Efficacy of Definitive Radiotherapy for Patients with Clinical Stage IIIB or IIIC Lung Adenocarcinoma and Epidermal Growth Factor Receptor (EGFR) Mutations Treated Using First- or Second-Generation EGFR Tyrosine Kinase Inhibitors

Background The effectiveness of definitive radiotherapy (RT) for patients with clinical stage IIIB or IIIC lung adenocarcinoma and epidermal growth factor receptor (EGFR) mutations who received first- or second-generation EGFR tyrosine kinase inhibitors (TKIs) is unclear. Methods Taiwan Cancer Registry data were used in this retrospective cohort study to identify adult patients diagnosed with EGFR-mutated stage IIIB or IIIC lung adenocarcinoma between 2011 and 2020. Patients treated with first- or second-generation EGFR TKIs were classified into RT and non-RT groups. Propensity score (PS) weighting was applied to balance covariates between groups. The primary outcome was overall survival (OS), and the incidence of lung cancer mortality (ILCM) was considered as a supplementary outcome. Additional supplementary analyses were conducted to assess the robustness of the findings. Results Among 270 eligible patients, 41 received RT and 229 did not. After a median follow-up of 46 months, PS-weighted analysis showed the PS-weighted hazard ratio of death for the RT group compared to the non-RT group was 0.94 (95% CI: 0.61–1.45, p = 0.78). ILCM rates did not differ significantly between the two groups. Supplementary analyses yielded consistent results. Conclusion The addition of definitive RT to first- or second-generation EGFR TKI treatment does not significantly improve OS of patients with EGFR-mutated stage IIIB or IIIC lung adenocarcinoma. NCT03521154NCT05167851.

Nevertheless, the role of EGFR TKIs for the treatment of locally advanced stage NSCLC remains uncertain.For patients with stage IIIB to IIIC disease, who are not eligible for surgical resection, the NCCN suggests defnitive concurrent chemoradiotherapy (dCCRT) [3]; however, this approach has been the subject of debate in the literature [4], and our previous study reported comparable outcomes between patients treated with EGFR TKIs alone and those treated with dCCRT [5].
Currently available EGFR TKIs can be classifed as frst (e.g., geftinib or erlotinib), second (e.g., afatinib or dacomitinib), and third (e.g., osimertinib) generation [6].A comprehensive National Health Insurance (NHI) system has been implemented in Taiwan since 1995.First-or secondgeneration EGFR TKIs have been approved by the NHI as a frst-line treatment for patients with at least stage IIIB lung adenocarcinoma and EGFR mutations [7].Consequently, many patients with lung adenocarcinoma and EGFR mutations in Taiwan have been treated with frst-or secondgeneration EGFR TKIs as their frst-line treatment, rather than the guideline-recommended dCCRT [3,5].
Drug resistance is a common issue in patients treated with EGFR TKIs [6].A randomized controlled trial (RCT) including patients with oligometastatic lung adenocarcinoma and EGFR mutations receiving frst generation EGFR TKIs, SINDAS, reported that the addition of radiotherapy (RT) signifcantly improved overall survival (OS) [8].Another RCT for patients with unresectable stage III NSCLC with EGFR mutations showed that erlotinib plus RTsignifcantly improved progression-free survival relative to dCCRT [9].A search conducted on PubMed using the terms, "((non-small-cell lung) OR (lung adenocarcinoma)) AND ((locally advanced) OR (stage III)) AND (EGFR mutation) AND ((radiation) OR (radiotherapy)) AND (geftinib OR erlotinib OR afatinib OR dacomitinib OR osimertinib OR (tyrosine kinase inhibitor * ))," in June 2023 yielded no relevant RCTs, only the observational KINDLE study [10,11].Given the aforementioned controversy and the limited literature available on this topic, we designed this retrospective cohort study to investigate the efcacy of defnitive RT for patients with clinical stage IIIB or IIIC lung adenocarcinoma and EGFR mutations treated with frst-or second-generation EGFR TKIs using a populationbased approach.

2.
1. Data Sources.Tis retrospective cohort study used deidentifed databases obtained from the Health and Welfare Data Science Center, Ministry of Health and Welfare, which included comprehensive information from the Taiwan Cancer Registry (TCR), the death registry, and reimbursement data for the entire population of Taiwan, provided by the Bureau of NHI.Te study was approved by the Central Regional Research Ethics Committee of China Medical University, Taichung, Taiwan (CRREC-108-080 (CR-3)).

Study Design, Population, and Intervention.
Adult patients aged 18-75 years [12] diagnosed with EGFR-mutated stage IIIB or IIIC lung adenocarcinoma [13] were identifed according to the American Joint Committee on Cancer 7th or 8th edition clinical staging criteria, [14,15] between 2011 and 2020.Tese patients had a good performance status (PS) (Eastern Cooperative Oncology Group (ECOG) PS 0-2) and were treated with frst or second-generation EGFR TKIs [6], as indicated in the TCR, because the NHI reimbursed geftinib, erlotinib, afatinib, and dacomitinib for patients with lung adenocarcinoma during the study period (2011-2020), but not other EGFR TKIs.Patients who underwent surgery, received other systemic therapies (including chemotherapy), had multiple treatment records, or had prior cancers were excluded to ensure data quality.
Patients were classifed into an RT group (treated with defnitive RT) and a non-RT group (without RT).Patients included in the RT group received external beam RT with equivalent doses in 2-Gy per fractions, assuming an alpha/ beta ratio of 10 (EQD2 (10)), [16] within the range of 31.25-70Gy, as per the literature [3,8].Te explanatory variable of interest was the receipt of RT (RT group vs. non-RT group), and the primary outcome was OS.Te incidence of lung cancer mortality (ILCM) was considered as a supplementary outcome.Information on OS and ILCM was obtained through linkage with the TCR or death registry records.Te index date was defned as the date of diagnosis, and OS/ILCM was calculated from the index date to the date of death or December 31, 2021 (the censoring date of the death registry).A study fowchart is shown in Figure 1, as suggested by the STROBE guideline [17].

Covariates.
To account for potential nonrandomized treatment selection, the following covariates were collected, based on recent relevant studies and our clinical research experiences: [5,[8][9][10]18] patient demographics (age, sex, residency region, and socioeconomic status (SES)), patient characteristics (body mass index (BMI), comorbidity, smoking, and performance status), and disease characteristics (clinical T-and N-stage, tumor size).
Patient residency region was classifed as northern or non-northern Taiwan, based on previous studies reporting geometric diferences in disease characteristics [19].Sex was classifed as male or female.SES was classifed as higher (income greater than the minimum wage) or not.Comorbidity was determined using the modifed Charlson comorbidity index score and classifed as ≥1 or <1 [20].Smoking status was classifed as yes or no.Clinical T-stage was classifed as (1-2) or (3-4), while the clinical N-stage was classifed as (0-2) or 3. ECOG PS was classifed as (0-1) or 2. Age (years), BMI (kg/m 2 ), and tumor size (mm) were treated as continuous variables.

Statistical Analysis and Supplementary Analyses.
In the primary analysis (PA), a propensity score (PS) approach was employed, using a logistic regression model with the aforementioned covariates to balance measured potential confounders [21].PS weighting (PSW) was used, with the overlap weight as the framework [22][23][24][25].After verifying covariate balance between groups using standardized difference after PSW, [26,27] the hazard ratio (HR) of death was compared between the RT and non-RT groups [28].Point estimates were calculated using a Cox proportional 2 Canadian Respiratory Journal hazards model in the weighted sample throughout the entire follow-up period [28,29].Te 95% confdence interval (95% CI) was estimated using the bootstrap method [30,31].
To assess the robustness of the fndings regarding potential unmeasured confounders, the E-value was applied, as suggested in the literature [32,33].Te incidence of lung cancer mortality between the RT and non-RT groups was evaluated using the competing risk approach in the weighted sample [34].Te International Classifcation of Diseases, Ninth Revision, Clinical Modifcation (ICD-9-CM) and ICD-10-CM coding (5080/5081 and J700/J701, respectively [35]) were used to estimate the rate of radiation pulmonary toxicity within 6 months after RT in the RT group.
Supplementary analyses (SA) included four additional analyses, to evaluate the robustness of the fndings.In SA-1, an alternative analytical approach, PS matching (PSM), was applied by constructing a 1 : 1 PS-matched cohort (without replacement) within the primary study population.Te HR of death was compared using a robust variance estimator [28].Te subdistribution HR was used to assess ILCM with the clustered Fine-Gray model [36].In SA-2, the RTdose was limited to at least EQD2 (10) 50 Gy [37] for the RT group in the PA, to explore its impact in the RT and non-RT groups.In SA-3, impact was evaluated by limiting patients to those with common mutations (Exon 19 deletion or Exon 21 L858R [3,9]), which were recorded in the TCR since 2019.In SA-4, patients who received palliative RT, according to TCR records, were excluded in the PA.
Statistical analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC, USA) and R version 4.3.0(R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria).

Study Population and Descriptive Data. As shown in
Figure 1, our study population consisted of 270 eligible patients with unresected clinical stage IIIB or IIIC lung adenocarcinoma and EGFR mutations who received RT (41 patients) or not (229 patients) between 2011 and 2020.Patient characteristics are described in Table 1.All covariates achieved balance (standardized diferences approximately 0) after PS weighting via the overlap weights, although patient residency region was imbalanced before PSW [26].

Primary Analysis.
After a median follow-up of 46 months (range, 12-127 months), 160 deaths were observed (23 and 137 patients in the RT and non-RT groups, respectively).In the unadjusted analysis, the 5-year OS rates were 39% (RT group) and 36% (non-RT group) (log-rank test, p � 0.86) (Figure 2).Te overlap weight-adjusted OS curve is shown in Figure 3. Te 5-year PSW-adjusted OS rates were 39% (RT group) and 37% (non-RT group).Relative to the non-RT group, the PSWadjusted HR of death for the RT group was 0.94 (95% CI � 0.61-1.45,p � 0.78), which could be explained by an unmeasured confounder associated with the selection of treatment (RT or non-RT) and survival with risk ratios of 1.26 (E-value)-fold each, but not by weaker confounding factors.No signifcant diference in ILCM was detected between groups (HR � 0.97, p � 0.92).Of patients in the RT group, 15% [6/41] developed radiation pulmonary toxicity.
In SA-4 (supplementary material Table S1), we again found no signifcant diference in OS between the RT group, which excluded those who received palliative RT, and the non-RT group (PSW-adjusted HR � 0.78, p � 0.32); further, comparison of ILCM generated a similar result (HR � 0.79, p � 0.51).

Discussion
In this retrospective cohort study based on a nationwide cancer registry, we investigated the efectiveness of defnitive RT for patients with locally advanced stage lung Confirmed eligible: patients a with unresected clinical stage IIIB or IIIC lung adenocarcinoma and EGFR mutations, acceptable performance status, and treated with definitive radiotherapy or without radiotherapy.[n = 345] Step 3.
Step 4. Included in the initial study population: patients in step 3 without missing data regarding co-variables used in primary analysis [age, sex, residency, social-economic status, body mass index, comorbidity, and smoking, clinical T-& N-stage, tumor size, and performance status] b .
Step 5.The final study population in the primary analysis with complete outcome information c .[n = 270] Canadian Respiratory Journal   Canadian Respiratory Journal adenocarcinoma and EGFR mutations treated with frst or second generation EGFR TKIs.Our fndings suggest that the OS of patients with EGFR mutations is similar whether or not RT is added to EGFR TKI treatment.Our results were consistent across diferent analyses, including of ILCM and SA.
To the best of our knowledge, this is the frst populationbased study to explore this specifc topic.Te outcomes of our investigation, which showed no statistically signifcant efect of RT on OS, align with the fndings reported in the KINDLE study, [10,11] as the 95% confdence intervals overlapped; however, it is important to note that the KINDLE study reported a numerically longer median OS for the RT group (42.6 months) compared to the non-RT group (25.4 months), whereas our study found similar OS values in both groups.
Given the nonrandomized and registry-based nature of our study, our fndings should be interpreted with caution.It will be essential to conduct further prospective studies, particularly RCTs, to investigate the efect of RT in patients treated with third generation EGFR TKIs.Although we are not aware of any ongoing RCTs specifcally addressing this question, two related trials, namely, the LAURA trial (NCT03521154) and the ABLATE trial (NCT05167851), [38,39] may ofer additional insights in the future.Te LAURA trial is designed to examine the role of consolidative osimertinib in locally advanced NSCLC with EGFR mutation treated with dCCRT, while the ABLATE trial is to investigate the role of upfront RT for oligometastatic NSCLC with EGFR mutation treated by lazertinib (another third generation EGFR TKI).Furthermore, geftinib was the frst EGFR TKI approved by NHI for locally advanced NSCLC, followed by erlotinib, afatinib, and dacomitinib; therefore, our results are not applicable to other frst or second generation EGFR TKIs, such as icotinib.
Several limitations should be acknowledged in our study.First, the potential for unmeasured confounders always exists in this type of nonrandomized study.To evaluate the robustness of our fndings to potential unmeasured confounders, we employed the E-value, which provides a measure of the minimum strength of associations that an unmeasured confounder would need to have with both the treatment and the outcome to explain away the observed efect.Second, although EGFR TKIs are indicated for treatment in the context of common mutations (exon 19 deletion and exon 21 L858R), which account for around 85% of patients with EGFR mutations, [40] and are also indicated for some less common mutations (such as EGFR S768I, L861Q, and/or G719X), [3] the potential inclusion of insensitive rare EGFR mutations, such as exon 20 insertions, [3,41] may limit the clinical implications of our fndings;      Canadian Respiratory Journal however, the incidence of exon 20 insertions among patients with EGFR mutations in Taiwan is low (3%-4%) [41].Finally, due to data limitations within the cancer registry, we neither included other endpoints, such as progression-free survival or quality of life, nor investigated the efect of postprogression treatment.

Conclusion
Our study suggests that the addition of RT to frst or second generation EGFR TKI treatment did not signifcantly impact the OS of patients with clinical stage IIIB or IIIC lung adenocarcinoma and EGFR mutations; however, further research is needed to validate our fndings, particularly in the context of third generation EGFR TKIs.RCTs investigating the role of RT in this patient population are eagerly awaited to provide more defnitive evidence.

Figure 1 :
Figure 1: STROBE study fowchart and the number of individuals at each stage of the study.a Only those treated (class 1-2) and with a single record were included, to ensure data consistency.b Te exact numbers are not reported because of a health and welfare data science center database center policy to avoid numbers ≤2 in single cells.c Without missing information in the TCR and death registry regarding survival status.

1 Figure 2 :
Figure 2: Kaplan-Meier overall survival curve (in years) showing no signifcant diference between the RT and non-RT groups in the primary analysis before adjustment for covariates.RT, radiotherapy.

Figure 3 :
Figure 3: Overall survival curve (in years) showing no signifcant diference between the RT and non-RT groups in the primary analysis after overlap weights adjustment.RT, radiotherapy.
BMI, body mass index; ECOG PS, eastern cooperative oncology group performance status; PS, propensity score; RT, radiotherapy; SD, standard deviation.† Rounded.‡ Te exact numbers are not reported because of a Health and Welfare Data Science Center (HWDC) database center policy to avoid numbers ≤2 in single cells.

Figure 4 :
Figure 4: Kaplan-Meier survival curve (in years) showing no signifcant diference between the RT and non-RT groups in the frst supplementary analysis (propensity score-matched analyses).RT, radiotherapy.

Table 1 :
Characteristics of the patient study population included in the primary analysis.

Table 2 :
Characteristics of patients included in the frst supplementary analysis.

Table 3 :
Characteristics of patients included in the second supplementary analysis.

Table 4 :
Characteristics of patients included in the third supplementary analysis.