Next-Generation Sequencing Identies Potential Actionable Targets In Pediatric Sarcomas

BACKGROUND Pediatric bone and soft-tissue sarcomas represent 13% of all pediatric malignancies. International contributions to introduce next-generation sequencing (NGS) approaches into clinical application are currently developing. We present the results from the Precision Medicine program for children with sarcomas at a reference center. RESULTS Samples of 70 pediatric sarcomas were processed for histopathological analysis, RT-PCR and NGS with a consensus gene panel. Pathogenic alterations were reported and if existing, targeted recommendations were translated to the clinic. Seventy pediatric patients with sarcomas from 10 centers were studied. Median age was 11.5 years (range 1-18). Twenty-two (31%) had at least one pathogenic alteration by NGS. Thirty pathogenic mutations in 18 different genes were detected amongst the 22 patients. The most frequent alterations were found in TP53, followed by FGFR4 and CTNNB1. Eighteen actionable variants were detected and six patients received targeted treatment observing a disease control rate of 78%. Extrapolating the results to the whole cohort, 23% of the patients would obtain clinical benet from this approach. CONCLUSIONS Pediatric sarcomas have a different genomic landscape when compared to adult cohorts. Incorporating NGS targets into pediatric sarcomas’ therapy is feasible and allows personalized treatments with clinical benet in the relapse setting.


Introduction
Pediatric sarcomas account for over 20% of all pediatric solid malignant cancers and represent 13% of all pediatric malignancies 1 . They also contribute substantially to cancer-related mortality and morbidity.
With more than over 70 histologic subtypes, sarcomas can arise from a primitive mesenchymal cell from almost every tissue in the human body and are classi ed into two main groups: soft tissue sarcomas (STS) and bone sarcomas (BS). The highest incidence rates in children are reported amongst rhabdomyosarcoma (RMS), osteosarcoma and Ewing's sarcoma (EWS). Although each subtype has a different phenotype and genetic pro le, they are classi ed into two molecular groups: a genetically complex group with a high mutational burden and complex karyotype, and a genetically simple group containing a single and disease-speci c translocation, ampli cation or mutation with a silent genomic background 2 . Most pediatric sarcomas are included in the second group as they are mostly characterized by chromosomal translocations that result in hybrid genes acting as drivers that are critical for sarcomagenesis 3 .
Pediatric RMS protocols, currently classify this tumor based on the presence of PAX/FOXO1 translocation and distinguish between fusion positive or fusion negative RMS 4 . The genetic pro le of EWS is dominated by the driving reciprocal chimeric translocation between EWSR1 and a variety of ETS partner transcription factors 5 . These gene fusions act as oncogenic transcription factors that trigger transcriptomic and epigenetic disregulations that explain tumors' biology 6,7 . In contrast to EWS and RMS, osteosarcoma shows an extremely complex and unstable genome but without a remarkable repetitive pattern 8 . In most clinical settings, sarcomas involving translocations are detected by uorescence in situ hybridization (FISH) and reverse transcriptase polymerase chain reaction (RT-PCR). Translocations are used by clinicians mostly as diagnostic markers 9 . However, the resulting chimeric proteins of these translocations are not easily druggable and hinder the development of inhibitors. Table 1 shows the most frequent fusion transcripts in pediatric sarcomas.
Both STS and BS display a highly aggressive behavior. During the last decades, addition of systemic chemotherapy has improved outcome of localized tumors resulting in the survival of 2 out of three patients. However, metastasic and relapsed sarcomas still have very poor survival rates. Despite the knowledge gained in cancer biology, aetiology and in the implementation of novel diagnostic techniques and omics, scarce improvement has been observed in advanced stage STS and BS.
During the last years, precision and quality criteria for the diagnosis of pediatric cancers including sarcomas, has experienced an increased demand. New techniques have been introduced that complement pathological diagnosis including immunochemistry, FISH, RT-PCR and next-generation sequencing (NGS). These demands have been gradually assumed by clinicians, pathologists, geneticists and molecular biologists in tertiary reference hospitals. In addition, precision medicine programs have been developed in order to expand our knowledge of tumor biology and defeat cancer with more precise pharmacological targets. 10,11,12,13 We present the results in pediatric sarcomas from the Precision Medicine program for children and adolescents with solid tumors in relapse/progression carried out at a national reference centre for pediatric sarcomas. This program has received samples from collaborative centres, providing a national perspective 14 . Since September 2019, these studies are routinely carried out at diagnosis in every pediatric sarcoma.

Study subjects
A total of 70 sarcoma samples from pediatric patients treated at a reference institution for pediatric sarcomas or at other Spanish center from February 2015 to March 2020 were included. Thirty patients were analyzed at diagnosis and forty patients were studied at relapse or refractory disease.
The program was approved by the Ethics committee of the center. Parents signed the informed consent and were informed about the possibility of nding germline mutations and accepting or refusing to be informed. Consent was also required when performing NGS studies at diagnosis. Every procedure was performed according to the Declaration of Helsinki.

Study samples
Fresh tumor samples were requested. Para ned-embedded tumors and/or pretreatment biopsies were only used if fresh samples were unavailable. Peripheral blood samples were simultaneously collected in 45 cases. All tumor samples were reviewed by a board-certi ed pathologist to con rm histology and estimate tumor cell content. Immunochemistry techniques (p-AKT, PDL1, p-EGFR, c-KIT, PTEN, Her2neu, p53) and FISH (NTRK1 / 3, ALK, BRAF) were also performed. Only samples with > 30% tumor cell content were considered for further genomic testing, the rest were excluded from the study. The selected tumor material and peripheral blood samples were sent to a biobank for DNA extraction and subsequently to the laboratory for sequencing analysis. In some cases, based on previous literature and according to the sequencing results obtained for each tumor type, studies were completed with SNP array analysis. For NGS data analysis, variant calling was based on the genome version GRCh37 (hg19). Genetic variants detected in both, blood and paired tumor samples were classi ed as germline variants, whereas variants detected exclusively in tumors were categorized as somatic variants.
Variant annotation was carried out applying an algorithm of lters in order to discard non clinically relevant variants: those with an allelic frequency < 5%, changes in non-coding regions (excluding those variants in exon splicing sites +/− 10 nucleotides), synonymous variants (excluding those coding variants nearby splicing sites +/− 4 positions), variants with high frequency in the general population (MAF > 0.01), polymorphic changes (SNPs) without clinical relevance found in healthy population or described as benign by several sources or our genomic database. The remaining variants were classi ed according to international recommendations as pathogenic, likely pathogenic, benign, likely benign or of uncertain signi cance 13 based on literature or disease databases (ClinVar, COSMIC, HGMD, St Jude PeCan or CiVIC).
Pathogenic or likely pathogenic variants were reviewed and approved by the pediatric molecular tumor board (PMBT) committee, and further con rmed using direct Sanger sequencing. Actionable variant was referred as a genomic change that suggests an alteration with biological activity that could be targeted with a speci c therapy already used in vivo. Targeted therapies were preferentially recommended to be administered within clinical trials but also as compassionate use basis if trials weren't available. Median time between biopsy/surgery and molecular tumour board recommendation was 5 weeks.

Pediatric molecular tumor board discussion
The PMTB was created in November 2014 and composed by pediatric oncologists, pharmacologists, geneticists, pathologists, molecular biologists and bioinformatics. The PMBT established the consensus gene panel for the NGS analysis. After the completion of pathological and genomic studies, results were discussed in periodical meetings in the PMTB and a nal report was transferred to the corresponding physician. The work ow was based on previous pilot studies 14,15 .

Clinical characteristics
A total of 70 pediatric and adolescent patients with STS and BS from 10 Spanish cooperating sites were included in a 5-year period from February 2015 to March 2020. Patients' median age at study entry was 11.5 years with a range of 1-18 years. Forty-one per cent of the patients were female and 59% were male. Distribution of tumor type is shown in Figure 1. The most frequent tumors were EWS (n=22), RMS (n=16) and osteosarcoma (n=13). Thirty patients were studied at diagnosis (43%), 22 patients at rst relapse (31%), 15 patients at second or successive recurrences (21%) and 3 patients when found to be refractory to rst line treatment (4%). Somatic variants described as pathogenic or likely pathogenic using international system classi cations 15 were reported. The main characteristics of the selected patients are detailed in Table 2.

NGS results
Twenty-two out of 70 patients (31%) had at least one pathogenic or probably pathogenic alteration identi ed by NGS as with a mean of 1,4 mutations per patient. Most of the cases had one unique mutation. A total of 30 different pathogenic or likely pathogenic mutations in 18 different genes were detected amongst the 22 patients. Mutations were detected in relapsed or refractory sarcomas (57%) and also at rst diagnosis (43%).
Diagnostic sarcoma fusion genes detected by FISH or RT-PCR were only used for diagnosis but were not considered for precision medicine recommendations as no targeted treatments are available for these alterations to date. Overall, TP53 was the most frequently affected gene (27%) and preferentially identi ed in EWS, RMS and angiosarcoma. Three embryonal rhabdomyosarcomas harbored alterations in FGFR4 whilst the two aggressive bromatosis and an embryonal RMS had CTNNB1 mutations. Identi ed gene and variant alterations are shown in Table 3. Including information obtained by complementary techniques (immunochemistry and FISH) up to 27 patients had an identi ed alteration (39% of the cases). A summary of the molecular alterations spotted in these 27 patients is shown in Figure 2.

Clinical translation
After discussion of the biological results in the PMTB, 18 actionable variants (26%) were identi ed and formal recommendations were submitted to the respective physicians. RMS was the tumor in which more actionable variants were observed (39%), particularly embryonal histology (28%). Two osteosarcoma patients presented actionable alterations. Despite the number of EWS cases included in the study (22), only one patient had an actionable variant. This result points out the di culties to implement a precision medicine strategy in EWS tumors.
Six patients out of the whole cohort received targeted treatment (9%), observing clinical bene t in ve of them (78%): A thirteen-year-old female with a radio-induced abdominal malignant nerve sheath tumor, 10 years after neuroblastoma treatment was found to have a mutation in ATM. After radical surgery and standard chemotherapy, she underwent disease progression. Targeted treatment with PARP inhibitor olaparib and temozolamide was administered, resulting in disease stabilization during one month with a clear disease control. She received treatment during two months before a subsequent progression.
A twelve-year-old female with malignant perivascular epithelioid cell kidney tumor with lung metastasis was found to have positive p-AKT with immunochemistry and targeted treatment with sirolimus and sorafenib was initiated after observing no response to classic sarcoma chemotherapy. A slight response was observed in tumor size and the disease was stabilized according to RECIST 1.1 criteria for a 5-month period. This achievement had not been possible with the previous schedules administered. On the third place, a nine-year old female with rst local and metastasic osteosarcoma relapse with positive mTOR immunochemistry was treated with an oral mTOR inhibitor during a two-month period after failure of standard treatments. Unfortunately, progression was observed after the third month.
Another twelve-year-old male affected by a mediastinal myo broblastic in ammatory tumor with ALK translocation detected by FISH. Disease progression was observed after standard chemotherapy (IVA regime) and surgery. Targeted treatment with ALK inhibitor ceritinib was started and a very good partial response was observed. Finally, a thirteen-year-old female with a stomach GIST with positive c-KIT diagnosis (CD-117) by immunohistochemistry is currently receiving imatinib after radical surgery and has achieved complete response.
Future treatment options were available in 12 patients (17%) that are at the moment in complete response or receiving other standard treatments. Altogether, implementing NGS with complementary diagnostic techniques such as immunohistochemistry and FISH in a precision medicine approach for targeted treatment of sarcomas, a disease control rate of 23% would potentially be achieved. The summary of the recommendations and clinical responses are shown in Table 4.

Discussion
Genetic variation is one of the main characteristics of pediatric sarcomas. This is mostly explained because despite being originated from a mesenchymal cell, they constitute different histologic entities with different genomic landscapes that explain their unequal behaviours. Beside pathology, chromosomal segmental aberrations, 16 changes in ploidy and speci c gene alterations are routinely used in order to guide intensity of treatment in pediatric oncology protocols.
It is worth noting important differences spotted when comparing adult with pediatric NGS studies in sarcomas. 17 Epidemiologically, sarcomas represent less than 1% of all solid malignant cancers in adult population while they represent 20% of all pediatric solid malignant cancers. Therefore, the rst main difference lies in the fact that the magnitude of the problem is proportionally much higher in pediatric population. Furthermore, adult type cancers such as epithelial neoplasms arise after accumulation of multiple sequential mutations directly linked to environmental exposures, and arise within differentiated adult tissues 18,19 . Mesenchymal tumors, such as sarcomas appear both in adult and pediatric population. However, speci c histologic subtypes and clinical progression are age-dependent, suggesting differential pathogenetics and underlying molecular mechanisms for tumor initiation and clinical behavior in the different age subgroups 18 .
In this study, we found that the overall mutational load in our cohort was relatively low when compared to adult studies. This might be explained by the fact that adult sarcomas are mostly driven by mutagenic exposure from environmental factors, whereas most of pediatric cancers contain a relatively small number of mutations 20 and frequently display unique gene rearrangements. Although this restricts the targeted treatment to available drugs, it also makes them attractive candidates for drug discovery 15 .
In order to improve outcome, international efforts amongst cooperative groups have been carried out developing genomic precision medicine programs. These programs aim to bring NGS approaches into the clinical practice and require the identi cation of patients that might bene t from targeted therapies. Once these targets are identi ed, in pediatric population it is important to communicate these results, as well as possible toxicities observed by compassionate use basis as dosing is more complex when compared to adult population. Hence, the importance of promoting pediatric phase I clinical trials in order to titrate infant dosing.
In this study, we conclude that the most frequent somatic mutation observed in pediatric sarcomas occurs in TP53 (27% of the pathogenic mutations detected by NGS). This information correlates with adult sarcoma cohorts such as the study presented by Groisberg et al. 21 Xiaosheng et al 22 compared overall survival (OS) time between TP53-mutated and TP53-wildtype cancers in 20 adult cancer types.
They reported that patients with TP53 mutations had lower survival compared with those without TP53 mutations in colon, lung and pancreas adenocarcinoma, acute myeloid leukemia and other epithelial cancers. In pediatric oncology, the clinical signi cance of somatic TP53 mutations remains unrecognized and no routine testing or therapy intensi cation is considered. Recent studies suggest that mutation in TP53 in localized EWS is not a reliable prognostic marker 23 . In order to target TP53, small molecules that reactivate mutant p53 by restoring wild-type conformation have been identi ed by various approaches. APR-246 alone is currently being tested in prostate or ovarian cancers or in combination with azacitidine in myeloid malignancies in adult phase I-II trials. No studies are currently recruiting pediatric population.
Mutations in Fibroblast Growth Factor Receptor 4 (FGFR4) have also been described in pediatric sarcomas, most outstandingly in RMS. Higher FGFR4 expression in RMS has been associated with advanced-stage cancer and poor survival 24 . FGFR4 pathogenic mutations appear in 33% of the embryonal RMS studied in our cohort and all of them received a targeted recommendation therapy. FGFR4 codi es for a cell surface tyrosine kinase (TK) receptor that is involved in normal myogenesis and muscle regeneration. It has been reported that human embryonal RMS cells have increased FGFR4 mRNA expression compared to normal human myoblasts, and FGFR4 pathway blockade decreases proliferation. 25 In fact, over-expression and mutational activation of FGFR4 has been reported in RMS, promoting tumor progression. FGFR4 signaling is also a common mechanism of oncogenesis in fusion positive RMS (usually alveolar subtype) 25 .
Alterations in FGFR4 are clinically relevant because they are actionable targets in patients with RMS. New generation of multikinase inhibitors are under current development such as ponatinib (AP-24534), an orally administered TK inhibitor that was initially developed as an inhibitor for BCL-ABL. Ponatinib recently received FDA approval for the treatment of adult patients with Philadelphia chromosome positive acute lymphoblastic leukemia and chronic myeloid leukemia resistant to other TK inhibitors. Inhibition pro le of ponatinib includes other TK such as c-KIT, PDGFR, FLT3, SRC and FGFR 26 . Moreover, inhibition of FGFR family members with ponatinib has been demonstrated in preclinical models with bladder cancer, endometrial cancer, breast, lung and colon cancer. Samuel Q. Li et al 26 tested a panel of RMS cell lines over-expressing FGFR4, all of them exhibiting sensitivity to ve different TK inhibitors including ponatinib, cediranib, nintedanib, dovitinib and danusertib. They observed that ponatinib resulted to be the most powerful FGFR4 inhibitor, inhibiting both, mutated and wild-type FGFR4 cell growth. It also inhibited tumor development expressing FGFR4 in vivo 26 . Currently, ponatinib is being tested in clinical trials including pediatric patients (NCT03934372) 25,27 . Erda tinib is also being tested in a phase II trial for tumors with FGFR mutations. (NCT03210714).
The CTNNB1 gene provides instructions to form the protein beta-catenin. The relationship between the Wnt/beta-catetin signaling pathway and desmoid-type bromatosis (DTF) has been widely studied and it has been reported that the vast majority of DTF tumors (up to 85%) harbor a mutation in exon 3 of the CTNNB1 gene (beta-catetenin) 28 . These mutations lead to an abnormally stable beta-catenin protein that is more resistant to proteolytic degradation and accumulates within the cells. Excess of beta-catenin promotes an uncontrolled proliferation of cells, allowing the formation of DTF 29 .
Therapeutic options targeting Wnt/betacatenin signaling pathway are limited and have not been tested in pediatric population. Accumulation of beta-catenin in the nucleus triggers transcription of Wnt-speci c genes responsible for the control of cell fate decisions. The development of drugs targeting mutated or altered beta-catenin signaling, or its interaction with CBP, TCF, GSK3β or APC (which are essential to complete its function) has been di cult due to the toxicity of the new compounds. Several of them are currently in Phase 1 clinical trials, such as the PRI-724 molecule (NCT01302405, NCT02413853, NCT01764477, and NCT01606579) that prevents the interaction of beta-catenin with CBP. Despite these and other approaches, there are no clinical trials available for pediatric patients with Wnt/beta-catenin inhibitors. 30 All DTF studied in our cohort harbored mutations in CTNNB1.
In the study, a patient with malignant nerve sheath tumor and ATM mutation was treated with PARP inhibitors in combination with olaparib. The ataxia telangiectasia gene (ATM), localized in 11q22-q23, plays an important role in maintaining genomic integrity. It regulates the double-strand DNA breaks repair and activates different checkpoints in the cell cycle. ATM is associated with some types of leukemia and lymphoma and it has also been described in neuroblastoma with 11q deletion. Poly ADP-ribose polymerase (PARP) is a protein that signals DNA damage and contributes towards DNA repair 31 . PARP catalyzes the addition of ADP-ribose to DNA, helicases, topoisomerases and histones. It also has a critical role in transcription, cellular replication, gene regulation, differentiation, spindle maintenance and protein degradation. PARP inhibition produces persistent single strand DNA breaks leading to double strand DNA breaks and nally produces DNA damage leading to apoptosis and cell cycle arrest.
Preclinical studies show that ATM mutated neuroblastoma cells also succumb to apoptosis when treated with PARP inhibitors and neuroblastomas with 11q deletion are extremely sensitive to conventional chemotherapy combined with PARP inhibitors. The patient in the study managed a short period of stable disease but progressed rapidly afterwards 31 . Other mutations considered as uncertainly signi cant in ATM have been detected but no recommendations were issued because no previous clinical evidence was found. Currently, early phase trials with PARP inhibitors are recruiting pediatric patients with diverse malignancies.
Recent studies in RMS have revealed recurrent mutations in the RAS pathway, particularly affecting NRAS. Dolghik et al 32 demonstrated that PIK3CA played a critical role in the activation of the PI3K/AKT/mTOR pathway in NRAS mutant RMS. They noted that NRAS-mutated RMS cells particularly relied on PIK3CA to prevent cell death upon NRAS silencing or MEK inhibition. Their data showed that speci c PIK3CA knockdown was su cient to cooperatively trigger cell death together with pharmacological MEK inhibition. In addition, pharmacological inhibitors of MEK or NRAS knockdown synergize with the PIK3CA speci c inhibitor BYL719 to trigger cell death in NRAS-mutated RMS cells. All this data supports the rationale for the combination of MEK and PIK3CA speci c inhibitors in NRAS mutated RMS. This recommendation is a future option for one of the patients studied in our cohort.
In this study, a patient diagnosed with c-KIT positive (CD-117) GIST was treated with imatinib and so far, has maintained complete response after surgery. Another patient with ALK + myo broblastic in ammatory tumor received treatment with ceritinib obtaining a partial response. Both of these rare sarcomas have a classical alteration that has been widely reported before.
In conclusion, we have observed that the incorporation of NGS results together with ancillary studies into pediatric sarcoma clinical practice is feasible and allows personalized treatments with acceptable disease control rates in the relapse setting. At the moment, as the integration NGS as a routine diagnostic technique has been limited this is di cult to estimate, although the situation is changing and sequencing studies are gradually becoming widespread 33,34,35 . Further investigations are required to con rm this hypothesis.
In this study, up to 23% of the patients would obtain clinical bene t by implementing this precision medicine approach complementing routine diagnostic techniques. Although the understanding of pediatric sarcomas' biology has improved in a relatively short period of time, outcomes in high-risk tumors remain poor and regarding new therapeutic strategies, very few advances have been highlighted. This emphasizes that strong, international efforts are still required in order to improve implementation of new diagnostic techniques, impulse pediatric drug development and access to clinical trials in childhood. Finally, we would like to stress the importance of treating childhood, adolescent and young adult sarcomas and other types of cancers in specialized units, with all the available expertise and distinct requirements involving this particular population.

Declarations
Ethics approval and consent to participate Informed consent for all the patients was obtained from the parents or legal guardians. All study actions have been done under the appropriate ethics code and the study has been approved by the Ethics and Investigation Committee Hospital U i P La Fe (CEIm). The study was performed according to the Declaration of Helsinki.

Consent for publication
Consent for publication has been obtained from all patients or, if patients are under 18, from a parent and/or legal guardian.

Availability of data and materials
The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.