AUTHORS
Cynthia K. Hahn, MD, PhD, Oscar Lin2, Elinor G. Sterner3, Sam Sadigh, MD4, Erin Parry, MD, PhD5, Kun Huang, PhD6, Abner Louissaint Jr, MD, PhD7, Jeremy M Simon, PhD8, Catherine J. Wu, MD9
AFFILIATIONS
1 Broad Institute, Cambridge, United States, Dana-Farber Cancer Institute, Boston, United States, Harvard Medical School, Boston, United States
2 Dana-Farber Cancer Institute, Boston, United States
3 Dana-Farber Cancer Institute, Boston, United States
4 Brigham and Women's Hospital, Boston, United States, Harvard Medical School, Boston, United States
5 Broad Institute, Cambridge, United States, Dana-Farber Cancer Institute, Boston, United States, Harvard Medical School, Boston, United States
6 Dana-Farber Cancer Institute, Boston, United States
7 Harvard Medical School, Boston, United States, Massachusetts General Hospital, Boston, United States
8 Dana-Farber Cancer Institute, Boston, United States, Harvard T. H. Chan School of Public Health, Boston, United States
9 Broad Institute, Cambridge, United States, Dana-Farber Cancer Institute, Boston, United States, Harvard Medical School, Boston, United States
ABSTRACT
HLA-E, a non-classical HLA class I molecule, is a well-established mediator of immune evasion in cancer. Through high-affinity engagement of the inhibitory NKG2A/CD94 receptor on NK cells and subsets of CD8+ T cells, HLA-E suppresses antitumor immune responses. However, whether this critical immune axis shapes or is shaped by the tumor microenvironment and how this differs among disease states remains unexplored.
We found that HLA-E is highly expressed in diffuse large B-cell lymphoma (DLBCL) relative to other cancer types across large-scale transcriptomic and proteomic datasets, with comparable HLA-E mRNA expression across germinal center and activated B-cell subtypes. Using immunohistochemistry, we discovered that DLBCL cases with mislocalized or absent classical HLA class I (HLA-I; 16 of 43, 37%) could still retain HLA-E expression (11 of 16, 69%). To better quantify the differences in HLA protein levels and their association with the lymphoma microenvironment (LME), we performed co-detection by indexing (CODEX) to compare the expression of HLA-E and HLA-I across 32 DLBCL cases and 4 reactive lymph nodes. Compared to normal B cells, median HLA expression was not significantly changed in the overall cohort, but the HLA-E to HLA-I ratio was increased within the 17 of 32 DLBCL cases transformed from indolent lymphomas (p=0.035), a subset associated with poor clinical outcome.
We leveraged our spatial proteomic panel to determine if DLBCL HLA-E expression levels were spatially associated with distinct cellular networks within the LME. We identified 10 cellular neighborhoods (CNs), classified as either immune-rich or lymphoma-rich, which were further subdivided by presence of HLA-Elow and HLA-Ehigh B cells. HLA-Elow CNs were either predominantly DLBCL-rich or associated with dendritic cells, whereas the HLA-Ehigh CNs preferentially clustered with more diverse immune components. Moreover, DLBCLs dominated by HLA-Ehigh CNs displayed a higher proportion of immune-rich CNs and immune cell infiltrate overall. Together, these findings suggest that HLA-E may act as a counter-regulatory mechanism under conditions of increased antitumor immune pressure.
To further assess the potential role of HLA-E in DLBCL, we characterized the presence of the NKG2A receptor in the LME. The NKG2A+ CD8+ T cells co-expressed the cytotoxic marker granzyme B (GZMB) and made up 6.9% of the entire CD8+ population, similar to the amount of PD-1+ CD8+ T cells (4.9%). In contrast, 38.6% of NK cells were NKG2A+, primarily found on CD56+ CD16- and CD56+ CD16+ subsets. These NKG2A+ cells were in closest proximity with HLA-Ehigh lymphoma cells, other T cells, and macrophages.
We validated these findings using an existing single-nucleus RNA-sequencing dataset comprising 217 large cell lymphoma and 15 reactive lymph node controls (Li et al., Cancer Cell 2025). Compared to reactive lymph nodes, KLRC1 expression was enriched within the DLBCL microenvironment. Both CD56bright and CD56dim NK cell subsets exhibited higher KLRC1 expression in treatment-naive and relapsed/refractory (R/R) DLBCL compared with control (control n = 15; treatment-naive n = 98; R/R n = 77; Wilcoxon rank-sum test, p < 0.005). Within the CD8+ T cells, KLRC1 expression was primarily found in exhausted T cells that co-expressed markers of activation/exhaustion (ENTPD1, LAG3, HAVCR2, TNFRSF9, CTLA4) and cytotoxic potential (GNLY, GZMB, PRF1, IFNG). This phenotype is consistent with observations in solid tumors, where NKG2A is acquired during chronic antigen stimulation as a late inhibitory receptor on tumor infiltrating lymphocytes. However, unlike solid tumors, KLRC1+ CD8+ T cells in DLBCL lacked co-expression of PDCD1 (PD-1) and the tissue-residency marker, ITGAE, highlighting features unique to the lymphoma microenvironment. Consistent with this, KLRC1 was enriched in the TEX archetype, an LME subtype which has been associated with poor outcomes to CAR-T cell therapy.
These findings highlight the relevance of the HLA-E/NKG2A axis as a critical and potentially targetable mechanism of immune escape in DLBCL.
DISCLOSURES
Cynthia K. Hahn, MD, PhD: No relevant conflicts of interest to declare.