Genomic Analysis of Congenital Myeloid Sarcoma Identifies Significant Bone Marrow Involvement in the Absence of Morphologic Blasts

Myeloid sarcoma is a tumor mass consisting of blasts at extramedullary sites, most frequently the skin and soft tissues. Skin involvement, also termed leukemia cutis, is a well-described entity in congenital leukemia and is present in approximately 50% of patients diagnosed within the first 4 weeks of life. Most cases of congenital leukemia cutis harbor KMT2A rearrangements, as do up to 80% of cases of infantile leukemia, and are typically accompanied by bone marrow involvement. Here we describe a rare CIC-rearrangement as the driver mutation for a case of congenital cutaneous myeloid sarcoma in a newborn female who was noted at birth to have a violaceous, nodular rash.

Skin biopsy of the nodular rash shortly after birth with immunohistochemical staining was consistent with myeloid sarcoma. Bilateral bone marrow aspirates and biopsies were also performed, but failed to reveal an aberrant blast population by morphology and flow cytometry. Cytogenetics of the cutaneous myeloid sarcoma was significant for t(10;19)(q23;q13.2) and fluorescence in situ hybridization performed on the tumor confirmed a CIC-rearrangement. CIC-rearranged sarcomas are a new entity of undifferentiated small round cell sarcoma characterized by fusion events involving the CIC gene but are not a known to be driver mutations in myeloid neoplasms. Despite the absence of a blast population in the bone marrow, 75% of analyzed metaphases carried the t(10;19)(q23;q13.2). To further explore the underlying genomic events, whole genome, exome, and transcriptome sequencing was performed on both sarcoma and bone marrow specimens.

Sequencing revealed an in-frame CIC-NUTM2A fusion gene present in both the skin and the bone marrow which has been previously described in a single case of undifferentiated soft tissue sarcoma. There were no large-scale chromosomal losses or gains and no copy neutral loss of heterozygosity events. Two exonic single nucleotide variations (SNVs) were detected, both of which were limited to the skin sarcoma and not predicted to be pathogenic. In vitro and in vivo modeling demonstrated that the CIC-NUTM2A fusion protein was highly leukemogenic. Colony forming assays performed using transduced murine bone marrow revealed that the CIC-NUTM2A fusion conferred self-renewal in contrast to CIC, NUTM2A, and the reciprocal NUTM2A-CIC product which failed to serially replate (p<0.001). Transplantation of CIC-NUTM2A modified murine bone marrow cells resulted in a fully penetrant myeloid leukemia with a median survival of 21 days in primary transplants and 16 days in secondary transplants. Transcriptome analysis of these tumors revealed a distinct gene expression profile when compared to several classic myeloid associated fusion genes including AML1-ETO and MLL-AF6.

To further understand the acquisition of a malignant phenotype by morphology between the bone marrow and the sarcoma, targeted deep sequencing was performed for all tier 1, 2 and 3 mutations identified by whole genome and exome sequencing. A shared ancestral clone was identified in both specimens along with three subclones specific to the sarcoma. Four SNVs acquired in the sarcoma were found to be present in regulatory regions of genes that were also differentially expressed between the sarcoma and the bone marrow including C6orf120, SMURF1, TJAP1, and PID1. PID1 was found to be downregulated in the malignant sarcoma specimen and has been previously shown to be a regulator of the AKT/PI3K pathway. Low PID1 expression has been associated with poor outcomes in other malignancies, including pediatric glioblastoma. A genome wide CRISPR screen of our CIC-NUTM2A positive murine leukemia cells revealed an enrichment for PID1 deficient cells as well as PTEN confirming the cooperativity between CIC-NUTM2A and the AKT/PI3K pathway (p=0.009 and p=0.005 respectively).

In conclusion, we describe a rare fusion gene, CIC-NUTM2A, which leads to an aggressive myeloid malignancy in both humans and mice. Targeted deep sequencing demonstrated the clonal evolution from the bone marrow and acquisition of a cooperating mutation targeting the AKT/PI3K pathway in a subset of extramedullary cells that led to morphologic transformation. A thorough interrogation of the bone marrow in patients with myeloid sarcoma is warranted even in the absence of morphologic and immunophenotypic blasts by flow cytometry to optimally track residual disease during treatment.