Checkpoint-blocked T cells checkmate AML

In this issue of Blood, Penter et al report the multimodality immune profiling of patients enrolled in a prospective trial to treat relapsed myeloid malignancies with checkpoint CTLA-4 blockade ipilimumab. They show distinct cytokine and cellular changes in responders and add to our understanding of graft-versus-leukemia (GVL) effects.¹ 

Allogeneic hematopoietic stem cell transplantation (HSCT) has improved outcomes for high-risk acute myeloid leukemia (AML), largely attributed to graft-versus-leukemia (GVL) effects.²  The seminal study from Horowitz et al demonstrated the importance of GVL, with increased relapse in those with impaired or absent T-cell function.³  However, some of this benefit is counterbalanced by the deleterious effects of life-threatening graft-versus-host disease (GVHD), the off-target T-cell attack of healthy tissues that constrains efforts to indiscriminately enhance T-cell immunity.⁴  Understanding the drivers of GVL remains the ultimate yet elusive goal in improving the dismal outcomes of relapsed AML after transplant.

The reasons for insufficient T-cell immunity in AML relapse after HSCT are poorly understood. Does relapse occur because the infused anti-AML T cells are absent or because their activity is constrained and insufficient to maintain remission? The anti-AML T cells could be absent at the time of relapse, either because they were lacking in the infused product, or because GVHD prophylaxis agents eradicated them. Alternatively, it is possible that these anti-AML T cells are present but in scarce numbers or with insufficient activation or homing signals to elicit effective GVL. Finally, is the biology of robust AML immunity and GVHD identical? Could there be nuances that could be exploited clinically as has been suggested in preclinical models?⁵ 

In this issue, Penter et al provide data that begin to address these key questions. They interrogated tissue and blood samples to expose the lymphocyte and cytokine/chemokine alterations after CTLA-4 blockade for patients with relapsed AML after HSCT. Their data support that some patients have anti-AML-specific T cells at the time of relapse and that these could be harnessed to mediate effective GVL using checkpoint inhibitors. These agents inhibit the signals that curtail T-cell responses, enhancing circulating T-cell activity. This group had previously reported the clinical response using ipilimumab to block CTLA-4,^6,7  and now present immune profiling of these patients. They query the biology of improved GVL after checkpoint blockade and whether GVL findings were distinct from GVHD.

As expected, ipilimumab led to a global reduction in naive T cells and enrichment of effector memory T cells, reflecting the expansion and maturation of circulating T cells following checkpoint blockade; their newer modalities of immune profiling confirm prior studies.⁸  However, these modern approaches revealed specific cellular and milieu changes that were linked to AML responses. In nonresponders, flow cytometry by time of flight showed evidence of simultaneous CD8⁺ T-cell activation and exhaustion. In complete responders, clonotype analysis confirmed increased CD8⁺ T-cell infiltration in tumor sites. Furthermore, AML complete response after ipilimumab was associated with upregulation of genes of activation and T-cell receptor signaling absent in nonresponders, using differential gene expression analysis. A similar genetic expression profile was observed in the tissues affected by GVHD, consistent with CTLA blockade mediating either effect. Thus, this work supports the hypothesis that many relapsed AML patients have circulating T cells that could support robust GVL if galvanized through checkpoint blockade, albeit with increased GVHD. Therefore, donor lymphocyte infusions may not be required. Similarly, new donors may not be needed to treat relapsed AML, but rather a new approach to the anti-AML T-cell number and activation after transplant is needed.

The authors also show evidence of enhanced T-cell honing to tumor sites after ipilimumab exposure in responders (see figure). There was increased T-cell receptor diversity in disease sites in responders, consistent with new T-cell clones mobilized to these sites. If this finding is confirmed, it suggests that ipilimumab may induce epitope spreading, which is the recruitment of other T-cell clones to mediate tumor eradication. Epitope spreading could be beneficial even in non-HSCT settings, potentially inducing resting antileukemia T cells to participate in antitumor immunity, thereby helping to overcome tumor evasion mediated via downregulation of the target antigens, as has been seen in many immunotherapy settings.⁹ 

Penter et al also report that responders had marked upregulation of proinflammatory cytokine and chemokines involved with T-cell proliferation, activation, infiltration, effector function, as well as inflammation or activation of other cellular subsets, using differential protein expression. Few of these cytokines have been tested clinically to evaluate whether they elicit antileukemia responses. If confirmed in other studies and supported by preclinical work, some of these may offer novel targets to evaluate for in vitro or in vivo T-cell stimulation to enhance GVL, although the in vivo administration may be similarly plagued by GVHD akin to checkpoint blockade. It is also possible though that this inflammatory milieu was sparked by robust GVL and that the nonresponders were nonresponders due to either lack of anti-AML T cells or rapid exhaustion of antitumor immunity.

Although these data include few patients and early data, this article provides important insights for the field. The multimodality approach led to internal confirmation of some findings through independent methods, increasing the confidence of such data in a complex clinical setting. These studies could provide a roadmap to interrogate samples for future leukemia immunotherapy trials. Using this approach, this work adds to others that have supported the concept that GVL may be present but insufficient in AML relapse, and may be enhanced through checkpoint blockade.^8,10  Penter et al offer an early glimpse into the mechanisms that mediate GVL and GVHD after AML relapse, which may promote novel therapies if confirmed in larger studies. Such translational studies that identify the biology underlying AML relapse and that of successful immunotherapy will likely be critical to promote GVL and avert severe GVHD. That said, most would agree that a good strategy would be a move like the Queen’s gambit, sacrificing a small pawn, a little GVHD, to control the board and checkmate AML.