Primary cutaneous diffuse large B-cell lymphoma, leg type (DLBCL-LT) is an aggressive subtype of cutaneous B cell lymphoma (CBCL) with a poor prognosis. It is most often diagnosed on the legs of the elderly, frequently relapses, and has a tendency towards extracutaneous dissemination. Current treatment strategies are ineffective and revolve around surgical excision, multiagent chemotherapy (e.g. R-CHOP) and radiation therapy, with estimated 5-year survival rates of ~50%. Identifying genetic mutations and altered pathways specific for DLBCL-LT is crucial for the diagnosis and development of effective targeted therapies for this condition.

We performed whole exome capture and massively parallel sequencing on 13 tumor samples of DLBCL-LT (4 matched and 9 unmatched) and 5 samples of cutaneous diffuse large B cell lymphoma, not otherwise specified (DLBCL-NOS) (1 matched and 4 unmatched). This is the largest whole exome cohort of cutaneous DLBCL to date. We applied algorithms routinely used by our lab to identify putative cancer promoting mutations and to compare the genetic landscape with DLBCL, both nodal and extranodal.

Broadly, the landscape of mutations in DLBCL-LT is similar to that of activated B-cell like (ABC) DLBCL and most analogous to primary central nervous system lymphoma (PCNSL), a type of extranodal ABC DLBCL. Similar to PCNSL, we found frequent disease-defining oncogenic gene mutations in MYD88 (85%), CD79B (38%), and TBL1XR1 (38%), and loss of function mutations in tumor suppressor genes KMT2D (15%), CDKN1B (15%), and CD58 (8%). We observed frequent mutations in PIM1 (54%), OSBPL10 (31%), BTG2 (31%), and IGLL5 (15%), which are either known or proposed targets of activation-induced cytidine deaminase (AID) and aberrant somatic hypermutation (aSHM) also seen in PCNSL. Importantly and similarly to PCNSL, we found alterations at the PDL-1 and PDL-2 locus, but instead of copy number gains, we discovered frequent PD-L1 and PD-L2 translocations (40%) that may translate into overexpression of PD-L1 .

This spectrum of mutations differed significantly compared to our DLBCL-NOS cohort, which did not have the above oncogenic mutations and lacked any PD-L1 or PD-L2 chromosomal alterations. Finally, we report novel oncogenic mutations in DLBCL-LT including MYD88 p.S243N, CD79B p.L199P, CCND3 p.T283P and p.P284S, BRAF p.V600E, and CARD11 p.R113Q and novel loss of function mutations in KMT2D, CDKN1B, CD58, MSH2, PIK3R1, MSH2, and BMF .

Collectively, we identified recurrent mutations in multiple key cancer-related biological pathways, including epigenetic modulators (TBL1XR1, HIST1H1E, and KMT2D), NF-κB signaling pathway (MYD88, CD79B, and CARD11), MAPK signaling pathway (BRAF), PI3K signaling pathway (PIK3R1), JAK/STAT signaling pathway (PIM1), cell cycle regulators (CCND3, CDKN1B and BTG2), and immune escape (CD58). All of the DLBCL-LT samples in our cohort contained one or more gene mutations that can be targeted by agents that are currently FDA-approved or under clinical trials, such as BRAF inhibitors, JAK/STAT inhibitors, PIM inhibitors, PI3K inhibitors, NF-kB inhibitors, MYD88 inhibitors, and PD-1/PDL-1 inhibitors.

Altogether, our data identified targetable putative driver gene mutations in DLBCL-LT. These mutational signatures bear broad similarities to PCNSL. Because of the relative rarity of this disease, we hope these efforts will allow for its inclusion, when relevant, in trials involving genotypically similar DLBCL.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.