Might broader be better?

Amy J. Johnson

In this issue of Blood, Erdmann et al examine the importance of phosphoinositide 3-kinase (PI3K) in diffuse large B-cell lymphoma (DLBCL) and rationalize dual inhibition of the p110 α and δ isoforms.1

DLBCL is the most common form of non-Hodgkin lymphoma; it has an aggressive course that requires immediate treatment. Two main subtypes, activated B cell-like (ABC) and germinal center B cell-like (GCB), were identified on the basis of cell of origin and subsequently further characterized by gene expression profiling; ABC was found to have a worse prognosis. Both the gene signature and next-generation sequencing analyses point to a dependence upon the nuclear factor κB (NF-κB) pathway. Upstream of NF-κB, the B-cell receptor (BCR) signaling pathway has gain-of-function mutations in CD79A, CD79B, and MYD88 and increased PI3K pathway signaling.

In cells of hematopoietic origin, the major catalytic isoform expressed is p110 δ (PIK3CD), which is mutated in some patients with lymphoma.2 On the basis of these findings, it is unsurprising that PI3K inhibitors have been tested in many lymphomas (reviewed in Blachly et al3). In 2014, the US Food and Drug Administration approved a PI3K p110 δ–specific inhibitor, idelalisib, for use in relapsed/refractory chronic lymphocytic leukemia and indolent lymphomas. However, this agent has failed in DLBCL phase 1 studies.

Erdmann et al screened ABC and GCB cell lines with various pharmacologic agents including idelalisib, ibrutinib, the dual p110 α/δ PI3K inhibitor AZD8835, and the AKT inhibitor AZD5363. PTEN-deficient DLBCL cell lines, like most GCB DLBCLs, were sensitive to AKT inhibition. ABC cell lines that were NF-κB–pathway dependent were sensitive to dual p110 α/δ inhibition, using both pharmacologic agents and genetic short hairpin RNA (shRNA) knockdown. AZD8835's activity was confirmed in vivo using cell lines OCI-Ly10 and TMD8 and appropriate patient-derived xenografts (PDXs).

These results suggest that dual inhibition of the p110 α and δ subunits of PI3K in ABC DLBCL models that are BCR and NF-κB dependent should be tested in the clinical setting. The article encourages the molecular characterization of each patient’s lymphoma, especially ABC versus GCB, BCR and NF-κB pathway mutations, and PTEN deficiency. This is not the first description of such an idea; Paul et al4 reported the preclinical success of copanlisib in CD79B- and MYD88-mutated ABC DLBCL models. Several clinical trials are already under way with such agents, including copanlisib and buparlisib.

Of course, these findings should be considered with some caution, given that the results are from panels of cell lines. Although well characterized, these cell lines may have mechanisms of sensitivity and resistance that differ from those of primary tumor cells, and due to a variety of factors, even the PDX models may not be entirely relevant. It is hoped that, as clinical trials are carried out using the prospective molecular characterization proposed here, the role of these models in drug development for DLBCL will be confirmed.


  • Conflict-of-interest disclosure: The author declares no competing financial interests.


  1. 1.
  2. 2.
  3. 3.
  4. 4.