Learn more: PMC Disclaimer | PMC Copyright Notice
First-line treatment of EGFR-mutated non-small cell lung cancer with brain metastases: a systematic review and meta-analysis
Abstract
This systematic review and network meta-analysis evaluates first-line treatment options for patients with EGFR-mutant non-small cell lung cancer (NSCLC) and brain metastases. We analyzed 24 randomized controlled trials (RCTs) involving 2,682 patients, comparing various EGFR tyrosine kinase inhibitors (TKIs) and combination therapies. Direct comparisons showed that the addition of bevacizumab or chemotherapy to first-generation (1G) EGFR-TKIs improved overall survival (OS) compared to 1G TKIs alone, with HRs of 0.704 (95% CI: 0.433–0.973) and 0.682 (95% CI: 0.464–0.899), respectively. However, third-generation (3G) TKI monotherapy did not significantly improve OS compared with 1G TKIs, with an HR of 0.855 (95% CI: 0.511–1.198). Indirect comparisons suggested that the combination of 3G TKIs with chemotherapy provided the most significant improvements in OS and progression-free survival (PFS), significantly outperforming 3G TKIs, with HRs of OS 1.69 (95% CI: 1.14–3.4) and PFS 2.13 (95% CI: 1.28–3.54). Intracranial PFS was best with 1G TKIs plus bevacizumab. Our findings suggest that 3G EGFR-TKIs in combination with chemotherapy may be the most effective strategy for patients with EGFR-mutant NSCLC and brain metastases, though further head-to-head trials are needed for validation.
Introduction
Lung cancer is the leading cause of cancer-related deaths worldwide, with non-small cell lung cancer (NSCLC) being the most predominant type. Approximately 30% to 40% of patients with advanced NSCLC experience brain metastasis (BM)1. BM not only has a significant impact on prognosis but also complicates the selection of treatment strategies. The advent of targeted therapies, particularly epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKIs), has revolutionized the treatment landscape for EGFR-mutant NSCLC, demonstrating improved outcomes over conventional chemotherapy (CT). Among them, third-generation (3G) TKIs have a greater ability to penetrate the blood–brain barrier (BBB) than first (1G) or second-generation (2G) TKIs2–4. Several phase III studies, including FLAURA, AENEAS, and FURLONG, have shown that 3G TKIs significantly improved intracranial progression-free survival (iPFS) in NSCLC patients with BM compared to 1G TKIs, with hazard ratios (HRs) ranging from 0.32 to 0.482–4. However, Despite these improvements, these patients still face a poor prognosis due to the development of secondary resistance to EGFR-TKIs. In particular, given the high risk of brain progression with subsequent treatment lines, there is a pressing need for effective first-line therapeutic strategies that can achieve better intracranial disease control5.
In recent years, the therapeutic armamentarium for EGFR-mutant NSCLC with BM has expanded to include not only different generations of EGFR-TKIs with varying intracranial antitumor activity, but also novel EGFR-TKI-based combination strategies, including combinations of EGFR-TKIs with CT, anti-angiogenic agents such as monoclonal antibodies (mAbs) against vascular endothelial growth factor (VEGF) or VEGF receptor (VEGFR), and bispecific antibodies (BsAbs) against EGFR and mesenchymal epithelial transition factor (MET). The development of these combination therapies has the potential to increase intracranial antitumor efficacy and prolong survival in this high-risk patient population6–8.
Despite the advances in therapies, the multiplicity of options adds a new layer of complexity to clinical decision-making for those NSCLC patients with BM. Questions regarding the relative merits of monotherapy versus combination strategies, the impact of different generations of EGFR-TKIs, and the role of anti-angiogenic agents in modifying the disease course remain unresolved. Therefore, determining an optimal first-line treatment strategy to optimize survival outcomes and control disease progression remains a pressing clinical challenge.
This study aims to address this critical knowledge gap by conducting a comprehensive network meta-analysis (NMA) that integrates data from multiple randomized controlled trials (RCTs) to compare the efficacy of different first-line treatment strategies in patients with EGFR-mutant NSCLC-related BM. Ultimately, our findings are intended to guide clinical practice by providing clinicians with suggestions for developing individualized treatments for patients with BM, and may contribute to improved survival in this challenging patient cohort.
Methods
This meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines9. The study was registered on the International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY202460015).
Data sources and search strategy
A comprehensive literature search of published articles and conference abstracts was conducted to assess the efficacy of different generations of EGFR-TKIs and their combination therapies in EGFR-mutant NSCLC with BM. The search was conducted in PubMed, EMBASE, and the Cochrane Library databases from inception to June 15, 2024 using the following strategy: (EGFR-TKIs OR gefitinib OR icotinib OR erlotinib OR afatinib OR dacomitinib OR osimertinib OR aumolertinib OR almonertinib OR furmonertinib OR lazertinib OR amivantamab OR zorifertinib OR AZD3759) AND (NSCLC OR lung cancer) NOT adjuvant NOT neoadjuvant. Conference abstracts were screened from major oncology conferences, including the American Society of Clinical Oncology, the European Society for Medical Oncology, the World Conference on Lung Cancer, and the European Lung Cancer Congress, with bibliographies of retrieved articles and related studies examined for supplementary relevant publications.
Inclusion and exclusion criteria
Two researchers independently screened titles and abstracts, and disagreements were resolved by a third researcher. After the initial screening, the full-text was reviewed with the following criteria for final inclusion: (1) Patients: untreated NSCLC patients with BM harboring EGFR activating mutations in exon 19 deletion or L858R. (2) Interventions: including different generations of EGFR-TKIs or combination regimens based on EGFR-TKI in combination with CT, anti-angiogenic inhibitors (e.g., VEGFR mAb or TKIs), or EGFR-Met BsAb. (3) Comparisons: comparisons between different generations of EGFR-TKIs or their combination therapies. (4) Outcomes: reporting at least one or more of the following outcomes in NSCLC patients with BM: progression-free survival (PFS), overall survival (OS), objective response rate (ORR), iPFS, intracranial ORR (iORR), intracranial disease control rate (iDCR), or intracranial duration of response (iDOR), with a focus on statistically reported and analyzable data. (5) Study design: prospective RCTs only, excluding observational studies, case reports, and retrospective analyses.
Data extraction
Two investigators independently extracted data from the main text, tables, and figures of each included study into a standardized Microsoft Excel spreadsheet. Extracted data included, but were not limited to, lead author, year of publication, geographic region, total number of patients, number of patients with BM, study design, patient age, gender, treatment regimen, follow-up duration, PFS, OS, ORR, along with their intracranial equivalents, safety, and corresponding HR values. Any disagreements encountered during the data extraction process were resolved through iterative discussions between the reviewers. If consensus could not be reached, a third investigator was consulted to arbitrate and finalize the data entry.
Assessment of study quality
Study quality was assessed using methods recommended by the Cochrane Collaboration10, focusing on random sequence generation, allocation concealment, blinding of participants and personnel, outcome assessment blinding, incomplete outcome data, selective reporting, and other biases. Each criterion was categorized as high, low, or unclear risk.
Main outcomes
Primary outcomes included OS, PFS, and iPFS. The secondary outcomes included iORR, iDCR, and iDOR.
Statistical analysis
Direct comparisons were statistically analyzed using Stata 12.0 software (StataCorp LP, College Station, TX 77,845, USA; 2011). Primary outcomes for the BM population (ORR, PFS, OS, and HR values) were summarized, with the I2 statistic used to evaluate heterogeneity between studies11. If I2 > 50%, a random-effects model (DerSimonian-Laird method) was used, otherwise, a fixed-effects model (inverse variance method) was used. Network meta-analyses were performed using R software (version 4.3.2) and the netmeta package (version 2.9.0) to generate network graphs, forest plots, adjusted funnel plots, node-split analysis comparing direct and indirect methods, and SUCRA rankings between interventions12,13. Funnel plots and Egger’s tests were used to detect publication bias14. Statistical significance was considered when the bilateral p-value was less than 0.05.
Results
Characteristics of the included studies
A PRISMA flowchart illustrating the systematic selection process for our meta-analysis is shown in Fig. 1. The PRISMA checklist for this review is provided as a supplementary file (Table S1). After an extensive database search and duplicate removal, a careful full-text review resulted in the inclusion of 24 relevant trials, including 18 phase III and 6 phase II RCTs. A total of 7,231 patients with EGFR-mutated NSCLC were enrolled, including 2,682 patients with BM. Among these, 197 patients with BM in eight studies had received previous radiotherapy, representing 7.3% of the total BM patients (Table (Table1).1). The evaluated regimens included three EGFR-TKI monotherapies and eight EGFR-TKI-based combination strategies7,8,15–40. Specifically, 10 trials compared the efficacy of 3G TKIs with 1G TKIs2–4,24–30,32–34. In addition, three trials each evaluated 1G TKIs + CT versus 1G TKIs6,35–37 and 1G TKIs + VEGF mAb versus 1G TKIs7,15–17 and two trials compared 3G TKIs + VEGFR mAb to 3G TKIs19,20. In addition, other independent studies evaluated 2G TKIs versus 1G TKIs22,23, 1G TKIs + VEGFR TKI versus 1G TKIs16, 2G TKIs + VEGF mAb versus 2G TKIs35, 3G TKIs + CT versus 3G TKIs38,39, 3G TKIs + EGFR-MET BsAb (amivantamab) versus 3G TKIs8, and 3G TKIs + VEGF mAb versus 3G TKIs18. The chemotherapy regimens used in the included trials align with guideline-recommended treatments for advanced NSCLC, including the combination of pemetrexed with platinum-based chemotherapy (details provided in Table Table1).1). A network diagram was drawn to clearly illustrate the directly comparable distributions of the different outcome measures (Fig. 2). Table Table11 summarizes the basic characteristics of the trials included in the study.
Table 1
Trial name (Year) | Phase | Comparison | Study group size (n/N) | Control group size (n/N) | Prior brain RT n1 (%) vs n2 (%) | OS HR (95% CI) | PFS HR (95% CI) | iPFS HR (95% CI) | iORR OR (95% CI) | iDCR OR (95% CI) | iDOR HR (95% CI) |
---|---|---|---|---|---|---|---|---|---|---|---|
2G EGFR-TKIs vs 1G EGFR-TKIs | |||||||||||
LUX-Lung 7 (2016/2017)22,23 | Phase 2b | afatinib vs gefitinib | 26/160 | 24/159 | – | 1.16 (0.61–2.21) | 0.76 (0.41–1.44) | NA | NA | NA | NA |
3G EGFR-TKIs vs 1G EGFR-TKIs | |||||||||||
AENEAS (2022)2,24 | Phase 3 | aumolertinib vs gefitinib | 56/214 | 59/215 | 4 (7.1%) vs 5 (8.5%) | NA | 0.38 (0.24–0.6) | 0.32 (0.18–0.58) | 1.75 (0.80–3.79) | 0.60 (0.10–3.77) | 0.19 (0.08–0.47) |
IBIO-103 (2023)26 | Phase 3 | befotertinib vs icotinib | 59/182 | 57/180 | – | NA | 0.48 (0.28–0.84) | 0.69 (0.36–1.33) | 2.33 (0.37–14.61) | 90% vs 100% | NA |
EVEREST (2023)34 | Phase 3 | zorifertinib vs gefitinib | 220/220 | 219/219 | – | NA | 0.72 (0.58–0.89) | 0.47 (0.35–0.62) | 1.66 (0.99–2.77) | NA | NA |
FLAURA (2018/2020)4,30,32 | Phase 3 | osimertinib vs gefitinib/erlotinib | 53/279 | 63/277 | 15 (28.3%) vs 16 (25.4%) | 0.83 (0.53–1.30) | 0.47 (0.30–0.74) | 0.48 (0.26–0.86) | 2.5 (1.2–5.2) | 2.5 (0.2–55.8) | NR (11.9-NC) vs 14.4 (7.0–18.7) |
FLAURA China (2021)31 | Phase 3 | osimertinib vs gefitinib/erlotinib | 17/71 | 21/65 | – | 0.95 (0.45–1.97) | 0.66 (0.30–1.38) | NA | NA | NA | NA |
FURLONG (2022)3,25 | Phase 3 | furmonertinib vs gefitinib | 63/178 | 58/179 | 0 (0) vs 1 (1.7%) | NA | 0.5 (0.32–0.8) | 0.40 (0.23–0.71) | 6.82 (1.23–37.67) | 100% vs 84% | 0.47 (0.15–1.41) |
LASER301 (2023)27 | Phase 3 | lazertinib vs gefitinib | 51/196 | 48/197 | – | NA | 0.42 (0.26–0.68) | NA | NA | NA | |
LASER301 Asian (2023)28 | Phase 3 | lazertinib vs gefitinib | 39/129 | 31/129 | – | NA | 0.33 (0.18–0.58) | NA | NA | NA | |
LASER301 Korean (2023)29 | Phase 3 | lazertinib vs gefitinib | 31/87 | 25/85 | – | NA | 0.28 (0.15–0.53) | NA | NA | NA | |
LASER301 subset analysis (2023)33 | Phase 3 | lazertinib vs gefitinib | 45/45 | 41/41 | 11 (24%) vs 11 (27%) | NA | NA | 0.42 (0.2–0.89) | 6.18 (0.61–62.83) | 1.21 (0.07–21.22) | NR (8.31-NR) vs 6.3 (2.79-NR) |
1G EGFR-TKIs + CT vs 1G EGFR-TKIs | |||||||||||
NEJ009 (2020/2022)6,35 | Phase 3 | gefitinib + CT (CabP) vs gefitinib | 50/170 | 38/172 | 17 (34%) vs 15 (39.4%) | 0.73 (0.47–1.16) | 0.32 (0.19–0.53) | NA | NA | NA | NA |
Vanita Noronha (2020)36 | Phase 3 | gefitinib + CT (CisP) vs gefitinib | 30/174 | 34/176 | 22 (73.3%) vs 31 (91.2%) | NA | 0.53 (0.29–0.98) | NA | NA | NA | NA |
GAP BRAIN (2023)36 | Phase 3 | gefitinib + CT (CisP/NedP) vs gefitinib | 80/80 | 81/81 | – | 0.65 (0.43–0.99) | 0.39 (0.27–0.58) | 0.36 (0.25–0.53) | 3.33 (1.56–7.14) | 1.25 (0.32–4.84) | NA |
3G EGFR-TKIs + CT vs 3G EGFR-TKIs | |||||||||||
FLAURA2 (2023/2024)38,39 | Phase 3 | osimertinib + CT (CisP/CabP) vs osimertinib | 116/278 | 110/279 | 16 (14.5%) vs 18 (16.4%) | 0.59 (0.4–0.87) | 0.47 (0.33–0.66) | 0.58 (0.33–1.01) | 1.19 (0.67–2.14) | 0.70 (0.26–1.88) | NR (23.8-NC) vs 26.2 (19.4-NC) |
3G EGFR-TKIs + EGFR-Met BsAb vs 3G EGFR-TKIs | |||||||||||
MARIPOSA (2023)8 | Phase 3 | amivantamab + lazertinib vs osimertinib | 176/429 | 176/429 | – | NA | 0.69 (0.53–0.92) | NA | NA | NA | NA |
1G EGFR-TKIs + VEGF mAb vs 1G EGFR-TKIs | |||||||||||
NEJ026 (2019/2020)15,16 | Phase 3 | erlotinib + bevacizumab vs erlotinib | 36/112 | 36/112 | – | 0.84 (0.43–1.63) | 0.78 (0.42–1.43) | NA | NA | NA | NA |
ARTEMIS-CTONG1509 (2021)7 | Phase 3 | erlotinib + bevacizumab vs erlotinib | 44/157 | 47/154 | – | 0.62 (0.38–1.01) | 0.48 (0.27–0.84) | NA | NA | NA | NA |
Lee Youngjoo (2022)17 | Phase 2 | erlotinib + bevacizumab vs erlotinib | 29/64 | 30/63 | 3 (10.3%) vs 12 (40%) | 1.27 (0.58–2.79) | 0.54 (0.31–0.95) | 0.18 (0.05–0.67) | NA | NA | NA |
2G EGFR-TKIs + VEGF mAb vs 2G EGFR-TKIs | |||||||||||
Takashi Ninomiya (2023)40 | Phase 2 | afatinib + bevacizumab vs afatinib | 13/49 | 19/50 | – | NA | 0.52 (0.20–1.33) | NA | NA | NA | NA |
3G EGFR-TKIs + VEGF mAb vs 3G EGFR-TKIs | |||||||||||
WJOG9717L (2022)18 | Phase 2 | osimertinib + bevacizumab vs osimertinib | 18/61 | 23/62 | – | NA | 0.83 (0.36–1.94) | NA | NA | NA | NA |
3G EGFR-TKIs + VEGFR mAb vs 3G EGFR-TKIs | |||||||||||
OSIRAM-1 (2023)19 | Phase 2 | osimertinib + ramucirumab vs osimertinib | 16/59 | 18/61 | – | NA | 0.65 (0.29–1.45) | NA | NA | NA | NA |
RAMOSE (2023)20 | Phase 2 | osimertinib + ramucirumab vs osimertinib | 40/93 | 24/46 | – | NA | 0.65 (0.32–1.31) | ||||
1G EGFR-TKIs + VEGFR TKI vs 1G EGFR-TKIs | |||||||||||
CTONG1706 (2021)21 | Phase 3 | gefitinib + apatinib vs gefitinib + placebo | 51/157 | 41/156 | – | NA | 0.91 (0.5–1.64) | NA | NA | NA | NA |
n/N: represents the number of patients with brain metastases as a proportion of the total number of patients. n1 (%) vs n2 (%): denotes the number and percentage of patients who received prior brain radiotherapy among those with brain metastases in the study and control groups, respectively
HR: Hazard ratio; OR: Odds ratio; CI: Confidence intervals; EGFR-TKIs: Epidermal growth factor receptor tyrosine kinase inhibitors; 1G/2G/3G EGFR-TKIs: First/second/third-generation EGFR-TKIs; VEGF mAbs: Vascular endothelial growth factor monoclonal antibodies; VEGFR mAbs: Vascular endothelial growth factor receptor monoclonal antibodies; EGFR-MET BsAb: Epidermal growth factor receptor-mesenchymal epithelial transition factor bispecific monoclonal antibodies; CT: Chemotherapy; CabP: Carboplatin plus pemetrexed; CisP: Cisplatin plus pemetrexed; NedP: Nedaplatin plus pemetrexed; RT: Radiotherapy.
Comparison of OS
This NMA of OS in patients with EGFR-mutant NSCLC-related BM included data from nine trials with six different treatment strategies7,16,17,23,31,32,35,37,39. In head-to-head comparisons, 1G TKIs plus VEGF mAb (bevacizumab) significantly reduced the risk of death by 29.6% compared with 1G TKIs, with an HR of 0.704 (95% CI: 0.433–0.973). The combination of 1G TKIs and CT significantly reduced the risk of death by 31.8%, with an HR of 0.682 (95% CI: 0.464–0.899). However, 3G TKIs did not confer a significant OS benefit over 1G TKIs, with an HR of 0.855 (95% CI: 0.511–1.198) (Table (Table22).
Table 2
Treatment regimen | Number of trials | Brain metastases cases n1 vs n2 | Pooled survival data | Pooled HR or OR (95% CI) |
---|---|---|---|---|
OS, months | ||||
1G EGFR-TKIs + VEGF mAb vs 1G EGFR-TKIs | 3 | 109 vs 113 | NA | 0.70 (0.43–0.97) |
2G EGFR-TKIs vs 1G EGFR-TKIs | 1 | 26 vs 24 | NA | 1.16 (0.61–2.21) |
3G EGFR-TKIs vs 1G EGFR-TKIs | 2 | 70 vs 84 | NA | 0.85 (0.51–1.20) |
1G EGFR-TKIs + CT vs 1G EGFR-TKIs | 2 | 130 vs 119 | NA | 0.68 (0.46–0.90) |
3G EGFR-TKIs + CT vs 3G EGFR-TKIs | 1 | 116 vs 110 | NA | 0.59 (0.4–0.87) |
PFS, months | ||||
1G EGFR-TKIs + VEGF mAb vs 1G EGFR-TKIs | 3 | 109 vs 113 | 18.18 (16.04 -20.32) vs 11.0 (9.69–12.31) | 0.55 (0.35–0.74) |
2G EGFR-TKIs + VEGF mAb vs 2G EGFR-TKIs | 1 | 13 vs 19 | 17.7 (9.8–21.9) vs 11.7 (4.2–18.7) | 0.52 (0.20–1.33) |
3G EGFR-TKIs + VEGF mAb vs 3G EGFR-TKIs | 1 | 18 vs 23 | NA | 0.83 (0.360–1.93) |
3G EGFR-TKIs + VEGFR mAb vs 3G EGFR-TKIs | 2 | 56 vs 42 | NA | 0.65 (0.28–1.03) |
1G EGFR-TKIs + VEGFR TKI vs 1G EGFR-TKIs | 1 | 51 vs 41 | NA | 0.91 (0.5–1.64) |
2G EGFR-TKIs vs 1G EGFR-TKIs | 1 | 26 vs 24 | 7.2 (3.7–17) vs 7.4 (5.4–12.8) | 0.76 (0.41–1.44) |
3G EGFR-TKIs vs 1G EGFR-TKIs | 9 | 589 vs 581 | 10.18 (9.46–10.90) vs 8.25 (7.70–8.79) | 0.46 (0.35–0.56) |
1G EGFR-TKIs + CT vs 1G EGFR-TKIs | 3 | 160 vs 153 | NA | 0.375 (0.27–0.48) |
3G EGFR-TKIs + CT vs 3G EGFR-TKIs | 1 | 116 vs 110 | 24.9 (22-NC) vs 13.8 (11–16.7) | 0.47 (0.33–0.66) |
3G EGFR-TKIs + EGFR-Met BsAb vs 3G EGFR-TKIs | 1 | 176 vs 176 | 18.3 (16.6–23.7) vs 13 (12.2–16.4) | 0.69 (0.53–0.92) |
iPFS, months | ||||
1G EGFR-TKIs + VEGF mAb vs 1G EGFR-TKIs | 1 | 29 vs 30 | NA | 0.18 (0.05–0.67) |
3G EGFR-TKIs vs 1G EGFR-TKIs | 6 | 501 vs 501 | 16.98 (14.13–19.83) vs 8.32 (7.64–9.01) | 0.43 (0.34–0.52) |
1G EGFR-TKIs + CT vs 1G EGFR-TKIs | 1 | 80 vs 81 | 15.6 (14.3–16.9) vs 9.1 (8–10.2) | 0.36 (0.25–0.53) |
3G EGFR-TKIs + CT vs 3G EGFR-TKIs | 1 | 118 vs 104 | NA | 0.58 (0.33–1.01) |
iORR | ||||
3G EGFR-TKIs vs 1G EGFR-TKIs | 6 | 502 vs 504 | 82% (74.6–89.4%) vs 54.7% (46.1–63.2%) | 1.82 (1.01–2.62) |
1G EGFR-TKIs + CT vs 1G EGFR-TKIs | 1 | 80 vs 81 | 85% vs 63% | 3.33 (1.56–7.14) |
3G EGFR-TKIs + CT vs 3G EGFR-TKIs | 1 | 116 vs 110 | 73% (64–81%) vs 69% (59–78%) | 1.19 (0.67–2.14) |
iDCR | ||||
3G EGFR-TKIs vs 1G EGFR-TKIs | 3 | 160 vs 170 | 94.4% (88.7–100%) vs 95.2% (89.9–100%) | 0.63 (-1.18–2.44) |
1G EGFR-TKIs + CT vs 1G EGFR-TKIs | 1 | 80 vs 81 | 95% vs 93.8% | 1.25 (0.32–4.84) |
3G EGFR-TKIs + CT vs 3G EGFR-TKIs | 1 | 116 vs 110 | 91% (84–95%) vs 93% (87–97%) | 0.70 (0.26–1.88) |
iDOR | ||||
3G EGFR-TKIs vs 1G EGFR-TKIs | 2 | 116 vs 117 | – | 0.22 (0.03–0.40) |
n1 vs n2: denotes the number of patients with brain metastases in the study and control groups, respectively
HR: Hazard ratio; OR: Odds ratio; CI: Confidence intervals; EGFR-TKIs: Epidermal growth factor receptor tyrosine kinase inhibitors; 1G/2G/3G EGFR-TKIs: First/second/third-generation EGFR-TKIs; VEGF mAbs: Vascular endothelial growth factor monoclonal antibodies; VEGFR mAbs: Vascular endothelial growth factor receptor monoclonal antibodies; EGFR-MET BsAb: Epidermal growth factor receptor-mesenchymal epithelial transition factor bispecific monoclonal antibodies; CT: Chemotherapy.
Indirect comparisons of OS showed that in patients with EGFR mutations and BM, the treatment regimens ranked in best-to-worst order in the NMA were 3G TKIs + CT, 1G TKIs + CT, 1G TKIs + VEGF mAb, 3G TKIs, 1G TKIs, and 2G TKIs with probabilities of 93.5%, 72.6%, 56.2%, 43.6%, 19.2%, and 15.1%, respectively. 3G TKIs + CT appeared to be the most effective in prolonging OS, significantly outperforming 3G TKIs and 1G TKIs with HRs of 1.69 (95% CI: 1.14–3.4) and 1.97 (95% CI: 1.07–1.98), respectively, but not significantly better than 1G TKIs + CT and 1G TKIs + VEGF mAb. In addition, 1G TKIs + CT also showed a better survival benefit compared to 1G TKIs, with an HR of 1.46 (1.07–1.98), but was not superior to 3G TKIs and 1G TKIs + VEGF mAb. 1G TKIs + VEGF mAb did not demonstrate a significant improvement in OS compared to 1G or 3G TKI monotherapy. Consistent with the results of the direct comparison, the indirect comparison also showed no improvement in OS with 3G TKIs compared to 1G TKIs (Table (Table3,3, Fig. 3A).
Table 3
1G EGFR-TKIs | |||||
0.86 (0.45–1.64) | 2G EGFR-TKIs | ||||
1.16 (0.79–1.70) | 1.35 (0.64–2.85) | 3G EGFR-TKIs | |||
1.46 (1.07–1.98)* | 1.69 (0.83–3.45) | 1.26 (0.77–2.05) | 1G EGFR-TKIs + CT | ||
1.97 (1.14–3.40)* | 2.28 (0.98–5.31) | 1.69 (1.15–2.50)* | 1.35 (0.72–2.52) | 3G EGFR-TKIs + CT | |
1.28 (0.90–1.82) | 1.49 (0.71–3.10) | 1.10 (0.66–1.86) | 0.88 (0.55–1.40) | 0.65 (0.34–1.25) | 1G EGFR-TKIs + VEGF mAb |
HR: hazard ratio; CI: confidence intervals; OS: overall survival; NSCLC: non-small cell lung cancer; 1G/2G/3G EGFR-TKIs: first/second/third-generation epidermal growth factor receptor tyrosine kinase inhibitors; VEGF mAbs: vascular endothelial growth factor monoclonal antibodies; CT: chemotherapy.
*Represents a statistically significant difference in HR values
In the consistency analysis, the results of the direct and indirect comparisons were not significantly different, and the 2 analyses showed consistent results (Figure S1).
Comparison of PFS
This NMA included 23 trials with 11 different treatment strategies in patients with EGFR-mutant NSCLC with BM6–8,15,17–22,24–31,35–37,40. In head-to-head comparisons, 3G TKIs significantly improved PFS and reduced the risk of disease progression by 54.3% with an HR of 0.457 (95% CI: 0.350–0.565) compared to 1G TKIs. In addition, both combination regimens, 1G TKIs + CT and 1G TKIs + VEGF mAb, significantly improved PFS in patients with BM compared to 1G TKIs, with HRs of 0.375 (95% CI: 0.267–0.484) and 0.55 (95% CI: 0.35–0.74), respectively. However, 3G TKIs + VEGFR mAb did not show statistically significant differences in PFS compared to 3G TKI monotherapy (Table (Table22).
The results of the indirect comparison showed that the top six NMA rankings for PFS were all combination regimens, in the following order: 3G TKIs + CT, 3G TKIs + VEGFR mAb, 3G TKIs + EGFR-Met BsAb, 1G TKIs + CT, 2G TKIs + VEGF mAb, and 3G TKI + VEGF mAb, with probabilities of 94.0%, 74.0%, 72.4%, 67.0%, 65.4%, and 58.3%, respectively (Fig. 3B). These data indicate that 3G TKIs in combination with CT improved PFS in patients with BM the most effectively. Specifically, compared to 1G, 2G, and 3G TKIs alone, 3G TKIs + CT significantly reduced the risk of disease progression by 77.68%, 70.67%, and 53.05%, respectively, with HRs of 4.46 (95% CI: 2.57–7.72), 3.39 (95% CI: 1.36–8.46), and 2.13 (95% CI: 1.28–3.54). In addition, 3G TKIs + CT was superior to 1G TKIs + VEGF mAb/VEGFR-TKI in terms of PFS improvement, with HRs of 0.38 (95% CI: 0.20–0.76) and 0.25 (95% CI: 0.10–0.60), respectively. However, 3G TKIs + CT did not significantly outperform the second-ranked 3G TKIs + VEGFR mAb in terms of PFS, with an HR of 0.72 (95% CI: 0.33–1.57). In contrast, 3G TKIs + VEGFR mAb showed significantly improved PFS compared to both 1G TKIs + VEGFR-TKI and 1G TKIs, with HRs of 0.34 (95% CI: 0.13–0.88) and 3.21 (95% CI: 1.72–6.00), respectively, but not compared to other treatments. Similarly, 3G TKIs + CT did not significantly outperform the third-ranked 3G TKIs + EGFR-Met BsAb in terms of PFS, with an HR of 0.68 (95% CI: 0.34–1.36). However, 3G TKIs + EGFR-Met BsAb also showed statistically significant differences in PFS improvement compared to both 1G TKIs + VEGFR-TKI and 1G TKIs, with HRs of 0.36 (95% CI: 0.15–0.86) and 3.04 (95% CI: 1.83–5.05), respectively, but was not significantly different from other treatments (Table (Table4).4). PFS was significantly improved in 1G TKIs + CT compared to 1G TKIs monotherapy and 1G TKIs + VEGFR-TKI but was not significantly different from other regimens. Finally, 2G TKIs + VEGF mAb was not significantly different from any of the other treatments in terms of PFS improvement.
Table 4
1G EGFR-TKIs | ||||||||||
1.32 (0.63- 2.73) | 2G EGFR-TKIs | |||||||||
2.11 (1.71–2.57)* | 1.59 (0.75–3.40) | 3G EGFR-TKIs | ||||||||
2.53 (1.77–3.61)* | 1.92 (0.85–4.43) | 1.21 (0.80–1.82) | 1G EGFR-TKIs + CT | |||||||
4.46 (2.57–7.72)* | 3.39 (1.36–8.46)* | 2.13 (1.28–3.54)* | 1.76 (0.92–3.39) | 3G EGFR-TKIs + CT | ||||||
3.04 (1.83–5.05)* | 2.31 (0.95–5.62) | 1.45 (0.91–2.31) | 1.20 (0.65–2.23) | 0.68 (0.34–1.36) | 3G EGFR-TKIs + EGFR-Met BsAb | |||||
1.72 (1.16–2.56)* | 1.31 (0.57–3.01) | 0.82 (0.53–1.28) | 0.68 (0.40–1.16) | 0.38 (0.20–0.76)* | 0.57 (0.30–1.08) | 1G EGFR-TKIs + VEGF mAb | ||||
2.53 (0.73–8.80) | 1.92 (0.70–5.27) | 1.21 (0.34–4.27) | 1.00 (0.27–3.66) | 0.57 (0.15–2.21) | 0.83 (0.22–3.20) | 1.47 (0.40–5.44) | 2G EGFR-TKIs + VEGF mAb | |||
2.52 (0.98- 6.47) | 1.91 (0.58–6.31) | 1.20 (0.48–3.02) | 0.99 (0.36–2.73) | 0.56 (0.20–1.62) | 0.83 (0.29–2.33) | 1.46 (0.52–4.07) | 0.99 (0.21–4.75) | 3G EGFR-TKIs + VEGF mAb | ||
3.21 (1.72–6.00)* | 2.44 (0.93–6.39) | 1.53 (0.85–3.77) | 1.27 (0.62–2.61) | 0.72 (0.33–1.57) | 1.06 (0.50–2.24) | 1.87 (0.89–3.91) | 1.27 (0.32–5.12) | 1.28 (0.43–3.82) | 3G EGFR-TKIs + VEGFR mAb | |
1.10 (0.54–2.22) | 0.84 (0.30–2.30) | 0.52 (0.25–1.09) | 0.43 (0.20–0.95)* | 0.25 (0.10–0.60)* | 0.36 (0.15–0.86)* | 0.64 (0.28–1.43) | 0.43 (0.10–1.86) | 0.43 (0.13–1.42) | 0.34 (0.13–0.88)* | 1G EGFR-TKIs + VEGFR TKI |
HR: hazard ratio; CI: confidence intervals; PFS: progression-free survival; NSCLC: non-small cell lung cancer; 1G/2G/3G EGFR-TKIs: first/second/third-generation epidermal growth factor receptor tyrosine kinase inhibitors; VEGF mAbs: vascular endothelial growth factor monoclonal antibodies; VEGFR mAbs: vascular endothelial growth factor receptor monoclonal antibodies; EGFR-MET BsAb: epidermal growth factor receptor-mesenchymal epithelial transition factor bispecific monoclonal antibodies; CT: chemotherapy.
*Represents a statistically significant difference in HR values
A node-split analysis that was performed to assess the concordance between the direct and indirect comparison results for PFS showed no inconsistencies (Figure S2A-2C). Taken together, these results suggest that combination therapies centered around 3G TKIs, complemented by approaches such as CT, EGFR-Met BsAb, and VEGFR mAb, are the optimal treatment strategies for improving PFS in patients with EGFR-mutant NSCLC and BM.
Comparison of iPFS
Nine trials provided data on iPFS with a total of five treatment strategies, six of which were 3G TKIs vs. 1G TKIs2–4,26,33,34 and the remaining three trials were 3G TKIs + CT vs. 3G TKIs38, 1G TKIs + CT vs. 1G TKIs37, and 1G TKIs + VEGF mAb vs. 1G TKIs17. The head-to-head analysis showed that 3G TKIs significantly reduced the risk of intracranial disease progression by 56.7% compared to 1G TKIs, with an HR of 0.43 (95% CI: 0.343–0.524).
Analysis of the results by NMA ranking showed that 1G TKIs + VEGF mAb ranked first for iPFS improvement with a probability of 84.6%, followed by 3G TKIs + CT with a probability of 77.6%, 1G TKIs + CT with a probability of 55.6%, 3G TKIs with a probability of 32%, and 1G TKIs with a probability of 0.2%. These results showed that 1G TKIs + VEGF mAb was the optimal treatment strategy for improving iPFS, significantly reducing the risk of intracranial progression by 82.01% with an HR of 5.56 (95% CI: 1.52–20. 34) compared to 1G TKIs alone. The risk of intracranial progression was significantly lower with other regimens (including 1G/3G TKIs + CT and 3G TKIs), and 1G TKIs + VEGF mAb showed a trend toward reduced intracranial progression but did not reach statistical significance. In addition, 1G/3G TKIs + CT or 3G TKIs performed better than 1G TKIs alone in improving iPFS, but the difference between them did not reach significance (Fig. 3C, Table Table55).
Table 5
1G EGFR-TKIs | ||||
2.21 (1.81–2.69)* | 3G EGFR-TKIs | |||
2.78 (1.91–4.04)* | 1.26 (0.82–1.93) | 1G EGFR-TKIs + CT | ||
3.80 (2.10–6.89)* | 1.72 (0.99–3.02) | 1.37 (0.68–2.76) | 3G EGFR-TKIs + CT | |
5.56 (1.52–20.34)* | 2.52 (0.68–9.36) | 2.00 (0.52–7.72) | 1.46 (0.35–6.08) | 1G EGFR-TKIs + VEGF mAb |
HR: hazard ratio; CI: confidence intervals; iPFS: intracranial progression-free survival; NSCLC: non-small cell lung cancer; 1G/3G EGFR-TKIs: first/third-generation epidermal growth factor receptor tyrosine kinase inhibitors; VEGF mAbs: vascular endothelial growth factor monoclonal antibodies; CT: chemotherapy.
*Represents a statistically significant difference in HR values
Consistency testing for iPFS showed no inconsistent results between direct and indirect comparisons (Figure S3). In conclusion, 1G TKIs in combination with VEGF mAb may represent a long-term effective strategy to inhibit intracranial disease progression, whereas combination regimens based on 1G or 3G TKIs also showed potential advantages in controlling intracranial lesions.
Comparison of iORR, iDCR, and iDOR
Eight studies provided data on at least one aspect of iORR, iDCR, or iDOR2–4,25,26,33,34,37,38. Six of these trials compared 3G TKIs with 1G TKIs2–4,25,26,33,34 and the remaining 2 trials compared 1G TKIs + CT with 1G TKIs37 and 3G TKIs + CT with 3G TKIs38 (Table (Table11).
In head-to-head analysis featuring iORR, 3G TKIs had an improved intracranial objective remission rate compared to 1G TKIs: 82% vs 54.7%, with an OR of 1.81 (95% CI: 0.011–2.621). For iDCR, 3G TKIs showed a similar rate of intracranial disease control compared to 1G TKIs (94.4% vs. 95.2%), with an OR of 0.630 (− 1.175–2.435), which was not statistically different. For iDOR, only three studies, FURLONG, AENEAS, and EVEREST, provided sufficient iDOR data, and the direct pooled results showed that 3G TKIs significantly prolonged the duration of intracranial treatment response, with an HR of 0.34 (95% CI: 0.21–0.49) (Table (Table33).
Indirect comparisons showed that NMA of iORR was associated with 1G TKIs + CT (probability 88.6%), 3G TKIs + CT (65.3%), 3G TKIs (45.8%), and 1G TKIs (0. 3%); and both 1G TKIs + CT and 3G TKIs + CT were superior to 1G TKIs, with ORs of 0.30 (95% CI: 0.14–0.64) and 0.41 (95% CI: 0.21–0.81), respectively. These results indicated that 1G TKIs + CT was optimal for improving iORR, but no significant improvement was seen compared to 3G TKIs + CT or 1G/3G TKIs as monotherapy. For iDCR, the NMA ranking showed that 1G TKIs + CT was the optimal strategy with a probability of 64.4%, followed by 3G TKIs (55% probability), 1G TKIs (51.9% probability), and 3G TKIs (28.7% probability), but none of these regimens differed significantly in improving iDCR (Fig. 3D and andE,E, Table Table6).6). In the consistency test, there were no inconsistent results between the direct and indirect comparisons for iORR and iDCR (Figure S4, S5).
Table 6
1G EGFR-TKIs | 0.80 (0.21–3.11) | 1.02 (0.26–3.95) | 1.46 (0.27–7.79) |
0.30 (0.14–0.64)* | 1G EGFR-TKIs + CT | 1.28 (0.19–8.68) | 1.82 (0.21–15.76) |
0.49 (0.35–0.70) | 1.63 (0.71–3.78) | 3G EGFR-TKIs | 1.43 (0.53–3.84) |
0.41 (0.21–0.81)* | 1.37 (0.50–3.80) | 0.84 (0.47–1.50) | 3G EGFR-TKIs + CT |
OR: odds ratio; CI: confidence intervals; iORR: intracranial objective response rate; iDCR: intracranial disease control rate; NSCLC: non-small cell lung cancer; 1G/3G EGFR-TKIs: first/third-generation epidermal growth factor receptor tyrosine kinase inhibitors; CT: chemotherapy.
*Represents a statistically significant difference in OR values
Safety
Only four studies provided toxicity data for patients with BM: three compared “3G TKIs vs 1G TKIs,” and one was the GAP study comparing “1G TKIs + CT vs 1G TKIs.” These studies reported varying treatment-related toxicities. A summary of safety data is provided in Table S2, which includes the rates of any ≥ grade 3 adverse events (AEs), the incidence of drug-related serious AEs, the mortality rate due to any treatment-related AE, and the rate of treatment interruptions due to any AE. In general, 1G or 3G TKIs as monotherapy were well-tolerated, with a lower incidence of any ≥ grade 3 AEs and a lower rate of treatment interruption due to any AE compared to chemotherapy combinations.
Assessment of study quality and publication bias
The quality assessment of the 24 included studies showed that the majority were of satisfactory methodological rigor. Notably, 15 studies were identified as having a high risk of performance and selection bias due to a lack of blinding. In addition, 15 studies only reported PFS, which introduces an unclear risk of attrition bias due to incomplete outcome data. A detailed visualization of the quality assessment is shown in Fig. 4.
Egger’s test was used to determine the presence of publication bias. The analysis detected evidence of publication bias for PFS, as shown by one study comparing 3G to 1G EGFR-TKIs that deviated from funnel plot symmetry. In response, a sensitivity analysis was performed focusing specifically on the HR for 3G versus 1G EGFR-TKI comparisons (Figure S6). Our results highlight the substantial impact of the EVEREST trial results on the pooled PFS HR. The EVEREST trial was unique as it was the only randomized controlled trial to initially recruit an EGFR-mutant BM population. This observation highlights the specific role of EVEREST and its impact on the interpretation of the PFS results. There was no evidence of publication bias for other outcome measures across all studies (Fig. 5).
Discussion
The quest to optimize first-line therapeutic strategies for patients with EGFR-mutant NSCLC complicated by BM remains a high priority in oncology research, due to the risk of rapid intracranial disease progression with subsequent treatment lines. Our comprehensive NMA, which pooled data from 24 RCTs involving 2,682 patients, provides important insights into the relative efficacy of different therapeutic regimens. Specifically, the analysis showed that 3G EGFR-TKI monotherapy did not significantly improve OS compared to 1G TKIs. The combination of CT with 3G TKIs has emerged as a leading strategy, significantly improving both PFS and OS compared to 3G TKIs alone, achieving the highest rank for both survival outcomes in this challenging clinical scenario. The addition of VEGF mAb bevacizumab or CT to 1G TKIs significantly improved OS and PFS compared to 1G TKIs alone, with OS rankings of second and third, respectively. Although the combination of VEGF mAb with 1G TKIs is ranked less prominent in terms of PFS, it has emerged as the leading regimen for delaying intracranial disease progression. This meta-analysis underscores a pivotal paradigm shift in clinical oncology, highlighting the critical importance of multi-targeted strategies that target not only the primary oncogene, but also the tumor microenvironment, especially in managing the complexities of BM in patients with EGFR-mutant NSCLC.
Our meta-analysis differs from the recent study by Landre et al.41 in that we provide a comprehensive assessment of treatment strategies for EGFR-mutant NSCLC with BM. Our findings, based on a larger dataset and NMA statistical approach, reinforce the importance of multi-targeted approaches and highlight the combination of CT with 3G TKIs as a leading strategy. The study by Landre et al. focused on CT plus TKIs versus chemotherapy alone, but did not provide OS data stratified by the presence of BM. In addition, we provide new insights into the role of VEGF-targeted therapies in controlling intracranial disease progression and the potential of 3G TKIs plus EGFR-MET BsAbs, including the MARIPOSA trial, in disease management. These findings underscore the evolving landscape of personalized medicine and the need for further research to optimize treatment options for this complex patient population.
Our study demonstrated a significant difference in iPFS and iORR between 3 and 1G EGFR-TKIs. This is consistent with previous evidence that 3G TKIs are superior to earlier generations of TKI in delaying intracranial progression, and supports the evolving understanding that 3G TKIs with high BBB permeability and broader mutation coverage are critical for this patient population4,42,43. However, the NMA showed that in monotherapy using 3G TKIs, the OS in patients with BM was not significantly improved compared with 1G TKIs. It is possible that half of the patients who fail to respond to 1G TKIs have a T790M mutation and will therefore benefit from subsequent 3G TKI therapy44. In addition, our previous studies suggest that 3G TKIs, including osimertinib and aumolertinib, may benefit T790M-negative or unknown populations, particularly those with BM45,46. Thus, the survival of patients with BM who have progressed on 1G TKIs may be compensated by subsequent treatment with a high proportion of 3G TKIs. Additionally, previous brain radiation may affect the response to subsequent treatments and survival. We found that the percentage of patients who had received previous brain radiation varied across the included studies, ranging from 0% to 91.2%, with an average of 7.4%. The subset analyses of the FLAURA, AENEANS, FURLONG, and LASER301 studies reported similar proportions of patients with prior brain radiotherapy in the 3G (0–28.3%) and 1G (1.7–27%) TKI groups. Future research needs to standardize the history of prior brain radiotherapy to clarify its specific impact on drug efficacy.
This finding reinforces the expectation that combination therapies will realize their full therapeutic potential in patients with BM. A very important finding in the results of this NMA is that the addition of CT to 3G TKIs significantly improves both PFS and OS compared with 1G, 2G, or 3G TKIs monotherapy. Compared to 3G TKI monotherapy, an HR of 2.13 for PFS and 1.69 for OS corresponds to a 53.05% reduction in the risk of disease progression and a 40.83% reduction in the risk of death, respectively. These findings underscore the potential of the combined approach in addressing the aggressive nature of EGFR-mutant NSCLC with BM. The efficacy of the regimen probably stems from the ability of CT to target rapidly dividing cells throughout the body, complemented by the 3G TKIs potent inhibition of EGFR signaling pathways and their enhanced penetration of the BBB, thus offering a dual-pronged attack on systemic and intracranial disease47,48. The high-ranking probability of 93.8% for CT plus 3G TKIs to optimize OS provides a strong rationale for their clinical use. However, the lack of statistically significant differences in the improvement of OS compared to other EGFR-TKI-based combination regimens highlights the complexity of treatment decision-making and the need for tailored approaches based on comprehensive patient profiling. This finding also calls for the discovery of biomarkers that can predict which subset of patients may benefit more from combination CT than other targeted therapies.
Notably, while the combination of 3G TKIs with CT demonstrated superiority in prolonging systemic disease control, it was not consistent across all endpoints. The NMA also considered other strategies, including mAbs targeting the VEGF/VEGFR pathway and BsAbs against EGFR and MET. However, these regimens did not demonstrate superiority to the combination of CT plus 3G TKIs, but they showed promise in certain contexts.
Studies such as NEJ026 and CTONG1509 have shown that bevacizumab in combination with 1G TKIs has potential therapeutic benefits for patients with EGFR-mutated NSCLC, particularly in the treatment of multiple BMs7,15,16,49. A key finding in this NMA was that the combination therapy of 1G TKIs plus VEGF mAb bevacizumab significantly improved OS, with a 29.6% reduction in the risk of death compared to 1G TKIs alone. This was despite being ranked third and seventh in terms of OS and PFS, respectively, and was similar to 3G TKI monotherapy. In terms of iPFS, the combination of 1G TKIs with VEGF mAb as the leading regimen reduced the risk of intracranial disease progression significantly by 82.01% compared to 1G TKIs alone. The comparative consistency of direct and indirect results in iPFS underlines the robustness of these findings. A retrospective real-world study50 involving 208 NSCLC patients with ≥ 3 BMs found that the combination of EGFR-TKIs with bevacizumab resulted in a significantly higher iORR of 66.1% compared to 41.6% in 149 patients on 1G TKIs alone (P = 0.001). Additionally, combination therapy prolonged median iPFS to 14.0 months, compared to 8.2 months with monotherapy (P < 0.001), demonstrating that the addition of bevacizumab resulted in a significant benefit in both iORR and iPFS. This combination regimen may inhibit tumor growth and metastasis by simultaneously targeting EGFR and VEGF/VEGFR pathways51,52, leading to better therapeutic outcomes. Additionally, the combination therapy has a distinct efficacy profile within the intracranial compartment, possibly due to enhanced anti-angiogenic properties and microenvironment modulation of VEGF inhibitors, which may significantly improve the control of brain lesions. However, it is important to note that despite the demonstrated efficacy in delaying intracranial disease progression, this regimen does not rank highest in terms of OS or PFS, highlighting the complexity of balancing systemic and intracranial disease management.
The use of 3G EGFR-TKIs with high BBB penetration in combination with bevacizumab/ramucirumab is controversial. Data from several phase II RCTs have shown that 3G TKI osimertinib in combination with VEGF/VEGFR mAb have not shown a consistent survival benefit compared to 3G TKI alone19,20,53,54. Validation studies from phase III RCTs are still lacking. The present NMA was unable to pool data on OS and intracranial efficacy of the 3G TKI plus VEGF/VEGFR mAb regimen in the BM subset of the population. However, this combination regimen seemed to benefit PFS compared to 3G TKI monotherapy and 1G TKI plus VEGF mAb, although it was not statistically significant.
Conversely, regimens with 1G TKIs in combination with CT ranked highest for iORR and iDCR. These findings illustrate the nuanced therapeutic landscape in which different combinations may excel in different aspects of disease management for patients with EGFR-mutant NSCLC and BM. Importantly, these results should be interpreted with the recognition that the comparative assessment of these endpoints was not possible for all therapeutic combinations due to the limitations of the available data, and caution should be exercised when inferring absolute superiority in the absence of head-to-head comparative trials.
The MARIPOSA trial evaluated the combination of amivantamab with lazertinib8. Although the OS results from the BM subgroup analysis had not been reported at the time of data synthesis in our study, the trial’s preliminary findings suggest a potential benefit of this combination, particularly in delaying disease progression. The addition of amivantamab to lazertinib may offer a synergistic effect through dual targeting of EGFR and MET, which is a relevant consideration given the role of MET alterations in resistance mechanisms to EGFR-TKIs. However, CT combined with 3G TKIs showed a more pronounced effect in delaying disease progression compared to 3G TKIs combined with EGFR-MET BsAbs. This indirect finding suggests that a two-pronged approach of CT and targeted therapy may offer advantages in systemic disease control. For intracranial outcomes, the data suggest that the combination of 3G TKIs with CT was more effective in delaying intracranial disease progression than the combination of 3G TKIs with EGFR-MET BsAbs. This highlights the importance of considering both systemic and intracranial efficacy when comparing treatment regimens in NSCLC with BM. For iORR and iDCR, the indirect comparison suggests a nuanced picture. The combination of 3G TKIs with CT appears to offer a competitive advantage in achieving higher rates of intracranial response and disease control, although definitive conclusions are limited by the indirect nature of the comparison. Further data from the MARIPOSA trial, including the OS results, will provide more insights into the comparative efficacy of this combination in the context of BM.
The inclusion of safety, adverse events, and quality of life (QoL) data provides important context for clinical decision-making. However, our study did not pool and compare the toxicities of various treatment strategies in the population of patients with BM, as the treatment-related toxicity data for this group are incomplete. Additionally, the studies did not report on the QoL of patients with BM, such as neurocognitive function, emotional state, social functioning, and related physical conditions. As a result, the current studies cannot definitively determine the specific impact of different treatment strategies on the QoL of patients with BM, which requires further investigation in future research. Despite this, the generally favorable adverse event profile of 3G EGFR-TKIs as monotherapy supports their use in patients who may not tolerate more aggressive combination therapies. However, the balance between efficacy and tolerability must be carefully considered, and individual patient factors should guide treatment selection. The limited availability of safety and QoL data highlights the need for further research to better understand the impact of different treatment regimens on patient-reported outcomes.
Our quality assessment of the 24 studies found that most were methodologically sound, although 15 had a high risk of bias due to inadequate blinding, which could influence the results. The limited reporting of PFS in 15 trials suggests attrition bias. Egger’s test revealed publication bias in PFS for the 3G vs. 1G TKI comparisons, with the EVEREST trial’s strong focus on EGFR-mutant BM populations particularly influencing the pooled PFS HR. Sensitivity analyses reinforced the pivotal impact of EVEREST. Other results showed no publication bias, confirming broader reliability. These findings underscore the need for rigorous methodology, comprehensive reporting, and complementary RCTs to advance our therapeutic knowledge and strategies.
The NMA has several limitations. First, the lack of head-to-head trials for some comparisons limited our conclusions, and indirect comparisons introduce uncertainty. Second, trial and attrition bias may have influenced the results, and the dominance of certain trials may indicate publication bias. Third, patient diversity and endpoint definitions varied between trials, affecting the generalizability of the findings. Fourth, confounding factors such as post-protocol therapy and censorship bias need to be considered. Post-protocol therapy can influence outcomes and may overestimate the benefits of certain treatments. Censorship bias, which occurs when patients are removed from the analysis for reasons unrelated to the primary outcome (e.g., switching to another therapy), can also affect the estimation of treatment effects. Fifth, while adenocarcinoma is the most common histology associated with EGFR mutations and BM, squamous cell carcinoma (SCC) can also occur, albeit less frequently. The effects of EGFR-TKIs in SCC are less well understood due to its rarity. Histologic transformation to SCC has been reported as a mechanism of resistance to EGFR-TKIs. Given the limited data on SCC in the included trials, we have focused primarily on adenocarcinoma. More research is needed to better understand the efficacy of EGFR-TKIs in SCC and the implications of histologic transformations. Sixth, the evaluation of previous brain radiation is a critical factor in interpreting the results of our meta-analysis. However, the variability in the percentage of patients who had received previous brain radiation across the included studies could introduce biases and limit the comparability of the results. Therefore, the results should be interpreted with caution, taking into account the differences in baseline characteristics between the studies. Future studies should aim to standardize eligibility criteria regarding previous brain radiation to reduce heterogeneity and improve the generalizability of the findings.
Conclusion
Our comprehensive meta-analysis of 24 RCTs demonstrated differences in the efficacy of various treatments for EGFR-mutant NSCLC with BM. The use of 3G TKIs as monotherapy failed to significantly improve OS compared to 1G TKIs, necessitating a combination approach. CT plus 3G TKIs was found to be optimal and significantly improved both PFS and OS. The combination of VEGF mAb with 1G TKIs also showed survival benefits, particularly for intracranial control. Indirect evidence suggests that 3G TKIs plus EGFR-MET BsAbs show promise in disease management; however, cautious interpretation is warranted for future updates. The superiority of combination therapies targeting multiple pathways signifies a critical shift in oncology, emphasizing the need for personalized medicine and biomarker development. Intracranial outcomes highlight the importance of tailored approaches, with CT combination regimens excelling in systemic control and VEGF-targeted therapies showing promise in selected contexts. The balance between intracranial and systemic disease management, together with quality-of-life considerations, is paramount in refining therapeutic strategies for this complex patient population. Ongoing research and head-to-head trials are essential to further the understanding of optimal combinations and improve personalized care.
Supplementary Information
Acknowledgements
We thank International Science Editing (http://www.internationalscienceediting.com) for the English language proofreading.
Author contributions
Concept and design (C.H., J.M.), literature screening (X.P., S.Z.), data extraction (X.P., S.Z., L.S.), data analysis (X.P., S.Z., L.H.), drafting of the manuscript (C.H., J.M.), critical revision of the manuscript for important intellectual content (C.H.). All authors have read and approved the final manuscript.
Funding
This study was supported by the Health Promotion for Wellness and Development Project—Special Programme for Oncology Research by the Oasis for Life Public Welfare Service Centre (Grant No. BJHA-CRP-062) and the 345 Talent Project of Shengjing Hospital.
Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
The online version contains supplementary material available at 10.1038/s41598-024-74496-0.