Atypical teratoid rhabdoid tumor is a rare, embryonal, central nervous system tumor seen predominantly in infancy and childhood. Outcomes are generally dismal, with median survival estimated at 6 months to a year. The purpose of this study was to evaluate proton beam therapy (PBT) outcomes in this population.
Sixteen patients with a diagnosis of atypical teratoid rhabdoid tumor were treated from November 2007 to January 2013 at the Indiana University Health Proton Therapy Center. All patients were treated with PBT. Fraction sizes of 1.8 Gy/fraction were used to deliver 28 to 33 fractions. Seven patients received craniospinal PBT. There were 12 male and 4 female patients. The median age at diagnosis was 18.5 months (range, 5 months to 39 years). Eight had metastatic disease at diagnosis. Fourteen patients underwent surgery. Fifteen patients received chemotherapy.
Median survival follow-up time was 3.18 years (or 38.2 months). The mean overall survival was estimated at 5.6 years (or 67.2 months; 95% confidence interval, 4.4–6.8 years). Patients of a median age of 18.5 months (range, 5.0–468 months) received a median radiation dose to the tumor bed of 54 Gy (range, 48.6–59.4 Gy). Eight patients also received craniospinal irradiation with a median dose of 36 Gy. Eleven patients showed no evidence of disease or stable disease, 3 patients were deceased, and 2 patients developed disease progression. Four patients suffered nausea and vomiting (common toxicity criteria [CTC] grade 2) as a result of treatment, and 4 patients also suffered moderate skin erythema (CTC grade 2). Two patients suffered from both weight loss and general fatigue during treatment.
The PBT was well tolerated in this heavily treated population. In the background of poor survival, these early outcome data are promising. Additional follow-up is necessary.
Atypical teratoid/rhabdoid tumor (ATRT) is a rare, embryonic tumor of the central nervous system seen predominantly in infancy and childhood . Its occurrence in adults is especially rare, and review of the literature reveals only a few case reports [2–4]. Recent data suggest that much like medulloblastoma, multiple genetic subtypes are likely to exist within ATRT. Together, all the ATRT subtypes account for approximately 1% to 2% of pediatric brain tumors . However, in patients under the age of 3 years, ATRT accounts for around 20% of malignant brain tumors .
Outcomes from this rare tumor are generally dismal, with median survival estimated at 6 months to a year [7, 8]. Prognostic factors include age at diagnosis and presence of germline mutations . The tumor suppressor INI1 (hSNF5) found at 22q11.2 is frequently mutated in patients with ATRT, making it a risk factor and useful in diagnosis [6, 8].
Management is usually multimodal with surgical resection, chemotherapy, and radiation therapy. The use of radiation therapy in patients under the age of 3 years has been avoided in the past because of concerns about long-term neurocognitive damage. Outcomes appear to be better in older patients, where radiation therapy tends to be a more-significant part of the multimodal approach to treatment.
In this single-center study of proton beam therapy (PBT) in patients with central nervous system ATRT, we report the outcomes of 16 patients treated during a period of 5 years at the Indiana University Health Proton Therapy Center (IUHPTC).
Materials and Methods
Sixteen patients with a diagnosis of ATRT were treated from November 2007 to January 2013 at IUHPTC. All 16 patients were included in the study. Institutional review board approval was obtained before patient records were reviewed. Patient names and medical charts were kept confidential.
Fourteen of the 16 patients had surgery for maximal resection. Fifteen patients received chemotherapy. Six patients received chemotherapy according to Children's Oncology Group (COG) ACNS0333 protocol. This consists of 5 rounds of chemotherapy. Two induction cycles are administered, consisting of cisplatin, cyclophosphamide, etoposide, methotrexate, and vincristine. Peripheral-blood stem cells are harvested to provide support for the following 3 rounds of treatment. Proton therapy follows the 2 induction rounds, as specified. Following radiation, 3 consolidation cycles were administered consisting of carboplatin, thiotepa, and stem cell rescue. Patients who presented with metastatic disease at diagnosis received all 5 induction and consolidation rounds before radiation treatment.
All patients were treated with image-guided 3D conformal proton therapy via uniform active scanning [10, 11]. Fraction sizes of 1.8 Gy/fraction were used to deliver between 28 and 33 fractions. Seven patients received craniospinal irradiation (CSI) to a dose between 23.4 to 36 Gy.
A relatively uniform method for treatment simulation was employed for this patient population. In all cases, 1-mm slices were obtained during computed tomography simulation, which was performed in the treatment position with the patient immobilized on an alpha cradle with a facemask. The magnetic resonance imaging (MRI) was fused. The radiation oncologist defined target volumes using the Eclipse (version 11; Varian Medical Systems, Palo Alto, California) treatment planning system. The range compensators and spread-out Bragg peaks incorporated smearing to accommodate any setup uncertainties per routine as part of the treatment planning process.
Custom-made, brass apertures and Lucite compensators were used for each field of radiation therapy. Two or more beams were used to treat the brain. The CSI was delivered using 2 posterior oblique beams to treat the brain and 2 posteroanterior spinal fields , and CSI treatments were feathered daily with 3 or 4 junctions. For all CSI treatments, 30-cm snout sizes were required for the spinal fields, and that was matched at the anterior spinal canal in all cases. Minimal dose at the junctions was 90% in all cases.
Most patients were treated on the active COG protocol at the time. The patients that were not treated on protocol were treated with 5- to 10-mm clinical target volume margins and 2- to 3-mm planning target volume margins, plus smearing based on specific beam tissue pathways and immobilization methodologies per the judgment of the treating physician. For example, beams traversing sinus cavities were avoided, and those passing bone anatomy that could have variable depth based on small rotational variations were given larger degrees of smearing. The brainstem dose was kept to a point dose maximum of 59.4 Gy during any single course of therapy. The beams were allowed to stop in critical structures via the use of our in-house, published range-modulation technique . Patient setup was verified on a field-by-field basis every fraction by a physician before the beam was delivered. Most of these patients had daily general anesthesia.
We report the median radiographic follow-up time, which is the median time between radiation therapy (RT) completion and the last MRI on follow-up. The corresponding radiographic survival status is the disease progression based on the MRI (Tables 1 and 2). Progression-free survival (PFS) was used to evaluate these parameters (Figure 1). The PFS for patients 7 and 10 was measured as the interval time between treatments because reirradiation was initiated within 4 weeks of recurrence (Table 3). Each patient was censored at the time of his or her last radiographic scan on follow-up. Quartiles were computed for the 75% survival interval using SPSS Version 22 (IBM, Armonk, New York) for Macintosh (Apple, Cupertino, California). Actuarial PFS could not be calculated because of the small radiographic follow-up time.
Median survival follow-up time is the median time between RT completion and a review of public health records to determine survival (Figure 2). Failure was defined as death from disease. The data were censored using the patient's last survival status based on review of public health records, the medical record, and direct communication by the families to our center. Families contacted the treating physician in all cases of death in this series.
The median survival follow-up was also used to compare the survival of patients presenting with metastatic disease to those without it. The factors were compared using SPSS software, significance was calculated using Log rank with data pooled over strata, and a third Kaplan-Meier curve was generated (Figure 3). Mean survival time for both groups is reported because the median overall survival (OS) between factors could not be calculated.
The IUHPTC was a quaternary referral center that received patients mainly from across the Midwest. Radiographic follow-up was setup outside of our department. Because of the center's recent closure, we could not extend the follow-up time to generate a median OS or actuarial PFS. The median survival follow-up time allowed extension of the follow-up for survival status only.
Sixteen patients with a diagnosis of ATRT were treated with proton therapy at the IUHPTC between November 2007 and January 2013.
There were 12 male and 4 female patients. The median age at the beginning of RT was 18.5 months (or 1.54 years), with a range of 5 months to 39 years. Five (31.5%) of these patients had metastatic disease at diagnosis. Of the 14 patients whose pathology was tested for INI-1 protein, 11 had the INI-1 mutation. Three had uncertain histologies, where the diagnosis remained ATRT and scant tissue. Two patients were not tested for the INI-1 mutation and the date of diagnosis may have made testing less certain for older cases . Testing was performed at the referring institutions if from out of state. Some patients from out of the country had central review in their country's preeminent center. Patients in our center had review either at Cincinnati Children's Hospital or at Johns Hopkins Hospital, if we were in doubt of the diagnosis and had access to the tissue.
Fourteen patients underwent surgery, 6 (35%) had subtotal resection and 8 (47%) had gross-total resections. Two patients (11%), both with metastatic disease at diagnosis, did not undergo surgery. One of these patients received a cranial dose of 54 Gy. The second patient received a palliative dose (30.6 Gy) to the brain and a boost of 18 Gy to the spine. All patients started radiation within 8 weeks of tissue diagnosis. Staging and treatment were per initial staging, and no progression was noted from diagnosis to simulation in our cohort.
Fifteen patients received chemotherapy. Six of those patients were treated according to the COG protocol, and the rest received the chemotherapy regimens as shown in Tables 1 and 2. One patient received chemotherapy concomitantly with radiation (patient 4 received vincristine during RT).
The median radiation dose to the tumor bed for all patients was 54 Gy. In addition, 8 patients received CSI irradiation with a median dose of 36 Gy. Of these, 4 were patients with metastatic disease. Patient 3 received both XRT and proton therapy for a combined dose of 50.4 Gy to the tumor bed (25.2 Gy from protons), and 46.8 Gy to a cervical spine metastasis. Patients 3, 14, and 15 also received a boost to spinal metastases to a dose of 46.8, 45, and 46.8 Gy, respectively. Patients 7 and 10 were treated twice. For the retreated patients, Table 3 shows the doses for treatments 1 and 2 and the interval time between treatments. Patient 7 received 59.4 Gy to the posterior fossa 18.4 months before primary recurrence and reirradition. Patient 10 received 50.4 Gy to the posterior fossa 8 months before reirradition for local recurrence and leptomeningeal spread.
The median follow-up time after completion of radiation was 3.18 years (or 38.2 months), at which point 13 of 16 patients (81%) were alive (Figure 1). Estimated from the curve, the mean OS was 5.6 years (or 67.2 months; 95% confidence interval [CI], 4.4–6.8 years). The median radiographic follow-up time was 1.38 years (or 16.6 months; range, 0.0–4.16 years); at which point, 11 patients had no evidence of disease or stable disease, 1 patient was suffering from distant relapse, 3 patients were dead from disease, and 1 patient developed a secondary acute myeloid leukemia (AML). The crude PFS of 75% was at a follow-up time of 1.42 years (Figure 2), just exceeding the median radiographic follow-up.
There were 6 ATRT failures. Two failures were marginal, and the patients were retreated (patients 7 and 10). Patient 7 had a subtotal resection, chemotherapy, and received 54 Gy with a boost to 59.4 Gy at 1.8 Gy per fraction. The patient developed recurrent disease 1.6 years later and was retreated with 36 Gy CSI with a boost to 59.4 Gy to the tumor bed. Patient 7 suffered tumor growth in the frontal lobe and base of the skull and died in hospice 8 months after retreatment from disease. Her plans are shown in Figure 3. Patient 10 was stable 9.6 months after retreatment and had similar dosing to patient 7, including salvage via craniospinal radiation to 36 Gy.
Four ATRT failures were widespread, one of which developed lung metastases and caused early withdrawal of treatment (patient 15). Patient 15 did not undergo surgery, but did receive chemotherapy and a palliative proton radiation dose of 30.6 Gy to the tumor bed, 48.6 Gy to the cervical spine, and 30.6 Gy CSI. Three patients developed distant relapse following RT. Patient 8 had a gross total resection, chemotherapy, and 50.4 Gy to the tumor bed with 36 Gy CSI. The patient developed metastasis and died of disease 5 months after RT, although the location of metastasis was not available to us. Patient 9 developed metastasis to the medulla 2.2 years following multimodal treatment and is currently undergoing further chemotherapy. Patient 11 was metastatic at diagnosis and developed secondary thecal nerve-root metastases following multimodal treatment but became stable following an additional 8 rounds of chemotherapy. The patient's MRI showed stable disease on the most recent follow-up. Of the 6 ATRT failures, 3 died of disease, 2 were stable on last follow-up, and 1 patient was still suffering from distant relapse. Patient 4 developed secondary AML 19 months after treatment with 54 Gy to brain and 23.4 Gy CSI.
The most common radiation-associated toxicities were nausea, vomiting, and skin erythema. Four patients suffered nausea and vomiting (CTC grade 2) as a result of treatment, and 4 patients suffered moderate skin erythema (CTC grade 2). Two patients suffered from both weight loss and general fatigue during treatment. Patient 4 had significant hearing difficulty in the left ear because of cochlear damage that was unavoidable because of tumor involvement adjacent to the cochlea.
Hematologic toxicity was managed either by the referring teams or our medical oncology team. In no case, did hematologic toxicity cause treatment delay, bleeding, or febrile neutropenia. We do not have complete access to those data because we were a quaternary referral center.
No radiation necrosis was seen in our patient population, but radiation-associated change was seen in some cases prompting a short course (< 4 weeks) of steroid treatment.
Although ATRT remains a dismal diagnosis, increasing evidence demonstrates promising outcomes for patients with ATRT treated with proton therapy. Our results are comparable to survival outcomes reported in the literature by groups treating with both proton and photon irradiation.
There are 3 other studies to our knowledge that look at ATRT treatment using proton therapy. Bernstein et al  reported 9 patients (median age, 2.3 years) without evidence of primary disease at a median follow-up time of 27.3 months. Only 2 of their patients suffered distant relapse, and only 1 died of disease.
Weber et al  reported on 15 children (median age, 1.45 ± 0.58 years) at a median follow-up of 33.4 months. Local failure was seen in 20% of patients. Distant failure was divided up into distant brain failure at 27% and distant spine failure at 13%. Median OS was 64.6% in their series. Six patients died from tumor progression in their series, and they did not find an association between young age and outcome.
McGovern et al  reported 31 patients (median age, 2 years) with a median follow-up time of 24 months. Radiation (planned course) completion was seen in 87% of their cohort. Progression was seen in 16% of their patients. Thirteen (42%) were known to have died. The PFS2 after radiation therapy was 45.9%, and the OS2 from the end of radiation was 53%. Similarly, at a median follow-up of 3.18 years (or 38.2 months), 81% of our patients were alive. All but one (94%) of our patients completed their course, and 2 completed full retreatment courses of at least 54 Gy. Our survival rates are in line with survival rates predicted by Chi et al , who described an estimated future OS2 rate of 70% ± 10%.
Although radiation therapy to young children is still debated, the literature reports increased OS rates in the treatment of ATRT, evidence that the benefits may far outweigh the risks. Buscariollo et al  looked at 144 patients in Surveillance, Epidemiology, and End Results (SEER) program, National Cancer Institute, Bethesda, Maryland) with ATRT from 1973 to 2008. Patients that did not receive RT had a median OS of 6 months compared with a median OS that was not yet reached at 34 months for those who received RT. This was shown to be a significant predictor of survival in a multivariate analysis (hazard ratio = 0.10; 95% CI, 0.01-0.73; P = 0.02). These data suggest RT has an important role in the multimodal approach to treatment to drastically increase survival times.
There is some evidence that survival rates are affected by the age of the patient receiving RT. A study by Bishop et al  reported on the SEER database with submissions of primary brain tumors in infants (< 12 months old) from 1973 to 2006. Infants (< 12 months) with ATRT had the lowest survival rate. Those infants treated with surgery alone had a 12-month survival of 24.4% . However, the 12-month survival increased to 100% with immediate adjuvant radiation therapy. We report a similar rate in this age group. Although only 4 of our patients were < 12-months old, all 4 patients treated with proton radiation show no evidence of disease or stable disease with a radiographic median follow-up time of 2.35 years (or 28.2 months), and a median survival follow-up of 3.89 years (or 46.7 months). Indeed, our data suggest that the youngest of the patients with ATRT may gain the most from early, adjuvant radiation therapy.
Some data suggests early metastatic disease may be the strongest predictor of failure. Buscariollo et al  reported a significant difference in median OS between patients who presented with metastatic disease at diagnosis (13% of the cohort; median OS, 3 months) and those who had focal disease (median OS, 6 months). Multivariate analysis showed metastatic disease present at diagnosis to be another independent factor of survival status (hazard ratio 2.83; 95% CI, 1.53-5.23; P = 0.001). These findings are consistent with other studies in the literature that show metastatic progression at presentation to be an independent risk factor [19, 20]. Metastatic disease at presentation in patient cohorts with ATRT is typically 13% to 33% [6, 8, 17, 21, 22]. At 31.5%, our patients were on the high end of that range. Of the 5 patients with metastasis in our cohort, 2 are dead of disease, 1 developed distant metastasis and treatment is unknown at the time of writing, and 2 have stable scans. The median survival follow-up times were compared between groups (Figure 3; P = 0.098). The median OS rate could not be calculated; however, the current median survival follow-up time of the 5 patients alone is 2.11 years (or 25.3 months), which exceeds some of the previous averages of ATRT survival. In addition, the mean OS for the group with metastases at diagnosis was estimated at 2.96 years (95% CI, 1.00–4.92 years), and the mean OS for the nonmetastatic group was estimated at 6.27 years (95% CI, 5.38-7.16 years).
We have one unusual case of an older individual with confirmed histology having long survival for metastatic disease without CSI. It is unclear whether adults have unusual survival in this disease or if they respond differently to systemic therapy. Our series used uniform active scanning techniques in all cases and our in-house method of modulating the end of the beam's increased biologic dose . No patients in our cohort needed either hyperbaric oxygen or bevacizumab for radiation necrosis, even in the case of retreatment.
The major limitations in this study are the short follow-up time and the size of the patient population studied. Because of the impending closure of IUHPTC, we could not extend the follow-up time. As it was, sufficient patients had not passed away to calculate more-optimal statistics. Hopefully, the current COG study will yield new insights because of the larger sample size. Lastly, because IUHPTC is a quaternary referral center, our study may have referral bias. More-palliative cases of ATRT in the referral region may have stayed at their home institution.
We present promising results as the fourth study looking at proton irradiation in ATRT. It will be interesting to see whether proton beam therapy can be evaluated on the COG study or within the pediatric proton consortium registry  as late effect and further genetic data evolve. These data suggest that PBT may be a good option for young children with ATRT.
ADDITIONAL INFORMATION AND DECLARATIONS
Conflicts of interest: The authors have no conflicts to disclose.
Acknowledgments: We wish to thank Stephanie Dickenson at the Indiana Statistical Counseling Center (Bloomington).