Diffuse midline glioma (DMG), H3 K27M mutant (WHO grade IV) is listed as a separate CNS tumor entity since 2016 [5], after large sequencing efforts had discovered H3 K27M mutations frequently appearing in gliomas located in midline structures [11]. Over time, we and others have observed single cases of DMG with concomitant mutations within FGFR1 or BRAF [1, 2, 4, 6, 7, 9, 10, 12,13,14]. FGFR1 and BRAF mutations are typical hallmarks of low grade glioma, such as pilocytic astrocytoma, ganglioglioma, or dysembryoplastic neuroepithelial tumor [3, 8]. So, the parallel occurrence of H3 and FGFR1/BRAF mutations within a single tumor may complicate the diagnostic decision towards a low grade or a high grade glioma. This dilemma, which has direct clinical implications, is particularly evident, if only small biopsies are taken and low-grade histology may not be respresentative and hence may not mirror the biology of the neoplasm. On the other hand, the presence of a MAPK pathway alteration, such as FGFR1 or BRAF mutations, may open up additional possibilities of targeted therapies, independent of the tumor classification.

In order to learn more about the frequency and impact on such mutations, we analyzed a series of 83 DMG, H3F3A K27M mutant. Details on clinical characteristics of patients are listed in Fig. 1a and Supplementary Table 1, online resource. One case (1.2%) displayed a BRAF (p.V600E) mutation and 9/83 cases (10.8%) showed FGFR1 mutations (p.K656E or p.N546K). Mutations within NF1, TP53, and ATRX were detected in 31.8%, 51.4%, and 35.2%, respectively. TP53 mutations were significantly associated with FGFR1 wild type status (FGFR1 WT, p = 0.009, Χ2-test, Supplementary Fig. 1a, online resource).

Fig. 1
figure 1

Clinical, histological, and molecular parameters of H3F3A K27M mutated DMG with and without additional mutations in FGFR1. a Overview on all 83 analyzed cases with 12% of cases harboring BRAF or FGFR1 hotspot mutations. Percentages of characteristics refer to cases with known attribute only. Representative images of FGFR1 WT (b, c) and MU cases (d, e) demonstrate comparable histomorphology in both groups. T-SNE analysis of DMG reveals FGFR1 and BRAF MU cases to harbor similar DNA methylation profiles as FGFR1 and BRAF WT cases (f). FGFR1 MU cases showed a significantly better prognosis than FGFR1 WT cases (p = 0.023, g), and multivariate analyses confirmed significance of FGFR1 status independent of age and localization. WT* = wild type for respective hotspot, MU = mutant, n. a. = not available, *WHO grade of initial diagnosis

Similar to FGFR1 WT cases, cases with additional FGFR1 mutation displayed features of a diffusely growing glioma with increased cellularity and signs of anaplasia, such as increased cell pleomorphism, mitoses, or vessel proliferation (Fig. 1b-e). Furthermore, all analyzed FGFR1 MU cases (and the BRAF MU case) matched to the methylation class”DMG, H3 K27M mutant” (Supplementary Fig. 1b, online resource, Fig. 1f, Supplementary Table 1, online resource).

Higher age (≥ 18 years), supratentorial tumor localization and FGFR1 MU status were associated with a significantly better prognosis of patients (p = 0.038, p = 0.034, and p = 0.023, Fig. 1g and Supplementary Fig. 2a, b, online resource). In contrast, TP53 MU status was associated with a significantly worse prognosis of patients (p = 0.002, Supplementary Fig. 2c, online resource). Including the latter factors in a multivariate cox regression analyses showed localization and TP53 status as significant variables (Supplementary Fig. 2d, online resource). FGFR1 and TP53 mutations occurred almost mutually exclusive and hence did not represent independent variables (see also Supplementary Fig. 1a, online resource). Thus, we performed a multivariate analysis including the independent variables age, localization, and FGFR1 status only (Fig. 1h). In this context, FGFR1 MU status was significantly associated with a better overall survival, independently of patient age, and tumor localization (p = 0.026). Interestingly, the single patient (#56) with an accompanying BRAF p.V600E mutation remained alive at 24.5 months after initial diagnosis. However, the prognosis for such diffuse midline gliomas with dual H3F3A p.K27M and BRAF p.V600E mutations remains to be defined.

Together, our results suggest that RAS-MAPK-pathway signaling might play an important role in DMG with implications for diagnosis, prognosis, and therapy of respective patients.