Purpose: Aurora A kinase (AAK) plays an integral role in mitotic entry, DNA damage checkpoint recovery, and centrosome and spindle maturation. Alisertib (MLN8237) is a potent and selective AAK inhibitor. In pediatric preclinical models, antitumor activity was observed in neuroblastoma, acute lymphoblastic leukemia, and sarcoma xenografts. We conducted a phase 2 trial of alisertib in pediatric patients with refractory or recurrent solid tumors or acute leukemias (NCT01154816). Patients and Methods: Alisertib (80 mg/m 2 /dose) was administered orally, daily for 7 days every 21 days. Pharmacogenomic (PG) evaluation for polymorphisms in the AURK gene and drug metabolizing enzymes (UGT1A1 Ã 28), and plasma pharmacokinetic studies (PK) were performed. Using a 2-stage design, patients were enrolled to 12 disease strata (10 solid tumor and 2 acute leukemia). Response was assessed after cycle 1, then every other cycle. Results: A total of 139 children and adolescents (median age, 10 years) were enrolled, 137 were evaluable for response. Five objective responses were observed (2 complete responses and 3 partial responses). The most frequent toxicity was myelosuppression. The median alisertib trough concentration on day 4 was 1.3 mmol/L, exceeding the 1 mmol/L target trough concentration in 67% of patients. No correlations between PG or PK and toxicity were observed. Conclusions: Despite alisertib activity in pediatric xenograft models and cogent pharmacokinetic-pharmacodynamic relationships in preclinical models and adults, the objective response rate in children and adolescents receiving singleagent alisertib was less than 5%.

Gain-of-function mutations in the ALK oncogene occur in 15% or more of newly diagnosed patients with high-risk neuroblastoma. This discovery positioned ALK as the first tractable molecular target for patients with this disease. However, crizotinib showed limited anti-tumor activity in this phase 2 trial for patients with relapsed ALK+ neuroblastoma. The preclinical mechanism underlying this observation revealed that two of the three hot spot mutations in ALK confer intrinsic resistance to crizotinib due to preferential affinity for ATP binding that could potentially be overcome by higher drug exposures. The observed responses occurred in patients with the most common both germline and somatic hot spot mutation at residue R1275.Despite limited activity and lack of objective responses in patients harboring de novo resistant ALK mutations, we conclude that, while this was possibly a limitation of the number of patients enrolled, this is more likely due to an inability to reach the higher concentrations of crizotinib needed to overcome the competing ATP affinity. Emerging data with the third generation ALK inhibitor lorlatinib shows promise for patients with ALK-driven neuroblastoma.

PURPOSE Novel biomarkers are needed to differentiate outcomes in intermediate-risk rhabdomyosarcoma (IR RMS). We sought to evaluate strategies for identifying circulating tumor DNA (ctDNA) in IR RMS and to determine whether ctDNA detection before therapy is associated with outcome. PATIENTS AND METHODS Pretreatment serum and tumor samples were available from 124 patients with newly diagnosed IR RMS from the Children's Oncology Group biorepository, including 75 patients with fusion-negative rhabdomyosarcoma (FN-RMS) and 49 with fusion-positive rhabdomyosarcoma (FP-RMS) disease. We used ultralow passage whole-genome sequencing to detect copy number alterations and a new custom sequencing assay, Rhabdo-Seq, to detect rearrangements and single-nucleotide variants. RESULTS We found that ultralow passage whole-genome sequencing was a method applicable to ctDNA detection in all patients with FN-RMS and that ctDNA was detectable in 13 of 75 serum samples (17%). However, the use of Rhabdo-Seq in FN-RMS samples also identified single-nucleotide variants, such as MYOD1 L122R, previously associated with prognosis. Identification of pathognomonic translocations between PAX3 or PAX7 and FOXO1 by Rhabdo-Seq was the best method for measuring ctDNA in FP-RMS and detected ctDNA in 27 of 49 cases (55%). Patients with FN-RMS with detectable ctDNA at diagnosis had significantly worse outcomes than patients without detectable ctDNA (event-free survival, 33.3% v 68.9%; P = .0028; overall survival, 33.3% v 83.2%; P < .0001) as did patients with FP-RMS (event-free survival, 37% v 70%; P = .045; overall survival, 39.2% v 75%; P = .023). In multivariable analysis, ctDNA was independently associated with worse prognosis in FN-RMS but not in the smaller FP-RMS cohort. CONCLUSION Our study demonstrates that baseline ctDNA detection is feasible and is prognostic in IR RMS.

Purpose: Rhabdomyosarcoma (RMS) is an aggressive soft-tissue sarcoma which primarily occurs in children and young adults. We previously reported specific genomic alterations in RMS which strongly correlated with survival; however, predicting these mutations or high-risk disease at diagnosis remains a significant challenge. In this study, we utilized convolutional neural networks (CNNs) to learn histologic features associated with driver mutations and outcome using H&E images of RMS. Patients and Methods: Digital whole slide H&E images were collected from clinically annotated diagnostic tumor samples from n=321 RMS patients enrolled in Children’s Oncology Group (COG) trials (1998-2017). Patches were extracted and fed into deep learning CNNs to learn features associated with mutations and relative event-free survival risk. The performance of the trained models was evaluated against independent test sample data (n=136) or holdout test data. Results: The trained CNN could accurately classify alveolar RMS, a high-risk subtype associated with PAX3/7-FOXO1 fusion genes, with an ROC of 0.85 on an independent test dataset. CNN models trained on mutationally-annotated samples identified tumors with RAS pathway with a ROC of 0.67, and high-risk mutations in MYOD1 or TP53 with a ROC of 0.97 and 0.63, respectively. Remarkably, CNN models were superior in predicting event-free and overall survival compared to current molecular-clinical risk stratification. Conclusion: This study demonstrates that high-risk features, including those associated with certain mutations, can be readily identified at diagnosis using deep learning. CNNs are a powerful tool for diagnostic and prognostic prediction of rhabdomyosarcoma which will be tested in prospective COG clinical trials.

<div>AbstractPurpose:<p>Aurora A kinase (AAK) plays an integral role in mitotic entry, DNA damage checkpoint recovery, and centrosome and spindle maturation. Alisertib (MLN8237) is a potent and selective AAK inhibitor. In pediatric preclinical models, antitumor activity was observed in neuroblastoma, acute lymphoblastic leukemia, and sarcoma xenografts. We conducted a phase 2 trial of alisertib in pediatric patients with refractory or recurrent solid tumors or acute leukemias (NCT01154816).</p>Patients and Methods:<p>Alisertib (80 mg/m²/dose) was administered orally, daily for 7 days every 21 days. Pharmacogenomic (PG) evaluation for polymorphisms in the AURK gene and drug metabolizing enzymes (UGT1A1*28), and plasma pharmacokinetic studies (PK) were performed. Using a 2-stage design, patients were enrolled to 12 disease strata (10 solid tumor and 2 acute leukemia). Response was assessed after cycle 1, then every other cycle.</p>Results:<p>A total of 139 children and adolescents (median age, 10 years) were enrolled, 137 were evaluable for response. Five objective responses were observed (2 complete responses and 3 partial responses). The most frequent toxicity was myelosuppression. The median alisertib trough concentration on day 4 was 1.3 μmol/L, exceeding the 1 μmol/L target trough concentration in 67% of patients. No correlations between PG or PK and toxicity were observed.</p>Conclusions:<p>Despite alisertib activity in pediatric xenograft models and cogent pharmacokinetic-pharmacodynamic relationships in preclinical models and adults, the objective response rate in children and adolescents receiving single-agent alisertib was less than 5%.</p></div>