Recurrence-Free Survival in Patients With Surgically Resected Non-Small Cell Lung Cancer

BackgroundStandard treatment for early-stage or locoregionally advanced non-small cell lung cancer (NSCLC) includes surgical resection. Recurrence after surgery is commonly reported, but a summary estimate for postsurgical recurrence-free survival (RFS) in patients with NSCLC is lacking.Research QuestionWhat is the RFS after surgery in patients with stage I-III NSCLC at different time points, and what factors are associated with RFS?Study Design and MethodsA systematic search was performed in MEDLINE, EMBASE, and Cochrane databases between January 2011 and June 2021. The primary outcome was RFS at 1, 2, 3, and 5 years postresection. Single-arm, random-effects meta-analyses were done to calculate effect estimates and 95% CIs. Analyses were stratified by stage/substage as per the AJCC Cancer Staging Manual, and RFS was estimated (1) after pooling studies, using 7th or 8th edition staging criteria; and (2) among studies using only the 8th edition. Meta-regressions were performed to assess associations between RFS and patient demographic/clinical characteristics of interest.ResultsData from 471 studies comprising 1,060 surgical study arms were extracted. RFS estimates from 60,695 patients staged with the 7th or 8th edition were analyzed. RFS ranged from 96% at 1 year postresection to 82% at 5 years for stage I, and from 68% at 1 year to 34% at 5 years for stage III. Estimates for patients staged using only 8th edition criteria were slightly higher. Older age, higher percentage of male patients, advancing stage, larger tumor size, and geographical region (North America/Europe vs Asia) were significantly associated with worse RFS.InterpretationThis study presents a comprehensive assessment of reported RFS from published clinical literature, offering estimates at multiple postsurgical time points and by geographical region. Findings can inform treatment decisions, clinical trial design, and future research to improve outcomes among patients with NSCLC. Standard treatment for early-stage or locoregionally advanced non-small cell lung cancer (NSCLC) includes surgical resection. Recurrence after surgery is commonly reported, but a summary estimate for postsurgical recurrence-free survival (RFS) in patients with NSCLC is lacking. What is the RFS after surgery in patients with stage I-III NSCLC at different time points, and what factors are associated with RFS? A systematic search was performed in MEDLINE, EMBASE, and Cochrane databases between January 2011 and June 2021. The primary outcome was RFS at 1, 2, 3, and 5 years postresection. Single-arm, random-effects meta-analyses were done to calculate effect estimates and 95% CIs. Analyses were stratified by stage/substage as per the AJCC Cancer Staging Manual, and RFS was estimated (1) after pooling studies, using 7th or 8th edition staging criteria; and (2) among studies using only the 8th edition. Meta-regressions were performed to assess associations between RFS and patient demographic/clinical characteristics of interest. Data from 471 studies comprising 1,060 surgical study arms were extracted. RFS estimates from 60,695 patients staged with the 7th or 8th edition were analyzed. RFS ranged from 96% at 1 year postresection to 82% at 5 years for stage I, and from 68% at 1 year to 34% at 5 years for stage III. Estimates for patients staged using only 8th edition criteria were slightly higher. Older age, higher percentage of male patients, advancing stage, larger tumor size, and geographical region (North America/Europe vs Asia) were significantly associated with worse RFS. This study presents a comprehensive assessment of reported RFS from published clinical literature, offering estimates at multiple postsurgical time points and by geographical region. Findings can inform treatment decisions, clinical trial design, and future research to improve outcomes among patients with NSCLC. Take-home PointsStudy Questions: What are the RFS estimates after surgery in patients with clinical stage I-III NSCLC at various time points, and what factors are associated with RFS?Results: For patients staged with the AJCC 7th or 8th edition, RFS ranged from 95% at 1 year postresection to 81% at 5 years for stage I, and from 68% at 1 year to 34% at 5 years for stage III, with slightly higher estimates for patients staged using only the 8th edition criteria. Older age, higher percentage of male patients, advancing stage, larger tumor size, and geographical region (North America or Europe vs Asia) were significantly associated with worse RFS.Interpretation: This study presents a comprehensive assessment of reported RFS from published clinical literature that can inform treatment decisions, clinical trial design, and future research to improve outcomes among patients with NSCLC. Study Questions: What are the RFS estimates after surgery in patients with clinical stage I-III NSCLC at various time points, and what factors are associated with RFS? Results: For patients staged with the AJCC 7th or 8th edition, RFS ranged from 95% at 1 year postresection to 81% at 5 years for stage I, and from 68% at 1 year to 34% at 5 years for stage III, with slightly higher estimates for patients staged using only the 8th edition criteria. Older age, higher percentage of male patients, advancing stage, larger tumor size, and geographical region (North America or Europe vs Asia) were significantly associated with worse RFS. Interpretation: This study presents a comprehensive assessment of reported RFS from published clinical literature that can inform treatment decisions, clinical trial design, and future research to improve outcomes among patients with NSCLC. Lung cancer is the second most common cancer worldwide and the leading cause of cancer-related death.1Sung H. Ferlay J. Siegel R.L. et al.Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries.CA Cancer J Clin. 2021; 71: 209-249Crossref PubMed Scopus (45055) Google Scholar The two primary histologic subtypes are non-small cell lung cancer (NSCLC) and small-cell lung cancer (SCLC).2Lemjabbar-Alaoui H. Hassan O.U. Yang Y.W. Buchanan P. Lung cancer: biology and treatment options.Biochim Biophys Acta. 2015; 1856: 189-210PubMed Google Scholar Surgical resection is part of the standard treatment for early-stage or locoregionally advanced NSCLC in patients fit enough to tolerate an operation; however, postoperative recurrence is common and ranges from 30% to 75%.3Uramoto H. Tanaka F. Recurrence after surgery in patients with NSCLC.Transl Lung Cancer Res. 2014; 3: 242-249PubMed Google Scholar,4Sasaki H. Suzuki A. Tatematsu T. et al.Prognosis of recurrent non-small cell lung cancer following complete resection.Oncol Lett. 2014; 7: 1300-1304Crossref PubMed Scopus (39) Google Scholar These estimates are derived from a few studies, most published more than 20 years ago. Two more systematic reviews included six studies5Stirling R.G. Chau C. Shareh A. Zalcberg J. Fischer B.M. Effect of follow-up surveillance after curative-intent treatment of NSCLC on detection of new and recurrent disease, retreatment, and survival: a systematic review and meta-analysis.J Thorac Oncol. 2021; 16: 784-797Abstract Full Text Full Text PDF PubMed Scopus (0) Google Scholar and 11 studies,6Hu J. Miao H. Li R. Wen Z. Surgery and subsequent risk of non-small cell lung cancer recurrence: a meta-analysis of observational studies.Transl Cancer Res. 2020; 9: 1960-1968Crossref PubMed Scopus (0) Google Scholar respectively, and neither published a summary estimate for recurrence-free survival (RFS). Compared with recurrence rates, RFS is reported at fixed time points, making it easier to meta-analyze than recurrence rates, which are often presented as the total number of events that occurred during study follow-up, with variable durations. RFS also captures deaths, which makes it a potential surrogate end point for overall survival (OS). With the implementation of the 8th edition of the American Joint Commission on Cancer (AJCC) TNM staging system as the standard classification for lung cancers in the United States in January 2018,7Detterbeck F.C. The eighth edition TNM stage classification for lung cancer: what does it mean on Main Street?.J Thorac Cardiovasc Surg. 2018; 155: 356-359Abstract Full Text Full Text PDF PubMed Scopus (100) Google Scholar comparisons between studies may be impacted.8Lim W. Ridge C.A. Nicholson A.G. Mirsadraee S. The 8th lung cancer TNM classification and clinical staging system: review of the changes and clinical implications.Quant Imaging Med Surg. 2018; 8: 709-718Crossref PubMed Scopus (155) Google Scholar A contemporary and comprehensive review of recurrence outcomes in patients with surgically resected NSCLC is lacking. The aim of this systematic literature review (SLR) and meta-analysis was to summarize and analyze RFS and recurrence rates after surgery in patients with NSCLC over time by stage, incorporating different staging classifications (7th and 8th edition), and to examine factors associated with RFS. Institutional review board approval was not required for this work because no participants were recruited. An SLR protocol was not registered, but the review followed predefined screening and extraction criteria. This SLR and meta-analysis were conducted on the basis of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines.9Page M.J. McKenzie J.E. Bossuyt P.M. et al.The PRISMA 2020 statement: an updated guideline for reporting systematic reviews.BMJ. 2021; 372: n71Crossref PubMed Scopus (22975) Google Scholar A systematic search was conducted by an information specialist and peer-reviewed by a second information specialist, using PRESS (Peer Review of Electronic Search Strategies) guidelines.10McGowan J. Sampson M. Salzwedel D.M. Cogo E. Foerster V. Lefebvre C. PRESS Peer Review of Electronic Search Strategies: 2015 guideline statement.J Clin Epidemiol. 2016; 75: 40-46Abstract Full Text Full Text PDF PubMed Google Scholar The detailed search strategy is presented in e-Tables 1 and 2. Example search terms used included key words and MeSH terms such as: "lung cancer," "cancer recurrence," "surgery," and "survival analysis." MEDLINE, EMBASE, and Cochrane databases were searched for English language studies conducted between January 2011 and June 2021, based on predefined PICOS (Population, Intervention, Comparator, Outcomes, and Study Design) criteria (e-Table 3). Studies enrolling adults with clinical stage I-III NSCLC who underwent lung surgery alone or in combination with other therapies (neoadjuvant or adjuvant) were eligible for inclusion. Resections of interest were wedge, segmentectomy, lobectomy, and pneumonectomy, with inclusion of all surgical approaches (video-assisted thoracoscopic surgery, robot-assisted thoracoscopic surgery, and thoracotomy). Eligible studies could be either single- or multiarm studies comparing distinct treatment methods. Studies that included a mix of surgery types were also included. Only studies including surgeries for primary lung cancer were included. Studies enrolling patients undergoing surgery for recurrent tumors or diagnostic purposes were excluded. Eligible studies must have reported one or more of the following recurrence outcomes: local, locoregional, regional, distant recurrence, local tumor recurrence, or RFS. Studies that reported disease-free survival (DFS) or progression-free survival (PFS) instead of RFS were also included, as these outcomes are often used interchangeably in practice. Included studies were single-arm, comparative observational studies, or randomized trials, whereas case reports, narrative reviews, and preclinical studies were excluded. Observational studies were included to capture all available evidence on surgery in patients with NSCLC. Studies focused on evaluating risk factors and predictors of recurrence outcomes were excluded. For a detailed list of inclusion and exclusion criteria, refer to e-Table 3. The search strategy was supplemented with manual screening of published reviews and meta-analysis reference lists for relevant publications to ensure completeness. To increase precision, included studies were restricted to those with a sample size of at least 100 surgically treated patients, or at least 50 in each arm if the study included multiple surgery arms. Studies with duplicate data were excluded if they overlapped with another publication. Screening of identified records was performed by two independent reviewers, using the DistillerSR platform (DistillerSR),11DistillerSR, version 2.35.https://www.distillersr.com/Date: 2021Date accessed: May 23, 2023Google Scholar with discrepancies resolved by a third reviewer. Records were screened for title and abstract eligibility, after which eligible studies were sought for retrieval. Records that could be retrieved and fit the PICOS criteria were then assessed for full-text eligibility. Only studies reporting staging data according to the 7th or 8th AJCC TNM classification were included in the meta-analysis summary RFS estimate. RFS was estimated from date of surgery to the date of recurrence or death from any cause at 1, 2, 3, and 5 years postresection, loss to follow-up, or end of study, whichever occurred first. Studies that censored non-cancer-related or all-cause death from RFS were not included in the meta-analysis, due to competing risk, which can occur for survival analysis of RFS if death is excluded as an event and instead censored. Censoring death in an RFS analysis is informative censoring, which may cause bias in the estimated RFS when using the Kaplan-Meier (KM) method because it violates the assumption of noninformative censoring (ie, a patient who is censored is representative of patients who remain uncensored).12Campigotto F. Weller E. Impact of informative censoring on the Kaplan-Meier estimate of progression-free survival in phase II clinical trials.J Clin Oncol. 2014; 32: 3068-3074Crossref PubMed Scopus (39) Google Scholar,13Leung K.M. Elashoff R.M. Afifi A.A. Censoring issues in survival analysis.Annu Rev Public Health. 1997; 18: 83-104Crossref PubMed Scopus (204) Google Scholar Studies in which the methods were unclear about handling competing risk were extracted and included in the meta-analysis. Extraction of key baseline characteristics (ie, geographical region, number of male patients, average follow-up duration, tumor stage) and outcome data (ie, RFS/DFS/PFS, local progression-free survival [LPFS]; and locoregional, distant, and overall recurrence rates) from identified studies was performed in Excel (Microsoft). Recurrence data from KM or cumulative incidence curves were digitized with DigitizeIt software (version 2.5.3; Bormisoft) to accurately estimate outcome rates at time points of interest. All analyses were performed with the meta package in R (version 3.6.1; R Foundation). Single-arm random-effects meta-analyses were conducted by multiplying the proportion of surviving patients without recurrence at each time point by the initial population sample size. Inverse variance 95% CIs were calculated. Study weighting in the meta-analyses was based on the initial population because complete censoring data were not available for all studies. Study arms were the unit of analysis, and studies could contribute multiple arms with distinct resection types. Results from the single-arm meta-analyses are presented as column graphs categorized by stage and time postresection. Heterogeneity was assessed with I2 values, and variability in study-level patient baseline characteristics was addressed by the use of multivariable meta-regressions. Meta-regressions were conducted to examine associations between RFS estimates and study-level patient age (1-year intervals), proportion of male patients, stages II and III vs stage I, average tumor size (1-cm intervals), study region (Asia, Europe, and North America), surgery type (lobectomy, sublobar resection, pneumonectomy, and mixed or unspecified surgical procedures), and use of adjuvant or neoadjuvant chemotherapy. Results for each meta-regression are reported as ORs with 95% CI, with OR < 1 indicating worse RFS and OR > 1 indicating better RFS. P values ≤ 0.05 were considered statistically significant. A summary of all meta-regression results by covariate and time intervals is presented as forest plots. Subgroup analyses, sensitivity analyses, and quality assessment of included studies were conducted to ensure robustness of the results (e-Table 4). Publication bias was also assessed. Refer to e-Appendix 1 for a detailed description of these analyses. The search identified 8,871 unique records, of which 771 were assessed for full-text eligibility. Data from 471 studies encompassing 1,060 surgical study arms were extracted (Fig 1). Included studies were conducted primarily in Asia (70%), followed by North America (18%) and Europe (10%). Most studies were retrospective (87%) and were conducted in single-center settings (75%). More than two-thirds (69%) of study arms included between 50 and 250 patients. The average follow-up duration was 31 to 50 months in 41% (n = 437) of the study arms and not reported in 16% (n = 171). RFS data from these study arms were included in the meta-analyses and meta-regressions. The average mean age was 64 years across all study arms. There were 523 and 255 study arms that used the 7th and 8th edition AJCC TNM classification, respectively. The most frequently reported disease stage was stage I, accounting for 32% (168 of 523) and 38% (97 of 255) of the 7th and 8th edition study arms, respectively. Few 8th edition study arms (n = 15) reported recurrence rates for clinical substages IA1, IA2, and IA3. Only pathologic staging was reported in 42% of 7th edition and 24% of 8th edition study arms, which were not included in this analysis. No staging information was reported in 7% of 7th edition study arms and 8% of 8th edition study arms. RFS estimates from 60,695 patients (n = 161 study arms) staged with AJCC 7th or 8th edition were analyzed, with stage I patients representing 79% of the sample. Death was included as an event in half of the study arms, while another 38% had unclear methods. Results for 3- and 5-year RFS are shown in Figure 2. RFS declined sharply with advancing stage at both the 3-year (87% for stage I, 58% for stage II, and 38% for stage III) and 5-year time points (81% for stage I, 50% for stage II, and 34% for stage III). A similar trend was observed at 1- and 2-year follow-ups (e-Fig 1). The highest RFS during follow-up was observed for stage IA (96% at 1 year and 86% at 5 years), whereas only 34% of stage III patients were alive without disease recurrence at 5 years (see forest plots in e-Figs 2-6). The unadjusted single-arm meta-analyses demonstrated high I2 values indicative of significant heterogeneity, and thus meta-regressions were performed to control for differences in study-level patient baseline characteristics. RFS estimates at 3- and 5-year follow-ups for 8th-edition-only study arms were similar for stage IA patients and overall stage I patients but higher for stage III patients than were pooled 7th and 8th edition results (Fig 3). This finding was also observed at 1- and 2-year follow-ups (e-Fig 7). Only six study arms reported RFS data for clinical substages IA1, IA2, and IA3; the small sample sizes precluded performing meta-analyses (Table 1, e-Table 5). Whereas RFS ranges for stage IA3 were lower than expected relative to stage IA1 and comparable to the single-arm meta-analysis results for stage IB, the median of the mean patient age in stage IA3 study arms was 67.8 years, compared with 58.4 years in the stage IA1 study arm and 64.4 years in the stage IB study arm.Table 1Recurrence-Free Survival by Time for Clinical Substages IA1, IA2, and IA3Clinical Stage3-Year RFS5-Year RFSStudy ArmsPatientsRFS Range (%)Study ArmsPatientsRFS Range (%)IA1455694.9-100455681.7-100IA2562675.8-100562662.5-97.9IA3347466.8-74.9347453.6-61.8RFS = recurrence-free survival. Open table in a new tab RFS = recurrence-free survival. Locoregional, distant, and overall recurrence ranges are shown in Table 2. Recurrence rates ranged widely, but summary estimates were not computed because rates were seldom reported by time point. Although distant recurrence is generally higher than locoregional recurrence, recurrence rate ranges should be interpreted with caution because of lack of time-standardized reporting.Table 2Proportion of Recurrence by Recurrence TypeClinical Stage (Staging Edition)Locoregional RecurrenceDistant RecurrenceOverall RecurrenceStudy ArmsPatientsRecurrence Range (%)Study ArmsPatientsRecurrence Range (%)Study ArmsPatientsRecurrence Range (%)IA (7th + 8th)295,2301-12.2379,9400-13.84612,3430-26.3IB (7th + 8th)230710.2-12.923072-6.3692512.7-34.5I (7th + 8th)7417,2490-27.58824,1950-3111429,1520-56.4II (7th + 8th)112235.2236728-46112244.3III (7th + 8th)61,3868.4-47.972,09630-62.8123,65232-65.4 Open table in a new tab Time-specific LPFS at 5 years was extracted for 56 study arms, but all but two study arms (96%) did not report results by stage, making estimates difficult to interpret. Therefore, meta-analyses of RFS by recurrence type were not computed. In the multivariable meta-regressions (Figs 4A, 4B), statistically significant worse odds of RFS were observed with increasing age, increasing tumor size, and North American/European study arms (vs Asian) at 5-year follow-up. The odds of RFS at 5 years from North America/Europe were 62% and 59% lower (P ≤ .001 and P = .03), respectively, compared with Asian arms. Similar findings were observed at 3-year follow-up, at which time an increased proportion of male patients and more advanced stage (stages II and III vs stage I) were additionally associated with worse RFS, whereas the OR estimate for European study arms (vs Asian) remained less than 1 but was not significant. Study arms treating patients with mixed approaches (vs lobectomy) at 5-year follow-up were associated with better RFS. Advancing stage was associated with 59% worse odds of RFS at 3 years and 46% worse odds of RFS at 5 years, and increasing tumor size was associated with 40% and 54% worse odds of RFS at the same time points. Similar findings on advancing stage and increasing tumor size associated with worse RFS were observed at earlier time points (e-Fig 8). Subgroup analyses demonstrated that RFS estimates across all time points for stage IA and I patients were consistently higher in Asian vs North American/European study arms (year 5: 87% vs 60% and 83% vs 71%, respectively; see e-Appendix 2 and e-Fig 9 for more information). Compared with the primary analysis, RFS point estimates in sensitivity analyses that included only study arms with no competing risk were slightly lower for each stage at each corresponding time point (e-Figs 10, 11). RFS point estimates from analyses of studies (n = 61) scoring ≥ 9, using the MINORS (Methodological Index for Non-Randomized Studies) instrument, were comparable to the primary analysis for most stages and time points (e-Fig 12, e-Table 6). Publication bias was detected in analyses of stage IA and I study arms, and there was a high level of agreement between the Egger's test and LFK index for most time points (e-Figs 13-16, e-Table 7). Refer to the online article for additional details on the subgroup, sensitivity, quality assessment, and publication bias analyses. To date, this is the largest SLR and meta-analysis analyzing RFS estimates after surgery in patients with NSCLC with early-stage or locoregionally advanced disease. This comprehensive assessment of RFS included 85 studies and recurrence data, extracted from published clinical literature, of more than 60,000 surgically treated patients with NSCLC. Overall, the 5-year RFS was 86% for stage IA, 62% for stage IB, 81% for stage I, 50% for stage II, and 34% for stage III patients. As expected, we observed a steep decline in RFS between stage I and stage III disease at all time points of interest. Several reviews have published summary estimates for the rate of postsurgical recurrence, but a summary estimate of RFS is lacking. Uramoto and Tanaka3Uramoto H. Tanaka F. Recurrence after surgery in patients with NSCLC.Transl Lung Cancer Res. 2014; 3: 242-249PubMed Google Scholar reported that 30% to 55% of patients with NSCLC develop recurrence and die of their disease despite curative resection. Sasaki and colleagues4Sasaki H. Suzuki A. Tatematsu T. et al.Prognosis of recurrent non-small cell lung cancer following complete resection.Oncol Lett. 2014; 7: 1300-1304Crossref PubMed Scopus (39) Google Scholar and Stirling and colleagues5Stirling R.G. Chau C. Shareh A. Zalcberg J. Fischer B.M. Effect of follow-up surveillance after curative-intent treatment of NSCLC on detection of new and recurrent disease, retreatment, and survival: a systematic review and meta-analysis.J Thorac Oncol. 2021; 16: 784-797Abstract Full Text Full Text PDF PubMed Scopus (0) Google Scholar presented broader recurrence ranges of 30% to 75% and 17.8% to 71%, respectively. The decision to focus on RFS in this study instead of overall recurrence was due to a lack of published studies meta-analyzing RFS, which is of interest to physicians and patients. Furthermore, Food and Drug Administration guidance has recommended the assessment of RFS as a clinical trial end point in certain contexts, stating it may serve as a surrogate end point for traditional/accelerated approval.14US Food and Drug AdministrationClinical Trial Endpoints for the Approval of Cancer Drugs and Biologics: Guidance for Industry.https://www.fda.gov/regulatory-information/search-fda-guidance-documents/clinical-trial-endpoints-approval-cancer-drugs-and-biologicsDate: 2018Date accessed: December 15, 2023Google Scholar Although disease recurrence is a simpler outcome that does not include non-cancer-related deaths and is less affected by the competing risk issues impacting RFS, the cumulative incidence of recurrence is infrequently reported in the NSCLC literature. In addition, studies that do report recurrence usually state the total number of recurrences during the follow-up period. Reporting in this way makes comparisons between studies and meta-analytic synthesis of data challenging because of variable or unknown follow-up durations. These limitations suggest that recurrence is less suitable than RFS for comparing treatment efficacy in the surgically treated NSCLC population. Similar to reporting of RFS, there should be an effort to standardize reporting of recurrence rates by time, as this information would be of high utility to both clinicians and patients. This SLR presents a robust summary of contemporary recurrence data, as represented by analyses of 8th edition-staged patients from more than 50 study arms. Patients classified using the 8th edition only had similar, but slightly higher RFS estimates compared with pooled 7th and 8th edition results, directionally aligning with the expected difference in disease severity. This may be explained by some 7th edition stage IB patients with more advanced disease being reclassified into 8th edition stage IIA, thus "decreasing" the overall disease severity of 8th edition stage I and IB cohorts.15Goldstraw P. Chansky K. Crowley J. et al.The IASLC Lung Cancer Staging Project: proposals for revision of the TNM stage groupings in the forthcoming (eighth) edition of the TNM Classification for Lung Cancer.J Thorac Oncol. 2016; 11: 39-51Abstract Full Text Full Text PDF PubMed Google Scholar Similarly, patients with more advanced stage IIB disease based on the 7th edition classification would be upstaged to 8th edition stage IIIA.15Goldstraw P. Chansky K. Crowley J. et al.The IASLC Lung Cancer Staging Project: proposals for revision of the TNM stage groupings in the forthcoming (eighth) edition of the TNM Classification for Lung Cancer.J Thorac Oncol. 2016; 11: 39-51Abstract Full Text Full Text PDF PubMed Google Scholar Multivariable meta-regressions indicated that increasing age, more advanced disease stage, and increasing tumor size were independently associated with worse RFS at all time points, consistent with previous studies.16Consonni D. Pierobon M. Gail M.H. et al.Lung cancer prognosis before and after recurrence in a population-based setting.J Natl Cancer Inst. 2015; 107: djv059Crossref PubMed Google Scholar, 17Dziedzic D.A. Rudzinski P. Langfort R. Orlowski T. Risk factors for local and distant recurrence after surgical treatment in patients with non-small-cell lung cancer.Clin Lung Cancer. 2016; 17: e157-e167Abstract Full Text Full Text PDF PubMed Google Scholar, 18Kent M.S. Mandrekar S.J. Landreneau R. et al.A nomogram to predict recurrence and survival of high-risk patients undergoing sublobar resection for lung cancer: an analysis of a multicenter prospective study (ACOSOG Z4032).Ann Thorac Surg. 2016; 102: 239-246Abstract Full Text Full Text PDF PubMed Scopus (0) Google Scholar, 19Schuchert M.J. Normolle D.P. Awais O. et al.Factors influencing recurrence following anatomic lung resection for clinical stage I non-small cell lung cancer.Lung Cancer. 2019; 128: 145-151Abstract Full Text Full Text PDF PubMed Scopus (0) Google Scholar, 20Varlotto J. Fakiris A. Flickinger J. et al.Matched-pair and propensity score comparisons of outcomes of patients with clinical stage I non-small cell lung cancer treated with resection or stereotactic radiosurgery.Cancer. 2013; 119: 2683-2691Crossref PubMed Scopus (89) Google Scholar, 21Wu C.F. Fu J.Y. Yeh C.J. et al.Recurrence risk factors analysis for stage I non-small cell lung cancer.Medicine (Baltimore). 2015; 94: e1337Crossref PubMed Scopus (39) Google Scholar Although study arms treating patients with mixed surgical approaches (vs lobectomy) were associated with better RFS at 5-year follow-up, the association was not significantly different at any other time point and therefore must be interpreted with caution. Most of the meta-regression data were from the Asian literature, representing 80% to 82% of included studies. Asian studies consistently reported higher RFS estimates at all time points compared with North American/European studies in our analyses and in the literature. For example, a Japanese trial (JCOG0802) enrolling 1,106 patients with peripheral stage IA disease reported a 5-year DFS of 88% (87% in our analysis).22Saji H. Okada M. Tsuboi M. et al.Segmentectomy versus lobectomy in small-sized peripheral non-small-cell lung cancer (JCOG0802/WJOG4607L): a multicentre, open-label, phase 3, randomised, controlled, non-inferiority trial.Lancet. 2022; 399: 1607-1617Abstract Full Text Full Text PDF PubMed Scopus (450) Google Scholar However, a North American/Australian trial (CALGB 140503) reported 5-year relapse-free survival of 63.6% and 64.1% after sublobar and lobar resection, respectively, in more than 1,000 patients with clinical stage IA disease (60% in our analysis).23Altorki N. Wang X. Kozono D. et al.Lobar or sublobar resection for peripheral stage IA non-small-cell lung cancer.N Engl J Med. 2023; 388: 489-498Crossref PubMed Scopus (144) Google Scholar Our analysis is aligned with these studies and others that demonstrated higher RFS estimates across all time points for stage I patients in Asian study arms. Improved RFS and relapse outcomes in Asian patients may be due to the higher prevalence of those who never smoked,24Zhou F. Zhou C. Lung cancer in never smokers—the East Asian experience.Transl Lung Cancer Res. 2018; 7: 450-463Crossref PubMed Scopus (87) Google Scholar younger age at diagnosis,25Zhou W. Christiani D.C. East meets West: ethnic differences in epidemiology and clinical behaviors of lung cancer between East Asians and Caucasians.Chin J Cancer. 2011; 30: 287-292Crossref PubMed Google Scholar and higher incidence of ground-glass opacities,26Lui N.S. Benson J. He H. et al.Sub-solid lung adenocarcinoma in Asian versus Caucasian patients: different biology but similar outcomes.J Thorac Dis. 2020; 12: 2161-2171Crossref PubMed Scopus (7) Google Scholar which are associated with better prognosis and lower recurrence rates than are pure solid tumors.22Saji H. Okada M. Tsuboi M. et al.Segmentectomy versus lobectomy in small-sized peripheral non-small-cell lung cancer (JCOG0802/WJOG4607L): a multicentre, open-label, phase 3, randomised, controlled, non-inferiority trial.Lancet. 2022; 399: 1607-1617Abstract Full Text Full Text PDF PubMed Scopus (450) Google Scholar,27Berry M.F. Gao R. Kunder C.A. et al.Presence of even a small ground-glass component in lung adenocarcinoma predicts better survival.Clin Lung Cancer. 2018; 19: e47-e51Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar, 28Hattori A. Matsunaga T. Takamochi K. Oh S. Suzuki K. Importance of ground glass opacity component in clinical stage IA radiologic invasive lung cancer.Ann Thorac Surg. 2017; 104: 313-320Abstract Full Text Full Text PDF PubMed Scopus (106) Google Scholar, 29Shigefuku S. Shimada Y. Hagiwara M. et al.Prognostic significance of ground-glass opacity components in 5-year survivors with resected lung adenocarcinoma.Ann Surg Oncol. 2021; 28: 148-156Crossref PubMed Scopus (20) Google Scholar Our systematic review revealed that stage I and IA patients enrolled in Asian studies were younger than their North American/European counterparts, but smoking status and incidence of ground-glass opacities could not be controlled in the meta-regressions. To our knowledge, our study is the most comprehensive and robust summary of postsurgical recurrence outcomes in patients with NSCLC presented to date. The SLR followed prespecified PICOS criteria and was conducted according to PRISMA 2020 guidelines. Summaries of RFS at different time points allow clinicians and patients to better understand how disease prognosis changes over time and provide a standardized measurement of recurrence to inform future research. In addition, presentation of separate analyses of pooled 7th and 8th edition results and more contemporary 8th edition results may inform how changes in staging edition can influence recurrence estimates. The analyses were further reinforced by using multivariable meta-regressions to control for confounding factors, by using inclusion criteria that considered issues of competing risk, and by conducting quality assessment and publication bias analysis of included studies. We excluded studies that used pathologic staging alone because treatment decisions are based on clinical staging. In addition, clinical trials typically enroll patients based on clinical staging before randomization. The limited number of 8th edition study arms resulted in 7th edition results being substantially better represented in the pooled 7th and 8th edition analyses. It was not possible to reclassify all patients staged using 7th edition criteria to the corresponding 8th edition classification, as few studies provided patient-level data. Specifically, the proportion of patients who were classified as stage IB or IIB, using 7th edition criteria, and subsequently upstaged to 8th edition stage IIA or IIIA, could not be calculated. Consequently, the cohort of patients included in the pooled 7th and 8th edition analyses is not completely representative of the overall population with disease categorized according to 8th edition staging. One aim of this study was to summarize locoregional, distant, and overall recurrence rates, but interpretation of the reported ranges is limited by most studies reporting recurrence rates over the course of follow-up. Standardizing the reporting of recurrence rates by time would allow robust summarization of recurrence estimates. In addition, we were unable to summarize LPFS because most of the studies did not report this outcome by stage. The sample sizes for some of the stage II and III analyses were small, which led to relatively wide RFS CIs. The analyses presented here highlight the need for more studies reporting multimodal treatments to improve RFS in those with locoregionally advanced NSCLC. Advancements in perioperative therapies such as immune-checkpoint inhibitors and molecular targeted therapies have been assessed in several pivotal practice-changing trials.30Felip E. Altorki N. Zhou C. et al.Adjuvant atezolizumab after adjuvant chemotherapy in resected stage IB-IIIA non-small-cell lung cancer (IMpower010): a randomised, multicentre, open-label, phase 3 trial.Lancet. 2021; 398: 1344-1357Abstract Full Text Full Text PDF PubMed Scopus (567) Google Scholar, 31Forde P.M. Spicer J. Lu S. et al.Neoadjuvant nivolumab plus chemotherapy in resectable lung cancer.N Engl J Med. 2022; 386: 1973-1985Crossref PubMed Scopus (668) Google Scholar, 32Spigel D.R. Faivre-Finn C. Gray J.E. et al.Five-year survival outcomes from the PACIFIC trial: durvalumab after chemoradiotherapy in stage III non-small-cell lung cancer.J Clin Oncol. 2022; 40: 1301-1311Crossref PubMed Scopus (360) Google Scholar, 33Herbst R.S. Wu Y.L. John T. et al.Adjuvant osimertinib for resected EGFR-mutated stage IB-IIIA non-small-cell lung cancer: updated results from the phase III randomized ADAURA trial.J Clin Oncol. 2023; 41: 1830-1840Crossref PubMed Scopus (55) Google Scholar We expect that the increased availability of these drugs in routine clinical practice will enable more personalized treatment plans tailored to patient-specific disease markers. It is hoped that this will lead to both reduced recurrence rates and higher RFS, particularly in later-stage patients. Competing risk was mitigated by excluding studies in which death was explicitly censored. Not accounting for competing risk in KM analysis may introduce bias and overestimate the cumulative incidence of an event in the presence of competing events34Austin P.C. Lee D.S. Fine J.P. Introduction to the analysis of survival data in the presence of competing risks.Circulation. 2016; 133: 601-609Crossref PubMed Scopus (1362) Google Scholar; in this study, the results would have been biased toward reporting higher RFS. As expected, in a sensitivity analysis that included only study arms clearly defining both recurrence and death as RFS events, RFS estimates were slightly lower at each corresponding time point and stage, compared with the primary analysis. Although it was beyond the scope of this study, assessing the correlation between RFS and OS to determine whether RFS serves as a good surrogate end point for OS is an important topic as it may reduce the time and cost of treatment development.35Li Y. Yuan S. Chen Y. et al.Evaluating relapse-free survival as an endpoint for overall survival in adjuvant immunotherapy trials.J Natl Cancer Inst. 2023; 115: 1085-1091Crossref PubMed Scopus (1) Google Scholar Previous studies have shown that depending on the cancer type and treatment, RFS correlates relatively strongly with OS.35Li Y. Yuan S. Chen Y. et al.Evaluating relapse-free survival as an endpoint for overall survival in adjuvant immunotherapy trials.J Natl Cancer Inst. 2023; 115: 1085-1091Crossref PubMed Scopus (1) Google Scholar, 36Araujo R.L.C. Herman P. Riechelmann R.P. Recurrence-free survival as a putative surrogate for overall survival in phase III trials of curative-intent treatment of colorectal liver metastases: systematic review.World J Clin Oncol. 2017; 8: 266-272Crossref PubMed Scopus (7) Google Scholar, 37Imai H. Onozato R. Ginnan M. Kobayashi D. Kaira K. Minato K. Post-progression survival is strongly associated with overall survival in patients exhibiting postoperative relapse of non-small-cell lung cancer harboring sensitizing EGFR mutations.Medicina (Kaunas). 2021; 57: 508Crossref PubMed Scopus (1) Google Scholar, 38Mauguen A. Pignon J.P. Burdett S. et al.Surrogate endpoints for overall survival in chemotherapy and radiotherapy trials in operable and locally advanced lung cancer: a re-analysis of meta-analyses of individual patients' data.Lancet Oncol. 2013; 14: 619-626Abstract Full Text Full Text PDF PubMed Scopus (186) Google Scholar Our SLR and meta-analysis presents a contemporary and thorough assessment of RFS and recurrence outcomes from the published clinical literature, providing estimates at multiple time points after surgery by stage and geographical region. RFS estimates for Asian study populations were better than for North American and European populations, consistent with previous findings from regional studies. The study also reveals the inconsistency in reporting of recurrence rates by time, highlighting a need for more standardized outcome reporting practices. These findings can inform treatment decisions, clinical trial design, and future research to improve outcomes among patients with NSCLC. The presented work was funded by Johnson & Johnson.