Background and Aims

While brain metastases in children with extracranial solid tumors are still relatively rare, there seems to be an increase in incidence related to prolonged patient survival following effective treatment for the primary tumor. The outcome post brain metastasis is dismal with a median survival of 2.7 months in one series. Stereotactic radiosurgery (SRS) is a radiotherapy option for adult brain metastases that delivers focused, high doses of radiation to selected targets and has better cognitive outcomes compared to traditional whole brain radiotherapy (WBRT) as well as excellent local control rates. There is limited data on the use of SRS in children with brain metastases. We aim to assess the role of SRS in pediatric patients with brain metastases in terms of progression-free survival, complications, and relapse.

Methods

We retrospectively reviewed the records of three male patients (ages 7, 11, and 16) with extracranial solid tumors who underwent SRS for a combined 29 brain metastases at Smilow Cancer Hospital.

Results

The initial diagnosis was metastatic Wilms tumor, metastatic rhabdomyosarcoma, and Ewing sarcoma. All three patients presented with neurological symptoms and developed brain metastases after the completion of primary therapy. There was no evidence of systemic disease at the time of CNS relapse. The three patients underwent surgical resection for their largest lesions, chemotherapy, hypofractionated radiotherapy to the resection cavity and/or WBRT, and SRS. One patient received SRS once while the other two underwent SRS twice for the development of new lesions. There was no detected reoccurrence of brain metastases following the last SRS in the three patients. The patients died 9, 36, and 62 months from the diagnosis of brain metastasis. Their cancer survival times were 26, 47, and 82.5 months, respectively.

Conclusions

In pediatric patients with extracranial solid tumors who develop brain metastases, stereotactic radiosurgery provided excellent local control and prolonged overall survival.

Background and Aims

Radiation therapy is indicated in patients with high risk neuroblastoma. Since 2013, we performed comparative dose planning to choose the optimal radiation modality, proton or photon RT, for the individual child. In this study, the dose comparison was evaluated retrospectively.

Methods

We found six patients with comparative dose plans. The dose plans were generated on diagnostic routine CT scans taken after surgery, that were transferred to our treatment planning system. The clinical target volume (CTV) was delineated according to the European SIOPEN protocol. The body, both kidneys and the liver were delineated in all patients, other selected organs at risk in individual patients. Proton and photon plans for 21 Gy (RBE) in 14 fractions were generated for all patients in collaboration with the MD Anderson Cancer Center (MDACC). For this study all dose volume histograms were reviewed, and a summation of mean doses was performed for the CTV and organs at risk.

Results

The mean CTV volume was 265 cc (range 62-523 cc). The target coverage was identical for both modalities. The mean dose to the body, liver, heart, pancreas and spleen were 1.9, 0.12, 2.6, 7.3 and 2.1 Gy (RBE) in the proton plans versus 3.3, 1.5, 8.8, 12.8 and 6.4 Gy with photons. The mean doses to the ipsi- and contralateral kidney were 13.4 and 3.8 Gy (RBE) with protons versus 13.2 and 5.8 Gy with photons.

Conclusions

All patients were selected to proton therapy based on lower doses to the body, heart or liver in their individual plan comparison. The minor difference in kidney doses by the different modalities was not seen as significant for the decision. Comparative dose planning was feasible with regard to timing of radiotherapy and was seen as a useful tool for choosing the optimal radiation modality in this patient group.

Background and Aims

Pulmonary metastases are common in pediatric sarcomas; however, little is known about safety and efficacy of lung SBRT for pediatric patients. We conducted a Phase I/II study to investigate the lowest dose level of SBRT with an acceptable safety profile and efficacy in pediatric patients with metastatic sarcoma.

Methods

The study was approved by the Dana Farber Cancer Institute Institutional Review Board. Patients with sarcoma and metastatic pulmonary lesions 3 cm in diameter and 21 years of age were enrolled. Dose levels I, II, and III were 24, 30, and 36 Gy in three fractions, respectively. Enrolled patients began at dose level II. Exclusion criteria included receipt of whole-lung/hemi-thorax irradiation >12 Gy within 6 months of consent. Primary endpoints were tolerability and safety per CTCAE grading, and disease response at six weeks post-SBRT per RECIST criteria. Secondary endpoints include rates of local failure (LF) and distant failure (DF) within the lung, but outside of the treatment volume.

Results

Five patients with median age of 13 years (range: 7-21) received SBRT at dose level II. Primary tumor histologies included Ewing sarcoma (n=3), anaplastic chordoma (n=1), and osteosarcoma (n=1). No grade ≥3 adverse events were observed, one patient developed grade 2 pneumonitis. OR rate at six weeks of 7/8 (87.5%) lesions achieving partial response (PR). With median follow-up of 2.1 years (range: 1.4-2.5), one-year LF-free and DF-free survival were 75% (n=8) and 40% (n=5). One patient developed widespread metastases and succumbed to disease 1.4 years after SBRT.

Conclusions

SBRT is effective in pediatric patients with sarcoma with acceptable toxicity; however, patients remain at risk of local and distant failure within the lung. Future prospective studies are needed to investigate whether higher doses of SBRT, possibly in combination with other therapy, would be safe and provide more durable response.

Background and Aims

Compared to adults, experience with hypofractionation in children is limited and consensus on fractionation schemes is lacking. The aim of this study was to assess the potential dosimetric benefit of hypofractionated stereotactic treatment in children with orbital bone metastases.

Methods

Twelve pediatric patients (mean age 5.4 years, range 1.6-12.8 years) with 14 metastatic lesions in the orbita were included. Mean gross tumor volume (GTV) was 4.7cc (range [0.1-11.5]cc). The planning target volume (PTV) was a 2mm-expansion of the GTV. Volumetric modulated arc therapy (VMAT) plans with 2 non-coplanar arcs, according to our institution protocol, were optimized for each singular lesion (n=14) using four fractionation regimens: conventional (20 fractions) and hypofractionated schemes (3/5/10 fractions). For the purpose of this planning study a prescription dose of 36Gy EQD2_α/β=10 for all plans was used. The plans were optimized such that 95% of the PTV received at least 95% of the prescribed dose whilst respecting organs-at-risk (OARs) constraints. Allowed maximum dose in the PTV was <107% for 20fx and <140% for hypofractionated schemes. Differences in the mean dose (D_mean) of the OARs in close proximity to the PTV were evaluated for the different fractionation schemes.

Results

For all plans (n=56), the PTV coverage was satisfactory (V_95% range 95.9-100%). For the ipsilateral eye, lens and lacrimal gland, D_mean was on average 5.5, 5.0, 5.3, 6.3 Gy EQD2_α/β=3, 1.8, 1.8, 1.9, 2.4 Gy EQD2_α/β=3, and 25.1, 17.1, 16.4, 17.5 Gy EQD2_α/β=3 for 3/5/10/20 fractionation schemes, respectively.

Conclusions

For children with orbital bone metastases, clinically acceptable radiotherapy plans can be achieved with less fractions. Hypofractionated schemes using 5 and 10 fractions showed a better dosimetric sparing on the OARs D_mean compared to the 3-fractions hypofractionated and the conventional 20-fractions scheme.

Background and Aims

Brachytherapy (BT) delivers highly conformal radiation with sparing of surrounding tissues, so it may limit late effects in children, adolescents and young adults (AYA). We aimed to characterize BT trends, specifically in rhabdomyosarcoma (RMS) patients, in the US.

Methods

The National Cancer Database (NCDB) was queried to identify all pediatric and AYA patients (<22yr) who were treated for solid tumor malignancies in the US from 2004-2016. We extracted data for patients with RMS treated with BT.

Results

99,506 pediatric and AYA patients in the NCDB were treated for solid tumor malignancies from 2004-2016. Of these, 22,586 (23%) received any radiation and 240 (1%) received BT. The most common histology receiving BT was sarcoma in 74 patients. BT was the sole radiation modality in 186 (79%), and BT types included high-dose rate (HDR) interstitial (42), HDR intracavitary (34), low-dose rate (LDR) interstitial (25), LDR intracavitary (12), and BT NOS (70). BT was used in combination with EBRT in 22%.

3,836 patients were treated for RMS and 2,531 received radiation. 37 (1%) were treated with BT (EBRT+BT in 3, BT alone in 34). Disease sites treated with BT included soft tissue (45%), vagina/uterus (32%), retroperitoneum (11%), and other (12%). All 38 patients with RMS treated with BT received multiagent chemotherapy and 29 (76%) underwent surgery. RMS subtypes were embryonal (71%) and alveolar (29%). Survival data were available for 33 patients with RMS treated with BT. With a median follow-up of 9.5yr, 89% were alive. No patient underwent surgery after BT, suggesting that surgical salvage after local failure was not required.

Conclusions

Despite its theoretical advantages, BT is rarely used to treat children and AYA in the US. Patients treated with BT for RMS experienced favorable survival, suggesting that this approach may not compromise oncologic outcomes. BT as a therapeutic option warrants further study in RMS.

Background and Aims

Pencil-beam-scanning proton-therapy (PBS) is becoming an increasingly used radiation modality as it reduces unnecessary dose to surrounding organs. However, PBS dosimetry is extremely sensitive to density changes particularly within the bowel. This study aims to quantify the dosimetric consequences of gastrointestinal density change and identify predictive characteristics in patients with abdominal neuroblastoma.

Methods

Twenty cases were double-planned with PBS and Intensity-Modulated-Arc-Therapy (IMAT) to 21Gy in 14 fractions. Cases were divided into unilateral (n=9) and midline (n=11) locations. Plans were recalculated after the gastrointestinal volume was simulated as air (Hounsfield Unit (HU) -700) and water (HU 0), then compared to original plans (recalculated-original, ∆D).

Results

Increased gastrointestinal air caused significant dose deterioration to PBS compared with IMAT plans for midline tumours (median/maximum CTV ∆D95% -2.4%/-15.7% PBS vs. 1.4%/0% IMAT, p=0.003). Minimal effect was seen for unilateral tumours (∆D95% CTV -0.5%/-1.9% PBS vs. 0.5%/-0.5% IMAT, p=0.008). D95% CTV was significantly decreased in PBS plans when PTV volume ≥400cm³ (median -4.1% ≥400cm³ vs. -0.2% <400cm³; p=0.001) and PTV extension ≥60% anterior to vertebral body (median -2.1% ≥60% vs. 0.0% <60%; p=0.002). ∆D2% was higher for IMAT compared to PBS. The median ∆D2% CTV-RA_PBS/PTV_IMAT for midline tumours was 3.3% IMAT vs. 0.5% PBS (p=0.037) whilst unilateral tumours was 1.4% IMAT vs. 0% PBS (p=0.038). Spinal canal ∆D0.1cc was significantly higher in IMAT plans only for unilateral tumours (median 2.4% IMAT vs. 0.0% PBS; 0.012). Minimal dosimetric effects were seen when gastrointestinal volume was simulated as water equivalent density.

Conclusions

Unilateral tumours have minimal dose degradation on PBS plans and increased hotspots on IMAT plans when gastrointestinal air is increased and highlights the suitability of treating tumours at this location with PBS. Tumour location, PTV volume and anterior extension of PTV are easily identifiable characteristics which can predict the dosimetric impact of gastrointestinal density changes on PBS plans.

Background and Aims

Head and neck rhabdomyosarcoma (RMS) often requires adaptive planning due to changes in tumor volume during proton therapy. To facilitate timely adaptive therapy, we determine whether MR-only proton therapy planning is feasible by introducing a transfer learning method to convert head and neck MRI to synthetic CT (sCT) for dose calculation and to specifically address the challenge of a small training dataset, commonly associated with pediatric studies.

Methods

Our transfer learning model, the transfer learning cycle-Generative Adversarial Network, was designed to transfer knowledge gained from converting a large number (n=125) of pediatric brain MRI studies to sCT and finetuning the well-trained model on head and neck data. Twenty-two patients (aged 2-22 years, 13 males) who received proton therapy to the head and neck were divided into age-symmetric training and testing groups. sCT generated from T1- or T2-weighted fat-suppressed (T1WFS and T2WFS) MR images was compared to the real CT in terms of peak signal-to-noise ratio (PSNR), structural similarity (SSIM) index, mean error (ME), and mean absolute error (MAE) in Hounsfield unit (HU). Dose accuracy was calculated with 3D gamma analysis (2%/2mm criteria) and DVH comparison.

Results

Each sCT generation took 45 seconds on a high-performance GPU cluster. The mean ± standard deviation of two-fold cross validation in PSNR, SSIM, ME and MAE were 27.1±1.9, 0.88±0.04, 20.1±33.0 HU, and 169.6±34.7 HU for T1WFS, respectively; 29.3±1.9, 0.91±0.03, -2.1±17.1 HU, and 158.0±30.1 HU for T2WFS, respectively. Since T2WFS produced more accurate sCT images, 5 patients with T2WFS-based sCT were randomly selected for dosimetric analysis. 3D gamma passing rate was 97.4%±1.1%. Deviations in CTV V95 and D95 were 0.75%±0.57% and 0.35%±0.21%, respectively.

Conclusions

Transfer learning facilitates MR-only pediatric head and neck proton therapy planning by generating a highly accurate synthetic CT, even when challenged by a small training dataset (n=11).

Background and Aims

The need to assess cystic changes in craniopharyngioma patients during radiation therapy is well recognized. However, the optimal timing and frequency are not yet established. We report the incidence of adaptive replanning due to significant tumor changes, the extent of change, and the efficacy of different surveillance imaging schemes for pediatric craniopharyngioma patients treated with proton therapy following limited surgery.

Methods

Between 2016 and 2019, 73 pediatric craniopharyngioma patients underwent weekly MRI during 6 weeks of 54-Gy (RBE) pencil-beam scanning proton therapy planned with a 3-mm CTV margin and robust optimization (3 mm, 3%). Simulation and weekly MRI included T1-, T2-weighted, and FLAIR 3D sequences. The surveillance MRI scheme is defined as adequate here if the significant tumor change (≥3 mm) can be detected within a week of occurrence.

Results

Twenty nine (40%) of 73 patients required 1 to 3 replanning (1 in 20, 2 in 8, 3 in 1) for significant GTV changes (range, 3-16 mm; median, 5.25 mm). Females had higher incidence of replanning than males, 54% vs. 26%. In the worst-case scenario, parts of the new GTV could receive <70% of the prescribed dose if the change was undetected. First replanning triggered by on-treatment MRI in week 1 to 6 occurred in 12, 5, 7, 0, 3, and 2 of 29 patients, respectively. MRI schemes of first week only, combined week 1 and mid-course scans, and every other week were adequate in 77%, 86%, and 90% of 73 patients, respectively.

Conclusions

To minimize the risk of tumor underdosage, weekly on-treatment MRI is recommended for proton therapy of pediatric craniopharyngioma. When this is not feasible, every other week or the combined week 1 and mid-course MRI could be adequate for the majority of the patients.

Background and Aims

Childhood cancer survivors (CCS) may have impaired health-related quality of life (HRQoL). Proton radiotherapy (PRT) might improve outcomes among CCS. We report mature HRQoL of CCS treated with PRT.

Methods

Pediatric, adolescent and young adult patients (<25 years-old) treated with PRT from 5/2004 to 2/2019 were prospectively assessed using both child self (CSR) and parent-proxy report (PPR) PedsQL 4.0 Generic Core Scales at baseline and annually thereafter. Associations between patient characteristics and HRQoL scores at follow-up were evaluated with paired t-tests for scores and published means from a healthy child population.

Results

From 506 patients, 361 had central nervous system (CNS), and 145 non-CNS tumors. Median age was 9.2 years (range, 0.8-24.0) and follow-up 6.2 years (range, 0.9-14.6). The most common diseases were medulloblastoma (n=77) and ependymoma (n=64). At baseline the PPR mean total core (TCS), physical and psychosocial summary scores were 60.2, 58.3, and 61.7, increasing over time, respectively, to 74.6, 77.8, and 73.5 (p<0.0001x3). CNS tumor PPR showed lower baseline (mean 58.8 vs 63.2; p=0.1), and significantly lower follow-up TCS (mean 71.9 vs 80.7; p=0.0001) than non-CNS PPR, respectively, 10.4 and 1.6 points lower than healthy children’s TCS of 82.3. PPR after craniospinal irradiation (CSI) (vs local-field) presented lower baseline TCS (mean 48.1 vs 63.8; p<0.0001), yet reaching similar scores at follow-up (71.7 vs 71.9, p=0.90). CSR correlated with PPR, albeit with slightly higher scores. Age, gender, and chemotherapy exposure did not correlate with HRQoL scores at follow-up.

Conclusions

HRQoL scores increased over time in childhood cancer survivors treated with PRT. Survivors of CNS tumors fare worse than non-CNS tumors. Surprisingly, among CNS survivors, CSI exposure did not correlate with impaired HRQoL compared with local field radiotherapy at last follow up. Encouragingly, childhood cancer survivors irradiated with protons for non-CNS tumors reported scores approaching those of the healthy control group.