Clinical, pathological, and molecular data on desmoplastic/nodular medulloblastoma: case studies and a review of the literature

The aim of this study was to better define the clinical and biopathological features of patients with desmoplastic/nodular medulloblastoma (DNMB) and to further characterize this subgroup. 17 children aged < 5 years, with initial DNMB treated according to the HIT-SKK protocol, were evaluated. A retrospective central radiological review, a pathological and immunohistochemical study, and array-CGH and sequencing of germline SUFU and PTCH1 genes were performed. 15 histologically reviewed cases were confirmed as DNMB including three cases of medulloblastoma with extensive nodularity. Median age at diagnosis was 26 months. Radiology showed five cases with a vermis location and one with T2 hyperintensity. All cases showed a SHH immuno-profile. A 9q deletion was found in 6 cases, a MYCN–MYCL amplification in 1 case, and a SUFU germline mutation in 1 case (/9). The presence of SUFU and PTCH1 germline mutations agreed with previous reports. At 3 years, progression-free survival and overall-survival rates were 72 ± 15% and 85 ± 10%, respectively. The rate of recurrence was relatively high (4 patients). This may have been because chemotherapy was delayed in two cases. Age > 3 years, and residual tumor may also have been an explanation for recurrence.


Introduction
Medulloblastoma (MB) is the most common childhood malignant central nervous system tumor. Clinical, pathological, and molecular variables are presently considered to stratify disease risk. The World Health Organization (WHO) classification of central nervous system tumors describes classic MB, desmoplastic/nodular MB (DNMB), MB with extensive nodularity (MBEN), anaplastic MB, and large-cell MB [1]. DNMB, including MBEN, represents 10% of all cases in children, reaching 57% in children aged < 3 years [2]. DNMB shows reticulin free nodules with a distinct neuronal immunophenotype, which are particularly intense in MBEN. Reticulin-rich zones have a high MIB1 index.
An international meta-analysis has shown that the histology of DNMB is a strong and independent favorable prognostic factor in young children, even in cases with metastatic disease [3]. Prolonged remission can be obtained when these patients are treated with intensive postoperative chemotherapy without irradiation [4]. Globally, progression-free survival (PFS) is higher in this histopathological sub-group [5]. An accurate diagnosis of DNMB is therefore crucial in young children, given the critical consequences related to treatment decisions.
Recent molecular expression and methylation profiles have led to a classification of MB according to molecular subgroups [6], which have been reduced to four by a consensus: i.e., WNT (the Wingless activated pathway), SHH (the Sonic Hedgehog activated pathway), group 3, and group 4 [7,8,9,10]. The SHH subgroup represents ~ 25 -30% of all cases of MB, but a much higher proportion of cases in early childhood DNMB (92% in those aged < 2 years) [11,12]. There is a consistent predisposition to PTCH1 or SUFU germline mutations in MB patients aged < 3 years, with MBEN strongly related to Gorlin's syndrome [11,13]. In agreement with the chromosomal location of the PTCH1 and SUFU genes, the loss of 9q and 10q are the most frequent chromosomal alterations found in this group [12,13]. The SHH pathway may also be activated through activating mutations of SMO or amplification at the GLI2 or SHH loci [14]. Comparative array genomic hybridization (array-CGH) can reveal MYC-MYCN amplification and most related chromosomal abnormalities [15].
In this study, we review our experience with patients aged younger than 5 years with DNMB and treated according to the HIT-SKK 92 trial [4]. Our aim was to assess the clinical, pathological, and biological data from this particular group with MB and to determine the relationships of these data with outcomes in comparison with literature.

Materials and methods
Patients' characteristics and treatments 17 children aged < 5 years and with newly diagnosed DNMB/MBEN after surgical excision were evaluated. Staging included pre-and postoperative cranial magneticresonance imaging (MRI) or computed tomography (CT), a spinal MRI, and CSF cytology. Available CT and/or MRI scans were centrally reviewed (14 patients). All children were treated according to the HIT-SKK 92 trial, which combined systemic chemotherapy with intraventricular chemotherapy (methotrexate) [4].
MYC and MYCN statuses were collected from fluorescence in-situ hybridization and/ or from array-CGH. The array-CGH techniques were performed on two different platforms [15]. Patients with a blood sample and whose parents had given their informed consent for genetic studies were assessed for germline mutations. SUFU gene mutational screening used procedures as previously described [11]. PTCH1-point mutation analysis was performed using Enhanced Mismatch Mutation Analysis (EMMA), a procedure based on hetero-duplex analysis [19].

Statistical analyses
PFS and overall survival (OS) were calculated using the Kaplan-Meier method. Differences between the groups were assessed using the log rank test. PFS was defined as the time from the date of diagnosis until the date of first progression, death from any cause, or last contact. OS was defined as the time from the date of diagnosis until death from any cause or last contact.

Results
Patients' characteristics and treatments (Table 1) Of the 17 potentially eligible patients, 2 were excluded (case 10 tumor data unavailable, case 13 reclassified as a classic/biphasic medulloblastoma). Median age at diagnosis was 26 months (range 6 -59), 4 patients were older than 3 years. The male/female ratio was 2 : 1.
Tumor location was midline in five cases, restricted to one cerebellar hemisphere in six cases, and affected both the midline and hemisphere in three. Five cases showed invasion of the dura mater. In six patients, tumor extension followed the cerebellum folia (gyriform pattern).The tumors were hypo-or isointense in T1 and usually isointense in T2. One patient had metastasis. Complete surgical resection was obtained initially or after a second surgery. In case 9, the second surgery was performed after onset of chemotherapy. Adjuvant chemotherapy began within a median delay of 23 days (range 14 -119) after the last surgery. 13 patients received three cycles of chemotherapy, and 2 patients received only two cycles because of neurotoxicity (case 17) or an early metastatic relapse (case 5). Two patients did not receive intraventricular methotrexate (cases 5 and 14).

Outcomes
The median follow-up was 37 months (range 15 -50). At 1 year, PFS and OS were 93 ± 8% and 100%, respectively. At 3 years, PFS and OS were 72 ± 15% and 85 ± 10%, respectively ( Figure 1). Four relapses occurred at 22 and at 360 days after the end of treatment. Two patients with local (case 9) or a metastatic relapse (case 14) died at 12 months and 15 months after diagnosis, when at 30 and 21 months of age, respectively. Case 5, who had a metastatic relapse, received sequential high-dose chemotherapy and craniospinal irradiation. Case 6, with bifocal recurrence, was enrolled in a phase I study on SHH inhibitors. Both patients were in complete remission at the last follow-up.

Pathologic findings (Table 2)
All 15 retained cases fulfilled the criteria for desmoplastic/nodular features: 3 typical MBENs and 12 DNMBs. Six of these 12 DNMB had large nodules reminiscent of MBEN, but these were rare and showed faint NeuN and synaptophysin expression. The combination of positive GAB1, YAP1, and Filamin A staining indicated an SHH profile in all tumors [18]. All tumors were positive for anti-P75NTR antibody. In case 5, immune staining with GAB1 and YAP1 was faint and focal. The inter-nodular MIB1 index was between 40%, and 80% in 12 cases, and did not exceed 20% in 3 cases (cases 4, 6, 8). P53 expression was low (≤ 10% in all cases). One tumor (case 2) showed posi- tive nuclei for beta-catenin (10%), without a CTNNB1 mutation after direct sequencing, and chromosome 6 monosomy.

Genetic features (Table 2)
From the tumor material, array-CGH, PTCH1, and SUFU direct sequencing were performed for 13, one, and three cases, respectively. Germline PTCH1 and SUFU sequencing was done for seven and nine cases, respectively. From array-CGH, we found evidence of a 9q deletion encompassing the PTCH1 locus in six cases. An inactivating mutation of PTCH1 on exon 21 was seen in the tumor DNA, which definitively proved the SHH subtype in one tumor (Figure 2, case 11). In two cases where 9q was deleted (6,16), PTCH1 germline mutations on exons 6 and 17, respectively, were found associated with a neutral variant, but were considered as polymorphisms. In another deletion case (1), an isolated silent mutation was found. PTCH1 germline alterations were found in case 7, without a 9q deletion, but both variants were considered to be polymorphisms. We did not observe any 10q deletions, where-as a germline SUFU mutation was observed in only one of the nine analyzed cases. No GLI2 or SHH amplifications were observed. One tumor (case 5) showed MYCN and MYCL amplifications, and a 9q deletion encompassing PTCH1; no chromothripsis was seen in that tumor. No tumors had chromosome 17q gain or isochromosome 17q, these being the most frequent chromosomal alterations in infant non-desmoplastic, non-SHH MB.

Discussion
This study reports on a relatively small number of cases but highlights several points. Because of the relatively low number of patients, our findings are discussed without making a definitive conclusion. Our findings led us to review the literature on the desmoplastic/nodular sub-group of MB, which was almost always linked to the SHH pathway at a young age.
DNMB was identified as an independent favorable prognostic factor in infants [3,5,20]. In the HIT-SKK 92 trial, the 3-year PFS and OS were 85 ± 8% and 95 ± 5%, respectively [4]. In the present study, at 3 years, the PFS and the OS were 72 ± 15% and 85 ± 10%, respectively, which was not as good as in prior published studies, with recurrence occurring in 4 patients. The delay in starting chemotherapy exceeded 4 weeks in 2 of these 4 recurring patients. T2 hyperintensity was noted in 1 of these 2 patients. Age at diagnosis was > 3 years in 2 patients who relapsed; there were 4 patients aged > 3 years in the series. In another relapsing patient, who died, there was a tumoral residue after initial surgery. Delay starting chemotherapy, age at diagnosis, and/or a residual tumor should be considered as potential outcome factors.
Location of the lesions was not restricted to the cerebellar hemisphere: 5/14 children presented with a midline tumor. Liu et al. [21] found that all patients with DNMB (12/12 cases) presented with a tumor involving the vermis. Although cerebellar hemispheric tumors are predominant in the SHH subtype [22], a recent study has shown that SHH subtypes were not exclusively hemispheric and that hemispheric MBs were not always SHH-activated [23]. In another study, nearly 80% of SHH tumors were located in the vermis when the group was restricted to infants [24]. Location of the tumor varies depending on age, pathology, and molecular subtype. A midline location is often seen, especially in younger patients with a combined desmoplastic/nodular and SHH classification.
DNMB can express the nodular phenotype to varying degrees. Desmoplasia may occur in any MB variant as a normal reactive phenomenon when tumor cells invade the leptomeninges. However, so far, subclassification has been difficult [20]. Indeed, one MB was reclassified as a classic MB with nodules (biphasic MB). The reticulin stain, delineating pale islands, remains a strong diagnostic feature [20] and immunohistochemistry may be helpful. The association between DNMB and activation of the SHH signaling pathway is not constant but is particularly elevated in infants [12]. Here, SHH immunohistochemistry was absent in excluded biphasic MB arguing again for the usefulness of this procedure.
The P53 index was low. TP53 gene status was not studied. A worse prognosis has been associated with a TP53 mutation in SHH MB patients aged > 5 years [25]: in contrast, all patients were aged < 5 years in our series.
MYC gene amplification was absent within our cohort. The only MYCN amplification was found in a case with a 9q deletion encompassing the PTCH1 locus, which strongly suggests this tumor belongs to the SHH subgroup. The association of MYCN and MYCL in this tumor is a striking and unique observation, although isolated MYCL amplifications have been sporadically observed in SHH tumors [26] and MYCN is one of the most frequently amplified genes in SHH tumors [9]. Multiple amplicons in a SHH setting usually associate GLI2 with MYCN and may indicate Li-Fraumeni syndrome, which is linked to a dismal prognosis. If though the status of MYCN amplification in MB as an outcome indicator remains globally controversial, it has been suggested that MYCN-amplified SHH tumors have a poorer outcome [2,22]. Retrospectively, one may wonder if tumors with multiple amplicons and/or MYCN-SHH tumors justify a more straightforward and intensive treatment.
Cases of MB are known to occur in predisposition syndromes, particularly in Gorlin's syndrome [13,27,28]. Germline SUFU mutations have been found to be responsible for a high proportion of desmoplastic MB in children aged < 3 years [11]. These mutations may cause Gorlin's syndrome with a higher risk of developing MB than PTCH1 mutations [28]. It has been shown that MBEN is strongly associated with Gorlin's syndrome, particularly in patients aged < 3 years [13,27]. In our series, the frequency of SUFU mutations (1/6 tested patients aged < 3 years) was in concordance with that reported in the literature. The absence of a deleterious germline PTCH1 mutation reinforces the hypothesis that desmoplastic MB as a stand-alone clinical sign is a low-evidence criterion for screening for a deleterious germline PTCH1 mutation [28].
This study confirms the prognostic clues for MB subtypes, especially desmoplastic types, depending on the complementary data. The efficiency of a targeted therapy, which can be particularly useful in these tumors, relies on combining criteria from pathology and genomics [10]. Further large international studies are needed to better comprehend this childhood disease.