Prognostic factors in adult patients with medulloblastoma

Abstract

Background

Medulloblastoma (MB) is a rare tumor in adults, with treatment strategies derived largely from pediatric data. Prognostic factors have not been uniformly defined in adults to date.

Methods

We retrospectively reviewed the medical records of 89 adult MB patients treated between 1995 and 2019 in our institution. Patient’s characteristics, disease features, and treatment modalities were analyzed for prognostic factors using univariate and multivariate analysis.

Results

Of the 89 patients, 66% were male. Most MBs were in the cerebellum (48%), with desmoplastic/nodular histology (43%), Sonic Hedgehog molecular type (79%), and M0 Chang´s stage (72%). Intermediate- and high-risk MBs were identified in ~46% and ~47% of cases, respectively. Complete/near complete tumor resection was achieved in 62% of cases. Surgery followed by chemoradiotherapy (CT/RT) was the most frequent treatment (76%) with carboplatin-based regimens used in 70% of cases. After the first-line treatment, complete response (CR) was achieved in 80% of patients. Median overall survival (mOS) was 124.4 months (95%CI 68.5-180.1) and the median progression free survival (mPFS) was 30.5 months (95%CI 13.5-47.5), the 5-year OS was 67% and the 5-year PFS was 51%. In multivariate analysis, Chang´s stage ≥ M2 metastatic classification (P = .001), RT without CT in first line setting (P = .005), and craniospinal RT < 30 Gy (P = .015) were associated with worse survival outcomes.

Conclusions

Chang’s stage ≥ M2, first-line treatment lacking CT, and first-line treatment with craniospinal RT < 30 Gy were significant predictors of poor survival. Chemoradiotherapy with craniospinal RT ≥ 30 Gy improved survival outcomes.

Key Points.

• We report a large series of adult MB patients as reference for prognostic factors.

• Chang´s ≥ M2, lack of CT, and craniospinal RT < 30 Gy predict poor survival.

• CT/RT leads to superior outcomes in adult medulloblastoma.

Importance of the Study.

This study represents one of the largest series of adult medulloblastoma (MB) patients, a rare and underexplored tumor in adults. Several important factors were taken into consideration: clinical, radiological, histological, molecular, and treatments. Key prognostic indicators included Chang’s M2 stage, absence of chemotherapy in first-line treatment, craniospinal radiotherapy doses below 30 Gy, all of which were associated with poorer overall survival and progression free survival. The findings underscore the critical role of a multimodal approach, in improving outcomes. Given the scarcity of prospective data for this population, the study provides essential insights that could guide treatment considerations and influence future clinical trials. Understanding the unique characteristic of adult MB compared with pediatric patients is crucial for improving survival rates and optimizing treatment strategies. This research study may serve as an important reference point for future studies.

Medulloblastoma (MB) is the most common malignant primary brain tumor in children, representing up to 25% of all central nervous system (CNS) tumors in this population. In contrast, adult MB is a rare entity, accounting for less than 1% of primary CNS tumors in adults, with an incidence rate of 0.6-1 case per million per year.^1–4

Although it is potentially treatable using multimodal approaches, the prognosis remains heterogeneous, with a median overall survival (mOS) of 65 months, a 5-year OS of 49%-84%, and a 5-year-progression free survival (PFS) of 50%-89% in different series.^5–10

To date, MB is classified according to histopathology and molecular features.¹¹ Histologically, it can be classified into four groups: (i) classic, (ii) desmoplastic/nodular, (iii) extensive nodular, and (iv) large cell/anaplastic. Initially, four molecular groups of MB were distinguished: (i) Sonic Hedgehog (SHH), (ii) Wingless (WNT), (iii) group 3, and (iv) group 4. Today, following the 2021 World Health Organization (WHO) classification, non-WNT/non-SHH MB includes both group 3 and group 4 tumors.¹²

According to tumor location, patients experienced initial clinical signs and symptoms leading to diagnosis of increased intracranial pressure (IIP) and/or cerebellar symptoms (>90%), gait disturbances (68%), nausea/vomiting (59%), and/or vestibular symptoms (41%).¹³ MRI is the standard for the assessment and follow-up of MB. However, if a CT scan is performed, the tumor will often (62%-97%) appear hyperdense.¹⁴

Due to its rarity, prognostic factors in adult MB are primarily derived from pediatric trials and retrospective series. In 1969, Chang et al. developed a staging system for MB, adapted from the TNM system, which categorizes disease based on tumor size, location, and metastatic spread. The tumor (T) categories include T1 (tumor < 3 cm in diameter, confined to the midline), T2 (tumor ≥ 3 cm in diameter without extension into the aqueduct of Sylvius or foramen of Luschka), T3a (tumor > 3 cm in diameter with extension into the aqueduct of Sylvius or foramen of Luschka), T3b (tumor > 3 cm in diameter with extension into the brainstem), and T4 (> 3 cm in diameter with extension beyond the aqueduct of Sylvius or foramen magnum). The metastatic (M) categories are defined as M0 (no evidence of metastasis), M1 (microscopic tumor cells present in the cerebrospinal fluid), M2 (gross nodular seeding in the cerebellar, cerebral, or spinal subarachnoid space), M3 (gross nodular seeding in the spinal subarachnoid space), and M4 (extraneural metastasis).¹⁵

In 2019, the first European practice guidelines for adult MB patients were published and the clinicobiological criteria of high-risk group were redefined. Nowadays, current therapeutic recommendations are based on chemoradiotherapy (CT/RT) instead of radiotherapy (RT) alone regardless the risk category.¹¹ In adults, the best CT regimen is debated. Furthermore, there is ongoing discussion about the appropriate dose and technique of RT.^16,17 Moreover, while novel molecular targeted therapies are under investigation, their clinical implementation has faced setbacks.^18,19

In the current study, we report the clinical, radiological, histomolecular patterns, therapeutic strategies, and clinical outcomes of our cohort of adult MB patients with a particular focus on prognostic factors.

Materials and Methods

We retrospectively reviewed the medical records of all adult patients (age 18 or older) treated at the Pitié-Salpêtrière hospital for a pathologically proven MB between 1995 and 2019. A total of 128 charts were reviewed, and 39 were excluded due to insufficient clinical information. A total of 89 patients were included in the present study. Of the 89 patients included in the study, 52 underwent a detailed histomolecular review by an experienced neuropathology team to ensure accurate histomolecular classification. This review was conducted using immunohistochemistry for molecular subgroup classification, and MYC status was assessed by fluorescence in situ hybridization (FISH). Cases were excluded from the histomolecular review if the neuropathology team determined that the material was insufficient or if formalin-fixed paraffin-embedded (FFPE) material was unavailable. For patients whose cases could not be reanalyzed, histomolecular data were obtained directly from their initial pathology reports in the medical records. These patients remained part of the study, but their histomolecular classification was based solely on the initial report without further re-evaluation. Data collection and evaluation were completed in June 2020.

Of the 89 patients, radiologists’ reports were used in 55 cases (61%) because digital or physical images were not available. Multifocal MB was defined as at least two infra-tentorial noncontiguous enhancing foci of disease without subarachnoid seeding evidence on MRI.

Extent of resection was classified on postoperative T1-weighted MRI with gadolinium as: (i) gross/near total resection when residual tumor was < 1.5 cm², or (ii) subtotal/partial resection when the tumor residue was ≥1.5 cm².

Time to diagnosis (TTD) was defined as the time between the first symptom to pathology diagnosis. Analysis of cardiovascular risk factors was included due to the age population of our series and was defined using the 2008 Framingham classification.²⁰ Metastatic spread was defined by Chang’s staging criteria,¹⁵ and clinical-risk stratification was analyzed according to 2 recognized scales: (i) the association of French Speaking Neuro-Oncologists (ANOCEF) recommendations, and (ii) the European Reference Network on Rare Adult Cancers (EURACAN) guidelines.^11,21

Response assessment was performed using the RANO criteria developed for MB and leptomeningeal tumors, as endorsed by the EANO-EURACAN guidelines for adult MB.²² Follow-up was defined as patients with clinical available information from diagnosis to last medical visit or death. PFS and OS were calculated from the date of surgery/biopsy to the date of first evidence of tumor progression and the date of death or last follow-up, respectively. Patients without relapse were censored at their last follow-up.

Chemotherapeutic regimens were classified into 6 subgroups: (i) carboplatin-based, (ii) temozolomide-based, (iii) cisplatin-based, (iv) bevacizumab-based, (v) SHH inhibitors-based, and (vi) others. We analyze survival by age groups, by ANOCEF-EURACAN risk scales, and molecular subgroups in uni- and multi-variate analysis.

For the statistical analysis, prognostic factors were grouped in patient-related, disease-related, and treatment-related factors; risk scales were also compared and analyzed. Survivals were compared using the log-rank test and plotted using Kaplan–Meier method. Those with a P-value < .2 in the univariate analysis were explored in the multivariate analysis using the Cox stepwise-regression analysis. A P-value < .05 was used as the threshold for statistical significance. Statistical analysis was performed using the SPSS v26.0 software.

Results

Patient’s Disease and Treatment-related Characteristics at Diagnosis

Based on our inclusion criteria, data from 89 adult MB patients diagnosed between 1995 and 2019 were retrospectively included in the present study. The status of patients at the last follow-up was: (i) alive for 48 patients, (ii) dead for 33 patients, and (iii) unknown for 8 patients. MB was more frequent in males with a sex ratio men/women of 59/30 (66%). Median age at diagnosis was 29 years [19-71]. Patients aged between 18 and 25, 26-40, and >40 years represented 32%, 50%, and 17% of patients, respectively. Only 11% of patients had cardiovascular risk factors. No patient received CNS radiotherapy before MB diagnosis ruling out radio-induced MB in our series. The Karnofsky performance status (KPS) at diagnosis was 90 [60-100]. TTD was 97 [0-785] days (Table 1).

Table 1.

Patient’s Disease and Treatment-related Characteristics at Diagnosis

Characteristics	Number of patients N = 89
Patient’s related characteristics
Age, median [range]	29 [19-71]
18-25 y (%)	29 (32)
26-40 y (%)	45(50)
>40 y (%)	15(17)
Sex, male	59(66)
KPS at diagnosis, median [range]	90 [60-100]
≥80 (%)	65(73)
<80 (%)	10(11)
unknown (%)	14(16)
CVRF (%)	10(11)
Time to diagnosis, median days [range]	45 [0-785]
Presenting symptoms (%)
Headaches	49(55)
Nausea and vomiting (IH)	51(57)
Cerebellar syndrome w/and wo/vestibular syndrome	66(74)
Others	11(12)
Disease related characteristics
Location
Lateral (%) [Right/left/NS]	43(48) [22/20/1]
Midline (%)	24(27)
Both lateral and midline (%)	5(6)
Cerebellopontine angle (%)	2(2)
Brainstem (%)	1(1)
Multifocal (%)	8(9)
Beyond posterior fossa (%)	5(6)
NA (%)	1(1)
Chang’s high-risk staging (%)
T stage ≥ T3b -T4	14(16)
M stage ≥ M1- M4	25(28)
M1/M2/M3/M4	9(10)/0/15(17)/1(1)
Positive CSF (%)	11(12)
Spinal metastasis (%)	16(18)
Linear leptomeningeal	8 (9)
Nodular leptomeningeal	4 (4)
Mixed pattern	2(2)
NA	2(2)
Anatomic location of spinal metastasis
One region (%) [Cervical/Thoracic/Lumbar]	11(12) [1/8/2]
Multiregional (%)	3 (3)
NA (%)	2(2)
Metastatic outside cerebrospinal axis (%)	1(1)
Histology (%)	89(100)
Desmoplastic/nodular	38(43)
Classic	29 (33)
LCA	4 (4)
NOS & NEC	17(19)
Genetic analysis (%)	52(58)
MYC, MYCN amp absent	51(98)
MYC, MYCN amp	1(2)
Molecular analysis (%)	52(58)
SHH [TP53wt/TP53mut]	41(79) [38/3]
WNT	4(8)
Non-WNT/non-SHH	7(13)
Integrated histomolecular analysis (%)
Desmoplastic/nodular	38(43)
SHH TP53wt/TP53mut	26(29)/2(2)
NA	10(11)
Classic	29(33)
SHH TP53wt/TP53mut	12(13)/1(1)
Non-WNT/non-SHH	5(6)
WNT	4(4)
NA	7(8)
LCA	4(4)
Non-WNT/non-SHH	2(2)
NA	1(1)
MYC, MYCN amp	1(1)
NOS/NEC	17(18)
Treatment-related characteristics
Extent of initial resection (%)
CRS/NTR	55(62)
PRS/STR	24(27)
Biopsy	8(9)
NA	2(2)
“Second look” surgical resection (%)	7(8)
CRS/NTR	59(66)
PRS/STR	24(27)
Biopsy	4(4)
NA	2(2)
Treatment at diagnosis (%)
Surgery only	4(4)
Surgery followed by RT	17(19)
Surgery followed by RT/CT	68(76)
Posterior fossa RT ≥ 50 Gy	63(71)
Posterior fossa RT < 50 Gy	8(9)
Craniospinal RT ≥ 30 Gy	53(59)
Craniospinal RT < 30 Gy	17(19)
CT Carboplatin-based regimensa	62(70)
Other CT-based regimen	6(7)
Risk scales (%)
ANOCEF
Intermediate/High-risk	39(44)/44 (49)
EURACAN
Intermediate/High-risk	43(48)/40(45)

Amp, amplification; CNS, central nervous system; CRS, complete resection; CSF, cerebrospinal fluid; CT, chemotherapy; CVRF, cardiovascular risk factors; IH, intracranial hypertension; KPS, Karnofsky Performance Status score; NA, not available; NC, non-classifiable; NEC, not elsewhere classified; NOS, not otherwise specified; NTR, near total resection; PCF, Posterior cranial fossa; PRS, Partial resection; RT, radiotherapy; SHH- sonic hedgehog; STR, Subtotal resection; WNT, wingless.

^aCarboplatin/etoposide (CARBO/VP-16). Carboplatin/etoposide with dose reduction craniospinal irradiation protocol regimen (RSMA protocol). Carboplatin/etoposide with thiotepa and temozolomide (CARBO/VP-16-TTP-TD). Ifosfamide, carboplatin and etoposide (IFO/CARBO/VP-16). Carboplatin/etoposide high dose chemotherapy by thiotepa, stem cell transplant, radiotherapy, and maintenance temozolomide (PNET HR5).

Cerebellar syndrome was the most frequent initial clinical symptom, followed by IIP. Other symptoms included were cranial nerve involvement, cervicalgia, radicular symptoms, dysesthesia, apathy, and tremor. Cerebellar hemisphere was the most frequent location (48%) followed by midline location (27%), and by both lateral and midline locations (6%). Multifocal MB was detected in 9% of patients and invasion beyond the posterior fossa in 6%. High-risk patients according to Chang’s criteria were 16% by T stage and 28% by M stage. Only one case of metastasis outside the cerebrospinal axis (bone, M4) at diagnosis was documented. CSF analysis was obtained in 81% of patients, 12% being positive. Spinal metastasis occurred in 18% of the patients, thoracic location and linear leptomeningeal radiologic pattern were the most frequent.

The histological subtype was desmoplastic/nodular, classic, large cell/anaplastic (LCA) variant in 38 (43%), 29 (33%), and 4 (4%) patients, respectively. The molecular group was available in 52 (58%) patients: (i) SHH group in 41 (79%) cases, with a higher proportion of TP53wt compared SHH TP53mut cases with 38 and 3 cases, respectively; (ii) WNT group in 4 (8%) cases, and (iii) the non-WNT/non-SHH group represented 7 (13%) cases. All desmoplastic histological types fell within the SHH molecular subgroup. The classic histological group was molecularly heterogeneous; the most frequent molecular group was SHH TP53wt, followed by the non-WNT/non-SHH. Interestingly, the four WNT MB cases fell within the classic group. The LCA group included two cases of non-WNT/non-SHH molecular subgroups and one MYCN-amplified case. The SHH molecular subgroup presented desmoplastic/nodular and classic histology in our series. The non-WNT/non-SHH group showed classic and LCA histology, while the WNT group displayed classic histology (Figure 1).

Figure 1.

Histological and Molecular Groups

All patients were initially operated, 8% of them underwent a second look surgery. After the initial surgery, 62% had complete/near complete resection (CRS/NTR) and 36% had incomplete resections, including partial resection (PRS), and biopsy in 27% and 9% of cases, respectively. After a second look surgery a total of 66% achieved complete resection or near total resection (CRS/NTR). A considerable number of patients 30 (34%) underwent either ventricular drainage or ventriculo-peritoneal shunting.

Surgery followed by systemic chemoradiotherapy (RT/CT) was the most common strategy in first-line setting (76%), and only 17 patients were treated with RT alone. Craniospinal RT ≥ 30 Gy and posterior fossa RT ≥ 50 Gy were the elected schemes in most patients, 59% and 71%, respectively. Carboplatin-based regimens were the most frequently used in 70% of cases.

In terms of risk, patients were distributed in intermediate/high-risk groups for 44%/49% and 48%/45% of patients, according to the ANOCEF and the EURACAN classification respectively.

Oncologic Outcomes

Our adult cohort had a mOS of 124.4 (95%CI 68.53-180.11) and a median PFS of 30.5 (95%CI 13.46-47.51) months. The median follow-up of our cohort was 52 months (range 0.2-248). According to RANO criteria,²² tumor CR, PR, SD, and PD was achieved in 80%, 4%, 6%, and 6% of the patients, respectively, 4% of patients were not evaluable because we lost follow-up or died before the date of the MRI. Of the 16 initial patients with medullar involvement, 56% achieved CR, 6% achieved PR, and 25% were in SD/PD. Progression/relapse was documented in 40% of patients with most of relapses being intracranial (40%), followed by medullar (21%) and metastatic outside the CNS (17%). The most frequent site of systemic metastasis was bone, lymph nodes, and peritoneum. PFS was around 30.5 [3-148] months. At relapse, carboplatin-based chemotherapies were favored, followed by temozolomide-based regimens and cisplatin-based regimens. Molecular-targeted therapies were particularly used for patients included in clinical trials (Table 2, Figure 2).

Table 2.

Oncologic outcomes

	Number of patients (%)
Response after first line of treatment
CR	71(80)
PR	4(4)
SD/PD	5(6)
NA	9(10)
First Relapse/Progression	35(40)
Intracranial	19(21)
Medullar	15(17)
Metastatic outside CNS	9(10)
Bone	6(7)
Bone marrow	1(1)
Lymph nodes	1(1)
Peritoneum	1(1)
Number of relapses. [range]	[1-9]
Time to first progression. median  [range months]	30 [3-148]
Type of strategies at progressions
Surgery	10(11)
RT, Radiosurgery or RT/Surgery	8 (9)
CT/RT	3(3)
CT ± target therapy alone	50(56)
CT/surgery	5(6)
Carboplatin-based	20(22)
Temozolomide-based	15(17)
Cisplatin-based	10(11)
Bevacizumab-based	8(9)
SHH driver-based	7(8)
Others	7(8)

CR, complete response; CT, chemotherapy; NA, non-available; PD, progression disease; PR, Partial response; RT: radiotherapy; SD, stable disease; SHH, sonic hedgehog.

Carboplatin-based: Carboplatin/etoposide (CARBO/VP-16). Carboplatin/etoposide with dose reduction craniospinal irradiation protocol regimen (RSMA protocol). Carboplatin/etoposide with thiotepa and temozolomide (CARBO/VP-16-TTP-TD). Ifosfamide, carboplatin and etoposide (IFO/CARBO/VP-16). Carboplatin/etoposide high dose chemotherapy by thiotepa, stem cell transplant, radiotherapy, and maintenance temozolomide (PNET HR5). Temozolomide-based: Temozolomide alone. Temozolomide/topotecan (TOTEM). Temozolomide/irinotecan (TEMIRI). Temozolomide/etoposide (TD/VP-16). Cisplatin-based: Cisplatin/etoposide. Cisplatin, lomustine, vincristine. (PACKER). Cisplatin, vincristine and fotemustine (POF). Cisplatin, vincristine, lomustine, procarbazine, hydroxyurea, aracytin, cyclophosphamide in 1-day chemotherapy regimen (8 in 1). Bevacizumab-based: Bevacizumab, thalidomide, celecoxib, fenofibric acid, etoposide, cyclophosphamide, cytarabine (MEMMAT). Bevacizumab/irinotecan. SHH driver-based: Vismodegib alone. Vismodegib/temozolomide. Others: Procarbazine, belustine, vincristine (PCV). Caryolysine, vincristine and procarbazine (PPOM). Cyclophosphamide.

Figure 2.

Kaplan–Meier Curves for OS and PFS (A-B: OS and PFS, C-F: OS and PFS for significant risk factors identified in the multivariate analysis)

Grade III–IV CT-associated Toxicities

Carboplatin-based regimens were the most used CT during the disease course, 82 patients were treated under this regimen, and this group had the most frequent hematologic toxicity (39%), followed by cisplatin-based regimens (31%), being neutropenia the most frequent hematologic toxicity. Gastrointestinal, infectious, and neuropathic toxicities were most frequently associated with cisplatin regimens occurring in 23%, 8%, and 15% of the cases, respectively.

Survival Outcomes (OS and PFS): Univariate and Multivariate Analysis for Patient’s Disease and Treatment Factors

Outcomes by univariate and multivariate analysis were grouped in patient’s disease, or treatment-related outcomes (Table 3, Figure 2). By univariate analysis, significant risk factors associated with worse OS were Chang’s M ≥ M2 and non-WNT/non-SHH molecular groups. For reduced PFS, significant risk factors were Chang’s T ≥ T3b, Chang´s M ≥ M2, spinal involvement, and incomplete resection. However, in multivariate analysis, significant independent prognostic factors associated with a worse OS, were Chang´s M ≥ M2, treatment after diagnosis lacking CT, and craniospinal RT dose < 30 Gy. The only significant risk factor for reduced PFS in multivariate analysis was Chang’s M ≥ M2. For the significant risk factors identified in the multivariate analysis, survival outcomes were calculated (Figure 2).

Table 3.

Univariate and multivariate analysis for risk factors and outcomes (OS and PFS)

		Univariate		Multivariate
		OS OR, [95%]CI, P	PFS OR, CI%, P	OS OR, CI%, P	PFS OR, CI%, P
Factor	Comparison
Patient and disease related characteristics
Age (years) KPS Sex Time to Diagnosis CVRF Location Chang (T) Chang (M) CSF spread Spinal involvement ANOCEF EURACAN Histology Molecular	<30 vs ≥30 ≥80 vs <80 Female vs male <1 month vs ≥1 month Absent vs present Lateral vs other* T1-T3a vs ≥T3b M0-M1 vs ≥M2 No vs yes (M1) No vs yes Intermediate vs high risk Intermediate vs high risk Non LCA vs LCA SHH/WNT vs non-SHH/non-WNT SHH vs others SHH TP53wt vs SHH TP53mut	0.70, [0.35-1.41], .318 1.95[0.71-5.23], .184 1.33 [0.65-2.69], .435 0.54[0.27-1.07], .07 0.81[0.49-1.35], .416 1.67[0.83-3.39], .153 0.83[0.29-2.36], .73 3.68[1.63-8.32], .002 1.20[0.82-1.72], .362 1.22[0.80-1.85], .363 0.91[0.45-1.82], .785 1.14[0.57-2.28], .719 0.72[0.10-5.26], .743 2.01[0.99-4.07], .05 1.59[0.79-3.19], .191 0.63[0.31-1.26], .191	0.75[0.37-1.54], .44 1.08[0.40-2.90], .871 1.63[0.80-3.3], .178 0.48[0.23-1.02], .057 .716[0.45-1.13], .157 1.14[0.55-2.36], .713 4.00[1.05-15.2], .042 8.11[2.88-22.8], .000 0.96[0.66-1.40], .845 1.87[1.12-3.10], .015 1.70[0.81-3.58], .157 1.70[0.81-3.58], .157 – 1.80[0.86-3.79], .120 1.07[0.53-2.14], .857 0.94[0.47-1.85], .844	– 2.00[0.55-7.27], .288 – 0.84[0.32-2.2], .727 – 1.67[0.57-4.88], .34 – 6.05[2.00-18.27], .001 – – – – – 5.37[0.53-53], .153 0.216[0.18-2.59], .227	– – 0.96 [0.365-2.55], .943 0.86[0.338-2.18], .749 0.751[0.41-1.36], .348 – 3.30[0.431-12.28], .33 9.51[1.77-51.06], .009 – 0.68[0.132-3.50], .645 – – 1.13[0.42-3.05], .806 –
Treatment-related characteristics
Extent of resection Treatment Craniospinal RT Posterior Fossa RT	Complete vs incomplete CT/RT vs surgery and/or RT ≥30 Gy vs <30 Gy ≥50 Gy vs <50 Gy	0.95[0.456-1.99], .889 1.93[0.96-3.88], .065 1.88[0.95-3.71], .068 1.65[0.83-3.31], .155	1.79[1.09-5.59], .029 0.67[0.33-1.38], .284 2.00[0.96-4.18], .064 1.51[0.70-3.23], .289	– 5.11[1.64-15.86], 0.005 3.73[1.28-10.83], 0.015 0.78[0.296-2.06], 0.621	0.70[0.128-3.85], 0.684 – 1.85[0.669-5.125], 0.23 –

Chang M, Chang’s metastasis classification; Chang T, Chang’s tumor classification; CSF, cerebrospinal fluid; CVRF, cardiovascular risk factors; HR, Hazard Ratio; KPS, Karnofsky performance scale score; NOS, not otherwise specified; OS, overall survival; PFS, progression free survival; TTD, time to diagnosis from the beginning of symptoms.

^*Other: midline, mixed, multifocal, cerebellopontine angle and beyond posterior fossa.

Survival Outcomes (OS and PFS): Subgroup Analysis

A subgroup analysis according to age group, ANOCEF-risk scale, EURACAN-risk scale, molecular groups, and type of surgical resection were done and graphed. Older age, high-risk stratification scale by ANOCEF, EURACAN, and WNT molecular group, although without statistical significance, had a tendency for better OS (Figure 3). In our series, patients aged 26-40 years demonstrated a longer median OS (138 months) compared to those aged 18-25 years (75 months). Likewise, the youngest group (18-25 years) also had a shorter median PFS than the 26-40-year-old group. Of note, the median OS for patients older than 40 years was not reached. For ANOCEF high-risk group and EURACAN high-risk group, the mOS were 156 months (95% CI 0-322.7) and 156 months (95% CI 14.9-297), respectively and for WNT molecular group it was 184 months (95% CI 120-247). The extent of resection did not seem to influence the OS, with 159 months (95% CI 112-206) for PRS/STR and 130 months (95% CI 100-159) for CRS/GTR. On the other hand, those achieving CRS/GTR had the longest PFS of 47 months (95% CI 32.8-61.5) compared to the PRS/STR and biopsied groups of patients (P = 0.08).

Figure 3.

Kaplan–Meier Curve for OS and PFS by Subgroup Analysis

Discussion

To our knowledge, our cohort is one of the largest series of adult MB patients. We are aware that other research teams have also contributed with valuable insights into this rare disease in the adult population.^6,23–27 Regarding the characteristics of our patients, a noticeable trend consistent with the literature is the slight male gender predominance. Median age also aligns with what has been reported by others.^6,28,29 Notably, a minority, albeit a significant 11% of our patients had cardiovascular risk factors, which we deemed important to analyze in the neuro-oncology setting. Additionally, it’s worth emphasizing that all our patients initially exhibited a favorable performance scale at initial diagnosis.

Clinical features vary according to the patient’s age of presentation and location in the posterior fossa.³⁰ In our series, cerebellar syndrome and IIP were the two cardinal clinical features, consistently with the literature.¹³ The most frequent tumor location was in the cerebellar hemispheres, aligning with reports that approximately 50% of adult medulloblastomas are located laterally in the cerebellum. This predilection for hemispheric cerebellar involvement is consistent with the SHH molecular subgroup of MB, which is known to exhibit a preference for lateral cerebellar localization.^6,31,32 It is worth noting the considerable percentage of patients that were classified as high risk based on Chang’s T and M criteria (16% and 28%, respectively), with 18% showing radiographic spinal metastasis at the time of diagnosis. According to ANOCEF and EURACAN risk scales, patients were similarly distributed.

According to the WHO 2021 classification of CNS tumors, MB is now defined by an integrated diagnosis that combines histological and molecular/genetic features. This integrated approach ensures a more accurate classification and better aligns with the biological behavior of MB subtype.¹² The most frequent histological subtype was desmoplastic/nodular in almost half of the cases, which is consistent with the literature. Molecular analysis was available in 52 (58%) patients, 41(79%) corresponding to the SHH category followed by the non-WNT/non-SHH group.³³ In adults, SHH-activated TP53wt MB represents the most frequent type.³⁴ The genetic data, encompassing MYC and MYCN amplification status in 52 (58%) cases, is important. However, it is essential to emphasize that MYCN amplification was observed in just one patient, making it challenging to draw any definitive conclusions about prognosis in this subgroup.

Regarding the therapeutical approach, current clinical guidelines recommend that GTR with a second look if needed.^11,35 This was the clinical scenario in 4 of our patients (4%). Adjuvant RT is known to be part of the management of these patients. Most of our patients received treatment after the year 2000 (90%), and for 76% of them, the treatment approach involved a combination of RT/CT. The conventional doses are approximately 36 Gy in the craniospinal axis plus a boost of 18-20 Gy in the posterior fossa (PF) (total dose of 54-56 Gy). Some recommendations changed through time regarding RT as well. A possible bias is the fact that RT alone was accepted in the years 2002-2005 for the intermediate-risk group. On the other hand, in recent years, it is now acceptable for the intermediate-risk group to receive reduced craniospinal doses (23.4 Gy),³⁶ an approach which has been modified from pediatric literature in which children treated with reduced RT doses had encouraging similar event-free survivals that those treated with 36 Gy.¹⁶ Evidence today supports the superiority of combination strategies over RT alone,³⁷ and current recommendations include in addition to RT, a systemic therapy, irrespective of the risk category.¹¹ In our series craniospinal RT ≥ 30 Gy and posterior fossa RT ≥ 50 Gy were the most frequent used doses in adults.

Oncologic outcomes revealed that a high proportion of patients achieved CR or PR (84%). Notably, PFS appears markedly variable across the literature. Majd et al.²³ reported a median PFS of 79.2 months in adult population, whereas Parakh et al.²⁴ described a significantly shorter time to recurrence of 18.4 months, while our cohort demonstrated a mPFS of 30.5 months. Remarkable, some progressions occurred more than 10 years after initial diagnosis. This finding aligns with existing literature which indicates that, compared to children, adults often experience late tumor recurrences, with approximately 20% occurring after 5 years.^6,38

The optimal treatment at relapses remains controversial.³⁹ The experience observed in our series indicates a predilection for CT ± molecular targeted therapy alone, also favoring carboplatin-based regimens. Recruitment for molecular target therapies in adults might be challenging leading to closure of clinical trials. This difficulty hinders the advances for this rare population, underscoring the need for continued efforts in research and clinical trial design. Current guidelines suggest at relapse a second surgery if feasible for CR,¹¹ however in our series, only 11% underwent a second surgery. In our series, up to nine progressions/recurrences were documented, eight of these were actively treated, with a preference for employing CT as the treatment modality.

It is well known that tolerance of CT is different in adults versus children.⁷ Many CT regimens are extrapolated from the pediatric experience; in our series the carboplatin-based schemes were privileged ones followed by cisplatin-based schemes. However, it might be worth noting that cisplatin-based regimens^7,19 have achieved notable survival outcomes, highlighting the importance of continued research to determine the most effective treatment for adult MB. Reports in children treated with carboplatin show higher rates of grade 3-4 treatment-related toxicities compared with our adult patients, particularly hematologic toxicity (42% neutropenia, 21% thrombocytopenia).⁴⁰ Tolerance to cisplatin-based chemotherapies appears to be worse in adults than in children.⁴¹ Reports around 30% of neuropathy and 53% of leukopenia have been described.⁴² Our series documented 15% with cases with neuropathy, 15% with neutropenia, and 8% with lymphopenia. The lower percentages of neuropathy cases may be explained by underreporting in medical records. Informed rates of toxicities in children and adult patients receiving vismodegib and sonidegib are around 11%; which is comparable to the 14% of events in our patients,⁴³ but it is important to note that this percentage is based on only 7 patients who received these treatments, which limits the generalizability of our findings. Neuroendocrine deficits and ototoxicity although documented in some of our patients, it is not showed in our series due to lack of access to audiology and endocrinology records, thus generating a potential bias in interpretation.

Survival rates and prognostic factors for adult MB remain difficult to establish due to the scarcity and heterogeneity of studies. The mOS of our series (124.4 [95% CI 68.53-180.11]) is comparable to adult mOS of large series, with mOS of 108,⁶ 105.6,²³ and 127 months.²⁵ Interestingly Parakh et al.,²⁴ in a large contemporary Australian dataset, reported a shorter mOS of 78 months, highlighting the diversity of prognostic factors. For example, their cohort most frequently received a craniospinal RT dose of 23.4 Gy, whereas in our cohort the majority received ≥ 30 Gy. Furthermore, the Australian study included survival between relapses, data which was not assessed in our series but worth exploring, especially given that we documented up to 9 relapses.

In our cohort, significant poor prognostic factors in multivariate analysis were Chang’s ≥ M2 for OS and PFS, and the use of lower doses of craniospinal RT (<30 Gy) and lack of strategies of combination of CT/RT for poor OS. It is appreciated that there is a tendency for poorer outcomes as well in the univariate analysis for the non-WNT/non-SHH group, Chang ≥ T3b, ≥M2, spinal involvement and incomplete resection. In terms of prognosis, our research aligns with several established prognostic factors reported in the available literature, however some associations could be confirmed in our study.^23–27

For the adolescents and young adults (AYA) patients, an increased extent of disease, a poor postoperative functional status, and lower doses/volume of RT were associated with poorer OS and PFS.¹⁰ For this group of patients, complete resection was not a clear predictor of better response. However, according to another adult meta-analysis, incomplete resection, was associated with poorer prognosis. Other risk factors reported in the literature include clinical and radiological signs of brainstem infiltration and no radiotherapy. Adults with MB should be treated with RT of the craniospinal axis, with a boost to the tumor bed. Pediatric trials have established the use of craniospinal RT with a reduced total dose of 23.4 Gy. This approach is currently under assessment in prospective trials involving adults who have M0–M1 disease.^41,44,45 In our series, the patients who did not benefit from RT were mostly those who experienced disease progression or death before the start of RT (protocol platinum-RT-platinum). It has been demonstrated that patients receiving first-line chemotherapy and radiotherapy survive significantly longer than patients treated with radiation alone,⁶ our study confirmed that as well. The prognostic value of molecular subgroups is clear in children, although this may not mirror the natural history seen in AYA or adults in general. Adult SHH group is clinically heterogeneous and the patients within this group demonstrate various outcomes; on the other hand, it has been described that the WNT-driven tumors have been observed to have greater OS and PFS,³⁴ these observations are comparable to what we found in our patients.

It is interesting to point out that there was not a statistically significant OS/PFS impact regarding the age category. According to the literature, it is controversial, but has been reported that youngers have more favorable prognosis.^6,46 In our series we observed for the group 26-40 years-old longer OS compared to the group between 18 and 25-years-old, however the youngest had shorter PFS compared to the group 26-40-years-old. Although not statistically significant, we were surprised by the tendency of a longer OS, both by ANOCEF and EURACAN risk stratification scales, for the high-risk group. On the other hand, the PFS tended to be worse in the high-risk group in both scales.

The literature focusing on adult MB patients is limited and heterogeneous. Our study is one of the largest retrospective cohorts including adult MB patients. Male gender, good KPS, lateral cerebellar location, Chang’s < M2/T3b, desmoplastic/nodular histology, and SHH TP53WT represented the predominant characteristics in our series. Chang’s ≥ M2 was associated with reduced OS and PFS. The use of CT/RT and craniospinal RT doses ≥ 30 Gy were associated with better survival. Interestingly, we observed a nonsignificant trend toward better OS in patients with Chang T > T3b compared to those with T1–T3a (OR = 0.83; P = .73; 95% CI 0.29-2.36), which may appear counterintuitive. However, this result must be interpreted with caution given the small number of patients in the >T3b subgroup (n = 15) and the lack of direct imaging review, as an important number of MRIs relied on radiology reports alone. Additionally, it is worth noting that patients classified as intermediate-risk according to the ANOCEF scale showed slightly worse OS than those in the high-risk group. We hypothesize that this may be partially explained by the treatment approaches used in earlier years of the cohort (2000-2005), where some patients did not receive chemotherapy and may have been treated with lower craniospinal RT doses (eg 23.4 Gy instead of 30 Gy). These treatment variations might have influenced outcomes in this subgroup.

This study has several limitations that should be acknowledged. First, the retrospective design may introduce selection and information biases. Second, the heterogeneity of the cohort, which includes patients with varying histological subtypes, prognostic groups, and systemic therapies, may limit the generalizability of our findings. This heterogeneity is further compounded by the variability in chemotherapy regimens, with differences in the number of cycles received by patients and the specific agents administered, which may influence treatment outcomes. Third, while a significant proportion of patients received a pathological review to confirm the histomolecular diagnosis, not all patients underwent this review, which could impact the accuracy of classification. Additionally, some patients did not have MRI images directly available for review, and we had to rely on the neuroradiologist’s report for assessment, which may introduce interpretation variability. Finally, the relatively small sample size for certain subgroups may reduce the statistical power of our analysis, particularly when assessing outcomes by specific histological or molecular subtypes. These limitations should be considered when interpreting our findings, and further research with larger, more homogeneous cohorts is needed.

Questions that remain unanswered in the adult population: de-escalation or intensification of RT and the role of molecular target therapies. Better characterization of the MB adult patients, and the better understanding of prognostic factors will permit to optimize in the future the standard of care. Larger cohorts and dedicated prospective trials are needed.

Funding

No funding was received for this study.

Conflict of Interest

A.I. reports outside the submitted work: travel funding: Carthera, Leo Pharma, Novocure; research grants: Transgene, Sanofi, Servier and Nutritheragene; consulting: Novocure, Novartis, Polytone Laser, Leo Pharma, and Boehringer Ingelheim. F.B. reports funding of research from Abbvie; service agreement for research contracted between his institution and Treefrog Therapeutics as well as Owkin; personal fees from Bristol Myers-Squibb and a next-of-kin employed by Bristol Myers-Squibb, all outside the submitted work. Y.G.-R. declares having received honoraria from ABBVIE and Roche as a speaker. Rest of the authors have any of interest to declare. Ytel Garcilazo-Reyes, Pierre Dal Col, Franck Bielle, Karima Mokhtari, Loïc Feuvret, Julian Jacob, Alberto Duran-Peña, Laurent Capelle, Julien Savatovsky, Franck Bourdeaut, Julien Masliah-Planchon, Bernardo Cacho-Díaz, Khê Hoang- Xuan, Ahmed Idbaih, and Florence Laigle-Donadey

Author Contributions

Y.G.-R.—conceptualization, methodology, data collection, writing—original draft. P.D.C.—data collection, analysis—review. F.B.—supervision, writing—editing. K.M.—supervision—review. L.F.—supervision—review. J.J.—supervision—review. A.D.-P.—data collection—review. L.C.—writing. F.B.—writing—review. J.M.-P.—writing—review. B.C.-D.—data analysis—review. K.H.-X.—supervision, writing—review. A.I.—writing—visualization, review and editing. F.L.-D.—writing, visualization, review and editing—original draft. L.C. passed away before the completion of this manuscript.

Ethics Statement

This study was approved by the local Institutional Review Board (IRB), and followed the STROBE guidelines. Data were retrospectively collected and anonymized. The requirement for signed informed consent was waived due to the retrospective nature of the study.

Data Availability

The data that support the findings of this study will be made available upon reasonable request. Previous presentation of the material—European Academy of Neuro Oncology (EANO) annual meeting 2023, oral presentation.