This retrospective study demonstrates that pediatric brain tumor patients treated with HDCT-AHSCR are most likely to have favorable outcome if patients achieved CR (either biopsy proven or by radiologic criteria, that is, minimal residual disease not excluded) at the time of AHSCR. Gross total resection, which has historically been one of the most prominent favorable prognosticators, [22
], was associated with CR in our study and independently predicted PFS. In contrast, patients with SD at the time of transplant and especially those who had progression prior to transplant had the poorest chance of survival when treated by HDCT-AHSCT. Our data also suggests that some of the patients with PR without overt progression prior to transplant still may be salvaged by HDCT-AHSCT. These findings are in concordance with multiple previous reports [8
]. Our 60% PFS and 69% OS rates are also comparable or superior with reported outcome data [6
]. These survival rates rationalize the use of this modality for pediatric brain tumor patients with poor prognosis, with estimated EFS/OS generally less or equal than 50–40% at best, if treated with nonmyeloablative therapies [1
]. In our study, it is not possible to conclude whether a portion of the included patients already were cured without the HDCT-AHSCR. International consensus meetings have concluded that the precise role of HDCT-AHSCR in pediatric brain tumors can only be determined in randomized trials.
The heterogeneity of tumor types and treatment regimens prevented the development of any strong conclusions based on the negative findings of our retrospectively studied small patient group. Particularly, subgrouping patients into embryonal tumors (compared with others) and/or further subdivision according to treatment regimens result in small numbers, which cannot be utilized for conclusive analyses. Nevertheless, only 25% (3/12) of patients with embryonal tumors experienced recurrence after AHSCR, compared to 60% (2/3) of patients with glial/mixed lesions (), an observation that concurs with well-established knowledge that medulloblastomas and PNETs have better responsiveness [7
] to this modality compared to glial tumors [6
]. In addition, we cannot exclude that more intensive induction with stronger single HDCT/AHSCR harbors some therapeutic advantage over less intensive induction and “milder” HDCT with ≥3 AHSCR for embryonal tumors, since 2 out of 3 recurrences in 12 embryonal tumors occurred among the 6 patients, who got ≥3 AHSCR. Yet, our one young patient with anaplastic ependymoma is in sustained CR 4.3 yrs post-triple tandem AHSCR regimen (no radiation), despite the literature which reports very poor outcome (5-yaer EFS 12%) for those treated with single AHSCR [20
While the aforementioned comparisons of clinical outcome between single and ≥3 AHSCR regimens in subgroups lack significant power, our toxicity data analyses are more robust (assuming that tumor type is unrelated to toxicity). There are no randomized studies between regimens with single versus ≥3 AHSCR reported in the literature. However, up to 10–20% toxic mortality has been documented in series with stronger thiotepa-based regimens and single-AHSCR [19
], whereas a larger medulloblastoma study with milder cyclophosphamide-based chemotherapy and up to 4 AHSCRs had no protocol-related deaths [12
]. Our data support these findings in that our only toxic death occurred in the single AHSCR group (1/9, 11%). In addition, regimens with ≥3 AHSCR were tolerated much better with significantly fewer toxic episodes (, ). It appears that repetitive administration of TT at doses comparable to that used in single transplant (which utilizes concomitant VP in HDCT) may decrease nonhematologic toxicity and allow administration of a greater cumulative drug dose. Whether this reduced toxicity is accompanied with decreased efficacy still remains to be seen.
The major papers on transplant in embryonal brain tumors in children are reported on the Head Start program byFangusaro et al., which is referenced above [8
]. The other approach to embryonal recurrent tumors is enhanced delivery by intra-arterial chemotherapy and osmotic blood-brain barrier disruption (IA/BBBD), as recently reviewed in Cancer by Jahnke et al. [29
]. Comparison of these two larger series with regards of efficacy demonstrates comparable disease-free and OS survival rates of around 40% and 50%, respectively, at about the 3-year mark. Our patients' outcome with an almost 60% PFS and 70% OS at 3 years may seem promising; however, many of the common critical objections apply. First, only 2 of our 12 patients with embryonal tumors had recurrent tumors; the remaining had primary tumors (). Second, our inclusion criterion for this study is the high-dose therapy/autologous rescue and not the intention to treat. Moreover, our study is nonrandomized and followup is rather short. Nevertheless, our survival curves apparently are reaching their plateau, and we anticipate ≥50% long-term survivorship. In general, small, heterogeneous, and selected study groups hamper studies in this field. According to newer data, medulloblastomas with particular histological features and molecular signatures may have a more favorable prognosis [30
], but the retrospective nature of our study does not allow us to perform these analyses.
Toxicities observed in our patients were similar to pediatric supratentorial PNET patients reported by Fangusaro et al. [8
] and for children with medulloblastoma [31
]. These mainly involved grade 3-4 hematopoietic toxicities, episodes of febrile neutropenia and, mucositis, often requiring narcotics. Transient liver transaminase elevations were also observed during chemotherapy. In contrast, in IA/BBBD study reported hematopoietic suppression and neutropenic fever to a lesser extent [29
], highlighting the advantage of less systemic toxicity by this modality. summarizes selected nonneurological toxicities of these three studies in general. In addition, regarding neurological toxicities, IA/BBBD was reportedly associated with neurotoxic episodes in approximately 30% of patients with a vast majority with reversible neurologic deficit lasting >24 to 48 hrs. This is comparable to the neurotoxicity observed for our 18 patients, a third of whom have experienced neurological adverse effects in one form or another, including grades 1 and 3 hearing loss (). In comparison, approximately 60% of patients sustained high-frequency hearing loss necessitating a reduction or elimination of cisplatin in study reported by Fangusaro et al. [8
]. Patients in the IA/BBBD study received heterogeneous sodium thiosulfate regimens, and an accurate comparison of ototoxicity was not feasible [29
Comparison of nonneurological toxicity of three studies.
A number of mechanisms by which dose-intensive chemotherapy might help include steep dose response curve of alkylating agents [1
], blood-brain-barrier penetration [33
], and overcoming intrinsic resistance of hypothetical subpopulation(s) of cells such as brain tumor initiating cells [35
]. While HDCT-AHSCR may offer these advantages, we conclude that further intensification of regimens using HDCT likely has reached the point of maximum clinical toxicity, while resulting in diminished return, necessitating the development of newer modalities to improve upon the efficacy of brain tumor therapy. Recent advances in the molecular pathology of malignant glioma and medulloblastoma provide more opportunities for targeting brain tumors [30
]. This knowledge might be used for proposed biological modifications following HDCT-AHSCR for pediatric brain tumors [38
], and pediatric brain tumor derived neurospheres may also provide an excellent predictive model for preclinical explorations for such important endeavors [39