AML relapse after a TIGIT race

In this issue of Blood, Gournay et al¹ dissect donor immune reconstitution after allogeneic hematopoietic stem cell transplantation (HSCT) with mass cytometry and identify T-cell immunoreceptor with Ig and ITIM domains (TIGIT) and CD161-expressing CD4⁺ T cells as early immune correlates of subsequent acute myelogenous leukemia (AML) relapse after HSCT.

The high relapse rate of AML after stem cell transplant has remained one of the most tenacious problems in malignant hematology. With the graft-versus-leukemia (GVL) effect at the core of successful long-term disease control, the early posttransplant course can be thought of as a delicate race between nascent donor immune reconstitution and recovering malignant cell populations that seek to escape GVL. AML relapse after HSCT thus often associates with donor immune cell dysfunction through upregulated immune checkpoint molecules or reduced antigen presentation, which leads to leukemic relapse through impaired recognition and killing of malignant cells.² 

Clinical studies with immune checkpoint blockade have demonstrated that relapsed AML after transplant can be successfully treated by closing loopholes in donor immunity, for example, using CTLA-4 blockade in cases of extramedullary AML, in which it induced long-term disease control through local CD8⁺ T-cell infiltration.³ However, such durable remissions following immune checkpoint blockade are the exception rather than the norm, and a deeper understanding is still lacking regarding the composition and interactions among donor immune cell populations that are needed to drive such effective GVL responses.

Mass cytometry is a systems immunology approach for the unbiased discovery of immune cell subpopulations and their dynamics at single-cell resolution that builds on the capability to capture dozens of surface markers. This permits extension of classical flow cytometry approaches to a much higher resolution and can identify rare, previously unknown cell subsets. The wealth of information provided per cell by mass cytometry requires dimensionality reduction techniques and unsupervised clustering for data analysis, which overcome the limitations of classical manual gating strategies to achieve the analytical depth and speed required for analysis of such data.

Gournay et al studied peripheral blood samples prospectively collected from 2 large cohorts of patients following allogeneic stem cell transplantation, which they analyzed using mass cytometry with a 45-antibody panel (see figure). The first cohort consisted of peripheral blood collected longitudinally at 3, 6, and 12 months after transplant from 37 patients with AML/MDS in remission, providing the opportunity to define immune cell kinetics during the crucial first year after transplant, when most cases of AML relapse occur. As comparator, the authors profiled peripheral blood from 20 healthy donors. As expected, based on longstanding characterizations in the field, immune reconstitution early after transplant (3 months) was dominated by innate cell types such as natural killer (NK) cells and monocytes, whereas adaptive immunity and professional antigen-presenting cells gradually recovered at later timepoints. Notably, however, they demonstrated that the circulating immune compartment after transplant, despite ongoing reconstitution, remained profoundly altered even after 12 months. Globally increased expression of immune checkpoint molecules, such as PD-1, LAG3, and TIGIT, were detected across T and NK cells. Moreover, higher TIGIT expression on several T-cell subsets was associated with subsequent AML relapse, suggesting that the immune checkpoint molecule is of high relevance for posttransplant relapse.

As follow-up to this finding, the authors evaluated a second cross-sectional cohort in which they compared circulating immune cells sampled 3 months after transplant from 20 patients with AML with sustained remission with cells from 20 patients who subsequently relapsed after stem cell transplantation. Through differential analysis of immune cell clusters associated with relapse, they again linked TIGIT⁺ CD4⁺ T cells with disease recurrence. TIGIT⁺ CD4⁺ T cells showed broad evidence of immune activation as they coexpressed other checkpoint molecules such as PD-1 or B- and T-lymphocyte attenuator and had high Ki-67 and human leukocyte antigen-DR isotype expression, suggesting that they represent T cells with ongoing antigen exposure. A second CD4⁺ T-cell population with high expression of CD161 (KLRB1) was also associated with AML relapse and shorter disease-free survival after transplant. CD161 expressing T cells could be another high-priority target for immune modulation, especially because the ligand C-type lectin domain family 2 member D (CLEC2D) is widely expressed by AML blasts, and recent work with glioma-infiltrating T cells has demonstrated that inhibition of CD161 may increase activation and cytotoxicity of tumor-infiltrating T cells.⁴ 

Overall, this study confirms and extends prior work to identify signatures of post-HSCT immune escape⁵ and makes a strong case for the further investigation and clinical testing of additional immune checkpoint molecules beyond classical CTLA-4 and PD-1 in the context of AML relapse after allogeneic stem cell transplantation. Such efforts are currently already underway in studies that test inhibitors of TIM-3 and CD47 in transplant-naïve AML and MDS.^6,7 Although therapeutic inhibition of TIGIT has not yet reached AML/MDS, it is under investigation in several solid tumors such as non–small cell lung cancer, in which signs of clinical activity have been documented.⁸ The observation by Gournay et al that TIGIT⁺ CD4⁺ T cells in peripheral blood of patients with AML after HSCT coexpress other immune checkpoint molecules suggests that combinatorial inhibition could be a relevant strategy. In fact, combined TIGIT and PD-1 inhibition has been documented to provide synergistic effects.⁹ An important question for the posttransplant setting is the amount of immune toxicity induced by these novel immune checkpoint inhibitors because nivolumab monotherapy already has been shown to generate high rates of graft-versus-host disease.¹⁰ 

More broadly, Gournay et al provide a compelling testimony to the power of large-scale longitudinal analyses of thoughtfully curated biospecimen collections with systems biology approaches to deeply dissect the immunological race between donor and recipient after stem cell transplantation. Intriguingly, such an approach could be conceived as a future clinical tool for real-time monitoring of donor immune reconstitution in the early posttransplant setting. This approach may aid clinicians in preemptively intervening with immune checkpoint blockade as dysfunctional immune cells arise to ensure that donor immune cells come out ahead in the race against AML.