Through single-cell RNA sequencing, two groups of researchers offer new insight into the respective roles of CD8+ T cells and tumor cells in driving immunotherapy resistance. Their studies pinpoint two distinct T-cell states, as well as a T-cell exclusion program promoted by tumor cells, that may influence patients' likelihood of response to immune checkpoint blockade.

In piecing together the puzzle that is resistance to immune checkpoint blockade, researchers have lately leveraged techniques such as single-cell RNA (scRNA) sequencing to better pinpoint molecular features driving immune evasion. Two recent, simultaneously published studies offer new insight into the respective roles of CD8+ T cells and tumor cells in stifling some patients' response to anti–PD-1 therapy.

“So far, PD-L1 expression has been used to predict response to immunotherapy, but it's not a great biomarker,” says one study's co–senior author, Nir Hacohen, PhD, of Massachusetts General Hospital in Boston and the Broad Institute in Cambridge, MA (Cell 2018;175:998–1013). Hacohen's team decided to closely examine tumor-infiltrating CD8+ T cells isolated from the biopsies of 32 patients with melanoma who had received an immune checkpoint inhibitor.

“We identified two distinct T-cell states associated with either good or bad clinical outcomes,” Hacohen says. T cells in one state, CD8_G, expressed a protein called TCF7, displayed memory and self-renewal features, and were enriched in treatment-responsive tumors; T cells dubbed CD8_B expressed exhaustion markers, chiefly TIM3 and CD39, and were clustered in nonresponsive tumors.

“CD8+ T cells are required for tumor killing when anti–PD-1 therapy is given, and we think cells in this memory- and stem-like state may serve as a source,” Hacohen explains. “We didn't realize such a subpopulation was necessary before—we thought simply reinvigorating exhausted T cells with checkpoint blockade would be sufficient.”

Finding ways to tweak the balance in favor of CD8_G T cells could well increase a given patient's likelihood of response to immunotherapy, Hacohen says. Additionally, because TCF7 expression “does seem to distinguish between these two T-cell states,” he adds, once prospective validation studies are carried out, “we could turn our findings into a practical predictor of clinical outcome, using simple two-color costaining of TCF7 with CD8.”

The second study, led by researchers at Dana-Farber/Brigham and Women's Cancer Center, also in Boston, as well as at the Broad, focused on exploring melanoma cells themselves for potential resistance mechanisms (Cell 2018;175:984–97).

“Conceptually, there are two key tumor cell–mediated resistance mechanisms to checkpoint blockade—immune evasion by downregulating antigen presentation, for instance, and excluding T cells from the tumor microenvironment,” explains co–senior author Benjamin Izar, MD, PhD. “The first is more readily measurable, but assessing the second systematically has been challenging.”

“We developed an algorithm from which we derived a signature, or program, specific to malignant cells—among other mechanisms, it induces resistance by promoting the exclusion of tumor-infiltrating T cells,” says first author Livnat Jerby-Arnon, PhD.

The team reported that this program can be intrinsic to tumor cells rather than acquired, and its overexpression was associated with poor survival in 473 patients with melanoma from The Cancer Genome Atlas. They also assessed the signature's predictive value in a separate cohort of 112 patients, analyzing biopsies obtained before and after checkpoint therapy. “We found that it could predict complete and partial responses, as well as progression-free survival,” Izar says. “It may be more robust than measuring PD-L1 expression or even tumor mutational burden alone.”

graphic

Two-color costaining shows TCF7+ and TCF7- CD8+ T cells from a patient with melanoma. TCF7 is in pinkish red, and CD8, which marks cytolytic T cells, is in green.

Diving deeper to see if their signature “encapsulated some biology that could be pharmacologically modulated,” Izar adds, the researchers discovered that CDK4/6 acts as a regulator of the program. In cell lines, the CDK4/6 inhibitor abemaciclib (Verzenio; Eli Lilly) repressed this resistance program; adding abemaciclib to checkpoint blockade also enhanced the latter's efficacy in a mouse model of melanoma.

“If we can better identify patients with high expression of this resistance program, it might be a good idea to combine these agents up front with immunotherapy,” says Charles Yoon, MD, PhD, another co–senior author. For now, though, clinical trials testing whether targeting CDK4/6 can reverse such resistance in people, as well as the potential toxicity associated with this combination, remain to be done.

To both teams, these findings offer different, albeit complementary, avenues to continue probing the why and how of immunotherapy resistance. “Digging into just one aspect will never be as effective,” Yoon observes. “That said, even if we had full information on T and tumor cells, we'd still need details on tumor architecture and microenvironment, for a more complete picture.”

“Both studies do suffer a bit from the fact that tumors get disaggregated during single-cell profiling,” Izar agrees, “but we've begun addressing this issue using multiplexed imaging approaches. There are a lot of efforts now not only to obtain high-resolution scRNA sequencing data, but to map it back to its spatial context within the tumor, which will provide a critical layer of information.”

Hacohen and Jerby-Arnon note that combining the two data sets could increase their predictive power. “Some patients may have one defect but not the other, so it'd be useful to know whether we're dealing with T-cell exclusion versus not enough memory-like T cells,” Hacohen says.

“Because every patient's tumor is a bit different,” he says, “it would be great to someday have a set of simple tests that could be done to determine therapeutic direction.” –Alissa Poh

For more news on cancer research, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/CDNews.