INTRODUCTION

Lung cancer remains the leading cause of cancer death.¹ Among lung cancer, non-small cell lung cancer (NSCLC) accounts for about 85% of cases and can be divided into two major subtypes: lung adenocarcinoma (LUAD) and nonadenocarcinoma. Recently, the molecular landscape of NSCLC has been profoundly interrogated, benefiting from the advances of high-throughput sequencing technologies.² The discovery of mutations and rearrangements including epidermal growth factor receptor (EGFR) mutations, anaplastic lymphoma kinase (ALK), and ROS proto-oncogene 1 (ROS1) rearrangements has led to the development of specific targeted agents and dramatically altered the therapeutic landscape, particularly regarding adenocarcinomas.³ Until now, other potential targets such as BRAF mutations have been probed and relevant efficacy data are emerging.

V-Raf murine sarcoma viral oncogene homolog B1 (BRAF), a serine/threonine protein kinase, together with V-Raf murine sarcoma 3611 viral oncogene homolog 1 (ARAF) and V-Raf-1 murine leukemia viral oncogene homolog 1 (CRAF), belongs to the RAF family. Physiologically, BRAF proteins are activated via rat sarcoma viral oncogene homolog (RAS) phosphorylation and then in turn phosphorylate mitogen-activated protein kinase kinase 1/2 (MEK1/2) and subsequently mitogen-activated protein kinase 1/2 (ERK1/2).⁴ Through the MEK–ERK pathway, BRAF plays a key role in regulating cell proliferation, differentiation and survival.⁵ Consequently, the mutation of BRAF may frustrate the negative feedback mechanism of the pathway and constitutively activate this signaling. BRAF mutations have been identified in about 3%–8% of all cancers, with a less frequent incidence in lung cancer.⁶ The majority of BRAF mutations occur in exon 15, corresponding to the substitution from valine to glutamate at codon 600 (V600E).

Pivotal phase II trials have investigated the efficacy of dual BRAF/MEK inhibition (dabrafenib in combination with trametinib) in patients harboring BRAF V600E mutations. The impressive results prompted the Food and Drug Administration (FDA) and European Medicines Agency (EMA) rapid approval of the regimen in clinical setting.^{7, 8} The encouraging outcomes urged us to further explore the BRAF mutations. Nevertheless, limited by the small number of patients, the impact of BRAF mutations on prognosis in LUAD remains unclear.^{3, 9, 10} In this study, therefore, we evaluated the data of 1604 stage I LUAD patients from the Shanghai Pulmonary Hospital and 431 patients from cBioPortal for Cancer Genomics, assessed the impact of BRAF mutation on prognosis and identified the clinical features of patients harboring BRAF mutations.

METHODS

Patient selection

This study was approved by the Ethics Committee of Shanghai Pulmonary Hospital (approval no. K23-208). Public data were accessed via cBioPortal for Cancer Genomics (https://www.cbioportal.org/datasets). We obtained the data from the cohort from the database of Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT).¹¹ This cohort included 604 LUAD patients who received an operation without neoadjuvant therapy. In addition, those patients were confirmed as having pathological stage I–III disease, and their predominant histological subtypes were identified and recorded. The flow chart of case selection is presented in Figure 1a. Finally, 431 patients with stage I LUAD from the database of MSK-IMPACT were included in cohort 1.

We also collected data from the Shanghai Pulmonary Hospital from 2015 to 2016. Patients were included in this study if they met all of the following criteria: (1) a clear pathological diagnosis of LUAD, (2) had undergone radical resection, (3) had no metastasis to lymph nodes or other organs, and (4) received genetic analysis (amplification refractory mutation system [ARMS] polymerase chain reaction [PCR]). Patients who met any of the following conditions were excluded from this study: (1) age < 18 years, (2) perioperative death (died within 1 month after operation), (3) with other malignant tumors, and (4) had missing clinical data. Preoperative staging was performed and strictly followed the guidelines. The pathological tumor-node-metastasis (TNM) stage was identified according to the eighth edition classification system. Detailed information concerning patient selection is presented in Figure 1b. A total of 1604 eligible patients were included (cohort 2).

RESULTS

Survival analyses

We first analyzed the survival outcomes by Kaplan–Meier methods and Cox regression to evaluate the prognostic significance between BRAF mutant and wild-type patients in cohort 1. As shown in Figure 2a, the survival curve revealed that the difference in RFS between patients with pathological stage I LUAD harboring BRAF mutations and BRAF wild-type was not significant (median survival time not reached; HR, 1.111; p = 0.885). The joint effect of gender, age, smoking history, surgery type, predominant pattern, tumor size, VPI, LVI, STAS, and adjuvant therapy (ACT) was examined utilizing stepwise Cox regression analysis. The results of the univariable analysis are shown in Table 2. In the univariable analysis, BRAF mutation status was excluded in the equation as the independent factor to predict RFS (V600, HR, 3.403; p = 0.230; non-V600, HR, 0.665; p = 0.688). To reduce potential bias and further confirm the results, PSM and OW were conducted. After adjustment, the results of PSM and OW revealed that patients with BRAF mutations had a similar trend of survival outcome with BRAF wild-type patients (Figure 2b,c, adjusted HR, 1.352; p = 0.742 and HR, 1.246; p = 0.764, respectively).

TABLE 2. Univariable survival analyses of patients in cohort 1 for RFS.

Variables	Univariable analysis
	HR	95% Cl	p-value
Gender
Male	1
Female	0.985	0.501–1.939	0.965
Age at surgery, years
≤65	1
>65	1.083	0.570–2.056	0.808
Smoking history
No	1
Yes	1.058	0.485–2.308	0.887
Predominant pattern
Lepidic	1
Acinar/papillary	3.725	0.879–15.774	0.074
Micropapillary/solid	11.888	2.696–52.434	<0.001
Tumor size, cm
<1	1
1–2	0.805	0.183–3.545	0.774
2–3	1.781	0.400–7.918	0.448
3–4	1.050	0.191–5.768	0.955
Class
Wild-type	1
V600	3.403	0.461–25.123	0.230
Non-V600	0.665	0.091–4.863	0.688

• Abbreviations: CI, confidence interval; HR, hazard ratio; RFS, recurrence-free survival.

To further explore the prognostic role of BRAF V600E mutation, we then analyzed the data of cohort 2. As shown in Figure 3a, no significant difference was found in pathological stage I LUAD patients harboring BRAF V600E mutation and without carrying BRAF V600 mutation (HR, 1.844, p = 0.226). After PSM and OW, similar results are presented in Figure 3b,c (adjusted HR, 1.144; p = 0.83 for PSM and adjusted HR, 1.466; p = 0.450 for OW, respectively). Cox regression was also performed to evaluate the prognostic role of BRAF V600E mutation. The results are shown in Table S2. In the univariable analysis, BRAF V600E mutation status was not significantly associated with RFS (HR, 1.844; p = 0.226). Only gender (HR, 0.747; p = 0.045), predominant pattern (all p < 0.01), tumor size (all p < 0.05), VPI (HR, 3.846; p < 0.001), LVI (HR, 5.958; p < 0.001) and ACT (HR, 1.563; p = 0.003) were significantly associated with RFS in the Cox regression analysis.

Concurrent oncogenic driver mutations

BRAF mutations included four groups, class 1 (V600 E/K/D/R), class 2 (K601, L597, G464, and G469), class 3 (G466, N581, D594, and D596), and class 4 (others). Next, we assessed the concurrent oncogenic driver mutations with BRAF mutations in cohort 1. Collectively, 41 patients were found to harbor BRAF mutations in the initial 604 patients, in which class 1 accounted for 22.0%, class 2 for 17.1%, class 3 for 26.8%, and others for 34.1% (Table S3). Most class 1 BRAF mutations were V600E mutations, with only one patient harboring the V600K mutation. The most frequent concurrent mutation was TP53 (15, 36.6%), followed by STK11 (10, 24.4%), SETD2 (8, 19.5%), MTOR (7, 17.1%), EPHA5 (6, 14.6%), FAT1 (6, 14.6%), and NOTCH2 (6, 14.6%) (Figure 4). Of note, 230 (38.1%) patients were found to harbor KRAS mutations while in the BRAF-mutant subgroup, only five KRAS mutations were observed and they all belonged to the non-V600 class, which may indicate that class 1 mutations were mutually exclusive from KRAS mutations. STK11 and EPHA5 mutations were also less likely to occur in BRAF class 1 mutations. Four patients carrying non-V600 BRAF mutations were found to concurrently harbor EGFR mutations. We also attempted to evaluate the prognostic role of concurrent BRAF with TP53 or STK11 mutations. However, no significant difference was observed (Figure S1).

DISCUSSION

This study aimed to evaluate the prognostic significance of BRAF mutations in patients with pathological stage I lung adenocarcinoma. The more we understand the clinical features and prognostic role of BRAF mutations, the more patients can be selected to undergo mutational screening and receive the appropriate therapeutic schedule. In cohort 1, the prevalence of BRAF V600E mutations was around 1.4%, while in cohort 2, the BRAF V600E mutation rate was estimated at 1.0%. The inconformity could be caused by ethnic differences, as the rate of BRAF mutations has been reported to be higher in Caucasian than in Asian populations.^{17, 18} The estimated BRAF mutation rate in lung adenocarcinoma was 5.6%, which is in line with previous studies.^{3, 6, 18, 19}

Former studies revealed that NSCLC patients harboring BRAF V600E mutations occurred predominantly in females with a never-smoking history, while non-V600 mutations were mainly found in male smokers.^{17, 20, 21} In cohort 2, most patients (93.8%) harboring BRAF V600E mutation never had a smoking history. In contrast, most patients (83.3%) harboring BRAF mutations in cohort 1 had a smoking history. The uneven distribution of BRAF mutation types between two cohorts may partly explain the reason. The predominant histological pattern of BRAF-mutant tumors in cohort 1 was acinar/papillary (54.2%), followed by micropapillary/solid (20.8%). When focusing on the subgroup of class 1 mutations, the proportion of micropapillary/solid accounts for 33.3% (2 in 6 patients), which was consistent with previous reports that BRAF V600E-mutant tumors were more likely to show a histological type characterized by micropapillary features.²⁰ In our cohort 2, a high proportion of VPI (18.8%) in the BRAF V600E mutation group was also observed. Limited by the small sample size, statistical significance was not reached. Dexter et al. also reported that the odds of group I mutations were higher (odds ratio:4.39, 95% CI:1.11–17.4) among tumors involving the pleural space.²² Nevertheless, the mechanism behind is of interest and needs to be further investigated.

It has been previously reported that BRAF V600E mutations are generally mutually exclusive from ALK rearrangements and EGFR mutations.^{2, 19} In our study, we found two patients with concomitant EGFR and BRAF V600E mutation, one with EGFR 19-Del and one with EGFR-L858R mutation in cohort 1. In cohort 2, two patients harboring concomitant EGFR and BRAF mutations were also observed, one with EGFR V292L mutation, and one with EGFR X901_splice mutation. However, these two patients belonged to the non-V600E BRAF mutation class. Li et al. reported five patients in the series with concurrent BRAF V600E plus EGFR mutations.²³ Liza et al. also reported a 16% rate of double mutation among patients carrying BRAF mutations.²⁴ Indeed, recent studies demonstrated that due to the intrinsic heterogeneity of intratumor, EGFR-mutated NSCLC could seldom carry additional BRAF mutations, which is considered a predictor of resistance to EGFR inhibitors and tumor rapid progress.^{25, 26} In addition, Kinno et al. pointed out that tumors with V600E BRAF mutations were mutually exclusive from KRAS mutations.¹⁷ Concurrent BRAF V600E plus KRAS mutations were also not found in this study, indicating such mutation combinations may be rarer than others.

In some other types of tumors, such as melanoma, papillary thyroid carcinoma and colorectal cancer, the association between BRAF mutation and poor survival has been documented.^{27, 28} However, the prognostic impact of BRAF mutation in NSCLC patients is sustained by scarce evidence and existing studies are often contradictory. Xi et al. analyzed 1680 patients with NSCLC and identified 28 patients harboring BRAF mutation. Patients with V600E-mutated tumors had a similar PFS to first-line chemotherapy compared to patients with non-V600E mutations (5.2 vs. 6.4 months; p = 0.561).¹⁸ Cardarella et al. screened 883 tumors and found advanced NSCLC patients with BRAF mutations and wild tumors had no difference in OS and showed similar PFS to platinum-based chemotherapy. Nevertheless, Marchetti et al. demonstrated that patients with V600E-mutated NSCLC had significantly shorter disease-free and overall survival rates (HR, 2.19; p = 0.11 and HR, 2.18; p = 0.014, respectively) in a retrospective study which included 1046 NSCLC patients.²⁰

In our study, we first analyzed 431 patients with radical resection LUAD and found 24 patients harboring BRAF mutations. Before and after PSM and OW, BRAF mutation patients showed a similar RFS trend compared with BRAF wild-type patients. Cox regression analysis also excluded BRAF mutations as an independent factor to predict RFS. Next, we further investigated the prognostic role of BRAF V600E mutation, and RFS was not significantly different between BRAF V600E mutation and without BRAF V600E mutation patients. Taken together, we conclude that the BRAF mutation status lacks prognostic significance in stage I lung adenocarcinoma patients.

There were some limitations in our study. First, despite two propensity-score methods being utilized in our study, our analysis was still limited by its retrospective nature. Second, because of the low incidence of BRAF mutations, the sample size of BRAF mutations was relatively small. Further large multicentric studies are required to research and verify our results. Third, the follow-up time was relatively short, and we will continue to observe these patients. Fourth, the genotype of BRAF in cohort 2 was detected by ARMS-PCR. Although ARMS-PCR can detect BRAF V600E mutations in NSCLC patients with high sensitivity, non-V600E BRAF mutations can hardly be detected. Limited by the technology, we could only analyze the prognostic role of V600E BRAF mutations in cohort 2, losing sight of class II and class III BRAF mutations.

In conclusion, in this study, we demonstrated that BRAF status may not be capable of predicting prognosis in this population. Further studies are required to confirm our findings.

AUTHOR CONTRIBUTIONS

Conception and design: Dong Xie, Shang-Shang Ma, and Lei-Lei Wu. Administrative support: Dong Xie. Provision of study materials or patients: Dong Xie, Shang-Shang Ma, Kun Li, Rang Rang Wang, and Le-Lei Wu. Collection and assembly of data: Shang-Shang Ma, Jia-Yi Qian, Zhi-Ye Huang, Yu'e Liu, and Ming-Jun Li. Data analysis and interpretation: Dong Xie, Shang-Shang Ma, Ming-Jun Li, Zhi-Ye Huang, Yu'e Liu, Rang-Rang Wang, and Lei-Lei Wu. Manuscript writing: All authors. Final approval of manuscript: All authors.

CONFLICT OF INTEREST STATEMENT

The authors declare no competing interests.