Genomic Profiling Program for High-Risk Pediatric Cancer Patients Makes Clinical Impact

A tumor genomic profiling program has shown feasibility and clinical impact for pediatric patients who are high-risk de novo and relapsed/refractory.

The use of a novel scoring system to quantitatively assess the effect of genomic profiling was shown to be feasible and clinically impactful across a wide range of settings among patients with pediatric cancers.1 The scoring system incorporated whole-exome sequencing (WES) of both tumor and germline DNA and transcriptome sequencing in combination with treatment recommendations offered by a molecular tumor board.

Somatic tumor alteration were foundin 121/127 (95.3%) tumor samples. Investigators reported cancer predisposition syndromes in cancer predisposition genes in 9 of 126 patients (7.1%).

Investigators conducted a prospective precision medicine trial, using whole-exome sequencing of tumor and germline tissue and whole-transcriptome sequencing of tumor tissue, to observe and identify the mutational landscape of 127 tumors from 126 patients. Patients were enrolled with any relapsed, refractory, or newly diagnosed high-risk leukemia, lymphoma, solid tumors, or brain tumors. Tumor genomic sequencing and germline sequencing were also conducted. A scoring system was developed to measure the impact of tumor and germline sequencing, encompassing therapeutically relevant genomic alterations, cancer-related germline findings, recommendations for treatment, and refinement of risk stratification or prognosis. Each patient case was assigned a clinical impact score based on these parameters and with each measure weighted from 0.5 to 1.0 points. The maximum possible score was 6.

Three of the 129 enrolled patients were not evaluable. Investigators noted that 1 patient had 2 de novo solid tumors sequenced, which resulted in successful sequencing in 127 samples from 126 patients.

Somatic Mutation Landscape

Investigators found reportable somatic alterations in 95.3% (121/127) of samples. Of these, 61 (50.4%) were from de novo patients and 60 (49.6%) from patients with relapsed/refractory disease. Low tumor mutational burden existed in95.3% of samples. In 7 of the 121 samples (5.8%), the only somatic alterations were chromosome gains/losses and/or chromosome loss of heterozygosity.

Reportable single-nucleotide variants or indels were found in 89of the 127 tumor samples (70.1%). Therapeutically targetable alterations (TTAs) were found in 69/127 samples (54.3%), with CDKN2A and CDKN2B deletions being most common across all tumor groups.Common TTAs in genes involved in epigenetic regulation included H3F3A, EZH2, and SETD2 mutations, as well as deletions of ATRX. The most common were H3F3A missense mutations which were unique to brain tumors. The most common potentially targetable genomic aberrations across all tumors clustered into 7 biologic pathways: cell cycle, DNA damage repair, epigenetic, metabolic, PI3K, RTK/RAS, and transcription

Additional significant alterations (ASAs) were found in 55 of the 127 samples (43.3%). TP53 was the most common somatic ASAs across all tumor types. TP53 was mostly found in solid tumor and brain tumor samples.

Of the 50 gene fusions identified, 6 were novel. PAX3/FOXO1 fusions were recorded among all alveolar rhabdomyosarcomas.Nine fusions found in 13 patients (18%) were potentially targetable therapeutically.The most common potentially targetable translocation was the PICALM/MLLT10 fusion found in T-cell acute lymphoblastic leukemia samples, occurring in 4 of 18 patients (22%).

There were no notable differences seen by race or ethnicity in the median number of single-nucleotide variants, fusions, copy-number alterations, or TTAs and ASAs.

Germline Sequencing

All evaluable patients consented to germline analysis.Known pathogenic or likely pathogenic (P/LP) variants in cancer predisposition genes were detected in 9 patients (7.1%). Of these 9, 8 patients had single-allele P/LP variants in autosomal dominant cancer predisposition genes, and 1 patient had biallelic mutation in PMS2. Confirmatory clinical testing has been completed in 4 of these patients. One patient with an apparent germline mutation in WT1 with mixed phenotype acute leukemia was not detected by clinical germline sequencing, which can be attributed to the contamination of saliva with DNA from leukemia cells. Twenty additional patients carried monoallelic known P/LP mutations in genes associated with autosomal recessive syndromes.

Clinical Impact of Tumor and Germline Sequencing

Of 127 samples, 108 (85.0%) indicated at least 1 impactful finding on the basis of the scoring system. Newly diagnosed patients had a median impactscore of 1 (range, 0.5-4), and patients with relapse/refractory disease had a median impact score of 1.5 (range, 0.5-3.5). The most common finding among de novo patients was a refinement of risk stratification/prognosis (0.5 point) in 62.3% of patients, followed by a tier 1-2 treatment consideration in 37.7%. Refinement of risk stratification/prognosis (0.5 point) in patients with relapse/refractory cases was noted in 42.4% of patients, followed by a tier 3-5 treatment consideration in 39.4%.

Overall, a recommendation for potential targeted therapy was made in 82/127 (64.6%) sequenced tumors. As a result, 20/82 patients (24.4%) have received 22 targeted agents. A recommendation for referral was made for 18 patients on the basis of germline sequencing (n = 10) or somatic sequencing findings, family history, and/or clinical features (n = 8). Eleven of these 18 patients are being evaluated to date.

The median time from sample submission to return of the results was 13 days (range, 8-27). White patients made up 52.4% of the patient population35.7% were Black, and 16.7% were Hispanic.The median age at enrollment was 12.1 years (range, 0.2-25.7).More than half of the patients (52.4%) had relapsed/refractory disease at enrollment. Solid tumors represented 42.5% of all tumors sequenced and half of the relapsed/refractory tumors; whereas, leukemia/lymphoma accounted for 39.3% of de novo tumors. The most common diagnoses among brain tumors, solid tumors, and leukemia/lymphomas were medulloblastoma (46.4%), T-cell acute lymphoblastic leukemia/lymphoma (40.0%), and nonrhabdomyosarcoma soft tissue sarcomas (24.1%).

In conclusion, this tumor genomic profiling program is feasible and clinically impactful for pediatric patients who are high-risk de novo and relapsed/refractory. Efforts should be made to expand genomic profiling to all pediatric cancer patients.

REFERENCE

Summers RJ, Castellino SM, Porter CC, et al. Comprehensive genomic profiling of high-risk pediatric cancer patients has a measurable impact on clinical care. JCO Precis Oncol. 2022;6:e2100451. doi:10.1200/PO.21.00451