Immunological biomarkers for precision stereotactic body radiation therapy in early-stage non-small cell lung cancer: a systematic review Authors: Stefania Volpe, Seungju Lee, Mateusz Bilski, Maria Giulia Vincini, Michela Onza, Gaia Piperno, Maria Alessia Zerella, Luigi Cornacchia, Lucia Palumbo, Paola Fulghieri, Alfonso Zappia, Roberta Noberini, Mariano Lombardi, Eltjona Mane, Konstantinos Venetis, Monica Casiraghi, Lorenzo Spaggiari, Tiziana Bonaldi, Nicola Fusco, Roberto Orecchia, Barbara Alicja Jereczek-Fossa Published: 2026 ABSTRACT SummaryBackgroundStereotactic body radiation therapy (SBRT) is a non-invasive alternative to surgery for the management of early-stage non-small cell lung cancer (ES-NSCLC). Despite excellent control rates, outcomes of SBRT are still heterogenous, and there is a clear need for biomarkers to guide treatment selection, estimate recurrence risk, or identify patients who may benefit from treatment intensification. Given the increasing role of immunotherapy in this setting, we conducted a systematic review to summarise the available evidence on prognostic and predictive immunological biomarkers in stage I–II NSCLC treated with curative SBRT.MethodsPubMed/MEDLINE, Scopus, Embase, and ClinicalTrials.gov were screened on 30th November 2025. Full-texts from database inception written in English were considered eligible if presenting the association between immunological biomarkers and oncological outcomes in patients treated with SBRT. Due to the heterogeneity of the included studies in terms of biomarkers, outcome definitions, and analytical approaches, neither a formal meta-analysis nor a GRADE assessment was performed. Consequently, statistical heterogeneity and publication bias were not formally assessed. Risk of bias was assessed with the Newcastle-Ottawa Scale (NOS).FindingsTen studies, for a total of 1019 patients, met the inclusion criteria. Eight were retrospective cohorts, one was a prospective observational study, and one was a phase I trial. Overall, the included studies showed moderate-to-high methodological quality, with most achieving adequate follow-up and higher NOS scores, demonstrating adequate selection criteria, outcome ascertainment, and control of confounders. Median sample size was 82 (interquartile range 61–117). The most frequently investigated biomarkers were neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), with elevated baseline levels consistently associated with poorer survival. CD44 emerged as a potential predictive biomarker, with higher expression correlating with increased risk of recurrence after SBRT and validated in an external surgical cohort. Immune-profiling studies identified prognostic signatures—such as PD-1+, CD4 T cells and Th17 subsets—but none demonstrated predictive value for SBRT response.InterpretationCurrent evidence highlights the predominance of prognostic, rather than predictive, biomarkers in ES-NSCLC treated with SBRT. While CD44 and other tumour microenvironment–related markers are promising, available data remain exploratory. Larger biomarker-driven studies are needed to enable personalised treatment allocation between surgery and SBRT and to guide rational use of systemic therapies in early-stage disease.FundingThe present work was conducted within the framework of the MONDRIAN trial. Research in contextEvidence before this studyThe prognostic and predictive role of immune-related biomarkers in early-stage non-small cell lung cancer (NSCLC) treated with stereotactic body radiation therapy (SBRT) remains poorly defined. We systematically searched PubMed/MEDLINE, Embase, and Scopus, as well as ClinicalTrials.gov, from database inception to 30th November 2025, without language restrictions. Reference lists of relevant articles were also screened to identify additional studies. Search terms included combinations of “SBRT”, “stereotactic body radiotherapy”, “NSCLC”, “lung cancer”, “biomarkers”, “immune”, “tumour microenvironment”, “NLR”, “PLR”, “CRP”, “lymphocytes”, and “cytokines”.A total of 10 eligible studies were identified, showing limited and largely based on small, retrospective, single-institution studies with heterogeneous methodologies. Most investigations have focused on systemic inflammatory markers, particularly neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and C-reactive protein (CRP), which consistently show an association with worse oncologic outcomes when elevated. However, the overall risk of bias is moderate to high, due to retrospective design, small sample sizes, inconsistent adjustment for confounders, and lack of external validation. Furthermore, substantial heterogeneity in biomarker definitions, thresholds, and endpoints precluded formal meta-analysis, and no robust pooled estimates are currently available. Evidence supporting predictive biomarkers to guide treatment selection remains particularly scarce and exploratory.Added value of this studyThis study provides a comprehensive and focused synthesis of immune-related biomarkers specifically in the setting of early-stage NSCLC treated with SBRT, a clinical context that has not been systematically appraised to date. By integrating data from both circulating biomarkers and tumour microenvironment analyses, this review highlights consistent signals supporting the prognostic role of systemic inflammatory markers while critically appraising emerging, more complex immune signatures.Importantly, this work distinguishes between prognostic and predictive biomarkers, emphasising the scarcity of robust predictors of treatment benefit. In particular, it identifies that evidence supporting biomarker-driven treatment selection—such as choosing between SBRT and surgery—is currently limited to isolated, hypothesis-generating findings without external validation. Additionally, this review underscores key methodological limitations across studies, including lack of standardisation, absence of histology-specific analyses, and limited consideration of biomarker dynamics over time, thereby outlining clear priorities for future research.Implications of all the available evidenceTaken together, current evidence suggests that simple, blood-based inflammatory biomarkers such as NLR and PLR may have reproducible prognostic value in patients undergoing SBRT for early-stage NSCLC, although their integration into routine clinical decision-making remains premature. Conversely, the role of biomarkers as predictive tools to guide treatment selection between SBRT and surgery is not yet supported by robust or externally validated evidence.Future research should focus on prospective, adequately powered, multi-institutional studies with standardised biomarker assessment, appropriate statistical methodologies (including correction for multiple testing), and independent external validation. Greater attention should also be given to histology-specific analyses and longitudinal biomarker dynamics. Integrating low-cost circulating markers with advanced tumour microenvironment profiling may provide a biologically informed framework for patient stratification. Ultimately, biomarker-driven approaches hold promise for refining treatment selection and personalising care, but their translation into clinical practice will require a higher level of evidence. Research in context Evidence before this study The prognostic and predictive role of immune-related biomarkers in early-stage non-small cell lung cancer (NSCLC) treated with stereotactic body radiation therapy (SBRT) remains poorly defined. We systematically searched PubMed/MEDLINE, Embase, and Scopus, as well as ClinicalTrials.gov, from database inception to 30th November 2025, without language restrictions. Reference lists of relevant articles were also screened to identify additional studies. Search terms included combinations of “SBRT”, “stereotactic body radiotherapy”, “NSCLC”, “lung cancer”, “biomarkers”, “immune”, “tumour microenvironment”, “NLR”, “PLR”, “CRP”, “lymphocytes”, and “cytokines”. A total of 10 eligible studies were identified, showing limited and largely based on small, retrospective, single-institution studies with heterogeneous methodologies. Most investigations have focused on systemic inflammatory markers, particularly neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and C-reactive protein (CRP), which consistently show an association with worse oncologic outcomes when elevated. However, the overall risk of bias is moderate to high, due to retrospective design, small sample sizes, inconsistent adjustment for confounders, and lack of external validation. Furthermore, substantial heterogeneity in biomarker definitions, thresholds, and endpoints precluded formal meta-analysis, and no robust pooled estimates are currently available. Evidence supporting predictive biomarkers to guide treatment selection remains particularly scarce and exploratory. Added value of this study This study provides a comprehensive and focused synthesis of immune-related biomarkers specifically in the setting of early-stage NSCLC treated with SBRT, a clinical context that has not been systematically appraised to date. By integrating data from both circulating biomarkers and tumour microenvironment analyses, this review highlights consistent signals supporting the prognostic role of systemic inflammatory markers while critically appraising emerging, more complex immune signatures. Importantly, this work distinguishes between prognostic and predictive biomarkers, emphasising the scarcity of robust predictors of treatment benefit. In particular, it identifies that evidence supporting biomarker-driven treatment selection—such as choosing between SBRT and surgery—is currently limited to isolated, hypothesis-generating findings without external validation. Additionally, this review underscores key methodological limitations across studies, including lack of standardisation, absence of histology-specific analyses, and limited consideration of biomarker dynamics over time, thereby outlining clear priorities for future research. Implications of all the available evidence Taken together, current evidence suggests that simple, blood-based inflammatory biomarkers such as NLR and PLR may have reproducible prognostic value in patients undergoing SBRT for early-stage NSCLC, although their integration into routine clinical decision-making remains premature. Conversely, the role of biomarkers as predictive tools to guide treatment selection between SBRT and surgery is not yet supported by robust or externally validated evidence. Future research should focus on prospective, adequately powered, multi-institutional studies with standardised biomarker assessment, appropriate statistical methodologies (including correction for multiple testing), and independent external validation. Greater attention should also be given to histology-specific analyses and longitudinal biomarker dynamics. Integrating low-cost circulating markers with advanced tumour microenvironment profiling may provide a biologically informed framework for patient stratification. Ultimately, biomarker-driven approaches hold promise for refining treatment selection and personalising care, but their translation into clinical practice will require a higher level of evidence. Introduction Lung cancer is the most frequently diagnosed cancer and remains the leading cause of cancer-related mortality worldwide, with an estimated 1.8 million deaths each year. Non-small cell lung cancer (NSCLC) constitutes about 85% of these cases.1 For patients with early-stage NSCLC (ES-NSCLC), stereotactic body radiation therapy (SBRT) has become a standard curative treatment, particularly for those who are medically inoperable or choose to forgo surgery.2, 3, 4, 5 SBRT is known for achieving high rates of local control while maintaining limited toxicity by delivering hypofractionated, high-dose radiation precisely to the tumour.6,7 Despite these clinical advantages, outcomes among ES-NSCLC patients treated with SBRT are heterogeneous, exhibiting significant variations in progression-free survival (PFS) and overall survival (OS). These discrepancies suggest that further refinement is needed to guide clinical decision-making in this setting. The increasing interest in combining SBRT with systemic therapies, such as immunotherapy, further highlights the necessity for reliable biomarkers to guide personalised treatment strategies.8 Biomarkers, particularly immune-related markers, have garnered significant attention as potential instruments to enhance prognostic accuracy and predict treatment response in NSCLC.9, 10, 11 However, their clinical relevance in early-stage disease treated with SBRT remains uncertain, and few biomarkers have been integrated into routine practice. This systematic review aims to collect and synthesise the existing evidence on immune-related biomarkers that may have prognostic and/or predictive value for patients with ES-NSCLC undergoing SBRT. By critically evaluating the literature, we aim to outline the current landscape of biomarker research in this context and identify future directions for precision radiotherapy. Methods Search methodology The methodology complied with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines. Of note, the present review was not registered. The research question was established by using the Population, Intervention, Outcome, and Study design (PICOS) framework (Supplementary Materials 1). All English-language publications were considered eligible if they met the following inclusion criteria: 1) clinical studies on ES-NSCLC (stage I-II) treated with SBRT, 2) report of the association between immunological biomarkers and oncological outcomes (e.g., OS, PFS), and 3) full-text availability. Case reports and case series were not included, as well as studies including stage III patients. References from all eligible records were screened for cross-reference. Toxicity outcomes were considered beyond the scope of this work. The search was performed on the NCBI PubMed, Embase, and Scopus databases, and on clinicaltrials.gov, from database inception to 30th November 2025. No publication year restrictions were applied. The complete search strategy is provided in Supplementary Materials 2. Evidence acquisition Article selection was performed by two authors (SL and SV) independently, and any discrepancies were resolved through discussion. For each study meeting the inclusion criteria, the following variables were retrieved: publication year, type of study, considered biomarker(s), endpoint of interest, sample size, patients’ age, stage, histotypes, SBRT schedule(s), follow-up duration, and study results. If multiple fractionation schemes were used, information was reported for each fractionation scheme, whenever possible. Statistical analysis Due to the heterogeneity of the included studies in terms of biomarkers, outcome definitions, and analytical approaches, neither a formal meta-analysis nor a GRADE assessment was performed. Consequently, statistical heterogeneity and publication bias were not formally assessed. Hence, a qualitative assessment was performed using the Newcastle-Ottawa Scale (NOS). Two reviewers (SL and SV) independently evaluated each study and any discrepancies were resolved through discussion. No automation tools were used in the risk-of-bias assessment process. The standard NOS for cohort studies was used for retrospective studies, while a modified approach for single-arm prospective or early-phase trials was adopted for prospective evidence. A predefined threshold of adequate follow-up (median ≥24 months) was adopted across all studies. Ethics Ethics approval was not required due to the study design (systematic review), as no individual patient data were collected or used. Role of the funding source The present work was conducted within the framework of the MONDRIAN trial (Multi-omics integrative modelling for stereotactic body radiotherapy in early-stage non-small cell lung cancer–AIRC 28282–NCT05974475). The sponsors did not play any role in the study design, collection, analysis and interpretation of data, nor in the writing of the manuscript, nor in the decision to submit the manuscript for publication. Results A total of 10 records met the inclusion criteria. The whole selection process is reported in Fig. 1. Publication years ranged from 2015 to 2025. Eight studies were retrospective,12, 13, 14, 15, 16, 17, 18, 19 one had a prospective observational design,20 and one was a phase I trial.21 Median sample size was 82 (interquartile range, IQR: 61–117, range: 20–286). All included patients were staged as cN0. The most frequently investigated immunological-based biomarker was the neutrophil-to-lymphocyte ratio (NLR), which was examined in five studies,12,16, 17, 18, 19 while other immune parameters were explored heterogeneously across the remaining publications. The main characteristics of the included studies, along with their results, are presented in Table 1.Fig. 1Study selection process per the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.Table 1Main characteristics of the included studies.Author, and yearStudy typeBiomarkerEndpoint of interestTypeSample sizeAge (Median, IQR)Staging (TNM 8th)HistotypesSBRT scheduleFollow-upMain resultsCannon et al.,19 2015RNRL/PLROSPrognostic59 (53% male, 47% female)70.0 (range: 48–89)IA 43 ptsIB 16 ptsAdeno (34 pts)SCC (19 pts)NOS (6 pts)34 Gy/1 fx: 3 pts54 Gy/3 fx: 33 pts48 Gy/4 fx: 1 pt60 Gy/5 fx: 22 ptsa17.0 monthsmOS- 23 months (NLR >2.98 or PLR >146) vs.mOS- NR (NLR <2.98 or PLR<146)Dong et al.,16 2023RNLR, CRPOS, PFSPrognostic148 (72% male, 28% female)75.5 (range: 47.2–89.0)T1a 4.7%T1b 48.0%T1c 34.5%T2a 9.5%T2b 3.4%Adeno (32.5%)SCC (18.2%)NOS (14.2%)50 Gy/4–5 fx: 89.2%60 Gy/8 fx: 3.4%70 Gy/10 fx: 5.4%Other: 2.0%52.8 monthsNLR ≥1.4, and CRP ≥2.9: worse OS, and PFS.Huang et al.,12 2023RNLRDF, PFSPrognostic286 (43% male, 57% female)76 (69–81)T1 82.2%T2 17.8%Adeno (55.9%)SCC (32.5%)NOS (11.5%)27 Gy/1 fx: 72.8%54 Gy/3 fx: 26.2%38.7 (range: 7.9–66.1) monthsPre-treatment NLR >3.28 was associated with:-lower OS and PFS-higher distant recurrences rateMcLaughlin et al.,15 2020RRILCorrelation of RIL with responsePrognostic40 (45% male, 55% female)69.17 (59–76)≤2 cm: 53.85%;>2 cm: 46.15%Adeno (55%)SCC (17.5%)Other (27.5%)Median total dose, IQR: 50.0 Gy (48–50)1–10 years (range)RIL not prognostic for PFS or OSMonjazeb et al.,21 2023Phase 1 trialTMETreatment responsePredictive20 (45% male, 55% female)76.0 (62.2–88.9)T1b 45%T1c 15%T2a 30%T2b 10%Adeno (65%)SCC (35%)50 Gy/4 fx for peripheral lesions50 Gy/5 fx for central lesions25.8 (range: 7.6–41.0) monthsThreefold higher levels of TME in responders and non-progressorsRutkowski et al.,20 2025PLymphocyte subpopulations + cytokines (subset)Treatment responsePredictive94 (69% male, 31% female)71.0 (range: 52–87)T1 57.4%T2 42.6%Nonquamous: 48.9%Squamous: 31.9%55 Gy/5 fx: 46%54 Gy/3 fx: 28%60 Gy/8 fx: 25%60 Gy/10 fx: 1%17.0 monthsbIncreases in PD-1+ lymphocytes and Th17 upregulation are independent favourable prognostic factorsShaverdian et al.,18 2016RNRL/PLROSPrognostic118 (NA)76.0Most frequent:T1a 54%T1b 26%Adeno: 63%Others: 37%54 Gy/3 fx for peripheral lesions50 Gy/4 fx for central lesions28.9 monthsNLRs >2.18 and PLRs >187.27: poor 3-year OSWang et al.,17 2020RNLRDFS, OSPrognostic69 (67% male, 33% female)74.0 (range: 43–91)T1 69.6%T2 30.4%Adeno: 40.6%SCC: 33.3%Unknown: 26.1%48/4 fx: 24.6%50/5 fx: 37.7%55/5 fx: 17.4%Other: 20.3%30.9 monthsPre-treatment NLR <1.99: better OS, and DFSYan et al.,13 2024RCD44RFSPredictiveDiscovery cohort44 (70.5% male, 29.5% female)Discovery cohort≥70 y: 66%Discovery cohortIA 52.3%IB 43.1%IIA 13.6Discovery cohortAdeno: 63.6%SCC: 36.4%Discovery cohort55/5 fx: 31.8%56/7 fx: 34.1%60/10 fx: 18.2%70/10 fx: 15.9%Discovery cohort36.4 monthsRFS higher:-after SBRT when low CD44 levels-after surgery when high CD44 levelsSBRT Validation cohort52 (48% male, 52% female)SBRT Validation cohort≥70 y: 84.6%SBRT Validation cohortIA 53.9%IB 34.6%IIA 11.5%SBRT Validation cohort adeno: 82.7%SCC: 17.3%SBRTValidation cohort55/5 fx: 5.8%56/7 fx: 26.9%60/10 fx: 40.4%SBRT Validation cohort31.7 monthsZhao et al.,14 2022RLRAssociation of LR with patient outcomePrognostic115 (63.5% male, 36.5% female)71 (64–78)Median diameter 20.2 (15.45–30.0) mmAdeno: 60.0%SCC: 19.1%Unknown: 20.9%50/5 fx: 33%60/10 fx: 67%28.0 (range 5.0–68.0) monthsLR within 6 months after SBRT -improved PFS in patients without lymphopeniaAbbreviation list: Adeno, Adenocarcinoma; CD44, Cluster of Differentiation 44; CRP, C-reactive Protein; DF, Distant Failure; DFS, Disease-Free Survival; fx, Fractions; IQR, Interquartile Range; LR, Lymphocyte Ratio; SCC, Squamous Cell Carcinoma; NLR, Neutrophil-to-Lymphocyte Ratio; OS, Overall Survival; P, Prospective; PD-1, Programmed Cell Death Protein-1; PFS, Progression-Free Survival; PLR, Platelet-to-Lymphocyte Ratio; pts, Patients; R, Retrospective; RFS, Recurrence-Free Survival; RIL, Radiation-Induced Lymphopenia; SBRT, Stereotactic Body Radiation Therapy; T1/T2, Tumor size categories in TNM staging; Th17, T-helper 17; TME, Tumor Microenvironment; TNM, Tumor–Node–Metastasis staging system; y, Years.aWith the exception of the 34 Gy/1 fraction, all other prescription doses for this study should be considered as median values.bThe 17-month follow-up applies only to the immunologic analysis subset; for the whole clinical cohort median follow-up time is 108 months. Study selection process per the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. Table 1Main characteristics of the included studies.Author, and yearStudy typeBiomarkerEndpoint of interestTypeSample sizeAge (Median, IQR)Staging (TNM 8th)HistotypesSBRT scheduleFollow-upMain resultsCannon et al.,19 2015RNRL/PLROSPrognostic59 (53% male, 47% female)70.0 (range: 48–89)IA 43 ptsIB 16 ptsAdeno (34 pts)SCC (19 pts)NOS (6 pts)34 Gy/1 fx: 3 pts54 Gy/3 fx: 33 pts48 Gy/4 fx: 1 pt60 Gy/5 fx: 22 ptsa17.0 monthsmOS- 23 months (NLR >2.98 or PLR >146) vs.mOS- NR (NLR <2.98 or PLR<146)Dong et al.,16 2023RNLR, CRPOS, PFSPrognostic148 (72% male, 28% female)75.5 (range: 47.2–89.0)T1a 4.7%T1b 48.0%T1c 34.5%T2a 9.5%T2b 3.4%Adeno (32.5%)SCC (18.2%)NOS (14.2%)50 Gy/4–5 fx: 89.2%60 Gy/8 fx: 3.4%70 Gy/10 fx: 5.4%Other: 2.0%52.8 monthsNLR ≥1.4, and CRP ≥2.9: worse OS, and PFS.Huang et al.,12 2023RNLRDF, PFSPrognostic286 (43% male, 57% female)76 (69–81)T1 82.2%T2 17.8%Adeno (55.9%)SCC (32.5%)NOS (11.5%)27 Gy/1 fx: 72.8%54 Gy/3 fx: 26.2%38.7 (range: 7.9–66.1) monthsPre-treatment NLR >3.28 was associated with:-lower OS and PFS-higher distant recurrences rateMcLaughlin et al.,15 2020RRILCorrelation of RIL with responsePrognostic40 (45% male, 55% female)69.17 (59–76)≤2 cm: 53.85%;>2 cm: 46.15%Adeno (55%)SCC (17.5%)Other (27.5%)Median total dose, IQR: 50.0 Gy (48–50)1–10 years (range)RIL not prognostic for PFS or OSMonjazeb et al.,21 2023Phase 1 trialTMETreatment responsePredictive20 (45% male, 55% female)76.0 (62.2–88.9)T1b 45%T1c 15%T2a 30%T2b 10%Adeno (65%)SCC (35%)50 Gy/4 fx for peripheral lesions50 Gy/5 fx for central lesions25.8 (range: 7.6–41.0) monthsThreefold higher levels of TME in responders and non-progressorsRutkowski et al.,20 2025PLymphocyte subpopulations + cytokines (subset)Treatment responsePredictive94 (69% male, 31% female)71.0 (range: 52–87)T1 57.4%T2 42.6%Nonquamous: 48.9%Squamous: 31.9%55 Gy/5 fx: 46%54 Gy/3 fx: 28%60 Gy/8 fx: 25%60 Gy/10 fx: 1%17.0 monthsbIncreases in PD-1+ lymphocytes and Th17 upregulation are independent favourable prognostic factorsShaverdian et al.,18 2016RNRL/PLROSPrognostic118 (NA)76.0Most frequent:T1a 54%T1b 26%Adeno: 63%Others: 37%54 Gy/3 fx for peripheral lesions50 Gy/4 fx for central lesions28.9 monthsNLRs >2.18 and PLRs >187.27: poor 3-year OSWang et al.,17 2020RNLRDFS, OSPrognostic69 (67% male, 33% female)74.0 (range: 43–91)T1 69.6%T2 30.4%Adeno: 40.6%SCC: 33.3%Unknown: 26.1%48/4 fx: 24.6%50/5 fx: 37.7%55/5 fx: 17.4%Other: 20.3%30.9 monthsPre-treatment NLR <1.99: better OS, and DFSYan et al.,13 2024RCD44RFSPredictiveDiscovery cohort44 (70.5% male, 29.5% female)Discovery cohort≥70 y: 66%Discovery cohortIA 52.3%IB 43.1%IIA 13.6Discovery cohortAdeno: 63.6%SCC: 36.4%Discovery cohort55/5 fx: 31.8%56/7 fx: 34.1%60/10 fx: 18.2%70/10 fx: 15.9%Discovery cohort36.4 monthsRFS higher:-after SBRT when low CD44 levels-after surgery when high CD44 levelsSBRT Validation cohort52 (48% male, 52% female)SBRT Validation cohort≥70 y: 84.6%SBRT Validation cohortIA 53.9%IB 34.6%IIA 11.5%SBRT Validation cohort adeno: 82.7%SCC: 17.3%SBRTValidation cohort55/5 fx: 5.8%56/7 fx: 26.9%60/10 fx: 40.4%SBRT Validation cohort31.7 monthsZhao et al.,14 2022RLRAssociation of LR with patient outcomePrognostic115 (63.5% male, 36.5% female)71 (64–78)Median diameter 20.2 (15.45–30.0) mmAdeno: 60.0%SCC: 19.1%Unknown: 20.9%50/5 fx: 33%60/10 fx: 67%28.0 (range 5.0–68.0) monthsLR within 6 months after SBRT -improved PFS in patients without lymphopeniaAbbreviation list: Adeno, Adenocarcinoma; CD44, Cluster of Differentiation 44; CRP, C-reactive Protein; DF, Distant Failure; DFS, Disease-Free Survival; fx, Fractions; IQR, Interquartile Range; LR, Lymphocyte Ratio; SCC, Squamous Cell Carcinoma; NLR, Neutrophil-to-Lymphocyte Ratio; OS, Overall Survival; P, Prospective; PD-1, Programmed Cell Death Protein-1; PFS, Progression-Free Survival; PLR, Platelet-to-Lymphocyte Ratio; pts, Patients; R, Retrospective; RFS, Recurrence-Free Survival; RIL, Radiation-Induced Lymphopenia; SBRT, Stereotactic Body Radiation Therapy; T1/T2, Tumor size categories in TNM staging; Th17, T-helper 17; TME, Tumor Microenvironment; TNM, Tumor–Node–Metastasis staging system; y, Years.aWith the exception of the 34 Gy/1 fraction, all other prescription doses for this study should be considered as median values.bThe 17-month follow-up applies only to the immunologic analysis subset; for the whole clinical cohort median follow-up time is 108 months. Main characteristics of the included studies. -lower OS and PFS lower OS and PFS -higher distant recurrences rate higher distant recurrences rate -after SBRT when low CD44 levels after SBRT when low CD44 levels -after surgery when high CD44 levels after surgery when high CD44 levels Abbreviation list: Adeno, Adenocarcinoma; CD44, Cluster of Differentiation 44; CRP, C-reactive Protein; DF, Distant Failure; DFS, Disease-Free Survival; fx, Fractions; IQR, Interquartile Range; LR, Lymphocyte Ratio; SCC, Squamous Cell Carcinoma; NLR, Neutrophil-to-Lymphocyte Ratio; OS, Overall Survival; P, Prospective; PD-1, Programmed Cell Death Protein-1; PFS, Progression-Free Survival; PLR, Platelet-to-Lymphocyte Ratio; pts, Patients; R, Retrospective; RFS, Recurrence-Free Survival; RIL, Radiation-Induced Lymphopenia; SBRT, Stereotactic Body Radiation Therapy; T1/T2, Tumor size categories in TNM staging; Th17, T-helper 17; TME, Tumor Microenvironment; TNM, Tumor–Node–Metastasis staging system; y, Years. With the exception of the 34 Gy/1 fraction, all other prescription doses for this study should be considered as median values. The 17-month follow-up applies only to the immunologic analysis subset; for the whole clinical cohort median follow-up time is 108 months. RoB assessment All studies except one19 met the predefined adequacy threshold for follow-up, supporting the validity of outcome estimates. Overall, methodological quality ranged from moderate to high. Five works12,13,16,18,20 achieved a NOS score ≥8/9, reflecting adequate selection criteria, robust outcome ascertainment, and good control of potential confounders, while two other studies14,17 demonstrated moderate-to-high quality with partial multivariable adjustment. The remaining three studies15,19,21 were rated as moderate quality (6/9), mainly due to limited control of confounders, and in one case, shorter follow-up. Details on the RoB assessment are provided in Supplementary Materials 3. Detailed results for selected immunological-based biomarkers Neutrophil/lymphocyte ratio Five retrospective studies encompassing 680 patients, published between 2015 and 2023, investigated the prognostic significance of the NLR in ES-NSCLC patients undergoing SBRT.12,16, 17, 18, 19 In all studies, pretreatment NLR was derived from peripheral blood samples obtained before SBRT. Across these investigations, elevated pretreatment NLR consistently correlated with inferior oncological outcomes, including OS, DFS, PFS, and distant metastasis-free survival. Proposed NLR thresholds for risk stratification varied between studies, yet consistently fell within a narrow range of 1.4–3.28. Cannon et al. identified a cutoff value of 2.98, above which the median OS declined to 23 months (vs. not reached for NLR <2.98), while Dong et al. reported that NLR ≥3.28 predicted significantly worse OS and a higher risk of distant recurrence.16,19 In addition to NLR, PLR was assessed in two of the included studies.18,19 Elevated PLR demonstrated prognostic relevance in univariate and multivariate models, with cutoff values ranging from 146 to 187.27, each associated with worse OS above those levels. Notably, PLR >146 was significantly associated with reduced median survival (23 months vs. not reached for PLR <146) in the cohort evaluated by Cannon et al.19 Furthermore, Shaverdian et al. observed that elevated CRP (≥2.9) in combination with NLR ≥1.4 portended worsened OS and PFS, highlighting the potential additive role of systemic inflammatory markers.18 CD44 One retrospective study by Yan et al.13 explored the predictive value of CD44 expression in guiding treatment selection between SBRT and surgical resection for patients with early-stage NSCLC. The analysis included 158 patients divided into three cohorts: a discovery SBRT cohort (n = 44), a validation SBRT cohort (n = 52), and a validation surgery cohort (n = 62). In the discovery cohort, high CD44 expression within lymphocyte-rich tumour regions was associated with significantly inferior recurrence-free survival (RFS) and increased recurrence rates. These findings were independently confirmed in the second SBRT validation cohort. Interestingly, the prognostic effect of CD44 appeared treatment-specific. While elevated CD44 expression predicted poor RFS following SBRT, it was associated with improved RFS among patients who underwent surgical resection, suggesting its potential role as a predictive biomarker to guide modality selection. The key clinical implication of the study was the interaction between CD44 expression and treatment modality: patients with low CD44 expression derived greater benefit from SBRT, whereas those with high CD44 expression experienced improved RFS following surgery. Lymphocyte recovery (LR) A study involving 115 patients was conducted to investigate LR following SBRT.14 Among the participants, 33 received SBRT at a dose of 50 Gy administered in 5 fractions, while the remaining patients were treated with 60 Gy over the course of 10 fractions. The results indicated that 100 of the 115 patients experienced a significant reduction in total lymphocyte counts post-SBRT, with values decreasing from 1700 cells/μL before treatment to 1100 cells/μL afterwards. By the six-month follow-up, 44 patients exhibited instances of LR. This finding suggests the potential utility of post-treatment LR as a prognostic biomarker, as a longer duration of radiation treatment was associated with a reduced likelihood of LR within six months post-SBRT in patients diagnosed with ES-NSCLC. Furthermore, LR within this timeframe correlated with improved survival outcomes in patients without lymphopenia. However, this advantageous prognostic association was not observed in those who developed lymphopenia. Radiation-induced lymphopenia Previous research has demonstrated that RIL is most pronounced in patients with low early lymphocyte counts, which are primarily influenced by low radiation doses administered to the heart and lungs when conventional or moderately hypofractionated radiotherapy is utilised.22 One selected study investigated the possible correlation between RIL and clinical outcomes in 40 patients with ES-NSCLC who underwent SBRT.15 These patients underwent four or five treatment fractions with a median dose of 50 Gy. Following treatment, 14 patients experienced RIL, while 26 did not. However, among those who exhibited a decline in lymphocyte counts following SBRT, no statistically significant differences were noted in clinical outcomes, including 2-year local failure (p = 0.265), 2-year nodal failure (p = 0.25), PFS (p = 0.089), and OS (p = 0.690). The authors acknowledged several limitations in the discussion section of their research, including the non-uniform timing of post- and pre-irradiation lymphocyte count measurements, which may have led to the oversight of time-dependent and self-resolving phenomena. Additionally, the ethnic homogeneity of the study population and the single-institution design may restrict the generalisability of the findings to broader population groups. Tumour microenvironment There was one study investigating the correlation between T cell functionality and clinical outcomes following SBRT in NSCLC.21 This prospective, phase 1 trial, which comprised a sample size of 20 patients, involved 6 patients receiving SBRT at a dose of 50 Gy in 4 fractions. At the same time, the remaining participants underwent SBRT with doses ranging from 13 to 50 Gy in 5 fractions. Notably, assessments were conducted on both tumour tissue and peripheral blood. The primary aim of this study was to demonstrate the tolerability, safety, and feasibility of incorporating full-dose atezolizumab with SBRT for medically inoperable patients with ES-NSCLC. Furthermore, the study highlighted that biomarkers indicative of functional adaptive immunity, such as T cell activation within the tumour and the response of circulating T cells to ex vivo stimulation, were strong predictors of treatment efficacy. Responding patients to the combination of atezolizumab and SBRT exhibited levels of proliferating and functional T cells that were three times higher than those in non-responders. Additionally, T cell functionality-specifically CD3+ T cells in the tumour microenvironment and CD8+ T cells in the blood-showed a stronger correlation with oncological outcomes than PD-L1 expression and T cell exhaustion. Immune phenotyping and immune signatures One prospective observational study by Rutkowski et al.20 evaluated systemic and intratumoural immune signatures in a cohort of 94 patients treated with SBRT. Immune profiling was performed on peripheral blood and tumour samples collected at predefined time points (namely, baseline, early post- SBRT and late post-SBRT). The panel of immune parameters included T-cell activation and exhaustion markers (PD-1 expression on CD4+ and CD8+ T cells), regulatory and effector CD4+ subsets (including Th17), and cytokine-associated signatures. Among the considered variables, higher proportions of PD-1+ CD4+ T cells were associated with shorter overall survival, while increased Th17 frequencies correlated with a higher risk of recurrence. No other circulating or tissue-based immune markers demonstrated significant associations with clinical outcomes, and none showed a measurable relationship with SBRT response. Discussion Our review reveals a clear upward trend in biomarker research in this setting, with publications steadily increasing over the past decade. We identified a set of potential predictive biomarkers (namely, CD44 and TME), as well as some possible prognostic biomarkers (namely, NLR, CRP, PLR, LR, and RIL). However, our findings should be interpreted with caution in light of the heterogeneity and relatively small sample sizes of the included studies. Substantial variability across investigations–in terms of study design, patient populations, biomarkers, endpoints, and analytical approaches–limits comparability and generalisability. These features, together with the limited number of studies available for each biomarker, also represent the main rationale for not performing a quantitative meta-analysis, as any pooled estimate would have been of questionable validity. Notably, predictive biomarkers could be a valuable support to candidate ES-NSCLC to either surgery or SBRT, a choice that today relies purely on clinical and operability evaluation.4,23 Importantly, biomarkers may guide patients’ selection towards more aggressive treatment strategies, including but not limited to the use of systemic therapies. In this regard, a recent narrative review from the IASLC ART Subcommittee has examined targeted therapies/drug combinations in NSCLC stages I-III. While prospective trials (e.g., LAURA and POLESTAR) have already focused on stage III, the substantial knowledge gap in stage I disease is far from being closed.24 Given that genomic profiling is hardly implemented in this stage, and that liquid biopsies are not yet clinically validated, the use of less sophisticated yet more accessible biomarkers may represent the most pragmatic strategy to date. Our work-focused specifically on stage I and II- reinforces this statement, showing a clear predominance of blood-based biomarkers. While this is also related to the retrospective nature of most studies, we may argue that blood-count-based descriptors represent an accessible first step in clarifying the mechanisms of SBRT sensitivity and guiding future observational and interventional trials. What emerges from our systematic review is the potential prognostic role of NLR and PLR, similar to what has been described—albeit with contrasting cut-offs—in other cancer types.25, 26, 27 AOrguably, population-specific thresholds may refine the prognostic utility of such markers. Indeed, neutrophil, lymphocyte, and platelet counts are influenced by several modifiers, including—but not limited to—age, smoking habits, concomitant pathological conditions, and certain medications. While most of these factors are difficult to control, they should be taken into account or at least reported. Additionally, literature so far has largely failed to consider biomarker kinetics over time. Not only may time-related modifications be informative in themselves, but adding the temporal dimension may also mitigate the effects of, and contribute to weighing, the above-mentioned confounders between different observation points (e.g., pre-SBRT, immediate post-SBRT, follow-up). In this regard, Zhao et al.14 highlighted the relevance of lymphocyte kinetics following SBRT, showing that prolonged radiation time appears to reduce the likelihood of returning to baseline lymphocyte counts within six months. Moreover, it sheds light on the importance of dose fractionation and radiation time as potential contributors to RIL. Indeed, the generalisability of these findings still needs confirmation in external, comparable cohorts. This setting-so far-has received limited attention, as most studies on RIL have focused mainly on brain and oesophageal cancers, whereas investigations in lung cancer are relatively scarce, and mainly centred on locally-advanced stages.28, 29, 30, 31, 32, 33 Overall, these works suggest that some RT-related factors may actually be associated with RIL, such as beam-on time and the number of treatment sessions,34 as well as the dose received by the heart, the thoracic vertebra, and the lungs.35 In addition to these factors, elucidating the biological mechanisms behind radiation-induced lymphopenia is paramount. Although radiation is increasingly recognised as a potent immunomodulator,36 its effects on circulating immune cells and the tumour microenvironment remain incompletely characterised.37 In mice models, tumour-secred factors, such as Galectin-1 have been associated with T-cell pro-apoptotic activities and intratumoural T-cell depletion, thereby contributing to immune suppression and impaired radiation response in NSCLC.38 Emerging evidence also indicates that tumour metabolic programmes may influence immune responses to RT. For example, a recent work by Sánchez-Castillo et al.39 has suggested that activation of the serine/glycine metabolic pathway supports redox homoeostasis and nucleotide synthesis during radiation-induced stress. In contrast, pharmacologic inhibition of this pathway enhanced radiosensitivity and reshaped the tumour immune microenvironment, partly through modulation of Galectin-1 expression and cytotoxic immune signatures. Taken together, these emerging lines of evidence suggest that integrating tumour metabolism, immune modulation, and radiation-induced systemic effects within a coherent mechanistic framework may help refine our understanding of treatment response and ultimately inform treatment selection in this clinical setting. Another compelling point is the recognition that NSCLC is not a monolithic disease. While the clinical implications of this distinction have been known for almost 20 years,40 to date, such knowledge has not yet translated into biomarker investigations. To the best of our knowledge, the only study discriminating between squamous cell carcinoma and adenocarcinoma in this setting was published in 2024 by Ohashi et al.41 Here, the authors state that a high pre-operative NLR reflects poor infiltration of B and T lymphocytes within the TME, suggesting that this simple blood-based biomarker can shed light on underlying poor immune competence and, hence, a higher risk of recurrence. Notably, however, this appeared to be true for squamous cell carcinoma, but not for adenocarcinomas. This finding is supported by other recent findings of imaging mass cytometry, showing that clinically matched adenocarcinomas and squamous cell carcinoma exhibit distinct patterns of immune cell aggregation within the TME, which correlate with prognosis in a subtype-specific manner.42 Moreover, adenocarcinomas exhibit higher expression of HLA I/II and enhanced antigen presentation by dendritic cells and cytotoxic T/NK cells compared to their squamous counterparts, which consequently display an immune-resistant phenotype.43 We can assume that the role of circulating blood cells may have been masked by limitations in experimental designs, and may therefore be more relevant than expected. Of note, comparable studies in the setting of ablative SBRT are lacking. While elegant in methods and sophisticated in techniques, the study by Monjazeb et al.21 did not implement simpler haematological ratios such as NLR or PLR, which would have been of particular clinical interest. In fact, such a design could help clarify the continuum between low-cost, easily accessible parameters and complex biological correlates. Bridging low- and high-tech biomarkers is also paramount to bring biomarker research from academic centres to smaller facilities, where its impact risks to be limited by high costs, need for specialised immunology and bioinformatic expertise. Nevertheless, in this rapidly evolving landscape, biomarkers are increasingly shedding light on the SBRT-immune system interplay. However, they may also serve as predictors of response to radiation or surgery, thereby supporting more informed decision-making within multidisciplinary oncology boards. This idea has been explored by Yan et al.13 in their recently published investigation on the potential role of lymphocyte CD44 for selecting candidates to either surgery or SBRT. However, some important concern remains. As with any high-dimensional assay, false discovery rate adjustments are necessary to mitigate the chance of pursuing spurious associations; in this study, such strategies could not be applied due to the limited sample size. Moreover, although the authors included internal validation cohorts, the evidence remains derived from a single study without independent external validation. Therefore, the findings should be interpreted with caution. While the observed interaction between CD44 expression and treatment modality is intriguing and suggests a potential predictive role, these results should be considered hypothesis-generating rather than practice-changing, and warrant confirmation in larger, prospective, and externally validated studies. Still, some limitations must be acknowledged. First, the currently available evidence is limited, with a relatively small number of studies and modest sample sizes across most cohorts. As a result, findings should be interpreted with caution. Moreover, substantial clinical and methodological heterogeneity exists across studies, including differences in investigated biomarkers, study design, patient characteristics, endpoints, and analytical approaches, which limits comparability and generalisability. These aspects, together with the limited number of studies available for each biomarker, precluded a meaningful quantitative synthesis and the application of certainty-of-evidence frameworks such as GRADE. Therefore, the available evidence was synthesised qualitatively, with methodological quality assessed using the Newcastle-Ottawa Scale. Additionally, variability in outcome definitions and biomarker thresholds may partially account for inconsistencies across studies, particularly for circulating blood cell count ratios. Another relevant limitation is the lack of histology-specific analyses: although histological subtypes were reported in most studies, only one study41 has stratified patients by histology. In light of recent biological findings, such granularity should become a priority in future research, especially when investigating immune-related biomarkers. Consistent with this, the prospective immune-profiling study by Rutkowski et al. did not identify any biomarker predictive of SBRT response.20 Although PD-1+ CD4 T cells and Th17 frequencies were associated with survival and recurrence, these parameters behaved as general prognostic markers and did not modify the effect of stereotactic ablative RT. This observation further supports the notion that most immune correlates described to date likely reflect baseline host immune status rather than treatment-specific sensitivity. Furthermore, the non-druggable nature of most biomarkers in radiation oncology is one of the major challenges in this area of study. As SBRT has achieved high local control rates (up to 90% for small lesions) regardless of histology, there is little incentive for pharmaceutical trials and no reimbursement pathways specifically dedicated to radiotherapy-related biomarkers. Finally, as in all systematic reviews, it is essential to recall the possibility of publication bias, which may inflate the apparent strength of reported evidence. Overall, these limitations reinforce the exploratory nature of the current evidence base and highlight the need for larger, prospective, and methodologically robust studies to validate the clinical utility of immunological biomarkers in early-stage NSCLC treated with SBRT. Despite the above-mentioned limitations, the novelty of the present work lies in its focus. While biomarker studies in NSCLC are abundant, none have systematically appraised their prognostic and predictive role in the context of curative SBRT for early-stage disease. By filling this gap, our work provides both a snapshot of current knowledge and a foundation for future research. What clearly emerges is the need for larger studies, considering both SBRT and surgery. Rather, these state-of-the-art treatment modalities should now be integrated within biomarker-driven discovery trials, where treatment allocation is informed not only by clinical parameters and technical operability, but also by tumour biology. Some ongoing trials, including the observational MONDRIAN trial, are already moving in this direction.44 Their results will clarify whether bridging simple blood-based indices with more sophisticated immune profiling can ultimately translate into biomarker-informed decision-making in ES-NSCLC. Looking forward, only prospective, multi-institutional efforts with standardised biomarker assessment will allow these exploratory signals to evolve into validated tools, eventually guiding guideline recommendations and multidisciplinary practice. Contributors SV and MB contributed to the conceptualisation of the study. SV and SL accessed and verified the underlying data. Formal analysis was performed by SV and MB, while funding acquisition was secured by SV. Investigation was conducted by MO and LC. The methodology was developed by SV, MB, and MGV. Supervision was provided by SV and GP. Validation was carried out by MAZ, LP, PF, AZ, RN, ML, EM, and MC. MGV contributed to data visualisation. The original draft of the manuscript was prepared by SV, SL, and MB, while review and editing were performed by SV, KV, LS, TB, NF, RO, and BAJF. All authors read and approved the final version of the manuscript. Data sharing statement Data supporting the findings of this study are available from the corresponding author upon reasonable request. Declaration of interests The Division of Radiation Oncology of IEO received research funding from AIRC (Italian Association for Cancer Research), Fondazione IEO-CCM (Istituto Europeo di Oncologia-Centro Cardiologico Monzino), IBA and Accuray Inc all outside the current project. BAJF received institutional research funding from Accuray, AIRC, and IBA; honoraria for lectures, educational activities, and manuscript review from Accuray, Astellas Pharma, Bayer, AstraZeneca, IBA, Ipsen, Tecnologie Avanzate, Recordati, Novartis, Daiichi Sankyo, and Elsevier; and advisory board/scientific committee fees from Ipsen, AstraZeneca, Bayer, Seagen, Astellas Pharma, Radformation Inc., and ESO. LP, AZ and PF are supported by a fellowship from AIRC within the MONDRIAN trial (Multi-omics integrative modelling for stereotactic body radiotherapy in early-stage non-small cell lung cancer–AIRC 28282–NCT05974475). The sponsors did not play any role in the study design, collection, analysis and interpretation of data, nor in the writing of the manuscript, nor in the decision to submit the manuscript for publication. All other authors declare no competing interests.