Sustained Complete Response to Palbociclib in a Refractory Pediatric Sarcoma With BCOR-CCNB3 Fusion and Germline CDKN2B Variant

Introduction

Genomic alterations in the Ewing sarcoma family of tumors (EFT) were discovered > 30 years ago with the identification of the reciprocal translocation, t(11;22)(q24;q12), otherwise known as EWS-FL1.^1,2 In the time since, multiple other fusion partners with EWS have been identified that fit a similar Ewing sarcoma phenotype.^3,4 When EWS fusions are not identified, tumors with histologic features of Ewing sarcoma have been labeled as primitive neuroectodermal tumors. In 2012, Pierron et al⁵ identified a subset of Ewing-like tumors harboring paracentric inversion on the short arm of chromosome X, resulting in the fusion of the BCOR and CCNB3 genes.⁵ Since that discovery, several small case series have further elucidated the clinical, morphologic, and genomic differences that make this diagnosis distinct from other round cell sarcomas, most notably Ewing sarcoma.^6-8

Though distinct from Ewing sarcoma, most BCOR-CCNB3–fused sarcomas (BCS) are treated with upfront compressed chemotherapy with vincristine, doxorubicin, cyclophosphamide, ifosfamide, and etoposide plus local control with surgery and/or radiation. BCS shares similar event-free and overall survival rates with the standard EWS-FLI1–fused Ewing sarcoma using this treatment strategy.^6-8 Despite the growing knowledge base related to BCS, little is known about potential drug targets related to this disease entity, especially with regard to treatment of disease recurrence. We highlight the treatment of a young patient who had multiply-relapsed disease with the US Food and Drug Administration–approved cyclin-dependent kinase 4/6 (CDK4/6) inhibitor palbociclib; the tumor harbored a BCOR-CCNB3 fusion and a germline variant in CDKN2B, and treatment resulted in a complete response and no evidence of disease 25 months into therapy.

Case History

Our male patient initially presented in 2010 at 1 year of age with a fixed mass on his back. Magnetic resonance imaging of the pelvis showed a large infiltrating presacral mass measuring 14 × 7.4 × 10.4 cm extending into the lower spinal canal, eroding the posterior right sacrum, and exerting a mass effect on both the rectum and bladder. A core needle biopsy was performed, which revealed a malignant, small, round, blue cell tumor along with small amounts of benign fibrofatty tissue and skeletal muscle. Tumor nuclei were round to oval with a fine-grained chromatin pattern and occasional small nucleoli or chromocenters. Immunohistochemical stains were positive for CD99, Fli1, and vimentin and were negative for NSE, synaptophysin, MYF4, GAF, CD45RB, and TdT—consistent with a primitive neuroectodermal tumor. No polymerase chain reaction–base fusion analysis or breakapart fluorescence in situ hybridization probe for EWSR1 was performed at the time. Three generations of family history were negative for malignancies on either side of the family, including melanoma or pancreatic cancer. A staging computed tomography scan of the chest and a bone scan showed no evidence of metastatic disease. The patient started chemotherapy per Children’s Oncology Group protocol AEWS0031, regimen B2, with ifosfamide, etoposide, vincristine, doxorubicin, and cyclophosphamide. Gross total resection was not feasible at the time per neurosurgery, and the patient received 57.6 Gy of proton beam radiation in October 2010. The patient remained in remission for > 2 years but then developed multiple local recurrences without metastases from 2013 to 2017 and underwent numerous surgeries, along with multiple different early-phase Children’s Oncology Group therapeutic studies, as outlined in the timeline in Figure 1A. After the most recent recurrence in October 2016, the patient was referred to our Pediatric Cancer Precision Genomics Program. Because of the findings outlined here in the Results, we chose to start palbociclib in February 2017. This patient has no evidence of disease on imaging 25 months into therapy (Fig 1B) and has had only hematologic toxicity that was grade 2 or less.

FIG 1.

Patient timeline and response imaging. (A) The timeline of treatments and disease recurrences and progression in our patient. Red circles represent time of recurrence or progression. Hash marks represent prolonged period of time progressed through therapy (not to scale of remaining timeline). Patient is still continuing with palbociclib therapy after the most recent scans in March 2019. (B) Patient imaging displayed with initial diagnostic magnetic resonance imaging from 2010; subsequent relapse scans were done via computed tomography imaging. Red lines outline tumor boundaries. There was concern for progression after 2 cycles of palbociclib, but there was a 2-month lag between the “before palbociclib” scan and actually starting drug, so interval progression likely occurred in this timeframe. No new baseline scan was performed on the day palbociclib started. Each cycle of palbociclib was 28 days. NED, no evidence of disease.

Results

Whole-genome sequencing, RNA sequencing (RNA-Seq) analysis, germline exome sequencing, and protein evaluation were performed at the Clinical Laboratory Improvement Amendment (CLIA)–approved laboratory, NantOmics (Culver City, CA). Somatic DNA changes were determined by comparing the whole-genome DNA sequence from the tumor with the patient’s germline sequence at 33X coverage. The mutational burden of the tumor was relatively low at 75,046 somatic mutations, with only 88 somatic mutations mapping to protein coding regions (Circos plot in Fig 2A). The tumor harbored an in-frame fusion of the second base in the last codon of BCOR exon 15 (chrX:39,911,366) and the first base of CCNB3 exon 5 (chrX:50,051,505) (Fig 2B). Additionally, an undescribed somatic mutation in the SMO gene (SMO N476S) was identified in the tumor, and germline sequencing revealed a CDKN2B N41D missense variant, which was heterozygous in both the germline and tumor genomes of this patient. RNA-Seq was also performed by NantOmics, and mRNA transcripts were ranked by abundance, which could be associated with increased pathway activity and sensitivity to a targeted drug. Overexpression of relevant tumor-promoting pathways is displayed in Table 1 and Figure 2C.

FIG 2.

Transcriptome analysis reveals activation of the cyclin-dependent kinase 4/6 (CDK4/6)-RB pathway. (A) Circos plot of patient’s tumor genome. (B) Identification of BCOR-CCNB3 fusion. This graphic illustrates the fusion of BCOR exon 15 to CCNB3 exon 5. With the exception of the destruction box in CCNB3, all functional domains from each encoded protein remain intact in the BCOR-CCNB3 fusion protein. (C) The CDK4/6 pathway is a gene regulatory program controlled by multiple tiers of protein kinases and transcriptional regulators. Increases in cyclin-D or CDK4 or 6 protein can lead to phosphorylation of the RB1-E2F tumor suppressor complex. Upon phosphorylation of RB1, the E2F1-3 transcription factors are released from the complex and are able to bind to the promoters of target genes, driving activation of transcription. These target genes include E2F1-3 themselves, the CDC25 genes, TOP2A, BUB1, and BUB1B. WT, wild type.

TABLE 1.

Overexpression of Relevant Tumor-Promoting Pathways

Discussion

Germline and somatic whole-genome DNA sequencing combined with RNA sequencing was used with the goal of developing a treatment plan, and it surprisingly provided our team with a more defined diagnosis of a recurrent BCS. The specific intrachromosomal fusion between BCOR and CCNB3 in our patient’s tumor is identical to previously described cases.^5,6 The BCOR gene itself can fuse to a number of 3′ partner genes in round cell sarcomas or additionally have internal tandem duplications, which have been reported to drive similar transcriptional patterns in a variety of sarcomas.^5,8,9 Similar to previous studies of mRNA transcripts in BCS,^5,8,9 both BCOR and CCNB3 transcripts were highly overexpressed in our patient’s tumor, as were the HOX-A, -B, and -C gene clusters (Table 2). Furthermore, analysis of our patient’s tumor (Table 1) matched the BCS-specific fingerprint of genes used in the Riggi_Ewing_sarcoma_progenitor signature, which can be used to distinguish between BCS and EWS.^5,10 Most notable for this patient was the observation that multiple genes in the CDK4/6–RB pathway (Table 1; Fig 2C) were overexpressed, which made palbociclib, which has known pediatric dosing information, an attractive drug to use in this case.

TABLE 2.

Comparison of Patient Transcriptome With Published BCS Reference Publications

Curiously, our patient was diagnosed at a very young age compared with the literature on BCS. In the several case series describing BCS, the median age of diagnosis is in the teenage years, with the youngest patient recorded at age 2.^5-8,11 Again, because of the age of presentation, one could be concerned about an inherited cancer syndrome. Alfaro-Cervello et al¹² published a case report of a congenital undifferentiated sarcoma with BCOR-CCNB3 fusion possibly similar to our patient case. This congenital tumor also harbored a SMARCB1/INI1 gene deletion common to malignant rhabdoid tumor, epithelioid sarcomas, and epithelioid malignant peripheral nerve sheath tumor that also, when found germline, is known to cause rhabdoid tumor predisposition syndrome.^12-16 In the case reported by Alfaro-Cervello et al,¹² INI1 germline analysis was not performed. Our patient’s tumor had functionally intact INI1, which precludes an effective comparison. Our patient also harbored a germline heterozygous missense variant, CDKN2B N41D. It is unclear what role this germline CDKN2B N41D variant could play in sarcomagenesis, as cancer risks associated with CDKN2A/B gene variants include melanoma, pancreatic cancer, and astrocytomas.^17,18 There is a recent short report from Jouenne et al¹⁹ that found an increased risk of soft tissue sarcoma development with germline loss of CDKN2A, though no data exist confirming this risk with CDKN2B variants. Additionally, little is known about this actual variant in CDKN2B. Sunita et al²⁰ showed that the specific CDKN2B N41D variant, which encodes p15(INK4B), is unable to bind to the CDK6 protein, leading to loss of function of CDKN2B, which could lead to dysregulated control of S-phase entry. Though this variant’s contribution to tumorigenesis is intriguing, CKDN2B was normally expressed in our patient’s tumor, and there are no data suggesting that this impaired binding to CDK6 leads to mRNA overexpression along multiple levels of the CDK4/6 pathway.

Despite discovering alterations of several key regulators of the CDK4/6 pathway in this tumor, none have been proven to serve as clinical biomarker for sensitivity to CDK4/6 inhibitors.²¹ In a preclinical Ewing sarcoma orthotopic xenograft model with CDKN2A deletion, palbociclib was able to greatly suppress growth despite doxorubicin resistance of this model.²² In other sarcoma subtypes, palbociclib reduced tumor burden in murine preclinical models.^23-25 Clinically, there is phase II evidence of palbociclib’s efficacy in adults with liposarcoma^26,27 and leiomyosarcoma.²⁸ Despite growing evidence in these sarcomas, there are no published data testing CDK4/6 inhibitors in BCS. Additionally, though phase I/II trials are underway, the only published response data for palbociclib in pediatrics is a case report of growing teratoma syndrome,²⁹ thus making our use of this drug in a child novel.

To summarize, 3 independent observations supported consideration of therapeutic inhibition of the CDK4/6-RB1 pathway for this patient: (1) the presence of the BCOR-CCNB3 gene fusion believed to drive entry into the cell cycle, (2) direct detection of an active CDK4/6-RB1 pathway, and (3) the presence of a germline CDKN2B variant. Using this information, our Precision Genomics team chose to place our patient with multiply-relapsed disease on palbociclib; the patient has now benefited from > 2 years of disease remission. The sustained complete response with palbociclib in our patient makes this case a novel and interesting application of palbociclib use and argues for additional research using CDK4/6 inhibitors in BCS.

AUTHOR CONTRIBUTIONS

Conception and design: Timothy F. Tramontana, Mark S. Marshall, Morgan R. Schmitt, Jacquelyn L. Carter, Jamie L. Renbarger, Michael J. Ferguson

Collection and assembly of data: Timothy F. Tramontana, Mark S. Marshall, Amy E. Helvie, Jennifer Ivanovich, Michael J. Ferguson

Data analysis and interpretation: Mark S. Marshall, Amy E. Helvie, Jennifer Ivanovich, Jamie L. Renbarger, Michael J. Ferguson

Provision of study material or patients: Amy E. Helvie

Manuscript writing: All authors

Final approval of manuscript: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Michael J. Ferguson

Consulting or Advisory Role: Bayer

No other potential conflicts of interest were reported.