Comparison of treatment regimens for unresectable stage III epidermal growth factor receptor (EGFR) mutant non-small cell lung cancer

Abstract

Background:

Durvalumab after chemoradiotherapy (CRT) failed to bring survival benefits to patients with epidermal growth factor receptor (EGFR) mutations in PACIFIC study (evaluating durvalumab in patients with stage III, unresectable NSCLC who did not have disease progression after concurrent chemoradiotherapy). We aimed to explore whether locally advanced inoperable patients with EGFR mutations benefit from tyrosine kinase inhibitors (TKIs) and the optimal treatment regimen.

Methods:

We searched the PubMed, Embase, the Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov databases from inception to December 31, 2022 and performed a meta-analysis based on a Bayesian framework, with progression-free survival (PFS) and overall survival (OS) as the primary endpoints.

Results:

A total of 1156 patients were identified in 16 studies that included 6 treatment measures, including CRT, CRT followed by durvalumab (CRT-Durva), TKI monotherapy, radiotherapy combined with TKI (RT-TKI), CRT combined with TKI (CRT-TKI), and TKI combined with durvalumab (TKI-Durva). The PFS of patients treated with TKI-containing regimens was significantly longer than that of patients treated with TKI-free regimens (hazard ratio [HR] = 0.37, 95% confidence interval [CI], 0.20–0.66). The PFS of TKI monotherapy was significantly longer than that of CRT (HR = 0.66, 95% CI, 0.50–0.87) but shorter than RT-TKI (HR = 1.78, 95% CI, 1.17–2.67). Furthermore, the PFS of RT-TKI or CRT-TKI were both significantly longer than that of CRT or CRT-Durva. RT-TKI ranked first in the Bayesian ranking, with the longest OS (60.8 months, 95% CI = 37.2–84.3 months) and the longest PFS (21.5 months, 95% CI, 15.4–27.5 months) in integrated analysis.

Conclusions:

For unresectable stage III EGFR mutant NSCLC, RT and TKI are both essential. Based on the current evidence, RT-TKI brings a superior survival advantage, while CRT-TKI needs further estimation. Large randomized clinical trials are urgently needed to explore the appropriate application sequences of TKI, radiotherapy, and chemotherapy.

Registration:

PROSPERO; https://www.crd.york.ac.uk/PROSPERO/; No. CRD42022298490.

Introduction

Approximately 25–30% of non-small cell lung cancer (NSCLC) patients are diagnosed as stage III, among which only one-third of stage IIIA patients are amenable to surgery.^[1] The PACIFIC study (a randomized, placebo-controlled, phase 3 trial evaluating the immune checkpoint inhibitor durvalumab in patients with stage III, unresectable NSCLC who did not have disease progression after concurrent chemoradiotherapy) prolonged the median overall survival (OS) to 47.5 months significantly and reduced the risk of death by 29% for unresectable stage III NSCLC patients through 1-year consolidation immunotherapy with durvalumab following concurrent chemoradiotherapy (CRT).^[2] However, only 43 (6%) patients in the PACIFIC study harbored epidermal growth factor receptor (EGFR) mutations and these patients failed to achieve significant survival benefits compared to concurrent CRT.^[3] Therefore, the most reasonable treatment for these patients remains controversial and needs to be explored urgently.^[4,5]

Tyrosine kinase inhibitors (TKIs) targeting EGFR mutations have achieved great success and are widely applicated in stage IV non-squamous NSCLC. TKI treatment obtained about 9.0–18.9 months progression free survival (PFS),^[6–9] which was even longer than that of stage III unresectable NSCLC treated by concurrent CRT with only 6.3–9.0 months.^[10–12] The latest phase III randomized controlled trial (RCT) ADAURA study^[13] further confirmed that osimertinib could improve disease-free survival (DFS) in NSCLC patients with stage IB (hazard ratio [HR] = 0.39, 95% confidence interval [CI] 0.18–0.76), stage II (HR = 0.17, 95%CI, 0.08–0.31), and stage IIIA (HR = 0.12, 95%CI, 0.07–0.20) in adjuvant therapy. Since TKI had led to significant survival benefits for patients with EGFR mutation in both first-line therapy for advanced NSCLC and as adjuvant therapy for early-stage NSCLC, could TKI benefit unresectable stage III EGFR mutant NSCLC patients, and how about its efficacy compared with the present treatments?

Several studies explored the efficacy of TKI combined with radiotherapy (RT) and/or chemotherapy, and the results showed great heterogeneity. The GGCP study reported a 28.8% overall response rate (ORR) and a median PFS of 9.9 months and a median OS of 24.0 months in a Caucasus population.^[14] The LOGIK0902 study obtained a 85.0% ORR and a 90.0% 2-year OS rate, but this study included only a small sample of 20 patients.^[15] The WJOG 6911L^[16] and RECEL^[11] studies combined RT and TKI, reaching a median PFS of 18.6 months and 24.5 months, respectively, while the WJOG 6911L study^[16] achieved a median OS of 61.1 months. Although the ORR of TKI combined with durvalumab reached 63.3% in unresectable stage III EGFR mutant NSCLC, its median OS was only 10.1 months, and serious adverse events (AEs) were worthy of attention.^[17] In the retrospective study REFRACT from China,^[18] RT with TKI or TKI monotherapy were both superior to CRT in terms of PFS. However, it was RT with TKI that could bring a significant OS benefit compared to CRT. The global multi-center retrospective study KINDLE^[19] focused on stage III NSCLC, with only 34 cases of inoperable EGFR mutation, achieving a median PFS and OS of 14.6 months (95%CI, 8.0–18.5 months) and 24.4 months (95%CI, 15.4–34.9 months) with TKI monotherapy.

At present, there is still a lack of large RCTs comparing various treatment regimens, especially TKI-containing therapies for the subset of inoperable stage III EGFR mutant NSCLC. Therefore, the roles and relationship of TKI, chemotherapy, RT, and immunotherapy in these patients were unknown. In order to address this clinical issue and identify the most reasonable treatment strategies for unresectable locally advanced NSCLC harboring EGFR mutation, we conducted pooled analysis and indirect comparison of various treatment regimens through meta-analysis and systematic reviews.

Methods

The study was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses and the Cochrane Guidelines (PRISMA) extension statement given in Supplementary Table 1, http://links.lww.com/CM9/C220.

Study selection

We searched PubMed, Embase, the Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov databases to find relevant articles from inception to December 31, 2022. Abstracts on NSCLC from several important conferences (American Society of Clinical Oncology, European Society of Medical Oncology, and World Conference on Lung Cancer) from 2015 to 2022 were inspected. For an outcome in the same trial, only the latest data were kept. The detailed search strategy is presented in Supplementary Table 2, http://links.lww.com/CM9/C220.

We included phase II/III prospective or retrospective studies published in English that met the following criteria: (1) trials that enrolled patients with histologically diagnosed NSCLC with EGFR mutation; (2) trials that enrolled patients who were clinical stage III and inoperable; (3) for network meta-analysis, trials with the control group where the hazard ratio (HR) or odds ratio (OR) of outcomes, such as OS, PFS, ORR, and AE, were available; and (4) for pooled analysis, trials where the median survival time and 95% CI were available.

We had four exclusion criteria: (1) trials that only focused on the radiation dose, RT method, or application of RT protectors; (2) trials that only studied the timing or regimens of chemotherapy in CRT; (3) trials that applied monoclonal antibodies targeting EGFR; and (4) trials that included surgical treatment.

Data extraction and risk of bias assessment

Trial details (study ID, first author, publication year, number of patients, patient characteristics), treatments, and survival outcomes were extracted into a spreadsheet. We assessed risk of bias of five studies including control group using the Cochrane Risk of Bias Tool, which was based on random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other sources of bias. Items were scored as low, high, or unclear risk of bias^[20] in Supplementary Figure 1, http://links.lww.com/CM9/C220. All investigators independently conducted study selection and data extraction.

Data synthesis and statistical analysis

According to the PRISMA network meta-analysis extension statement,^[21] we used R (version 4.1.2; The R Project for Statistical Computing Vienna, Austria) based on the JAGS package (version 4-16, https://cran.r-project.org/web/packages/rjags/index.html) and the GeMTC package (version 1.0-2, https://cran.r-project.org/web/packages/gemtc/index.html) including five studies^{[2,11,12,18,22]} containing control groups to perform network meta-analysis [Supplementary Figure 2, http://links.lww.com/CM9/C220]. The primary outcome was OS and PFS. Secondary outcomes were ORR and AE ≥grade 3. Fixed and random-effects models were considered and compared using deviance information criteria. We used non-informative uniform and normal prior distributions to fit the model, with four different sets of initial values. We tested the adequacy of convergence (reaching a stable equilibrium distribution) using visual inspection methods of trace plots and estimating the values of the Brooks–Gelman–Rubin statistic in Supplementary Figure 3, http://links.lww.com/CM9/C220.

In addition, we conducted a pooled analysis of 15 clinical studies that met the inclusion criteria using STATA (version 16.0, StataCorp LP, College Station, TX, USA) to get an overview of the efficacy and toxicity.^{[2,10–12,14–19,23–27]} Heterogeneity was assessed between studies using the Q test and I² statistic.^[28] The estimated I² values ≤25%, >25% and <50%, or ≥50% indicated low, moderate, or high heterogeneity, respectively.^[29] When the heterogeneity was low or moderate, the selection of either the fixed-effect model or the random-effects model had little effect on the results. Otherwise, the random-effects model should be selected.^[30] We selected the random-effects model for the pooled analysis. For binary outcomes, such as ORR and AEs, we performed meta-analysis by pooling risk ratio. For time-to-event outcomes, such as PFS and OS, survival data was extracted and pooled in Supplementary Figure 4, http://links.lww.com/CM9/C220.

Sensitivity analyses

In addition to the principal analyses, a sensitivity analysis for network meta-analysis was conducted to test the robustness and reliability of results by excluding a small sample study.^[12]

Subgroup analysis

Two studies were included for subgroup analysis by different mutant subtypes because of the limited survival data. Besides, one of them was presented by Kaplan–Meier (KM) curves without the median survival time and 95% CI. Then, we obtained the time-to-event outcomes of this study according to the theoretical basis introduced by Guyot et al^[31] using the IPDfromKM package from R,^[32] and calculated the median PFS according to the time-to-event outcomes using packages such as IPDfromKM (version 0.1.10, https://cran.r-project.org/web/packages/IPDfromKM/index.html), RcppHungarian (version 0.3, https://cran.r-project.org/web/packages/RcppHungarian/index.html), survival (version 3.7-0, https://cran.r-project.org/web/packages/survival/index.html), tableone (version 0.13.2, https://cran.r-project.org/web/packages/tableone/index.html), and ggplot2 (version 3.5.1, https://cran.r-project.org/web/packages/ggplot2/index.html).

Results

Study selection and study characteristics

We initially identified 1261 records through titles and abstracts for screening and reviewed 89 reports in full text [Figure 1]. Finally, 16 studies were deemed eligible for inclusion with a total of 1156 patients enrolled. Six treatments were included: TKI, radiotherapy plus TKI (RT-TKI), induction or consolidation therapy of CRT combined with TKI (CRT-TKI), CRT followed by durvalumab (CRT-Durva), CRT, and TKI combined with durvalumab (TKI-Durva) [Table 1].

Figure 1.

Flow chart of study selection of prospective or retrospective study involving the patients (≥18 years) with unresectable locally advanced EGFR-mutant NSCLC. EGFR: Epidermal growth factor receptor; NSCLC: Non-small cell lung cancer.

Table 1.

Key study features of studies on locally advanced unresectable NSCLC with EGFR mutation.

Study	Type	Year	Population	Sample size	Male/Female	Median ages (years, intervention/control)	Intervention arm	Control arm	Outcome
REFRACT[18]	Retrospective	2020	Asian	440	265/175	56 (RT-TKI), 63 (TKI)/58	RT-TKI, TKI	CRT	OS, PFS
RECEL[11]	Prospective	2020	Asian	41	17/24	59.5/59.0	RT-TKI	CRT	OS, PFS, ORR, AE
KINDLE[19]	Retrospective	2021	White, Black and Asian	71	NA	NA	TKI	CRT	OS, PFS
PACIFIC[2]	Prospective	2017	White, Black and Asian	43	NA	NA	CRT-Durva	CRT	OS, PFS
Aredo et al[12]	Retrospective	2021	White, Black and Asian	37	8/29	67.7 (CRT-Durva), 69.3 (CRT-TKI)/67.9	CRT-Durva, CRT-TKI	CRT	OS, PFS, AE
GGCP[14]	Prospective	2013	White	66	60/6	61.5	CRT-TKI	NA	ORR, AE
LOGIK0902[15]	Prospective	2021	White, Black and Asian	20	9/11	66	CRT-TKI	NA	ORR, AE
Yagishita et al[23]	Retrospective	2014	Asian	34	16/18	62	CRT	NA	OS, PFS, ORR
Ishihara et al[10]	Retrospective	2017	Asian	15	10/5	61	CRT	NA	ORR
Lim et al[24]	Retrospective	2016	Asian	22	8/14	62	CRT	NA	ORR
Hsia et al[22]	Retrospective	2018	Asian	199	77/122	70.1/60.1	TKI	CRT	OS, PFS
Nisaa Zia et al[25]	Prospective	2019	Asian	49	40/9	57	CRT-TKI	NA	OS, PFS, ORR, AE
Lee et al[26]	Prospective	2017	Asian	12	4/8	61	RT-TKI	CRT-TKI	OS, PFS, ORR, AE
WJOG6911L[16]	Prospective	2021	Asian	27	13/14	67	RT-TKI	NA	PFS, ORR, AE
Creelan et al[17]	Prospective	2020	White, Black and Asian	40	18/22	61	TKI-Durva	NA	PFS, ORR, AE
Park et al[27]	Prospective	2019	Asian	36	13/23	52	CRT	NA	ORR

AEs: Adverse events; CRT: Chemoradiotherapy; Durva: Durvalumab; NA: Not available; ORR: Objective response rate; OS: Overall survival; PFS: Progression-free-survival; RT: Radiotherapy; TKI: Tyrosine kinase inhibitors.

Network meta-analysis of comparison of treatment regimens

PFS

In network meta-analysis, all the regimens containing TKI were significantly longer than CRT alone in PFS, regardless of TKI alone (HR = 0.66, 95% CI, 0.50–0.87), RT-TKI (HR = 0.37, 95% CI, 0.28–0.50) or CRT-TKI (HR = 0.14, 95% CI, 0.03–0.75), with a sequentially decreasing HR. Furthermore, both RT-TKI (HR = 0.40, 95% CI, 0.21–0.76) and CRT-TKI (HR = 0.15, 95% CI, 0.03–0.74) were superior to CRT-Durva. In addition, the PFS of TKI only was significantly shorter than that of RT-TKI (HR = 1.78, 95% CI, 1.17–2.67) [Figure 2A]. In PFS ranking according to the surface under the cumulative ranking curve (SUCRA), CRT-TKI, RT-TKI, and TKI were in the top three, while CRT-Durva and CRT ranked last as shown in Figure 3.

Figure 2.

Efficacy and safety profiles of the Bayesian network meta-analysis in unresectable locally advanced EGFR-mutant NSCLC. (A) HRs and 95% CIs of OS and PFS. (B) OR and 95% CIs of ORR and AEs of grade 3 or higher (≥3 AEs). Data in each cell are HR and 95% CIs for the comparison of row-defining treatment versus column-defining treatment. HR less than 1.00 favors upper-row treatment. Significant results are highlighted in red. AEs: Adverse events; CI: Confidence intervals; CRT: Chemoradiotherapy; Durva: Durvalumab; EGFR: Epidermal growth factor receptor; HRs: Hazard ratios; NSCLC: Non-small cell lung cancer; OR: Odds ratio; ORR: Overall response rate; PFS: Progression free survival; RT: Radiotherapy; TKI: Tyrosine kinase inhibitor.

Figure 3.

Bayesian ranking profiles comparing efficacy and safety of different treatment. Ranking plots indicate the probability of each comparable treatment strategies (A, TKI; B, RT-TKI; C, CRT-TKI; D, CRT-Durva; E, CRT) being ranked from first to last on OS, progression-free survival, objective response rate, and AEs of grade ≥3. AEs: Adverse events; CRT: Chemoradiotherapy; CRT-Durva: CRT followed by durvalumab; Durva: Durvalumab; OS: Overall survival; RT: Radiotherapy; TKI: Tyrosine kinase inhibitor.

OS

In network meta-analysis, HRs of OS were all close to 1.0 between any two treatment regimens, and the upper limit of 95% CI for each group exceeded 1.00, so there was no statistically significant difference [Figure 2A]. Considering the OS ranking, RT-TKI ranked first with the cumulative probability of 63.6%, TKI ranked second with the probability of 43.6%, CRT ranked third (probability of 48.2%), while CRT-Durva ranked last (probability of 45.6%) [Figure 3]. CRT-TKI could not be ranked because of the lack of OS data in relative trials.

ORR and AE

The ORR of RT-TKI was significantly higher than that of CRT (OR = 2.48, 95% CI, 1.06–6.05), but not significantly greater than that of CRT-TKI (OR = 0.52, 95% CI, 0.01–10.01). The ORR of CRT-TKI was not significantly better than that of CRT (OR = 4.38, 95% CI, 0.22–209.62). [Figure 2B]. In addition, no significant difference was found in the toxicity between RT-TKI, CRT-TKI, and CRT [Figure 2B].

Meta-analysis of comparison of different treatment strategies

In order to disclose the role of different treatment strategies in these patients, we classified these six treatment regimens according to whether they applied chemotherapy or TKI. We pooled and compared the OS and PFS of the strategies with or without chemotherapy, and with or without TKI. As shown in Figure 4A, a significant difference was found in the PFS (HR = 0.41, 95% CI, 0.22–0.74) for chemotherapy-free (TKI, RT-TKI) vs. chemotherapy-containing (CRT-TKI, CRT-Durva, CRT) strategies. Although no significant difference was found in the OS (HR = 0.79, 95% CI, 0.60–1.03), the upper limit was just over 1.00. Moreover, the PFS treated by TKI-containing strategies (TKI, RT-TKI, CRT-TKI) was significantly longer than the TKI-free ones (CRT-Durva, CRT), with an HR = 0.37 (95% CI, 0.20–0.66). However, the PFS advantage from TKI was only translated into a marginal benefit of OS (HR = 0.79, 95% CI, 0.60–1.03) [Figure 4A].

Figure 4.

Forest plots comparing survival outcomes among treatment strategies or different EGFR mutations in non-small cell lung cancer patients. (A) Pooled results of comparisons on median PFS and OS according to treatment strategies with/without chemotherapy or EGFR-TKI. (B) Median PFS of subgroup analysis in different EGFR mutation sites. CRT: Chemoradiotherapy; CT-free: Treatment strategies without chemotherapy; CT-containing: Treatment strategies with chemotherapy; Durva: Durvalumab; TKI-free: Treatment strategies without EGFR-TKI; TKI-containing: Treatment strategies with EGFR-TKI; TKI: Tyrosine kinase inhibitor; RT: Radiotherapy.

In the above meta-analysis, we compared the advantages and disadvantages of various treatment regimens through trials using a controlled arm. In order to disclose the overall efficacy and AEs of these regimens, we conducted a pooled analysis of OS, PFS, ORR, and AE from all relative trials, including single-arm trials.

OS and PFS

From the integrated meta-analysis, the pooled OS of RT-TKI, CRT, TKI, and CRT-TKI were about 60.8 (95% CI, 37.2–84.3) months, 52.0 (95% CI, 44.2–59.8) months, 35.3 (95% CI, 13.6–56.9) months, and 15.2 (95% CI, 8.2–22.3) months, respectively [Supplementary Figure 4A, http://links.lww.com/CM9/C220]. The pooled PFS of RT-TKI, TKI, CRT, TKI-Durva, and CRT-TKI were about 21.5 (95% CI, 15.4–27.5) months, 15.6 (95% CI, 13.2–18.1) months, 11.9 (95% CI, 10.3–13.5) months, 10.1 (95% CI, 5.3–15.0) months, and 8.4 (95% CI, 6.6–10.1) months, respectively, in Supplementary Figure 4B, http://links.lww.com/CM9/C220.

ORR

The pooled ORR of RT-TKI, CRT, CRT-TKI, and TKI-Durva were about 76.8% (95% CI, 65.7–88.0%), 76.3% (95% CI, 67.0–85.5%), 65.2% (95% CI, 35.3–95.1%), and 63.3% (95% CI, 46.1–80.6%) [Supplementary Figure 4C, http://links.lww.com/CM9/C220].

AE

In the pooled analysis of the incidences of AE, CRT-TKI showed the highest incidence of myelosuppression. The overall incidence of AE in neutropenia, anemia, and thrombocytopenia were 59.5%, 80.0%, and 80.0%, respectively, among which grade ≥3 AE was mainly neutropenia (40.3%). The incidence of nausea and vomiting caused by CRT was the highest (52.0%). TKI-Durva had the highest incidence of diarrhea (70%), and the diarrhea grade ≥3 mainly occurred in CRT-Durva (8.3%). CRT-Durva had 53.8% fatigue, but all of them were grades 1–2 AE. By contrast, TKI-Durva had 12.5% fatigue and all of them were grade 3 or above. In terms of abnormal liver function, the overall incidence of AE in TKI-Durva was 50.0%, and the incidence of AE grade ≥3 was 27.5%, which was higher than in the other regimens. The incidence of AE grade ≥3 rash in RT-TKI and CRT-TKI was 7.3% and 10.1%. The overall incidence of radiation pneumonitis and radiation esophagitis in RT-TKI was 39.0% and 18.3%, of which 13.3% and 16.7% were grade ≥3 AE. The overall incidence of radiation pneumonitis and radiation esophagitis in CRT-TKI was 80.0% and 70.0%, but the incidence of AE grade ≥3 was low [Supplementary Figure 5A and 5B and Supplementary Tables 3 and 4, http://links.lww.com/CM9/C220].

Subgroup analysis of different EGFR mutations

In subgroup analysis of the mutant subtype, limited to the small sample size, our results needed to be further confirmed. The median PFS of RT-TKI, CRT, and CRT-Durva were about 21.1 (95% CI, 5.3–31.9) months, 12.0 (95% CI, 6.2–17.7) months, and 6.2 (95% CI, 1.1–not available [NA]) months, respectively, in the exon 19 deletion subgroup. The median PFS of RT-TKI, CRT-Durva, and CRT were about 24.5 (95% CI, 9.1–29.4) months, 6.1 (95% CI, 1.8–15.4) months, and 7.8 months (95% CI, 2.3–NA), respectively, in the L858R mutation subgroup of exon 21 [Figure 4B].

Sensitivity analysis

In the network meta-analysis comparing the outcomes of different treatment regimens, we removed a retrospective study with a small sample size^[12] and found that the overall trend of OS and PFS remained the same. In the PFS comparison, treatment regimens including TKI, such as TKI monotherapy (HR = 0.66, 95% CI, 0.50–0.87) or RT-TKI (HR = 0.37, 95% CI, 0.27–0.51), were still significantly longer than CRT. Compared with CRT-Durva, RT-TKI could bring significant PFS advantage (HR = 0.44, 95% CI, 0.20–0.99). The PFS of TKI monotherapy was significantly shorter than that of RT-TKI (HR = 1.80, 95% CI, 1.20–2.70). There was no significant difference in OS among the treatments in Supplementary Figure 6, http://links.lww.com/CM9/C220.

In the sensitivity analysis of the comparison of the efficacy of the treatment strategies, after we had removed a retrospective study with a very broad 95% CI,^[12] the difference in the PFS between the TKI-containing and the TKI-free groups showed little change (HR = 0.41, 95% CI, 0.22–0.74), while other results remained stable [Supplementary Figure 7, http://links.lww.com/CM9/C220].

Discussion

As revealed in the EGFR mutation subgroup of the PACIFIC trial,^[2,3] our study confirmed that standard CRT in combination with durvalumab consolidation did not improve the survival of patients significantly vs. CRT for inoperable stage III EGFR mutant NSCLC. The underlying suppressive immune microenvironment induced by EGFR mutation may explain the poor efficacy of immunotherapy,^[33–35] as in the advanced setting. The PFS of TKI monotherapy, RT-TKI, or CRT-TKI was significantly longer than that of CRT, and the PFS of RT-TKI or CRT-TKI was significantly better than CRT-Durva. These evidences initially reveal the essential role of TKI in the research population. To further distinguish the effects of chemotherapy and TKI on these patients, we conducted a meta-analysis comparing treatment strategies with or without chemotherapy or TKI and found that chemotherapy-free treatments had a significantly longer PFS (HR = 0.41, 95% CI, 0.22–0.74) and a superior OS (HR = 0.79, 95% CI, 0.60–1.03) in contrast to the chemotherapy-containing treatments. Chemotherapy-free treatment might have potential survival benefits. The reason lies in that TKI was commonly used in the chemotherapy-free group, but was rarely used in the chemotherapy group. Similarly, the PFS of patients treated with TKI was significantly better than those without TKI, and OS also showed a benefit trend, with the upper limit of 95% CI of 1.03. Combining the better PFS of RT-TKI vs. CRT (HR = 0.37, 95% CI = 0.28–0.50) in the above network meta-analysis and the longer PFS (21.5 months vs. 11.9 months) in the pooled analysis, we believe that TKI may be more suitable than chemotherapy for inoperable locally advanced NSCLC patients with EGFR mutation.

Despite being considered an important treatment option, TKI monotherapy, due to its significantly shorter mPFS, did not prove to be sufficient compared to RT-TKI (15.6 vs. 21.5 months in the pooled analysis), having an HR of 1.78 (95% CI, 1.17–2.67) in the network meta-analysis. Moreover, the combination of RT and TKI ranked first in the cumulative probability ranking of SUCRA values, and both its OS (60.8 months vs. 35.3 months for TKI only) and PFS were the longest in the pooled analysis. This highlights the indispensable role played by RT in these populations. Taken together, RT and TKI treatments are essential for these patients. Although there was no statistically significant difference in the OS in this network meta-analysis, which may have been affected by the time TKI was applied relative to disease progression,^[26] we believe that radiotherapy combined with TKI may be the most effective treatment for unresectable stage III EGFR mutant NSCLC based on current research data.

Basic studies have confirmed that EGFR mutant NSCLC is more sensitive to RT and that TKI has a synergistic effect on RT.^[36–38] In clinical trials, RT was regarded as an independent positive prognostic factor for OS in stage IIIB patients with EGFR mutation.^[22] It has been reported that the induction therapy of osimertinib can shrink the gross tumor volume (GTV) and facilitate efficacy of the subsequently administered RT in IIIA/B unresectable NSCLC.^[39] Even in advanced NSCLC with EGFR mutation, RT for primary or oligometastasic lesions combined with targeted therapy could improve the patient’s survival significantly.^[40,41] These survival benefits of RT-TKI in advanced NSCLC are consistent with our results in locally advanced EGFR mutant NSCLC. Furthermore, the function of RT-TKI needs confirmation through large phase III RCTs such as the ADVANCE trial,^[42] which explores the efficacy and safety of RT combined with peri-RT usage of another third-line EGFR-TKI almonertinib, compared with CRT. In addition, the appropriate sequence of RT and TKI, or which generation of EGFR-TKI is more suitable for RT, remains to be determined.

Although CRT-TKI showed a significant benefit in the PFS compared with CRT or CRT-Durva, no significant advantage existed when comparing with TKI or RT-TKI in the network meta-analysis. Notably, the aforementioned analysis of the OS in the network meta-analysis lacks CRT-TKI data from qualified studies. CRT-TKI had an even shorter PFS (8.4 months) and OS (15.2 months) than RT-TKI (PFS 21.5 months, OS 60.8 months) or TKI (PFS 15.6 months, OS 35.3 months) in the pooled analysis. Our network meta-analysis included only one study,^[12] including eight patients with an extremely significant PFS benefit (HR = 0.15) compared with CRT. Also, in the pooled analysis, only four studies reported the HR of OS; two of them (PACIFIC and RECEL)^[2,11] were prospective studies with few EGFR mutant cases, and the other two were retrospective real-world studies.^[12,18] Therefore, the data of CRT-TKI mainly came from one study,^[25] with markedly shorter OS (12.9 months) and PFS (7.4 months) than the other studies, which might lower the final pooled outcomes. However, due to the fact that other studies containing the CRT-TKI regimen had too small sizes, or did not report complete OS and PFS, or included a homogeneous population,^{[14,15,26,43]} we still included this study for integrated analysis, intending to give an overview of OS and PFS for the current treatment measures in unresectable stage III EGFR mutant NSCLC. In addition, the CRT-TKI regimen itself has some limitations. Considering the previous analysis of the efficacy comparison of treatment strategies with or without chemotherapy, the role of chemotherapy in such population may be very limited. Moreover, CRT followed by TKI may reduce patient compliance due to increasing toxicities and result impaired survival efficacy. Obviously, we still lack sufficient research results to conduct a comprehensive analysis of the efficacy of CRT-TKI, which needs further large RCTs, such as the ongoing phase III clinical trial LAURA (NCT03521154)^[44] and POLESTAR (NCT04951635),^[45] all of which apply the third generation TKI following CRT. However, the radiation pneumonia after third generation TKI in combination with RT was more serious and might affect the completion of treatment.^[46,47]

As RT-TKI played an important role in inoperable stage III EGFR mutant NSCLC, the administration sequence of TKI and RT or CRT become an issue worth exploring. From the current limited results, the median OS of TKI maintenance after sequential^[25] or concurrent^[14] CRT was 12.9 months and 24.0 months, less than 39.3 months of TKI induction therapy followed by RT with or without TKI maintenance therapy,^[26] or 2-year OS rate of 90% from TKI induction followed by CRT.^[15] Given the small size of the current phase II clinical trial on application orders, the suitable order of TKI and RT or CRT needs to be verified by large prospective RCTs.

Based on the pooled analysis of AEs, we found that TKI together with RT or CRT had a higher probability of myelosuppression, abnormal liver function and rash compared with CRT-Durva. In particular, the incidences of radiation pneumonitis and radiation esophagitis in TKI with RT or CRT were higher than those in CRT. In addition, the higher incidence of serious AEs in TKI with or after durvalumab administration was worth noting.^[17,48]

Our research has some limitations. Firstly, some of the included studies had small sample sizes, which might bring bias to the results. The differences between the two analyses may related to their partly different in included trials, for only five studies^{[2,11,12,18,22]} met the inclusion criteria of network meta-analysis. Especially for the CRT-TKI treatment, there was only one trial with eight patients could be included in the PFS comparison and no qualified study was included in the OS comparison for the network meta-analysis, which was bound to affect the final results. Moreover, studies included in pooled analysis were also affected by heterogeneity bias in single-arm study populations. Studies such as REFRACT,^[18] RECEL,^[11] and Hsia TC^[22] were performed in Asia, while PACIFIC^[2] and Aredo et al^[12] enrolled global populations. In terms of gender, the ratio of male to female in the REFRACT^[18] was 1.5:1.0, while the proportion of women was higher than that of men in other studies. Moreover, in the study of Aredo et al^[12] the median age of patients was more than 65 years old, which was also one of the sources of heterogeneity. Secondly, due to the poor availability of prospective studies, the included studies contained some retrospective studies with relatively limited quality and strictness. Thirdly, different studies with the same treatments might not have the exactly same order of treatments. Given the size of included studies, we merged the same treatment regimens of various application sequences to make the complex issue more intuitive for comparison. Our study aims to explore the optimal combination of different treatments; the comparison of RT dose and mode was not included.

In summary, although limited by the sample size and research quality, our study provides a comprehensive summary of treatment options for unresectable stage III NSCLC with EGFR mutation, which remains a clinically difficult issue. Based on the available evidences, the combination of RT and TKI brings the best survival advantage, superior to CRT with or without immunotherapy, or TKI only. Overall efficacy of CRT plus TKI was hard to evaluate to date, while TKI with immunotherapy had serious AEs instead of satisfactory survival. Furthermore, the role and status of chemotherapy, and the timing of application among TKI, RT, or chemotherapy, need to be explored. We provide a valuable basis for optimization of treatment regimens and evidence to perform related large prospective RCTs for these patients.

Conflicts of interest

None.

Supplementary Material

cm9-138-1687-s001.pdf ^{(792.8KB, pdf)}