Amy Carpenter Aquino
September 2020—Acute myeloid leukemia was one of the first diseases for which T cells were incorporated into the therapeutic paradigm, in the form of allogeneic stem cell transplant and donor lymphocyte infusion. Why then are there no approved immune therapies, or more specifically checkpoint inhibitors, for this T-cell–sensitive disease?
That’s the question Ivana Gojo, MD, addressed in a session at last year’s AMP annual meeting. “We are just grasping whether there is a role for several novel immunotherapies including checkpoint inhibition in the therapeutic armamentarium for AML,” said Dr. Gojo, co-director of the Leukemia Drug Development Program and associate professor of oncology, Johns Hopkins School of Medicine.
It’s not yet known which patients are likely to benefit and what the best time is to incorporate immunotherapy. Nor is it known what clinical endpoints are meaningful. With the population aging (median age for AML is 68, with one-third of patients age 75 or older) and five-year survival only 27 percent, improving the outcomes through rational integration of novel interventions is the goal, she says.
“We are learning, in a way, from solid tumors that different mechanisms can contribute to the sensitivity and resistance to immune checkpoint inhibitors.”
T-cell response first requires T-cell receptor (TCR) recognition of antigen, which is presented in the context of major histocompatibility complex on the surface of antigen-presenting cells. “However, effective activation of naive T cells requires a second costimulatory signal that strongly amplifies T-cell signaling. Besides costimulatory signals, activation of T cells can be further modulated or inhibited by upregulation of checkpoint inhibitory receptors,” Dr. Gojo says.
As with solid tumors, altered antigen expression and presentation as well as upregulation of coinhibitory receptors contribute to the immune evasion of AML. “There is also uniqueness to AML in that the disease is already disseminated, or systemic at presentation, which makes it different from a solid tumor.” The bone marrow microenvironment in AML is also highly immunosuppressive in terms of cellular or metabolic composition, she says.
Multiple T-cell inhibitory and costimulatory receptors provide a balance of activation and inhibition of T-cell responses. “In AML we recognize that there are multiple inhibitory pathways that are altered and may ultimately lead to T-cell and NK-cell dysfunction.”
“It’s never a question of one unique inhibitor or stimulator,” Dr. Gojo says. “This is linked by the balance of multiple signals that ultimately lead to effective or ineffective T-cell responses.”
AML is a heterogeneous disease, which makes it a difficult disease to target with one therapy, she says. “And how do we link all this complexity of the immune system with cytogenetics, molecular complexity, epigenetics, and clonality in AML?”
Studies from the solid tumor field have shown that tumors with increased neoantigen load are most sensitive to immune checkpoint inhibition, but this is not the definition of absolute sensitivity, she says. “In other words, there is a good number of patients who have highly mutated tumors that do not respond.” The reverse is also true.
AML is a disease with very low mutational burden (Alexandrov LB, et al. Nature. 2013;500[7463]:415–421), so the question may be more one of quality than quantity of mutations, she says, “because T cells that are specifically reactive to different fusion proteins in AML, or to different mutations, have been detected in many patients with leukemia.”
Several common translocations and gain-of-function mutations (FLT3 ITD, NPM1, IDH1R132H) in AML have been found to produce leukemia-specific immunogenic proteins and induce CD4+, CD8+ T-cell, and/or humoral responses (Greiner J, et al. Blood. 2012;120[6]:1282–1289). Then, too, there is the role of leukemia-associated antigens (LAAs) in AML. “Many different LAAs are upregulated in AML, but how they relate to response to checkpoint inhibition is not fully understood,” Dr. Gojo says.
The next issue: “To activate T cells, we need to have T cells.”
In solid tumors, she says, there is a tumor milieu spectrum including those that have a high population of T cells and are “inflamed” and then those that are “deserted” or “excluded” with no detectable T cells or T cells in the periphery of the tumor, respectively.
It has been reported that the number of T cells in the peripheral blood of AML patients is equal to the number of T cells in healthy controls, Dr. Gojo says.
“Recently, data from MD Anderson suggested that distribution of CD3 T cells in patients with acute myeloid leukemia is similar to that of healthy controls. So T cells are present in the bone marrow” (Williams P, et al. Cancer. 2019;125[9]:1470–1481).
Also known is that T cells are relevant because a high percentage of T cells—CD3 and CD8 T cells—and NK cells in the bone marrow seem to be associated with better response to chemotherapy or other interventions and better survival (Ismail MM, et al. Int J Hematol. 2017;105[4]:453–464; Williams P, et al. Cancer. 2019;125[9]:1470–1481). And increased recovery of lymphocytes post-treatment is associated with reduced relapse risk (Behl D, et al. Leukemia. 2006;20[1]:29–34).
“So neoantigen load in AML is low, there are mutations in AML that may induce T-cell responses, and there are T cells in the bone marrow,” she says.
[dropcap]T[/dropcap]he next question when considering immune checkpoint inhibition is T-cell function, Dr. Gojo says. “We have known for many years that the T cells in AML patients have increased expression of activation markers but that their activation signals are perturbed and that they have difficulty in forming immunologic synapse upon target cell recognition” (Le Dieu R, et al. Blood. 2009;114[18]:3909–3916).
In murine models of AML, upregulation of checkpoint inhibitory receptors has been associated with progressive disease, she says. “And different interventions—whether genetic ablation or inhibition of checkpoint receptors in combination or with Treg [regulatory T-cell] depletion—led to the cure of some mice” (Zhou Q, et al. Blood. 2011;117[17]:4501–4510; Zhou Q, et al. Blood. 2010;116[14]:2484–2493).
Several small studies revealed that there is upregulation of coinhibitory receptors on T cells in AML patients and that T cells may have been functionally impaired. “In terms of AML blasts, upregulation of PD-L1 is clearly dependent on interferon gamma, so the data are more controversial,” she says.
“But this upregulation is also related to chemotherapy administration and exposure to hypomethylating agents.” And upregulation of inhibitory immune checkpoint molecules on AML blasts with corresponding changes on T cells is one of the mechanisms of AML relapse post-transplant.
The Johns Hopkins team tried to understand what exactly happens in T cells in AML. “We are fortunate to be able to collect a significant number of specimens in AML patients at the time of diagnosis and then serially at the time of response assessment,” Dr. Gojo says. “We found that in patients with acute myeloid leukemia, there is an increased frequency of terminally differentiated effector cells, which are enriched for antigen-experienced and senescent cells. We also found that there is a downregulation of CD27 and CD28 and upregulation of multiple inhibitory receptors, but also CD57, which is actually a ‘senescence marker’” (Knaus HA, et al. JCI Insight. 2018;3[21]:e120974).
They also studied CD8+ T-cell dysfunction in AML “because the sole expression of inhibitory receptors does not define T-cell function in AML.” Different studies suggested that cytokine expression in AML patients is not that different than in healthy controls, she says. “When we look at the global cytokine expression in T cells, you really don’t see the difference.” However, “there is a dysfunction present in CD8+ T cells in AML. There are different T-cell subpopulations, and there is an important distinction between exhaustion and senescence,” which are the dominant processes of deranged T-cell function (Knaus HA, et al. JCI Insight. 2018;3[21]:e120974).
As an example of what is functionally relevant in AML, Dr. Gojo notes that bispecific T-cell engager (BiTE) antibodies are antibodies that are designed to target AML cells by binding to CD33 or CD123 or other leukemia markers, but that they are also engineered to bind to CD3 and activate T cells.
“In in vitro studies, we observed that we can reactivate CD57-negative T cells, but CD57+ T cells—senescent cells—cannot be reactivated,” she says. “And they are the ones that have increased cytokine expression, while the remaining cells are exhausted.”
The prevailing theme of the study, she says, is that “T cells in AML have increased expression of inhibitory receptors, and downregulation of inhibitory receptors occurs in patients who achieve remission after chemotherapy, while ongoing and increased expression of inhibitory receptors—particularly in the bone marrow—occurs with disease persistence and relapse” (Knaus HA, et al. JCI Insight. 2018;3[21]:e120974).
For example, increase in T cells co-expressing PD-1 and TIM3, and PD-1 and LAG3, occurs with disease progression (Williams P, et al. Cancer. 2019;125[9]:1470–1481).
“Correlating these findings on T cells with AML genetic signatures has been challenging,” Dr. Gojo says. “And I think that the difficulty stems from the heterogeneity of this disease.”
However, the MD Anderson group found that PD-L1 upregulation might be increased on leukemia blasts in patients whose leukemia carries TP53 mutation or has an adverse karyotype, she says (Williams P, et al. Cancer. 2019;125[9]:1470–1481).
If dysfunctional T cells are there, can they be reactivated? Dr. Gojo and her colleagues at Johns Hopkins performed gene expression studies on CD8+ T cells. “We noticed that in AML patients at diagnosis, they have altered gene expression and cluster separately from healthy control CD8+ T cells,” she says. In patients who responded to induction chemotherapy—whose specimens were serially collected and who were treated with the same chemotherapy, the “response to therapy correlated with upregulation of costimulatory, and downregulation of apoptotic and inhibitory, T cell signaling pathways, indicative of restoration of T cell function,” Dr. Gojo and coauthors wrote.
“So yes, we can reactivate T cells in AML,” she says, “but whether these reactivated T cells induce durable antileukemia responses remains unknown” (Knaus HA, et al. JCI Insight. 2018;3[21]:e120974).
[dropcap]F[/dropcap]rom a clinical perspective, the most interesting data came initially with the introduction of the anti-CTLA-4 antibody ipilimumab (Yervoy, Bristol-Myers Squibb) in patients relapsing after allogeneic stem cell transplant, Dr. Gojo says, where ipilimumab was given to potentially reactivate graft-versus-leukemia effect. While the reactivation of the immune system can be associated with adversity such as development of graft-versus-host disease, she says, in this study of 28 patients (12 AML), administration of ipilimumab led to remissions in five patients with AML. “Interestingly, four of the five patients who responded to therapy had extramedullary disease.” The achievement of complete remission was associated with systemic changes such as decreased activation of Tregs, expansion of effector T cells, and specifically infiltration of cytotoxic CD8+ T cells at the tumor site (Davids MS, et al. N Engl J Med. 2016;375[2]:143–153).
“Many studies of novel immunotherapeutic strategies are ongoing in AML in the post-transplant setting, and inhibition of PD-L1/PD-1 pathway appears to be associated with increased risk of graft-versus-host disease,” she says.
Other studies in AML are examining the combination of hypomethylating agents—the standard therapy for AML in older patients—and checkpoint inhibitors. Hypomethylating agents have “multiple immunologic effects such as the upregulation of checkpoint molecules, induction of inflammatory milieu, and induction of leukemia-associated antigens, among others,” Dr. Gojo says.
In MD Anderson’s phase two, single-arm study of anti-PD-1 antibody nivolumab (Opdivo, Bristol-Myers Squibb) and azacitidine in 70 relapsed/refractory AML patients, the treatment combination led to an overall response rate of 33 percent, including 22 percent with complete remission or complete remission with incomplete count recovery. While the single-arm nature of this study limits the interpretation of clinical benefit, “the authors concluded there is improvement in overall survival compared with their historical controls,” Dr. Gojo says (Daver N, et al. Cancer Discov. 2019;9[3]:370–383).
“We learned from this study that the only predictors of response to therapy for this combination was the presence of T cells in the bone marrow, and the higher percentages of CD3+ and CD8+ T cells in the bone marrow were associated with improved responses.”
In nonresponders to therapy, CTLA-4 was significantly upregulated on CD4+ effector T cells after four doses of nivolumab.
In terms of genomic predictors, the study found that only the presence of the ASXL1 mutation in the leukemia cells was associated with improved responses. Patients with earlier relapses fared better than patients with advanced disease, which correlated with the presence of T cells in the bone marrow, Dr. Gojo says. “In other words, the extensive prior therapy may lead to a depletion of T cells in the bone marrow and may limit responses.”
“This same group conducted, in the context of the same study, double-checkpoint inhibition in a separate cohort of patients and showed improved survival when ipilimumab and nivolumab were combined with azacitidine” (Daver NG, et al. Blood. 2019;134[suppl 1]:830).
At Johns Hopkins, Dr. Gojo and colleagues are seeing similar results with a combination of a different anti-PD-1 antibody, pembrolizumab (Keytruda, Merck), and azacitidine (Gojo I, et al. Blood. 2019;134[suppl 1]:832). “It will require a randomized study,” she says, noting they saw improved responses in newly diagnosed patients. “So it seems that earlier introduction of a checkpoint inhibitor might be more beneficial clinically.”
Chemotherapy has also been combined with checkpoint inhibitors. Anthracyclines, an AML therapy mainstay, can induce immunogenic cell death, leading to cell surface translocation of calreticulin, and pretreatment calreticulin exposure on AML blasts correlates with superior overall survival post-treatment (Fucikova J, et al. Blood. 2016;128[26]:3113–3124).
An MD Anderson study of 44 patients (AML, 95 percent, and high-risk MDS) looked at the responses among newly diagnosed AML patients, median age 54, who received nivolumab in combination with cytarabine and idarubicin. “Since it’s not a randomized study, I cannot comment on the improvement of outcomes,” Dr. Gojo says of the phase two, single-arm study. “There was some experience of grade III–IV graft-versus-host disease in patients who proceeded subsequently to transplant”—five of 19 (26 percent). “But in terms of predictor response, what we know is that, at baseline, a higher frequency of PD-1 and TIM3 co-expressing CD4+ effector cells was observed in the BM of nonresponders, and that TP53 mutation was enriched among nonresponders” (nonresponders 40 percent, responders 12 percent) (Ravandi F, et al. Lancet Haematol. 2019;6[9]:e480–e488).
Can we improve the selection of patients who might benefit from checkpoint inhibition post-transplant? For use of checkpoint inhibitors such as anti-PD-1 in the post-transplant setting, Dr. Gojo says, “there is a lot of caution.” A recent study identified that a transcriptional signature of AML blasts at the time of post-allogeneic transplant relapse is highly enriched in immune-related processes, she says.
That was not the case, however, for AML relapses solely after chemotherapy. “So there is a difference between these two AML relapse situations.”
Further analysis of the biological processes revealed that AMLs relapsing post-allogeneic transplant can be divided into two groups: those who have downregulation of HLA class II molecules and those who have upregulation of inhibitory molecules.
“Why is this relevant? Because clinically, downregulation of HLA II molecules will limit antigen presentation. Therefore, administering checkpoint inhibitors is unlikely to be of benefit in this situation. Since this is epigenetic change, administration of interferon gamma to create the inflammatory environment and upregulate the expression of HLA II molecules may be beneficial. However, in the small percent of AML patients who don’t have downregulation of HLA II molecules but have upregulation of inhibitory receptors, the treatment with checkpoint inhibitors might be beneficial in reactivating those T cells.” Knowing the status, then, she says, might be relevant in selecting those patients who might benefit from checkpoint inhibition versus those who don’t (Toffalori C, et al. Nat Med. 2019;25[4]:603–611).
Amy Carpenter Aquino is CAP TODAY senior editor.