Evaluation of age-dependent treatment strategies for children and young adults with pineoblastoma: analysis of pooled European Society for Paediatric Oncology (SIOP-E) and US Head Start data

Abstract

Background: Pineoblastoma is a rare pineal region brain tumor. Treatment strategies have reflected those for other malignant embryonal brain tumors.

Patients and methods: Original prospective treatment and outcome data from international trial groups were pooled. Cox regression models were developed considering treatment elements as time-dependent covariates.

Results: Data on 135 patients with pineoblastoma aged 0.01-20.7 (median 4.9) years were analyzed. Median observation time was 7.3 years. Favorable prognostic factors were age ≥4 years (hazard ratio [HR] for progression-free survival [PFS] 0.270, P < .001) and administration of radiotherapy (HR for PFS 0.282, P < .001). Metastatic disease (HR for PFS 2.015, P = .006), but not postoperative residual tumor, was associated with unfavorable prognosis. In 57 patients <4 years old, 5-year PFS/overall survival (OS) were 11 ± 4%/12 ± 4%. Two patients survived after chemotherapy only, while 3 of 16 treated with craniospinal irradiation (CSI) with boost, and 3 of 5 treated with high-dose chemotherapy (HDCT) and local radiotherapy survived. In 78 patients aged ≥4 years, PFS/OS were 72 ± 7%/73 ± 7% for patients without metastases, and 50 ± 10%/55 ± 10% with metastases. Seventy-three patients received radiotherapy (48 conventionally fractionated CSI, median dose 35.0 [18.0-45.0] Gy, 19 hyperfractionated CSI, 6 local radiotherapy), with (n = 68) or without (n = 6) chemotherapy. The treatment sequence had no impact; application of HDCT had weak impact on survival in older patients.

Conclusion: Survival is poor in young children treated without radiotherapy. In these patients, combination of HDCT and local radiotherapy may warrant further evaluation in the absence of more specific or targeted treatments. CSI combined with chemotherapy is effective for older non-metastatic patients.

Keywords: high-dose chemotherapy; pediatric; pineoblastoma; radiotherapy; treatment.

Fig. 1

Therapy courses for (A) 57 patients <4 years old at diagnosis and (B) 78 patients ≥4 years old at diagnosis. CR: complete remission; CTX: chemotherapy; locRT: local radiotherapy; LFU: lost to follow-up; NA: not available; PD: progressive disease, Tx: therapy; PR: partial remission; SD: stable disease; RT: radiotherapy; y: years

Fig. 2

Forest plot of Cox proportional hazard models of the entire group (A) or age-defined subgroups (B, C) for both PFS (black boxes/first line) and OS (gray boxes/second, dashed line). Therapy elements were modeled as time-varying covariates. (A) Selected model by variable selection algorithm in 130 patients with complete data for staging, start of HDCT and start of radiotherapy. Gender was only selected in the model for PFS, but not in the model for OS. HDCT or postoperative, pre-RT chemotherapy were not selected as influencing factor for PFS and OS models by the variable selection algorithm. (B) Cox proportional hazard models in patients <4 years old at diagnosis who started therapy with chemotherapy and with complete data on chemotherapy strategy, staging, starting dates of HDCT and radiotherapy. The model contains most important risk factors as defined in (A) plus type of chemotherapy (intensified, aiming at HDCT versus conventional). (C) Cox proportional hazard model in patients ≥4 years old at diagnosis with available data on staging and starting dates of HDCT and radiotherapy. The model contains most important risk factors identified in (A) plus HDCT. In each age group (<4 years and ≥4 years), the model is being presented with the best model fit (lowest Akaike information criterion) of the models described within the text.

Fig. 3

PFS (A) and OS (B) according to age. Univariate 5-year PFS/OS estimates were 10 ± 4%/12 ± 4% for patients <4 years old at diagnosis versus 63 ± 6% / 66 ± 6% in patients ≥4 at diagnosis.