Introduction

Germline mutations in the proofreading exonuclease domain of the polymerase ε gene (POLE) were described in association with a familial predisposition to colorectal cancer (CRC) [1, 2]. Polymerase proofreading-associated polyposis syndrome (PPAP), caused by autosomal dominant mutations of polymerases genes POLD1 and POLE [3], predispose to colorectal cancer, with or without polyposis. Rohlin et al. described a family carrying the mutation of POLE c.1089C>A, p.Asn363Lys and suggested that POLE mutations can predispose carriers to a broad spectrum of tumours [4]. We report a family with the same mutation displaying a high incidence of CRC and glioblastoma. We support the hypothesis that this mutation is associated with brain tumours.

Subjects and methods

Family report

The family we report (Fig. 1) has been followed up in the Medical Oncogenetics Department of the Institut Universitaire du Cancer de Toulouse-Oncopole (IUCT-O). Four members, patient III:3, III:4, IV:15 and IV:19, developed a brain cancer (glioblastoma) at age 45, 52, 30 and 37 respectively. Ten members have developed at least one colorectal (I:1, II:8, II.10, III:5, III:3, III:4, III:8, III:18, III:16, and IV:15). The main phenotypic information relating to the family members is summarized in Table 1. Informed written consent for genetics germline analysis was collected from all participating subjects in the context of medical care.

Fig. 1
figure 1

Pedigree of the family carrying POLE c.1089C>A mutation reported in the current study. Age at diagnosis is bracketed when this information is available. POLE +: POLE c.1089C>A mutation carriers, POLE −: wild type POLE

Full size image
Table 1 Summary of phenotypic characteristics of POLE c.1089C>A, p.Asn363Lys pathogenic mutation carriers and their siblings presenting symptoms
Full size table

Analysis of tumour tissue

Somatic MMR status was determined by assessing microsatellite instability (MSI) and by immunohistochemistry (IHC) on colorectal cancer tissues of patient III:3 and III:4 and the glioblastoma of patient III:3. MSI status was established in all cases on the DNA from paired tumour and normal tissue samples as recommended by the National Cancer Institute workshop [5]. Expression of MLH1, MSH2 and MSH6 was assessed by IHC as previously described [6].

Germline mutation analysis

Analysis of MLH1 (NM_000249.3), MSH2 (NM_000251.2), MSH6 (NM_000179.2) and APC (NM_000038.4) genes was performed in the Laboratoire d’Oncogénétique of the IUCT-O using Sanger sequencing. The screening method of these two genes is described in Supplementary Material.

Mutation screening of the exonuclease coding region of POLD1 (NM_001308632.1) codons 268–471 and POLE (NM_006231) codons 304–517 was performed in the Laboratoire d’Oncogénétique Biochimie Génétique et Moléculaire of the Hôpital Cochin, Paris using custom made Ion Torrent Ion AmpliSeq panel (Life Technologies) on DNA extracted from peripheral blood samples. Mutations were confirmed using Sanger sequencing.

Analysis of BMPR1A and MUTYH was performed in the Laboratoire de Génétique Moléculaire of the CHU de Rouen by Sanger sequencing.

Results

All tumours tested (Table 1) were micro-satellite stable (MSS) with maintained expression of the MMR proteins MLH1, MSH2, MSH6 (patients III:3, III:5, III:4) and PMS2 (patient IV:15) except the sigmoid polyp of patient IV:15 which presented an isolated loss of MSH6 expression (checked on two occasions).

Germline DNA sequencing of MLH1, MSH2 (patients III:3, III:4, III:5) and MSH6 (patients III:5, III:3, IV:15) did not reveal any mutation or large rearrangement of these genes. APC (patients II:10, III:5, III:4), MUTYH (patients II:10, III:5, III:4, IV:5, IV:15), BMPR1A (patient IV:15) and POLD1 (patient IV:5) were also sequenced and no mutation was detected.

Finally, a c.1089C>A, p.Asn363Lys mutation of POLE was detected in patient III:4. She developed a rectal adenocarcinoma (MSS) diagnosed at age 37 years and a right temporal glioblastoma at 52 years of age. Her brother, patient III:3, was a carrier of the familial mutation and developed an adenocarcinoma of the caecum, a synchronous cancerous sigmoid polyp and a glioblastoma at 42 years of age. Patient IV:15 (distant cousin of patient III:4 and patient III.3), carrying the familial mutation, developed a colonic adenocarcinoma at 23 years of age, associated with multiple polyposis, and a glioblastoma at 30 years of age.

The mutation status of another distant cousin (IV:19) was not tested, but this patient also developed a glioblastoma diagnosed at 37 years of age. His father (III:18) is a carrier of the familial mutation.

Glioblastoma pathology

Glioblastoma biopsies were performed, before initiating any treatment, for patient III:3, patient III:4, patient IV:19 and patient IV:15. Pathological characteristics of these glioblastoma biopsies are summarized in Table 2 and their histological appearances are illustrated in Fig. 2. All were giant cell glioblastomas.

Table 2 Summary of pathological characteristics glioblastoma tissue of patient III:3, III:4, IV:15 and IV:19
Full size table
Fig. 2
figure 2

Resected giant cell glioblastomas of patient III:3 (1.c), III:4 (1.b), IV:19(1.a) and IV:15 (1.d) stained with haematoxylin and eosin (H&E). These reveal numerous multinucleated giant cells with atypical mitoses. Histological descriptions are summarized in Table 2

Full size image

Discussion

Glioblastoma is the most frequent primary brain tumour. In European and North American countries, glioblastoma incidence is approximately 3 cases per 100,000 people per year [7, 8].

In POLE, only the germline mutation p.Leu424Val was associated with glioblastoma with an incidence ranging from 4% [2] to 6.4% [9].

The current family presents an unusually high frequency of early onset glioblastoma, which occurred in three mutation carriers (III:4, III:3 and IV:15) and a suspected carrier (IV:19) who had a high probability (50%) of carrying the familial mutation via paternal transmission.

Rohlin et al. reported, in a Norwegian family, a patient displaying a glioblastoma of early onset (at age 28 years) who carried the same mutation [4]. His grandfather developed a CRC and a glioma at 65 and 68 years of age respectively. Of his seven children, five were carriers of a POLE familial mutation. Even though the mutation status was not determined for this person, we can therefore hypothesise that he is most likely the transmitter of the familial mutation [4].

Compiling information from the two families, POLE mutation p.Asn363Lys appears to be associated with a high rate of primary grade IV glioma with a relative risk of glioblastoma of at least 17.4% (4/23) in the carrier population. Glioblastoma incidence in carriers of the p.Asn363Lys mutation seems to be more than twice as high as in families with a p.Leu424Val mutation, or any other mutation of the proofreading exonuclease domain of POLE.

We can put forward three hypotheses to explain these findings. Firstly, the incidence of glioblastoma associated with mutations of the exonuclease domain of the polymerase ε may have been overlooked or underestimated in previous studies. Secondly, the p.Asn363Lys POLE mutation may be associated with a higher relative risk (20%) of developing glioblastoma than the other POLE mutations. Thirdly, the increased incidence of glioblastoma could be explained, in our family, by synergistic mutations located in unexplored genes, completely segregating with the familial POLE mutation.

Tumour tissue from patients (III:3, III:4, IV:19, IV:15) revealed giant cell glioblastoma histology with a high proportion of multinucleated giant cells, a rare variant which account for only 2–5% of glioblastomas and which was previously associated with somatic mutation of POLE [10]. The occurrence of this rare histological type in all four of our patients’ glioblastomas in the same family allows us to propose a link between germline POLE mutation and the occurrence of glioblastoma with giant cell morphology. A hypermutated tumour phenotype has been described in glioblastoma tissue from a POLE germline mutation carrier [11] and in a case of glioblastoma bearing a somatic POLE mutation (a patient without a germline POLE defect) [10]. Unfortunately, the rate of glioblastoma mutations in patients III:3, III:6, IV:15, IV:19 has not been tested in the context of medical care, so we cannot provide this information.

Among the POLE mutated population of the two families the incidence of colorectal cancer was particularly high, 69.5% (16/23) with a mean age at diagnosis of 45 (ranging from 23 to 67). These findings are consistent with previous observations [2]. There are colorectal follow-up recommendations for carriers of the POLE germline mutation [9], but no recommendation relating to the increased susceptibility to brain tumours has been proposed so far. The clinical history of glioblastoma is usually short (< 3 months in > 50% of patients), so the benefit of neurological follow up with MRI imaging for POLE mutation carriers should be carefully assessed. Given the short clinical history of such tumours, enhancing patient vigilance to early symptoms (unusual headaches, sensory or motor problems) would not help to improve their early care and would be stressful for patients and their families.

We report the second family in the literature bearing the POLE c.1089C>A, p.Asn363Lys mutation. Carriers have an increased risk of brain cancer, with an estimated incidence of at least 17.4% and a colorectal cancer risk of 69.5%. Even though these results could be improved with more patients, with functional testing and glioblastoma sequencing (to assess mutational burden and second hit of POLE), practitioners treating patients bearing such a mutation should be informed about this increased relative risk of high grade glioma so that they can adapt patient management and follow up accordingly.