Data enrollment in the study
In all the 20 CNS/Brain studies (6164 samples), 4674 samples with mutation data were queried; 90 samples (90 patients) with BRAF mutations, including 53 samples (53 patients) with BRAFV600E and 37 samples (37 patients) with BRAFnon−V600E, are shown in Table 1. The cancer types of 20 CNS/Brain studies included diffuse glioma, glioblastoma, oligodendroglioma, embryonal tumor, encapsulated glioma, and miscellaneous neuroepithelial tumor. The scheme for the final enrolled and investigated data is shown in Fig. 1. Ninety patients with BRAFV600E or BRAFnon−V600E were enrolled in this study, and data from 52 non-redundant patients were investigated. The integrated data of major patient characteristics, including sex, age, diagnosis age, cancer type, data of co-occurring mutations, copy number alterations, and overall survival time and status, were collected for further analysis.
Table 1
The CNS/Brain projects of TCGA database enrolled in the study retrieved by cBioPortal
Project | All Samples | Samples with mutation data | Samples with BRAFV600E | Samples with BRAFnon−V600E | References |
Diffuse Glioma | | | | | |
Brain Lower Grade Glioma (TCGA, Firehose Legacy) | 530 | 286 | 1 | 1 | https://www.cancer.gov |
Brain Lower Grade Glioma (TCGA, PanCancer Atlas) | 514 | 512 | 1 | 2 | (42–47) |
Glioma (MSK, Nature 2019) | 91 | 91 | 2 | 1 | https://www.cancer.gov |
Glioma (MSKCC, Clin Cancer Res 2019) | 1004 | 1004 | 22 | 19 | (48) |
Low-Grade Gliomas (UCSF, Science 2014) | 61 | 61 | 2 | 0 | (49) |
Merged Cohort of LGG and GBM (TCGA, Cell 2016) | 1102 | 812 | 5 | 2 | (50) |
GLIOBLASTOMA | | | | | |
Brain Tumor PDXs (Mayo Clinic, 2019) | 95 | 83 | 2 | 1 | https://www.cbioportal.org |
Glioblastoma (Columbia, Nat Med. 2019) | 42 | 32 | 1 | 1 | (51) |
Glioblastoma (TCGA, Cell 2013) | 543 | 257 | 3 | 0 | (52) |
Glioblastoma (TCGA, Nature 2008) | 206 | 91 | 0 | 0 | (53) |
Glioblastoma Multiforme (TCGA, Firehose Legacy) | 604 | 290 | 5 | 1 | https://www.cancer.gov |
Glioblastoma Multiforme (TCGA, PanCancer Atlas) | 592 | 397 | 5 | 3 | (42–47, 54) |
OLIGODENDROGLIOMA | | | | | |
Anaplastic Oligodendroglioma and Anaplastic Oligogastrocytoma (MSKCC, Neuro Oncol 2017) | 22 | 22 | 0 | 0 | (55) |
Embryonal Tumor | | | | | |
MEDULLOBLASTOMA | | | | | |
Medulloblastoma (Broad, Nature 2012) | 92 | 92 | 0 | 0 | (56) |
Medulloblastoma (ICGC, Nature 2012) | 125 | 125 | 0 | 0 | (57) |
Medulloblastoma (PCGP, Nature 2012) | 37 | 37 | 0 | 0 | (58) |
Medulloblastoma (Sickkids, Nature 2016) | 46 | 46 | 0 | 1 | (59) |
Encaspulated Glioma | | | | | |
PILOCYTIC ASTROCYTOMA | | | | | |
Pilocytic Astrocytoma (ICGC, Nature Genetics 2013) | 96 | 96 | 4 | 3 | (60) |
Miscellaneous Neuroepithelial Tumor | | | | | |
Pheochromocytoma and Paraganglioma (TCGA, Cell 2017) | 178 | 178 | 0 | 1 | (61) |
Pheochromocytoma and Paraganglioma (TCGA, Firehose Legacy) | 184 | 162 | 0 | 1 | https://www.cancer.gov |
Major characteristics of cohorts with BRAFV600E and BRAFnon−V600E
The study populations were divided into two groups, BRAFV600E and BRAFnon−V600E. The major demographic characteristics and clinical data of the two groups are summarized in Table 2. The patients’ ages ranged from 20 to 85 years and were divided into early adulthood, midlife, mature adulthood, and late adulthood (aged 20–35, 35–50, 50–80, and > 80 years, respectively). The two groups had comparable male patients, diagnosis age, cancer type, and overall survival status. Glioblastoma multiform was the most common cancer type in both cohorts (74.07% vs. 56.00%; P = 0.175; Table 2).
Table 2
The major characteristics of cohorts including BRAFV600E and BRAFnon−V600E
Variables | BRAFV600E (n = 27) | BRAFnon−V600E (n = 25 ) | Univariate analysis |
Number | % | Number | % | Odds Ratio | 95% Confidence Interval | P Value |
Male | 16 | 59.26 | 18 | 72.00 | 0.566 | 0.177–1.809 | 0.337 |
Diagnosis Age | | | | | | | |
Ages 20–35 | 9 | 33.33 | 6 | 24.00 | 1.583 | 0.469–5.350 | 0.459 |
Ages 36–50 | 9 | 33.33 | 8 | 32.00 | 1.062 | 0.333–3.390 | 0.918 |
Ages 51–80 | 7 | 25.93 | 11 | 44.00 | 0.445 | 0.139–1.433 | 0.175 |
Age 80+ | 2 | 7.41 | 0 | 0.00 | 1615474843 | 0.000- | 0.999 |
Cancer type detailed | | | | | | | |
Glioblastoma Multiform | 20 | 74.07 | 14 | 56.00 | 2.245 | 0.698–7.219 | 0.175 |
Astrocytoma | 3 | 11.11 | 6 | 24.00 | 0.396 | 0.087–1.794 | 0.229 |
Oligoastrocytoma | 1 | 3.70 | 0 | 0.00 | 1553341195 | 0.000- | 1.000 |
Oligodendroglioma | 0 | 0.00 | 3 | 12.00 | 0.000 | 0.000- | 0.999 |
Gliosarcoma | 0 | 0.00 | 2 | 8.00 | 0.000 | 0.000- | 0.999 |
Other glioma | 3 | 11.11 | 0 | 0.00 | 1682786295 | 0.000- | 0.999 |
Overall survival status | | | | | | | |
Deceased | 14 | 51.85 | 11 | 44.00 | 1.371 | 0.460–4.087 | 0.572 |
Co-occurring mutations of the BRAFV600E and BRAFnon−V600E cohorts using univariate and multivariate logistic regression analysis
Available co-occurring gene mutations of the BRAFV600E and BRAFnon−V600E cohorts were retrieved, and differences between the two groups were compared; the results are summarized in Table 3. The mutation frequencies of KRAS, HRAS, RAF1, MAP3K1, MAP2K1, MAP2K2, MAP2K4, MDM2, MDM4, CDKN2A, and CDKN2B were comparable between the two groups. In contrast, the BRAFnon−V600E group exhibited a significantly higher mutation frequency of TP53 (56.00% vs. 7.41%; P = 0.001), IDH1/2 (36.00% vs. 3.70%; P = 0.015), and ATRX (32.00% vs. 7.41%; P = 0.037) than the BRAFV600E group. The variables with P < 0.10 were analyzed using multivariate logistic regression analysis, and the BRAFnon−V600E group exhibited a significantly higher TP53 mutation frequency (56.00% vs. 7.41%; P = 0.031) than the BRAFV600E group (Table 3).
Table 3
The co-occurred mutations of BRAFV600E and BRAFnon−V600E cohort using univariate and multivariate logistics regression analysis
Gene | BRAFV600E (n = 27) | BRAFnon−V600E (n = 25 ) | Univariate analysis | Multivariate analysis |
Number | % | Number | % | Odds Ratio | 95% Confidence Interval | P Value | Odds Ratio | 95% Confidence Interval | P Value |
KRAS | 0 | 0.00 | 1 | 4.00 | 1817409198 | 0.000- | 1.000 | | | |
HRAS | 0 | 0.00 | 1 | 4.00 | 1817409198 | 0.000- | 1.000 | | | |
RAF1 | 0 | 0.00 | 2 | 8.00 | 1896426989 | 0.000- | 0.999 | | | |
MAP3K1 | 1 | 3.70 | 6 | 24.00 | 8.211 | 0.911–73.959 | 0.060 | | | |
MAP2K1 | 0 | 0.00 | 2 | 8.00 | 1896426989 | 0.000- | 0.999 | | | |
MAP2K2 | 0 | 0.00 | 4 | 16.00 | 2077039084 | 0.000- | 0.999 | | | |
MAP2K4 | 0 | 0.00 | 2 | 8.00 | 1896426989 | 0.000- | 0.999 | | | |
TP53 | 2 | 7.41 | 14 | 56.00 | 15.909 | 3.078–82.224 | 0.001 | 12.186 | 1.251-118.721 | 0.031 |
MDM2 | 1 | 3.70 | 3 | 12.00 | 3.545 | 0.344–36.561 | 0.298 | | | |
MDM4 | 0 | 0.00 | 4 | 16.00 | 2077039084 | 0.000- | 0.999 | | | |
CDKN2A | 0 | 0.00 | 3 | 12.00 | 1982628216 | 0.000- | 0.999 | | | |
CDKN2B | 0 | 0.00 | 1 | 4.00 | 1817409198 | 0.000- | 1.000 | | | |
IDH1/2 | 1 | 3.70 | 9 | 36.00 | 14.625 | 1.690-126.537 | 0.015 | 5.498 | 0.512–59.020 | 0.159 |
ATRX | 2 | 7.41 | 8 | 32.00 | 5.882 | 1.110–31.170 | 0.037 | 0.665 | 0.048–9.188 | 0.761 |
Co-occurring copy number alteration in the BRAFV600E and BRAFnon−V600E cohorts using heatmap and univariate logistic regression analyses
There were no available copy number data for five patients with BRAFV600E and five patients with BRAFnon−V600E. The copy number alterations of the available co-occurring genes included BRAF, RAF1, MAP3K1, MAP2K1, MAP2K2, MAP2K4, MAPK1, MAPK3, TP53, MDM2, MDM4, TP53BP1, IDH1, IDH2, ATRX, CDKN2A, and CDKN2B. The homozygous deletion (HD) copy number was frequently retrieved in these two genes, including CDKN2A and CDKN2B (Fig. 2), and the HD of both CDKN2A (77.27.00% vs. 60.00%; P = 0.032) and CDKN2B (77.27.00% vs. 60.00%; P = 0.032) was more frequent in the BRAFV600E cohort than in the BRAFnon−V600E cohort (Table 4).
Table 4
The HD of CDKN2A/2B of BRAFV600E and BRAFnon−V600E cohort using univariate logistics regression analysis
Variables | BRAFV600E (n = 22) | BRAFnon−V600E (n = 20) | Univariate analysis |
Number | % | Number | % | Odds Ratio | 95% Confidence Interval | P Value |
CDKN2A | 17 | 77.27 | 12 | 60.00 | 0.193 | 0.043–0.867 | 0.032 |
CDKN2B | 17 | 77.27 | 12 | 60.00 | 0.193 | 0.043–0.867 | 0.032 |
Cross over analysis using Kaplan–Meier survival curves and the log rank (Mantel-Cox) test
The cross over Kaplan–Meier survival curves and the log rank (Mantel-Cox) test were performed to explore the difference between the overall survival of glioma patients with BRAFV600E and BRAFnon−V600E. The estimated mean survival time was 51.394 months for patients with BRAFV600E, 89.958 months for patients with BRAFnon−V600E, 44.550 months for patients with BRAFV600E and IDH1/2WT, and 93.821 months for patients with BRAFnon−V600E and IDH1/2WT. There was no difference between the survival of BRAFV600E and BRAFnon−V600E (51.394 vs. 89.958, chi-square 1.130, P = 0.288). In addition, there was no difference between the survival of BRAFV600E and IDH1/2WT and BRAFnon−V600E and IDH1/2WT (44.550 vs. 93.821, chi-square 0.007, P = 0.935), which excluded the side effect of IDH1/2. We also evaluated the survival of BRAFnon−V600E and IDH1/2WT with mutations in the G-loop and activation segment. The estimated survival time of these two subgroups was 12.250 months for patients with BRAFnon−V600E and IDH1/2WT with mutations in the G-loop, and 34.800 months for patients with BRAFnon−V600E and IDH1/2WT with mutations in the activation segment. In addition, there was no difference between the BRAFV600E and IDH1/2WT cohorts and those of the BRAFnon−V600E and IDH1/2WT cohorts. As shown below, BRAFV600E and IDH1/2WT vs. BRAFnon−V600E and IDH1/2WT with mutations in the G-loop (44.550 vs. 12.250, chi-square0.122, P = 0.727), and BRAFV600E and IDH1/2WT vs. BRAFnon−V600E and IDH1/2WT with mutations in the activation segment (44.550 vs. 34.800, chi-square 0.145, P = 0.703). Since the estimated mean survival of BRAFnon−V600E and IDH1/2WT with mutations in the G-loop was the shortest, we compared the BRAFnon−V600E and IDH1/2WT with mutations in the G-loop with the remaining BRAFnon−V600E and IDH1/2WT patients. There was no difference between them (12.250 vs. 95.100, chi-square 0.008, P = 0.927) (Fig. 3).