The addition of chemotherapy to radiation therapy improves survival in elderly patients with stage III non-small cell lung cancer

Abstract

Introduction

Elderly patients represent the majority of lung cancer diagnoses but are poorly represented in clinical trials. We evaluated the overall survival (OS) of elderly patients with stage III non-small cell lung cancer (NSCLC) treated with definitive radiation compared to those treated with definitive chemoradiation.

Methods

We conducted a comparative effectiveness study of radiation therapy vs. chemoradiation in elderly (≥70 years old) patients with stage III NSCLC not treated surgically diagnosed from 2003–2014 using the National Cancer Database. Two cohorts were evaluated: patients treated with definitive (≥59.4 Gy) radiation (n=5,023) and patients treated with definitive chemoradiation (n=18,206). Chemoradiation was further defined as concurrent (radiation and chemotherapy started within 30 days of each other) or sequential (radiation started>30 days after chemotherapy). We compared OS between the treatment groups using the Kaplan-Meier method and Cox proportional hazards regression before and after propensity score matching (PSM).

Results

Treatment with chemoradiation was associated with improved OS compared to radiation before PSM (HR=0.66, 95%CI 0.64–0.68, p<.001) and after PSM (HR=0.67, 95%CI 0.64–0.70, p<.001). Relative to concurrent chemoradiation, sequential chemoradiation was associated with a 9% reduction in the risk of death (HR=0.91, 95%CI 0.85–0.96, p=.002).

Conclusions

We found that definitive chemoradiation resulted in a survival advantage compared to definitive radiation in elderly patients. Sequential chemotherapy and radiation was superior to concurrent chemoradiation. While prospective trials are needed, this analysis suggests that chemoradiation should strongly be considered for elderly patients and the optimal sequencing of chemotherapy and radiation remains an unanswered question for this patient population.

Introduction

Stage III non-small cell lung cancer (NSCLC) represents 30% of all new lung cancer diagnoses and is a heterogeneous disease requiring a multi-disciplinary treatment approach.¹ Nearly 70% of all lung cancer diagnoses and >70% of lung cancer deaths in the U.S. occur in patients ≥65 years of age.² However, the elderly are under-represented in clinical trials making treatment decisions in this population challenging.^3–5

Overall, the available data guiding decision-making in the elderly is limited. A multicenter retrospective review, based on the Netherlands Cancer Registry, reported no improvement in overall survival (OS) for patients ≥70 years old treated with concurrent chemoradiation (CRT) compared to sequential CRT or radiation therapy (RT) alone.⁶ In contrast, two additional studies, including subset analysis of 2 prospective trials and a Surveillance Epidemiology and End Results-Medicare (SEER) analysis demonstrated a survival benefit for CRT.^{7, 8} Given the conflicting and scant data in this patient population, the optimal treatment strategy for stage III NSCLC in the elderly needs to be further defined.

The objective of this study was to compare OS in elderly patients treated with RT alone compared to CRT using the National Cancer Database (NCDB). We hypothesized that patients that received CRT would have improved OS compared to those treated with RT alone.

Materials and Methods

The NCDB, a combined effort of the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society, is a nationwide hospital-based database that contains de-identified hospital registry data from more than 1,500 accredited facilities, representing more than 70% of newly diagnosed cancer cases in the U.S.⁹ The NCDB collects data on patient demographics and comorbidities, tumor characteristics and staging details, primary therapies administered, and OS. The CoC's NCDB and the hospitals participating in the CoC NCDB are the source of the de-identified data and have not verified and are not responsible for the statistical validity of the data analysis nor the conclusions presented in this study.

Patient Selection

Patients diagnosed with stage III NSCLC from 2003–2014 were collected from the NCDB participant user file with additional inclusion and exclusion criteria summarized in Figure 1. We defined elderly as patients≥70 years old, as used previously in numerous studies.^10–12 The transition to the American Joint Committee on Cancer (AJCC) 7^th edition occurred in 2010, consequently, our patient cohort consisted of patients staged using both AJCC 6^th and 7^th editions. Patients with clinical T4N0M0 disease based on the AJCC 6^th edition were excluded as these patients could be either clinical stage II (T3 N0 M0) or stage IV (M1a – malignant pleural or pericardial effusion) based on the AJCC 7^th edition. Since the focus of this study was on patients not treated surgically, those treated with any type of surgical procedure were excluded. Additional exclusion criteria included: patients with unknown RT or chemotherapy details, with regional or total RT doses ≥80 Gy, treatment with more than 45 fractions of RT, treatment with immunotherapy, no treatment, treatment with chemotherapy alone, and treatment with palliative doses of radiation therapy (<59.4 Gy). After exclusion, patients were categorized as having received definitive RT alone (≥59.4 Gy) or definitive CRT (≥59.4 Gy). CRT patients were considered to have received concurrent CRT (CCRT) if chemotherapy was delivered within 30 days prior to or after initiation of RT while sequential CRT (SCRT) was defined as RT delivered >30 days after initiation of chemotherapy as defined in a prior study.¹³ Details regarding radiation treatment technique (e.g., 3D conformal radiation therapy versus intensity-modulated radiation therapy) were available for 35% of the patient population. Rather than exclude patients with unknown treatment technique, we chose not to include this variable in our analysis. Patients with unknown demographic data were excluded. For the urban/rural code, the 2013 classification codes were used while income was determined using the 2008–2012 code. Finally, patients with less than 1 month of follow-up were excluded from the analysis to limit immortal time bias.¹⁴

Figure 1.

Study flow diagram for analytic cohorts. NCDB, National Cancer Database; SBRT, stereotactic body radiation therapy.

Study Variables

We dichotomized the following baseline covariates: gender (male vs. female), race (white vs. non-white), median income (≥$48,000 vs. <$48,000), primary insurance payor (private vs. non-private), county location (metropolitan vs. urban/rural), facility type (academic vs. community/comprehensive community/integrated network programs), chemotherapy agents used (multi-agent regimen vs. single-agent regimen), and clinical stage group (IIIB vs. IIIA). The Charlson-Deyo score, a measure of comorbidity, was dichotomized as 0 (no comorbities) or 1 (≥1 comorbidity). The variables age and distance to the nearest facility were analyzed as continuous variables.

Statistical Methods

The primary objective of this study was to evaluate OS in elderly patients treated with CRT vs. RT alone. We also sought to evaluate the impact of number of chemotherapy agents used (multi-agent CRT vs. RT alone; single-agent CRT vs. RT; multi-agent CRT vs. single-agent CRT). Additional analysis included comparison of OS in CRT patients treated with CCRT vs. SCRT. Differences in patient characteristics between CRT and RT patients were tested using the χ² test for categorical variables and the t-test for continuous variables. Survival was estimated using the Kaplan-Meier method and the log-rank test was used to compare survival curves. Cox regression analysis was used to test the association between treatment and demographic variables with OS on univariate analysis. Variables with p≤.10 on univariate analysis were included in the multivariate model.

Propensity score-matching (PSM) was performed to reduce potential selection bias. Logistic regression was used to identify predictors of treatment with definitive CRT vs. definitive RT alone. Patients treated with CRT were matched with those receiving radiation alone using a 1:1 nearest available neighbor match without replacement using an algorithm described by Coca-Perraillon.¹⁵ The caliper size was calculated as 0.20*standard deviation of the propensity score as described by Rosenbaum et al.¹⁶ Common support of the propensity score distributions was evaluated graphically and balance was evaluated by computing the standardized difference of the covariates across the two groups.¹⁷ Following PSM, OS was estimated using the Kaplan-Meier method and Cox regression was used to perform univariate and multivariate analysis. The exact same approach was used to perform PSM for patients receiving CCRT vs. SCRT. All statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc., Cary, NC).

Results

Patient Characteristics

We identified 5,023 elderly patients treated with RT alone and 18,206 patients treated with CRT. The median radiation dose was 64.8 Gy (IQR, 61.2–66.6 Gy) over a median of 34 fractions (33–36) in the CRT group and 64.8 Gy (IQR, 60.0–66.6 Gy) over a median of 33 (30–35) in the RT group. The radiation was delivered with conventional fraction sizes with a median daily fraction size of 1.8 Gy (1.8–2.0 Gy) in the CRT group and 2.0 Gy (1.8–2.0 Gy) in the RT group. Several imbalances in characteristics existed between RT and CRT patients (Table 1). Most notably, CRT patients were more likely to have stage IIIB disease (36.7% vs. 32.5%, p<.001), but were also more likely to have no comorbidities (62.9% vs. 58.7%, p<.001) and be of younger age (75.8 years vs. 79.4 years, p<.001) compared to RT patients. Most of the CRT patients received multi-agent chemotherapy (86%, N=15,715). In the CRT cohort, the majority were treated with CCRT (87%, N=15,840) with a median time between the start of radiation and chemotherapy of 0 days (interquartile range (IQR), −1 [RT first] to 0 days). The median time from the start of chemotherapy to radiation in the SCRT group was 82 days (IQR, 56–112 days).

Table 1.

Patient Characteristics

	Radiation therapy alone (N=5,023)	Chemoradiation (N=18,206)	pa
Age, mean (SD)	79.4 (5.4) years	75.8 (4.4) years	<.001
Gender			<.001
Male	2750 (54.7%)	10540 (57.9%)
Female	2273 (45.3%)	7666 (42.1%)
Race			.005
White	4398 (87.6%)	16197 (89.0%)
Non-white	625 (12.4%)	2009 (11.0%)
Charlson-Deyo Score			<.001
0	2949 (58.7%)	11450 (62.9%)
>1	2074 (41.3%)	6756 (37.1%)
Median income			<.001
≥$48,000	2522 (50.2%)	9825 (54.0%)
<48,000	2501 (49.8%)	8381 (46.0%)
Primary insurance payor			.74
Private	491 (9.8%)	1751 (9.6%)
Non-private	4532 (90.2%)	16455 (90.4%)
County location			.045
Metropolitan	4062 (80.9%)	14489 (79.6%)
Non-metropolitan	961 (19.1%)	3717 (20.4%)
Distance to closest facility, mean (SD)	15.9 (41.4)	20.4 (74.7)	<.001
Facility type			.98
Academic	1243 (24.8%)	4502 (24.7%)
Non-academic	3780 (75.2%)	13704 (75.3%)
Clinical stage group			<.001
IIIA	3390 (67.5%)	11523 (63.3%)
IIIB	1633 (32.5%)	6683 (36.7%)
Days to start of radiation, mean (SD)	47.9 (40.2)	55.5 (46.5)	<.001
Radiation therapy dose, median (IQR)	64.8 (60.0–66.6) Gy	64.8 (61.2–66.6) Gy	.67
Elapsed days of radiation, median (IQR)	50 (45–55) days	51 (47–56) days	.51
Days to start of chemotherapy, median (IQR)	N/A	36 (25–54) days	N/A
Type of chemotherapy
Multi-agent	N/A	15715 (86.3%)	N/A
Single-agent/unknown	N/A	2491 (13.7%)	N/A
Type of chemoradiation
Concurrent	N/A	15840 (87.0%)	N/A
Sequential	N/A	2366 (13.0%)	N/A

^aThe p values are from the χ² test for categorical variables and the t-test for continuous variables.

SD, standard deviation; IQR, interquartile range; N/A, not applicable.

Predictors of Receiving CRT vs. RT

We identified both clinical and demographic variables associated with elderly patients receiving definitive CRT treatment (Supplementary Table 1). On univariate analysis, younger age, male sex, white race, higher income, living in a non-metropolitan county, stage IIIB, increased distance from the treating hospital, and a Charlson-Deyo score<1 were associated with higher odds of receiving CRT. All of these covariates, except for living in a metropolitan county, were independently associated with higher odds of receiving CRT on multivariate logistic regression analysis and were used to calculate propensity scores. Following PSM, the propensity score distributions between the two groups showed nearly ideal common support (Supplementary Figure 1A). The covariates were well-balanced between the two treatment groups following PSM with standardized differences between the covariates well below 10% (Supplementary Table 2).

Predictors of Receiving CCRT vs. SCRT

Supplementary Table 3 demonstrates that amongst the CRT patients, factors associated with receipt of CCRT compared to SCRT included: males, non-academic treatment facilities, stage IIIA disease, and higher comorbidity index. All of these factors remained significant on multivariate logistic regression analysis and were used to calculate propensity scores. Supplementary Figure 1B and Supplementary Table 4 demonstrate that PSM resulted in well balanced groups.

Survival Outcomes in the CRT vs. RT Cohorts

At the time of analysis, 19,041 of the 23,229 patients had died. The median follow-up for all elderly patients was 15.5 months (IQR 8.3–28.8 months) and the median follow-up for survivors was 30.7 months (IQR 19.1–49.8 months). We used Cox univariate and multivariate analyses to identify patient and treatment factors associated with OS (Supplementary Table 5). Factors independently associated with improved OS included younger age, female sex, non-white race, treatment at an academic facility, higher income, living in a metropolitan county, stage IIIA disease vs. IIIB, closer distance to the treatment hospital, Charlson-Deyo score<1, longer time to start of RT, and CRT vs. RT alone.

Prior to PSM, the median OS was 18.1 months (95%CI, 17.8–18.5 months) for patients treated with CRT and 12.2 months (95%CI, 11.7–12.6 months) for patients treated with RT (p<.001, Figure 2A). With PSM, a CRT therapy match was successfully identified for 4,718 of the 5,023 patients treated with RT (caliper size=0.03 based on propensity score SD=0.13). In the matched cohort, the survival advantage of CRT over RT alone persisted with a median OS of 17.2 months (95%CI, 16.6–17.8) and 12.2 months (95%CI, 11.8–12.6), respectively (p<.001; Figure 2B).

Figure 2.

Overall survival of patients treated with definitive chemoradiation (CRT) compared to those treated with definitive radiation therapy (RT) alone. Panel A represents the unmatched patient population while panel B represents propensity-score matched elderly patients. All curves represent actual survival as estimated by Kaplan-Meier.

On univariate analysis in the matched cohort of elderly CRT vs. RT patients, factors associated with worse OS included: older age, male gender, white race, non-academic treatment facility, non-metropolitan location, stage IIIB, Charlson-Deyo score>0, and treatment with RT alone (Supplementary Table 6). After adjustment for confounders, CRT corresponded to a 33% reduction in the risk of death (HR=0.67, 95%CI 0.64–0.70, p<.001, Table 2). The benefit of CRT was greater for elderly patients treated with multi-agent chemotherapy (HR=0.64, 95%CI 0.61–0.67, p<.001, Figure 3B, Table 2) compared with single-agent chemotherapy (HR=0.83, 95%CI 0.75–0.92, p<.001, Figure 3A, Table 2). As demonstrated in Figure 3C and Table 2, for elderly patients treated with CRT, multi-agent chemotherapy resulted in a 21% decrease in the HR for death compared to single-agent chemotherapy (HR=0.79, 95%CI 0.71–0.88, p<.001).

Table 2.

Multivariate Analysis of Overall Survival for Chemoradiation vs. Radiation Therapy Alone and Concurrent Chemoradiation vs. Sequential Chemotherapy and Radiation after Propensity Score Matching

	CRT vs. RT Cohorts (N=9,436)			CCRT vs. SCRT Cohorts (N=4,732)
Variable	HR	95% CI	p-value	HR	95% CI	p-value
Age	1.01	1.01–1.02	<.001	1.02	1.01–1.02	<.001
Female vs. male	0.84	0.81–0.88	<.001	0.82	0.77–0.88	<.001
White vs. non-white	1.18	1.10–1.26	<.001	1.12	1.01–1.24	.03
Academic vs. non-academic	0.95	0.90–1.00	.05	0.90	0.84–0.97	.005
Private vs. non-private insurance	NS	NS	NS	NS	NS	NS
Median income (≥$48,000 vs. <$48,000)	0.96	0.91–1.00	.07	0.96	0.90–1.03	.23
County location (Metropolitan vs. non-metropolitan)	0.97	0.91–1.02	.23	0.98	0.90–1.06	.58
Clinical stage IIIB vs. stage IIIA	1.18	1.12–1.23	<.001	1.16	1.09–1.24	<.001
Distance to closest facilitya	NS	NS	NS	NS	NS	NS
Charlson-Deyo score (1 vs. 0)	1.08	1.03–1.13	<.001	1.15	1.08–1.23	<.001
CRT vs. RT alone	0.67	0.64–0.70	<.001	N/A	N/A	N/A
SCRT vs. CCRT	N/A	N/A	N/A	0.91	0.85–0.96	.002
Single-agent CRT vs. RT alone	0.83	0.75–0.92	<.001	N/A	N/A	N/A
Multi-agent CRT vs. RT alone	0.64	0.61–0.67	<.001	N/A	N/A	N/A
Multi-agent vs. Single-agent chemotherapyb	0.79	0.71–0.88	<.001	0.74	0.65–0.96	.002
Days to start of radiationc	1.00	1.00–1.00	<.001	N/A	N/A	N/A

Abbreviations: CRT, chemoradiation; RT, radiation therapy; CCRT, concurrent chemoradiation, SCRT, sequential chemotherapy and radiation; NS, p-value>.10 on univariate analysis; N/A, not applicable

^alog of distance (miles) used for analysis

^bin patients receiving chemoradiation

^cHR is 0.996 (0.995–0.997)

Figure 3.

Overall survival of patients treated with either single (sCRT) or multi-agent chemoradiation (mCRT) compared to those treated with radiation therapy alone (RT). Panel A is a comparison of sCRT vs. definitive RT alone and panel B represents mCRT vs. definitive RT alone. Panel C is a comparison of mCRT vs. sCRT. All analysis was performed using the propensity-score matched patients and all curves represent actual survival as estimated by Kaplan-Meier.

Survival Outcomes in CCRT vs. SCRT Patients

Elderly patients treated with definitive CRT were further subdivided into those treated with concurrent vs. sequential CRT based on the timing of initiation of radiation and chemotherapy. Of the 18,203 patients treated with definitive CRT, 15,840 received CCRT and 2,366 received SCRT. The median OS was significantly higher in patients treated with SCRT compared to CCRT: 20.0 months (95%CI, 19.1–20.9) vs. 17.8 months (95%CI, 17.4–18.2), p<.001 (Figure 4A). PSM identified a CCRT match for all 2,366 SCRT patients (caliper size=0.004 based on propensity score SD=0.02). Supplementary Table 6 demonstrates the univariate OS analysis in the matched CCRT vs. SCRT cohorts. On multivariate analysis, SCRT corresponded to a 9% reduction in the risk of death (HR=0.91, 95%CI 0.85–0.96, p=.002, Table 2, Figure 4B). We performed a sensitivity analysis where the definition of CCRT was restricted to patients who received chemotherapy within 14 days of the initiation of RT, and the survival benefit of SCRT over CCRT persisted.

Figure 4.

Overall survival of patients treated with definitive concurrent chemoradiation (Con CRT) and definitive sequential chemoradiation (Seq CRT). Panel A represents the unmatched patient population while panel B represents the propensity-score matched elderly patients. All curves represent actual survival as estimated by Kaplan-Meier.

Discussion

Treatment of the elderly with locally advanced NSCLC is challenging and, with an aging population, will remain an issue for the U.S. healthcare system for the foreseeable future.¹⁸ To our knowledge, our study represents the largest reported cohort of elderly stage III NSCLC patients not treated surgically. We found that combined modality therapy with radiation and chemotherapy results in improved OS compared to radiation alone in the elderly and that sequential therapy appears superior to concurrent therapy.

We found a significant OS benefit with the addition of chemotherapy to definitive RT with a 33% reduction in risk of death. Prior studies evaluating combined modality therapy in the elderly have reported conflicting results. For example, a retrospective study using the Netherlands Cancer Registry also evaluated elderly (≥70 years old) patients with unresectable stage III NSCLC and reported no survival benefit with concurrent CRT and increased toxicity.⁶ In contrast, a second analysis of patients aged ≥65 years compared concurrent CRT to RT alone and demonstrated a survival benefit (13.7 vs. 10.5 months, p=.05) for CRT compared to RT alone.⁸ However, the rates of grade ≥3 toxicity were significantly higher in the CRT group (89.9%) vs. RT alone group (32.4%). Similarly, Davidoff et al. found that CRT had a significant survival benefit when compared to RT alone (12.0 vs. 7.6 months) in patients ≥66 years old.⁷

In general, limited prospective data of CRT vs. RT alone in the elderly with stage III NSCLC exist. Two trials from Japan have evaluated CRT vs. RT in the elderly, the first of which was stopped early due to 4 deaths in the CRT arm.¹⁹ In a subsequent phase III trial that completed accrual, Atagi et al. found that CRT improved OS compared to RT alone in 197 patients >70 years of age (HR=0.68).¹⁰ The chemotherapy in this trial included single-agent low-dose carboplatin (30 mg/m² × 20 days). There were higher rates of grade 3–4 hematologic toxicity and grade 3 infection in the CRT group, although rates of grade 3–4 radiation pneumonitis and late lung toxicity were similar between the groups. In a recent meta-analysis of the 243 patients treated on both Japanese trials and an additional 164 elderly patients from the Auperin et al. 2006 analysis of CRT using platinum compounds, the use of CRT was associated with a 34% reduction in the HR for death, similar to the 33% reduction seen in our study that included over 9,000 patients in the matched cohort.²⁰ Additionally, we found a modest, but significant, 15% reduction in the risk of death when non-standard single agent chemotherapy is used.

In this NCDB analysis, we found that survival in elderly patients treated with SCRT had a 9% reduction in risk of death compared to CCRT. This finding is in contrast to the results of the 2010 Auperin et al. meta-analysis which demonstrated an OS benefit with concurrent vs. sequential CRT for patients with locally advanced NSCLC.²¹ The elderly were under-represented in that analysis with only 15% of patients ≥70 years old. Our results were similar to those from the Davidoff et al. SEER-Medicare database study that found an increased mortality risk with concurrent CRT compared to sequential CRT.⁷ However, this finding from our analysis must be taken with caution - the NCDB does not collect duration of chemotherapy treatment, so it is not possible to determine if patients in the sequential CRT group received combined modality therapy at the time of radiation.

The toxicity of combined modality therapy is a central issue in the management of elderly patients.^{12, 22–24} Several studies have found significantly higher rates of toxicity in patients receiving CRT vs. RT alone.^{6, 8, 10, 19, 25} In a recent pooled analysis of stage III NSCLC patients treated with CRT on 1 of 16 U.S. National Cancer Institute cooperative group studies, the 832 elderly patients (≥70 years old) experienced more toxicity, higher rate of treatment-related death, and worse OS compared to the 2,768 non-elderly patients.¹² Given the strict eligibility criteria and close follow-up required on clinical trials, the rates of adverse-events/deaths and survival outcomes are likely worse in the non-clinical trial elderly population. Nonetheless, our results demonstrate that elderly patients treated with CRT had superior survival compared to those treated with RT alone. This underscores the importance of identifying and incorporating tools such as the comprehensive geriatric assessment (CGA) or Vulnerable Elders Survey-13 into the treatment decision-making process.^{18, 26, 27} Completion of the CGA can help predict risk of toxicity with treatment, can be used to fine-tune treatment recommendations,²⁸ and its use has been endorsed by the International Society for Geriatric Oncology and the European Organization for Research and Treatment of Cancer.^{27, 29}

This study has several limitations. The NCDB is a retrospective database with inherent weaknesses that include incomplete data, selection bias, and unmeasured confounders. One major limitation of the NCDB is that performance status is not captured. Instead, the Charlson-Deyo score, which measures the number of comorbidities each patient has, is collected. We fully recognize that there is no single variable or group of variables in any dataset that can accurately reflect the treatment decision for an individual patient. Therefore, while PSM was utilized to minimize treatment selection bias, imbalances likely remain in unmeasured variables between the treatment cohorts. The NCDB does not collect several key chemotherapy details, including specific agents used and the duration/number of cycles delivered. This makes it impossible to determine if patients received consolidation chemotherapy. Also, while an OS benefit was observed with definitive CRT compared to definitive RT alone, there is no available data in the NCDB regarding critically important endpoints of toxicity, quality of life, and cause of death.

Nonetheless, we feel that there are several strengths to be noted. This is by far the largest analysis of CRT vs. RT (and CCRT vs. SCRT) in elderly patients to date. While chemotherapy details are limited in the NCDB, the RT data is much more complete. The RT doses captured in the NCDB are those that were delivered, not intended, and we ensured patients received definitive doses of RT (≥59.4 Gy) to be included in this analysis. While the exact chemotherapy regimen is not reported, the NCDB does collect if 1 or >1 agent was used. We feel that the analysis of survival by number of chemotherapy agents used provides a useful framework for how to approach the elderly patient with stage III NSCLC.

Treatment of the elderly with stage III NSCLC should involve a multidisciplinary discussion. All patients not eligible for surgery should first be considered for CRT, with either concurrent or sequential radiation. Based on our findings, multi-agent chemotherapy is preferred over single-agent regimens in suitable patients. When multi-agent chemotherapy is not feasible, a single-agent regimen is supported by our results as well as those of Atagi et al.¹⁰ When chemotherapy is contraindicated or not recommended, then definitive RT alone should be considered.

In conclusion, we found that definitive CRT is superior to definitive RT in elderly patients with stage III NSCLC not treated surgically. We also found that in patients that received CRT, sequential chemotherapy and RT resulted in improved OS compared to concurrent CRT. While the optimal methodology for deciding on appropriate therapy in these patients is unknown, future clinical trials in stage III NSCLC should prioritize inclusion of the elderly in order to help further tailor therapeutic decisions for this expanding patient population.