non-progression to surgery in this cohort compared with prior clinical trials may reflect a variety of factors. We found no statistically significant difference in pre-operative staging in our cohort and similar rates of PET-CT, brain MRI, and mediastinal staging were found in surgery and no surgery groups. However, only 54.8% of the patients who did not undergo resection received all three staging modalities of brain MRI, pre-operative PET CT, and invasive mediastinal staging. A study conducted between 2014 and 2019 in a non-VHA cohort found that among patients with an indication for mediastinal staging, 38% did not undergo appropriate mediastinal evaluation.¹² These authors also found that guideline-concordant staging and guideline-concordant therapy were associated with improved long-term survival.¹² Rates of invasive mediastinal staging in our cohort were also incomplete, particularly in the non-surgical group (69.2% vs 88.1% in the surgical group), highlighting persistent gaps in guideline-concordant staging within the VHA.

Another difference between the surgery and no surgery group was the type of immunotherapy given. Given the nature of this dataset, these findings may reflect temporal or institutional practice variation rather than causal relationship with progression to surgery. In particular, the earlier publication and clinical adoption of nivolumab-based neoadjuvant therapy following the CheckMate 816 trial compared with later pembrolizumab-based perioperative strategies may partially explain the increased utilization of nivolumab in this cohort.⁴

Interestingly, we did not identify a clear clinicopathological indicator for patients who failed to progress to surgery. Identifying such patients remains challenging, particularly given that 30% of our cohort did not progress to surgery compared with approximately 20% in randomized trials.^3–6 In clinical trials, dropout from surgery was less common due to disease progression than we found in our cohort. For example, in the CheckMate 816 trial, progression accounted for 9% of patients who did not proceed to surgery, compared with 42% in our cohort.⁴ This higher rate of progression is notable given that our cohort included fewer patients with cN2 and stage III disease and may reflect differences in pre-treatment staging completeness as discussed above. Alternatively, the higher rate of progression may reflect referral patterns in which patients with borderline resectability or higher risk of R1 resection were selected for induction therapy in this cohort. In addition, the high rate of progression likely highlights poor tumor biology in this subset of patients. This highlights the need for improved methods to detect micrometastatic disease at diagnosis. Emerging approaches such as circulating tumor DNA (ctDNA) may improve identification of patients at elevated risk of early metastatic progression.^13,14 Overall, these findings suggest that more rigorous pre-induction staging and improved methods to identify aggressive tumor biology prior to treatment allocation would be useful.

Finally, given that 30% of the population did not progress to surgical resection, we evaluated outcomes in this population. Intermediate-term OS in this cohort was acceptable (2-year OS 65%). Notably, only 13 patients in this cohort received radiotherapy. Among patients in the non-surgical arm, the 2-year OS was 77% in those who underwent radiotherapy with or without chemotherapy compared with 55% in those who received chemotherapy alone. This difference in survival between these two groups likely reflects differences in disease-status (systemic multi-site disease in systemic therapy alone versus potentially curable in those who received radiation). However, the higher survival rates among these patients are still enouraging. A recent study evaluated patients with borderline resectable or unresectable stage III NSCLC with T4 and/or N2–N3 disease treated with neoadjuvant chemoimmunotherapy.¹⁷ They found that 75% of the patients were able to undergo surgery, with a 25% complete pathological response rate and a 2-year event-free survival of around 65%.¹⁷ This study suggests there may be a growing population of patients with locally-advanced NSCLC who fail to undergo resection. In patients who are borderline resectable and do not undergo surgery, definitive chemoradiation remains the standard of care. The PACIFIC trial, which evaluated definitive chemoradiation followed by adjuvant durvulamab for unresectable NSCLC, reported a 2-year OS 66.3% which is fairly high given the advanced nature of disease in this cohort.^15,16 A direct comparison between our patients who did not undergo resection and the patients in the PACIFIC trial is not possible, as there are multiple differences between our cohort and the PACIFIC trial participants. However, as induction chemoimmunotherapy and surgery are considered in more advanced patients, we must understand the survival in these patients and how to treat those patients who fail to progress to surgery following induction chemoimmunotherapy. There is currently no direct comparison possible between patients with borderline resectable NSCLC and whether a trial of chemoimmunotherapy and resection is worthwhile over definitive chemoradiation. In these more advanced patients, as a community we will need to develop a consensus into how to appropriately manage, screen, and treat patients with locally-advanced, potentially resectable NSCLC.

There are several limitations to our study. First, although all patients were evaluated by a surgeon prior to initiation of chemoimmunotherapy, the reasons for referral and whether therapy was initiated in the setting of potentially unresectable disease cannot be determined from the available data. It is possible that in some cases systemic therapy was initiated with a degree of clinical uncertainty regarding ultimate resectability rather than as part of a strictly predetermined surgical pathway. Additionally, while there were differences in immunotherapy agents between the surgery and no-surgery groups, it is unclear whether this reflects true differences in treatment effect or underlying confounding factors, including institutional practice patterns and timing of clinical trial adoption. Given the small sample size and multiple unmeasured confounders, these observations should be interpreted cautiously. Importantly, key biologic variables, including PD-L1 expression and driver mutation status (e.g., EGFR, ALK), were not available in this dataset. Furthermore, there is inherent selection bias and tumor biology between patients in different treatment pathways; therefore, patients who received surgery, radiation, or no local therapy cannot be directly compared, as these groups likely differ in underlying tumor biology, treatment decisions, and disease trajectory. Additionally, this study was performed within the VHA population, which consists predominantly of male patients with a high prevalence of smoking, potentially limiting generalizability to other healthcare settings. Despite this, prior work has demonstrated that surgical outcomes within the VHA are comparable to those observed in other national datasets.⁷ Finally, while all VHA medical centers were eligible for participation in this study, we cannot ascertain whether there were patients who were potentially eligible for induction chemoimmunotherapy and resection but did not meet our exact inclusion criteria (i.e. Veterans who received only 1 cycle of chemoimmunotherapy and either progressed or went on to resection).

In conclusion, current utilization of induction therapy within the VHA system may be lower than expected and preferentially applied to patients with larger, node-negative tumors; however, we have found similar OS in our cohort compared with previously published trials. While about 18–22% of patients did not proceed to surgery in the randomized controlled trials, approximately 30% of patients in our cohort did not progress to surgery, with 42% of these cases attributable to disease progression. Outcomes among patients who did not undergo surgery varied based on subsequent treatment strategy, likely attributable to differences in patient choice, tumor biology, and extent of disease. Given the considerable proportion of patients in this study who did not progress to surgery (30%) and newer studies which propose utilization of neoadjuvant chemoimmunotherapy and resection borderline resectable patients, this study with others highlights a few major gaps in our current knowledge and practice which can be addressed. For one, better mechanisms to predict early progression are needed AND understanding outcomes in patients who undergo a trial of potential resection versus those that undergo definitive chemoradiotherapy initially are desperately needed. Finally, as we increase utilization of neoadjuvant chemoIO, there will be a subsequent increase in patients who fail to undergo a resection. More work needs to be done to understand the appropriate treatment paradigm for this growing population of patients.