Brief Report: SARS-CoV-2 mRNA Vaccination Improved Survival in NSCLC Treated with Radiotherapy

Abstract

Stereotactic ablative radiotherapy (SABR) is standard for early-stage NSCLC, but strategies to enhance its systemic antitumor immune effects remain limited. Preclinical data suggest SARS-CoV-2 mRNA vaccines enhance innate immunity and antigen presentation, raising the possibility that vaccination could augment radiotherapy-induced immune priming. We evaluated the clinical impact of peri-SABR vaccination (defined as within 3 months before or after SABR) in 413 real-world patients with T1–3N0M0 NSCLC treated during the pandemic era, including 144 patients who received vaccination within this window. In time-varying analyses, vaccination was associated with improved recurrence-free survival (HR 0.69, 95% CI 0.49–0.98; P = 0.039) and a trend toward improved overall survival (HR 0.66, 95% CI 0.42–1.03; P = 0.069). In propensity score–matched time-dependent analyses, vaccination remained associated with improved recurrence-free survival (HR 0.61, 95% CI 0.40–0.91; P = 0.016) and overall survival (HR 0.57, 95% CI 0.34–0.94; P = 0.029), with directionally consistent findings observed after multivariable adjustment. These findings support further investigation into potential interactions between SARS-CoV-2 mRNA vaccination and SABR-induced antitumor immunity.

INTRODUCTION

Stereotactic ablative radiotherapy (SABR) is the standard of care for patients with early-stage non–small cell lung cancer (NSCLC)^1,2. SABR delivers high-dose, focal radiation that can release tumor antigens and stimulate local immune responses, prompting extensive investigation of its combination with immune checkpoint inhibitors^3,4. A phase II randomized I-SABR trial in early-stage NSCLC reported that adding programmed cell death 1 (PD-1) immunotherapy to SABR improved event-free survival significantly⁵. However, subsequent phase III evidence has not confirmed this benefit. In particular, KEYNOTE-867 (NCT03924869) showed that the combined immunotherapy/SABR arm did not cross the statistical significance boundary in recurrence-free survival as compared with SABR alone in early-stage NSCLC; the SWOG/NRG S1914 (NCT0421426200) was also closed prematurely following interim analysis and did not demonstrate a statistically significant improvement in recurrence-free survival. These findings temper prior enthusiasm for routine combination therapy and underscore the need to better define patient selection and optimize the timing, duration, and integration of immunotherapy with SABR.

SARS-CoV-2 mRNA vaccines are potent activators of innate immunity; preclinical data indicate they induce type I interferon signaling and enhance antigen presentation⁶. Retrospective analyses have also associated receipt of SARS-CoV-2 mRNA vaccination within defined windows of immunotherapy with improved survival across several advanced cancers⁶. Despite these observations, the potential interaction between SARS-CoV-2 mRNA vaccination and radiotherapy, and specifically peri-SABR vaccination in early-stage NSCLC remains poorly characterized.

To address this gap, we performed a retrospective study evaluating clinical outcomes after peri-SABR SARS-CoV-2 mRNA vaccination in patients with early-stage NSCLC, with the goal of generating preliminary evidence on whether an established mRNA vaccine platform might augment SABR-mediated antitumor immunity.

METHODS

Patient cohorts

The first retrospectively identified cohort is real-world patients with treatment-naïve or secondary primary lung or isolated parenchymal recurrent early-stage (T1–3N0M0, AJCC 9th edition) NSCLC who were treated with the standardized SABR program, with predefined staging workup, SABR procedures (clinical indications, dose regimens, normal tissue dose volume constraints, quality assurance criteria), follow-up images, and clinical visit schedule in an NCI-designated Cancer Center during the pandemic era between December 2020 and December 2022. The second one is the phase II randomized I-SABR trial cohort, conducted from July 2017 to April 2022, including both the SABR-only and SABR plus immunotherapy (Nivolumab) arms, which served as a supportive observational cohort. Detailed eligibility criteria were the same as the I-SABR study and have been previously described in the prospective I-SABR trial (NCT03110978)⁵.

SARS-CoV-2 mRNA vaccination and clinical variables

SARS-CoV-2 mRNA vaccination was recorded and categorized as peri-SABR vaccination, defined as occurring within 3 months before (pre-SABR) or after SABR (post-SABR). The 3-month window was selected based on prior work from our group⁶ and the expected kinetics of vaccine-induced innate and adaptive immune activation, providing a biologically plausible interval for interaction with radiotherapy while maintaining adequate sample size. For patients who received multiple doses, classification was based on the timing of the first dose during the 6 months surrounding SABR. COVID-19 infection and influenza vaccine within the same time window were also documented. Other survival-related clinical variables included age, sex, smoker, histology, race, tumor size, and Eastern Cooperative Oncology Group (ECOG) performance status.

Statistical analysis

The primary endpoint was recurrence-free survival (RFS), defined as the time from SABR to any recurrence (local recurrence, regional recurrence, or distant metastasis), death, or last radiologic assessment (censored). Follow-up imaging was performed according to a standardized institutional protocol, with a post-SABR chest CT imaging every 3 months for 2 years, every 6 months for an additional 3 years, and annually thereafter. This protocol was applied uniformly to all patients irrespective of vaccination status. The secondary endpoint was overall survival (OS), defined as the time from SABR to death or last follow-up (censored).

Survival analyses were performed to compare RFS and OS across vaccination groups, including vaccinated vs. unvaccinated, as well as pre-SABR, post-SABR, and vs. never vaccinated subgroups. To rigorously minimize bias inherent to this retrospective study design, we implemented a comprehensive statistical analysis pipeline.

To mitigate immortal-time bias, which arises when pre-vaccination person-time is erroneously attributed to the vaccinated group, follow-up for each patient was reformatted using a counting-process framework. Specifically, each patient’s follow-up was split into time-varying intervals such that vaccination status was updated at each event time according to the individual’s current exposure state. Hazard ratios were estimated using time-dependent Cox proportional hazards models fitted to this counting-process data structure.

Survival curves were visualized using Simon-Makuch plots, which appropriately accommodate time-varying treatment assignment by dynamically reallocating patients between the unvaccinated and vaccinated risk sets according to their contemporaneous vaccination status, rather than fixed baseline group membership. Given the inherently dynamic risk-set composition, conventional at-risk tables were not displayed.

To further address selection bias and baseline confounding, propensity score matching (PSM) was performed using 1:1 nearest-neighbor matching with a caliper width of 0.2, based on well-established, objective, and routinely measured prognostic factors in early-stage NSCLC, including age, race, ECOG performance status, tumor size, histology, and lung cancer history to balance baseline characteristics between vaccinated and unvaccinated patients. In the matched cohort, associations with survival outcomes were further evaluated using multivariable time-dependent Cox regression, adjusting for clinically relevant covariates. Exploratory E-value analyses were performed to assess robustness to unmeasured confounding.

RESULTS

A total of 413 patients treated with SABR in a real-world retrospective cohort (308 with treatment-naïve primary lung cancer and 105 with second primary lung cancer or isolated recurrence), along with 141 patients from the I-SABR trial, were included. Exploratory supportive analyses of the I-SABR cohort were restricted to patients treated between December 2020 and April 2022 (pandemic-era subset, n=41), corresponding to the period during which SARS-CoV-2 vaccination was available. Baseline characteristics were generally balanced across groups (Table 1). In the real-world cohort (n=413), 26 patients (6.3%) developed ≥grade 2 radiation pneumonitis. No specific COVID-19 vaccine–related toxicities were observed. Vaccinated patients demonstrated improved RFS (HR 0.69, 95% CI 0.49–0.98; P = 0.039) and a trend toward improved OS (HR 0.66, 95% CI 0.42–1.03; P = 0.069) compared with unvaccinated patients after adjusting for immortal-time (Figure 1a–b). Vaccine timing relative to SABR (pre- vs. post-SABR) did not significantly differ in outcomes (Figure 1c–d). After propensity score matching (138 pairs, balanced for age, race ,ECOG, histology, lung cancer history and tumor size; Supplementary Figure 1) vaccination was associated with improved RFS (HR 0.61, 95% CI 0.40–0.91; P = 0.016) and OS (HR 0.57, 95% CI 0.34–0.94; P = 0.029) (Figure 1e–f). These associations remained significant after multivariable adjustment, with E-values of 3.0 for RFS and 3.6 for OS, suggesting moderate robustness to potential unmeasured confounding (Figure 1g–h). In contrast, COVID-19 infection and influenza vaccination within the same peri-SABR window were not associated with RFS or OS (COVID-19 infection, RFS: HR 0.83, 95% CI 0.44–1.59, P = 0.579; OS: HR 1.03, 95% CI 0.47–2.22, P = 0.949; influenza vaccination, RFS: HR 0.80, 95% CI 0.44–1.45, P = 0.466; OS: HR 0.84, 95% CI 0.39–1.81, P = 0.652) (Supplementary Figure 2). Exploratory analyses stratifying vaccination exposure demonstrated a stronger association with improved outcomes among patients receiving booster exposure while a similar directional trend was observed for primary series exposure (Supplementary Figure 3).

Table 1:

Demographics and clinical characteristics of patients in the main and validation cohorts

Characteristic	Real-world SABR cohort (Pandemic-Era)			Exploratory I-SABR subset (Pandemic-Era)
	Vaccinated (N=144)	Unvaccinated (N=269)	P	Vaccinated (N=20)	Unvaccinated (N=21)	P
Age, N (%)			0.777			0.656
<=75 years	73 (50.7)	131 (48.7)		11 (55.0)	14 (66.7)
>75 years	71 (49.3)	138 (51.3)		9 (45.0)	7 (33.3)
Gender, N (%)			0.221			0.808
Male	73 (50.7)	118 (43.9)		5 (25.0)	7 (33.3)
Female	71 (49.3)	151 (56.1)		15 (75.0)	14 (66.7)
Smoker, N (%)			1.000			0.238
Yes	121 (84.0)	225 (83.6)		15 (75.0)	19 (90.5)
No	23(16)	44 (16.4)		5 (25.0)	2 (9.5)
Histology, N (%)			0.135			0.277
ADC	92 (63.9)	150 (55.8)		17 (85.0)	14 (66.7)
Non-ADC	52 (36.1)	119 (44.2)		3 (15.0)	7 (33.3)
Race, N (%)			0.035			0.230
White	115 (79.9)	237 (88.1)		16 (80.0)	110 (90.9)
Non-white	29 (20.1)	32 (11.9)		4 (20.0)	11 (9.1)
Tumor size, N (%)			0.146			0.184
< 3cm	118 (81.9)	236 (87.8)		19 (95.0)	16 (76.2)
≥ 3cm	26 (18.1)	33 (12.3)		1 (5.0)	5 (23.8)
ECOG, N (%)			0.342			0.488
0–1	132 (91.7)	237 (88.1)		20 (100.0)	19 (90.5)
2–3	12 (8.3)	32 (11.9)		0 (0)	2 (9.5)
Covid infection, N (%)			0.703			1.000
Yes	9 (6.2)	21 (7.8)		0 (0)	1 (4.8)
No	135 (93.8)	248 (92.2)		20 (100.0)	20 (95.2)
Cancer history, N (%)			0.246			0.132
Treatment-naive	102 (70.8)	206 (76.6)		12 (60.0)	18 (85.7)
SPLC/IR	42 (29.2)	63 (23.4)		8 (48.0)	3 (14.3)
Pneumonitis Grade2+, N (%)			0.144			NA
Yes	13 (9.0)	13 (4.8)		0 (0)	0 (0)
No	131 (91.0)	256 (95.2)		0 (0)	0 (0)
Vaccine category, N (%)			NA			NA
Primary series	63 (43.8)	NA		12 (60.0)	NA
Booster exposure	81 (56.2)	NA		8 (40.0)	NA
4Y-OS, (%)						0.023
95% CI	76.4 (69.0 – 84.7)	71.3 (64.8– 78.4)	0.047	93.8 (82.6– 100.0)	68.1 (49.8 – 93.1)
4Y-RFS, (%)						0.004
95% CI	60.6 (52.0 – 70.6)	51.2 (43.0 – 60.8)	0.028	93.8 (82.6– 100.0)	58.4 (40.0 – 85.5)

^*Pandemic-era was defined as patients treated between December 2020 and April 2022, corresponding to the period during which SARS-CoV-2 vaccination became available. Accordingly, 41 patients from the original 141-patient I-SABR cohort were eligible for the exploratory supportive analyses. No Grade ≥2 pneumonitis events were observed within the subset, although pneumonitis events were reported in the overall parent trial population.

^*CI: Confidence Interval, ADC: Adenocarcinoma, OS: Overall Survival, RFS: Recurrence-Free Survival, ECOG: Eastern Cooperative Oncology Group performance status, SPLC: Secondary Primary Lung Cancer, IR: Isolated Recurrence (IR)

^*P values were calculated using the χ² test or Fisher’s exact test, as appropriate, for categorical variables, and the Wilcoxon rank-sum test for continuous variables. P values for 4Y-OS and 4Y-RFS were calculated using the log-rank test.

Figure 1. Survival analyses in the real-world SABR cohort.

Simon-Makuch plots displaying time-varying recurrence-free survival (RFS) and overall survival (OS) stratified by (a–b) vaccination status, (c–d) vaccine timing relative to SABR initiation (pre-SABR, post-SABR, never vaccinated), (e–f) vaccination status in a propensity score-matched cohort and (g-h) multivariate confounder adjustment. Across all panels, immortal-time bias is addressed via time-varying Cox proportional hazards models and at-risk tables are omitted given the dynamic group membership inherent to the Simon-Makuch framework.

In the exploratory pandemic-era I-SABR subset, mRNA vaccination was similarly associated with significantly improved RFS in both the overall subset (HR 0.09, 95% CI 0.01–0.71, P = 0.022) and the SABR-only arm (HR 0.07, 95% CI 0.01–0.62, P = 0.018) (Figure 2a–b), although findings should be interpreted cautiously given the limited sample size and number of events. A similar directional trend with marginal significance was observed for OS (Figure 2d–e). In contrast, vaccination was not associated with improved outcomes in the SABR plus immunotherapy arm (Figure 2c,2f).

Figure 2. Survival analyses in the I-SABR trial cohort restricted to the COVID-19 pandemic period.

Simon-Makuch plots displaying time-varying recurrence-free survival (RFS) stratified by vaccination status in (a) the overall trial cohort, (b) the SABR-only arm, and (c) the SABR plus immunotherapy (Nivolumab) arm. Corresponding time-varying overall survival (OS) analyses are shown in (d–f) for the same subgroups, respectively. Hazard ratios were estimated using time-varying Cox models to address immortal-time bias across all panels; estimates were not obtainable in panels (c) and (f) due to no events in one group. At-risk tables are omitted given the dynamic group membership inherent to the Simon-Makuch framework.

DISCUSSION

mRNA nanoparticles are highly immunogenic and can provoke cytokine and chemokine responses that reshape systemic and intratumoral immune landscapes, thereby amplifying antitumor activity^7,8. Preclinical and emerging clinical data suggest that mRNA vaccines, including those targeting SARS-CoV-2, are also potent immune modulators that amplify antitumor activity by reshaping systemic immune landscapes⁶. On this basis, we hypothesized that the combination of local immune activation from SABR^3,9,10 and systemic immune activation from SARS-CoV-2 mRNA vaccination would synergistically augment priming of anti-tumor immune responses⁶, improving clinical outcomes for patients with early-stage NSCLC treated with SABR.

To our knowledge, this is the first study to suggest a potential synergistic interaction between radiotherapy and SARS-CoV-2 mRNA vaccination in cancer. In addition to the real-world cohort, exploratory analyses within the restricted pandemic-era subset of the prospective I-SABR trial demonstrated directionally consistent associations between SARS-CoV-2 vaccination and improved clinical outcomes following SABR. However, given the limited sample size and number of events in I-SABR, the observed hazard ratio should not be interpreted as a definitive estimate of the vaccination effect, and the effect size will require confirmation in larger cohorts. The observed survival benefit of SABR plus vaccination compared with SABR alone is consistent with vaccine-associated enhancement of antitumor immunity. Interestingly, mRNA vaccination did not appear to confer additional benefit when immunotherapy was already added to SABR. This lack of improved outcomes in the SABR and immunotherapy combination arm may reflect complex immune dynamics and warrants future mechanistic investigation.

In agreement with recent reports suggesting that COVID-19 infection does not provide a survival advantage in cancer patients¹¹, we did not observe an association between clinical outcomes and either COVID-19 infection or influenza vaccination within the same treatment window. However, this comparison should be interpreted with caution, as infection represents a heterogeneous exposure and may be incompletely ascertained due to asymptomatic cases, variability in testing, and diagnoses occurring outside our institution. Therefore, the absence of a similar association should not be overinterpreted as formal negative-control analysis or definitive evidence of specificity. Nonetheless, the observed association between SARS-CoV-2 vaccination and improved outcomes suggests a potentially distinct interaction between vaccination and SABR that warrants further investigation. While licensed SARS-CoV-2 mRNA vaccines provide a readily available immunologic stimulus with an established safety profile, offering a practical opportunity to investigate systemic immunomodulation in oncology, tumor-specific mRNA vaccines under active clinical development may provide even more robust antitumor responses. Future work should consider the use of tumor-specific RNA vaccines in combination with radiation.

Although we applied comprehensive statistical approaches to mitigate multiple sources of bias, including immortal-time bias, selection bias, and confounding, the findings should be considered hypothesis-generating and require validation in larger, ideally randomized, prospective studies. Residual and unmeasured confounding remains possible. Although we selected well-established, objective variables for propensity score matching to minimize subjectivity in this retrospective cohort, this approach may not fully capture all relevant confounders, including comorbidity, healthcare access, follow-up adherence, salvage treatment, among others. All patients were treated at a single center with relatively uniform access to care likely reducing, though not entirely eliminating, differences in access between groups. Exploratory E-value analyses suggested that moderately strong unmeasured confounding would be required to fully explain away the observed associations. Nevertheless, residual confounding and healthy-vaccinee bias cannot be excluded, as differences in healthcare engagement, follow-up behavior, and vaccination practices may still persist and influence outcomes.

RFS is also potentially susceptible to surveillance bias given its dependence on imaging follow-up; despite standardized imaging protocols, residual differences in surveillance intensity or outside imaging may still exist. The timing of vaccination relative to SABR may not be biologically equivalent. Although pre- and post-SABR vaccination were combined into a single exposure category and no statistically significant difference was observed between these groups, this comparison was limited by sample size and statistical power; therefore, timing-specific effects cannot be excluded.

Future mechanistic studies are needed to clarify causality and define the optimal timing of vaccination. It also remains unclear which patients derive the greatest benefit from vaccination alone or in combination with immunotherapy; emerging approaches such as digital twin models may help guide patient selection^12,13. Overall, the potential synergistic interaction between mRNA vaccination and radiotherapy warrants further investigation.

Data Availability:

During peer review, research data access is provided to editors and reviewers via a private preview link for evaluation purposes. Upon publication, the data will be available under controlled access via the repository (Data DOI) and can be obtained upon reasonable requests and approval, in accordance with institutional policy. Private Access Link for Preview (Please click “Go to published record” on the page to access the data): https://zenodo.org/records/19389276?preview=1&token=eyJhbGciOiJIUzUxMiJ9.eyJpZCI6IjU5NTUyZGYxLTFlMDctNDVkYS1hMDU4LTQ0YTEwNTI2ZDM4OSIsImRhdGEiOnt9LCJyYW5kb20iOiJiMjI1MWQ1ODBjNzlmZmUyMmEyZWM3ZWNkYjdkMzdjOCJ9.GAaMdQuGiCiD2BOjWdBndBc1tWO2L0nuicR2J4eFpFdXaveIm-MHbWjUYT-0gXNKasNAWJ3-D1qndRMSaUmCVg

Data DOI: https://doi.org/10.5281/zenodo.19389276

Code:

The code used in this study is available at https://github.com/WuLabMDA/VaxOnSABR.