Oncogenic KRAS hotspot mutations are rare in IDH‐mutant gliomas

Dear Sir,

Kirsten rat sarcoma viral oncogene homolog (KRAS) belongs to the RAS oncogene family that encodes intracellular signal transduction proteins (guanosine triphosphate hydrolases (GTPases) which regulate cell proliferation, motility and apoptosis 9. KRAS mutations account for up to 85% of RAS mutations in all human tumors. Cancer KRAS hotspot mutations at codons 12/13 and 61 usually prevent Ras protein from hydrolyzing GTP and thus the protein remains in the active state 9. While KRAS mutations are currently not druggable per se, the detection of KRAS mutations serves as a biomarker for targeted anti‐cancer therapies and predicts therapeutic response, for example in non‐small cell lung cancer 3. In contrast to the experimental data that indicated an essential role for Ras signaling in glioma maintenance, older studies in surgically resected cases did not provide any evidence for KRAS hotspot mutations in glioblastomas 5. In the TCGA sequencing data set, two KRAS G12R mutations were detected in 275 glioblastomas 4. Recently, whole exome sequencing of 11 oligodendroglial tumors identified two cases with activating KRAS mutations (Q61L, G12D) 7, and KRAS G12/13 hotspot mutations were detected in four out of eight IDH‐mutant 1p/19q‐codeletd tumors with CIC mutations 10. These data indicate that KRAS hotspot mutations might be frequently present in a glioma subset, especially with oligodendroglioma phenotype and/or recurrence after treatment with alkylating agents.

We therefore examined 186 brain tumor samples retrospectively for KRAS mutations (Table 1) including 54 oligodendrogliomas grade II and 33 anaplastic oligodendrogliomas grade III, 7 IDH‐mutant astrocytomas, grade II/III, 16 IDH wild‐type astrocytomas, grade II/III, 69 IDH‐wild‐type glioblastoma grade IV and 7 IDH‐mutant glioblastomas, grade IV. The formalin‐fixated, paraffin‐embedded samples were obtained from patients undergoing surgery for astrocytic and oligodendroglial brain tumors between 2000 and 2018 in the Department of Neurosurgery at the University hospital of Tuebingen. The study was authorized by the Tübingen University ethics board (permission number 788/2016BO2).

Table 1.

Epidemiological data of the tumors analyzed. M = male; F = female; IDH = isocitrate dehydrogenase; ND = not determined; TMZ = temozolomide; PCV = procarbazine, lomustine, vincristine.

Diagnosis	WHO Grade	cases	Gender (M/F)	Mean age (range)	Primary/recurrent	Radiotherapy (No/Yes/ND)	Chemotherapy TMZ/PC(V)/Other	IDH mutant	MGMT methylated/unmethylated	KRAS mutant
Oligodendroglioma	II	54	22/32	46.4 (17‐71)	43/11	22/25/7	2/19/0	54 (100%)	7/1	2 (4%)
Anaplastic oligodendroglioma	III	33	21/12	47.8 (31‐79)	15/18	13/14/6	7/11/3	33 (100%)	24/0	0
Astrocytoma	II	12	7/5	44.2 (15‐71)	10/2	6/2/4	2/0/0	4 (33%)	1/4	0
Anaplastic astrocytoma	III	11	5/6	50.0 (17‐67)	10/1	1/8/2	9/0/0	3 (27%)	3/6	0
Glioblastoma	IV	76	49/27	55.3 (23‐88)	55/21	15/52/9	45/0/17	7 (9%)	24/43	1 (1%)
All tumors		186	104/82	50.4 (15‐88)	132/53	57/101/28	65/30/20	101 (54%)	59/54	3 (0.2%)

For DNA extraction, tissue was selected from regions on paraffin blocks that presented sufficient (>60%) tumor content in microscopy. The region around KRAS codons 12/13 was amplified and PCR analysis was performed using the primer pairs KRAS‐forward, 5′‐nnn ggc ctg ctg aaa atg act gaa‐3′ and KRAS‐reverse biotinylated primer, 5′‐tta gct gta tcg tca agg cac tct‐3′. Pyrosequencing was performed on the Pyromark Q24 system according to the manufacturer’s instructions (Qiagen, Hilden, Germany). For pyrosequencing, we used starting primer KRAS‐PF2 (5′‐tgt ggt agt tgg agc t‐3′) with dispensation order “TACGACTCAGATCGTAG”. Pyrograms were analyzed with the PyroMarkQ24 software (Version 2.0.7 Build 3). Positive cases were confirmed by repeated pyrosequencing with a level of detection of 5%. NGS gene panel sequencing and target enrichment of further 75 glioblastomas, 9 astrocytomas and 1 oligodendroglioma were performed as described in the Supplemental File. The panel included 710 genes among them other members of the Ras/Raf pathway as well as EGFR and members of the pAKT/mTOR pathway (full list and coverage details available in Supplemental File). Sequencing was carried out on Illumina platforms as 100 bp paired‐end runs. Somatic variants were reported when blood and tumor samples had a minimum of 30 reads and at least 5 reads reported the same changes. The absolute detection limit was set to 5% of NAF.

KRAS mutation analysis results are depicted in Table 1. Initial pyrosequencing of 85 oligodendroglial and 15 astrocytic tumors identified one case with KRAS p.G12C and one case with p.G12R mutation which are considered pathogenic (Cosmic No 516 and 518). One oligodendroglioma specimen could not be evaluated because we observed weak KRAS pyrosequencing relative to the light unit signal repeatedly and therefore assumed insufficient sample quality. Both KRAS‐mutant tumors were diagnosed as oligodendrogliomas, IDH‐mutant and 1p/19q codeleted, WHO grade II (one female, 51 and one male 48 years, representative histology depicted in Figure 1) and, showed ATRX retention and TERT promotor hotspot mutations as expected. Next, we compared the frequency of KRAS mutations by panel sequencing of further 85 cases (75 glioblastomas, 9 astrocytomas and one oligodendroglioma). One glioblastoma, IDH wild‐type, grade IV WHO showed a pathogenic KRAS mutation Q61L (Cosmic no 553). The glioblastoma case was a 39‐year‐old female with IDH1/2 wild‐type status and TERT promotor hotspot mutation. No mutations were detected in HRAS and NRAS genes.

Figure 1.

A. Oligodendroglioma, IDH‐mutant and 1p/19q codeleted, WHO grade II. KRAS pyrosequencing displays a GGT>CGT nucletotide substitution in codon 12 resulting in p.G12R mutation. B. Oligodendroglioma, IDH‐mutant and 1p/19q codeleted, WHO grade II with KRAS codon 12 GGT>TGT substitution (p.G12C mutation).

In total, the detected KRAS hotspot mutation frequency among all 54 grade II oligodendrogliomas was 4% and among all grade IV tumors 1% (no significant distribution differences between astrocytomas and oligodendrogliomas). Thirteen of our IDH wild‐type cases had TERT hotspot mutations, EGFR amplifications or a 7+/10‐ chromosomal signature and would qualify for a designation of diffuse astrocytic glioma, IDH wild‐type, with molecular features of glioblastoma, WHO grade IV according to the cIMPACT‐NOW consortium (update 3) recommendations. Thus, the actual proportion of KRAS mutant glioblastomas in molecularly stratified cohorts could be even lower. KRAS‐mutant allelic frequencies ranged between 14% and 43%. As expected from the low number of cases, Kaplan‐Meier analysis (follow‐up data available in 138 cases) did not show significant differences between KRAS‐mutant and wild‐type tumors (mean survival: 818 days, vs. 769 days, Log‐rank: P = 0.7793).

Mutant allelic frequencies were lower as expected from the tumor cell content (Figure 1). Compared to the IDH mutation frequency determined in the two mutant oligodendrogliomas, the ratio of KRAS‐mutant alleles was 40%–79%. This suggests that KRAS mutations are spatially distributed and represent a subpopulation of the tumor tissue. All three mutant cases were primary tumors, indicating that KRAS mutations are not treatment related. Based on the recently reported data of combined 6 KRAS mutations in 19 oligodendroglial tumors analyzed with next‐generation sequencing 7, 10, we expected a higher mutation frequency in our oligodendroglioma cohort. In accordance with the low frequency in our study, the TCGA data set including oligodendroglial tumors reports only a single KRAS p.G12A mutation 2. The nonsignificant distribution among glial tumors indicates that KRAS mutations are not associated with a particular glial phenotype and occur only in low frequencies in gliomas. Of note, the sample cohort with previously reported high KRAS mutation frequencies in oligodendrogliomas mainly contained treatment‐recurring tumors (9 untreated patients including only one 1p/19q‐codeleted case and 11 previously treated samples, including 8 1p/19q‐codeleted cases). Different KRAS genotyping results may also result from interlaboratory discrepancies leading to false‐positive or negative reports. We have employed pyrosequencing to examine the KRAS hotspot regions directly. Depending on the dispensation used, pyrosequencing may be unable to produce pyrograms that distinguish between uncommon simple and complex mutations. Next‐generation sequencing methods cover a broader range of the genes, depending on the reading depths available. False‐positive results may occur from short reads with mutations generated at one end though amplification with mismatched primers.

Prevalence of KRAS mutation in brain tumors other than glioblastoma and pilocytic astrocytoma is poorly characterized. Ma et al (2017) used next‐generation sequencing to study KRAS, BRAF and EGFR mutations in diverse tumors from 822 patients including 29 brain tumors. The authors did not specify whether KRAS mutation was found in these 29 brain tumors 6. A different study focusing on unusual 1p/19q codeleted tumors with oligodendroglioma morphology reported two cases with KRAS p.G12D mutations in the absence of IDH1/2 mutations 1. Because IDH1/2 mutations were absent in these tumors, they cannot be classified as oligodendroglioma according to the current WHO classification (2016). Abnormalities of the mitogen‐activating protein kinase (MAPK) pathway mutation are common in pediatric low‐grade gliomas and glioneuronal tumors. Thus, occasional KRAS mutations are more likely seen in these tumors compared to oligodendrogliomas and glioblastomas (for a full list of references with KRAS mutation reports, see Supplemental file).

Limitations of our study are the restriction to the relevant KRAS codon 12/13 hotspot region, the possible intratumoral genetic heterogeneity by selecting regions of interest for DNA analysis and the further molecular evolution of relapsing tumors after continuous treatment. However, data from colorectal cancer indicate that KRAS mutations are robustly expressed with a discordance rate of only 2% in matched samples 8. The limit of detection in pyrosequencing is superior to conventional Sanger sequencing, but cannot detect possible intratumoral heterogeneity with only minor clones below our limit of 5%. Our study also included 29 recurrent oligodendrogliomas, yet the KRAS mutations were all identified in primary, untreated tumors indicating that these mutations are not an imperative response to previous alkylating treatments as suggested previously 10. The low number of KRAS mutations detected also limits a meaningful interpretation of survival analysis. Larger studies with high numbers of KRAS‐mutant cases are required to evaluate prognostic results. Taken together, KRAS hotspot mutations remain a rare event in diffuse gliomas. Targeted screening of oligodendrogliomas for KRAS mutations does not provide additional diagnostic benefit in clinical practice.

Supporting information