Abstract
Background
Cutaneous CD30+ T cell lymphoproliferative disorders (CD30+ T LPD) comprise a group of diseases that share overlapping dermato-histopathologic features with variable clinical outcomes. These diagnoses include lymphomatoid papulosis (LYP), primary cutaneous anaplastic large cell lymphoma (cALCL), CD30+ mycosis fungoides (MF), MF with large cell transformation, and nodal ALCL (nALCL) with secondary cutaneous involvement. The latter is associated with the presence of systemic lymphadenopathy, but primary cutaneous CD30+ T LPD may also lead to secondary lymph node involvement during disease progression. Ultimately, treatment decisions are influenced by severity of disease. Here we characterize the molecular features of these entities, and determine differences in the pathogenic pathways and immune microenvironment between two risk groups of cutaneous CD30+ T LPD.
Methods
Retrospective clinical data and skin biopsy samples from CD30+ T LPD were collected from 12 institutions in the United States, Finland, and Singapore from 1997 to 2019. Patients were grouped by treatment: those offered cytotoxic chemotherapy, romidepsin, pralatrexate, or alemtuzumab were assigned to group 1, and all others were assigned to group 0.
Sequencing was performed on skin biopsies using whole exome DNA, whole transcriptome RNA, and using a targeted gene panel.
Results
79 patients with skin biopsies showing CD30+ T cell LPD were included. 34 (58%) were male, with mean age of diagnosis of 62 years. There were 39 patients with cALCL, 5 with LYP, 17 with MF, 6 with transformed MF, 7 with nodal ALCL with cutaneous involvement, 4 with CD30+ T LPD NOS, and 1 with cALCL vs. nALCL with secondary cutaneous involvement.
In 74 patients with available clinical data, the median overall survival (OS) was 14.95 years [10.66, 30.58]. The median OS was 24.44yrs for group 0 (n=46) and 7.21 yrs for group 1 (n=28), p=1.48e-3. Groups 1 and 0 therefore represent adverse and low risk groups respectively.
Recurrent driver mutations were identified in 18 genes, but no differences in mutation frequency were found in any gene when comparing the two risk groups (n=38).
KDR and TET2 emerged as the most frequently mutated genes, occuring in 23% and 18% of patients respectively. Genes whose primary function is epigenetic regulation were affected in 18/38 (47%) cases (TET2, NCOR2, PRDM1 and BCOR). Mutations in tyrosine kinases or signaling genes that mediate cellular proliferation through intracellular signaling pathways (KDR, GNAS, EPHB1, TGFBR2, and JAK3) occurred in 17 cases (45%).
Comparing gene expression profiles between patients in the two risk groups (n=48), a distinct transcriptional signature emerged. The adverse risk group was characterized by greater expression of genes associated with cell cycle progression, DNA damage response, and immune activation, including CKAP2, SSBP1, PDE4D, BIRC3, IL2RA, B2M, NBN, ERO1A, CKAP5, IARS, POLR2B, CUL1, PRKDC, KIF23, BAZ1B, and DGKG. Gene Set Enrichment Analysis performed using the Molecular Signature Database showed enrichment in MYC driven transcriptional programs and G2M checkpoint gene set in the adverse clinical risk group.
Immune cell deconvolution using bulk transcriptomic data and the LM22 matrix also revealed immune cell compositional differences between the two risk groups. Resting dendritic cells were significantly higher in patients in the low risk group (p=3.2e-3).
Conclusions
This is the largest study in cutaneous CD30+ T LPD that incorporates whole exome DNA and whole transcriptome RNA sequencing with clinical and pathologic data. The presence of recurrent mutations in KDR, a gene involved in angiogenesis and VEGF signaling, is a novel finding in CD30+ T LPD. Gene alterations are very common in epigenetic and signal transduction pathways. Increased proliferative signaling and MYC-driven transcriptional programs are associated with high risk disease, and an increase in resting dendritic cells in lesional skin is associated with low risk disease. Future studies are needed for validation of these findings for prognostication.