The Emerging Profile of Immunotherapy Approaches in the Treatment of AML

Dr. Daver discusses immunotherapy approaches for patients with acute myeloid leukemia (AML), and his recent study results on the combination of azacitidine and nivolumab.

Immunotherapy has become a major pillar of oncology therapy, with more than 20 approvals by the US Food and Drug Administration (FDA) for both solid tumors and hematologic malignancies in the last few years. There are two distinct approaches to immunotherapy in leukemia treatment: 1) antibody-drug conjugates that target leukemia-specific antigens, such as CD33, CD123, and CLL1, in AML and deliver a toxic payload (bacterial or chemical) into the leukemic blasts, resulting in cytotoxic cell death; and 2) T-cell–based therapies that seek to enhance T-cell responses to leukemia blasts, such as bispecific T-cell engagers, immune checkpoint antibodies, and CAR T cells. These therapies are distinct in their mechanism of action, efficacy, and toxicities.

Q: For patients with AML, what is the rationale for using immunotherapy? Are there specific characteristics of AML that make it amenable to immunotherapy approaches?

DR. DAVER: AML is one of the first diseases for which we used immunotherapy in the form of allogeneic stem cell transplant. For almost 4 decades we have been using this modality to replace the patient’s immune system with a donor immune system, and we have seen that it is quite an effective approach for treating AML. In addition, in the last 4 to 5 years, my colleagues and I in the immunotherapy group at MD Anderson, including Drs. Padmanee Sharma and James Allison, have been interrogating AML patients’ blood and bone marrow samples to try to understand the nature of T cells in AML, and our data were published recently in the journal Cancer.[1]

What we found when we looked at more than 100 patients with AML is that the T-cell population, both in the absolute numbers and the percentage, was quite preserved compared with healthy donors who had no disease. Additionally, we saw that there was an upregulation of multiple different immune checkpoint co-receptors, the most striking being PD-1 [programmed cell death 1] and OX40. This upregulation was highest in patients with relapsed AML compared with new AML, but both sets of patients had higher expression of PD-1 and OX40 compared with healthy donors. This suggested that there is a favorable T-cell profile, as well as expression of inhibitory immune checkpoint receptors, in the setting of AML, which may be one of the ways that AML tries to escape from being killed by the host defense immune system. We could potentially activate these T cells by blocking the inhibitory immune checkpoint receptors, allowing them to fight the leukemia. This form of immunotherapy-T-cell–specific immunotherapy, which includes immune checkpoint antibodies, bispecific antibodies, and CAR [chimeric antigen receptor] T cells-enhances the patient’s own immune system.

Antibody-drug conjugates are also frequently used in patients with AML, targeting leukemia-specific antigens, most commonly CD33 and CD133, as well as newer antigens like CLL1 (CLEC12a). We usually develop an antibody that also carries a toxic payload that is bacterial or chemical, and these are delivered directly to the leukemia cells. It is important to note the difference between T-cell-based and antibody-based immunotherapy, but both types seem to have preclinical and clinical efficacy in AML and are being extensively evaluated.

Q: In regard to T-cell-based immunotherapy, your recently published study in Cancer Discovery[2] tested the combination of azacitidine, a chemotherapy agent, and nivolumab, an anti-PD-1 checkpoint inhibitor, in patients with relapsed or refractory AML. Can you tell us about this trial and the results? Is this among the first positive immunotherapy clinical trials in AML?

DR. DAVER: This trial was developed a few years ago as part of an ongoing strategic alliance collaboration between MD Anderson Leukemia Department chairman Dr. Hagop Kantarjian and Bristol-Myers Squibb to evaluate multiple immunotherapy combinations in different types of leukemias, with the best combinations being further studied in larger registrational trials. The conception of this trial came from preclinical data that our colleagues at MD Anderson had generated that showed that azacitidine-which we have used for many years both in frontline and relapsed AML, as well as in myelodysplastic syndromes (MDS)-upregulated the expression of PD-1 and its ligand PD-L1 on the surface of T cells in patients with AML and MDS.[3] We believed we could use azacitidine as an immune modulator in this situation and combine it with an anti–PD-1 antibody, which was nivolumab in our study.

The study was designed in the same way as most initial first-in-human trials in any given malignancy, to see if the combination was safe and effective. We enrolled a total of 70 relapsed AML patients, allowing for both de novo and secondary AML and patients with different types of cytogenetics. The study included patients 18 years of age and older, but the median age of patients in the trial was 69 years. This was due to the fact that many physicians tend to favor hypomethylating-based treatment approaches, whether in the frontline or relapsed setting, in elderly patients. One-third of patients had adverse cytogenetics, and about 22% had TP53 mutations. The primary endpoint of this study was to assess the response rate, which was defined by the European LeukemiaNet (ELN) criteria, and included complete remission (CR) and complete remission with incomplete blood count recovery (CRi), as well as partial response and hematologic improvement.


The overall response rate was 33% in all relapsed AML patients. What was interesting was that the CR/CRi rate was 23%, but we had a number of non-traditional CR/CRi responses, including hematologic improvements in 10% and stable disease in 10%. Hematologic improvement and stable disease are not typically responses we consider when we use high-dose intensive cytotoxic therapies in AML, but these responses have been shown to be meaningful and important for quality of life, survival, and transfusion independence in elderly AML patients treated with agents such as azacitidine or IDH [isocitrate dehydrogenase] inhibitors. With immunotherapy for solid tumors, we know that a number of responses could occur in the form of stable disease and partial responses that can sometimes last for many years and can be as durable or even more durable than complete responses. We showed that we had about 10% stable disease responses lasting for more than 1 year, in addition to the documented ELN response rate of 33%. This was one of the first larger studies using an immune checkpoint-specific approach in AML to show that you can get an encouraging response rate in relapsed disease.

To assess how these study results compared with clinical trial results at MD Anderson with other azacitidine-based regimens we have evaluated, we developed a rigorous historical control that looked at 10 different trials of azacitidine or azacitidine combinations that we have done over the last 10 years. We saw that the combination of azacitidine and nivolumab had better response rates and, in fact, better survival compared with other azacitidine/decitabine-based combinations that we had studied previously, suggesting that there is something specifically beneficial with this combination. However, at the end of the day, the most important thing that we discovered was that it was specific subsets that seemed to have what we would consider the most clinically important benefit; 34% is a reasonable response rate, and the overall survival was 6 to 7 months, but that is not really a home run. What we saw was that in first salvage (salvage 1) patients who made up half of the study, we had an overall survival of almost 11 months, which was almost double the overall survival that we had seen historically with other azacitidine combinations or with azacitidine alone in prior clinical trials at MD Anderson. Similarly, patients with no prior hypomethylating agent exposure had a very healthy overall response rate of 58% vs 26% in prior hypomethylating agent–exposed patients. These two populations are likely good groups in which to consider azacitidine and nivolumab, with expectations for encouraging response rates and survival outcomes.

Additionally, we found clear biomarkers in the bone marrow or blood that could help select patients. For example, we found that pre-treatment CD3 levels in the bone marrow could predict for a higher response rate. The patients who had more than 13% infiltrating CD3 in their bone marrow had a response rate of almost 56% compared with those who had lower CD3 in their bone marrow initially, in whom the response rate was 23%. We believe that the way these drugs need to be developed-which is already what is being done for solid tumors-is to use biomarkers to select the patients who are most likely to respond to these agents rather than exposing larger numbers of patients to see a benefit in only a small proportion of them. We believe that if we use CD3 as a selection biomarker, which is a relatively simple assay that can be done using flow cytometry or immunohistochemistry, then we could actually have response rates of 50% to 60%, which is an attractive response rate for relapsed AML.

Q: You're also leading and taking part in several other immunotherapy trials in AML. Can you tell us about some of those?

DR. DAVER: We are conducting a number of different immune checkpoint inhibitor trials. We have a parallel MDS study ongoing, looking at the combination of azacitidine and nivolumab, but also azacitidine with the anti–CTLA-4 antibody immune checkpoint inhibitor ipilimumab in both frontline high-risk MDS patients with high blasts and in MDS patients who have failed hypomethylating agents. My colleague, Dr. Guillermo Garcia-Manero, presented these data last year as an oral abstract at the 2017 American Society of Hematology (ASH) Annual Meeting,[4] and presented updated data at the 2018 meeting as well.[5] What we are seeing overall is that in the frontline setting, the combination of azacitidine with ipilimumab showed very encouraging response rates and survival. Even in patients who have failed azacitidine and decitabine (so-called post–hypomethylating agent MDS), who historically have dismal response rates and survival, we see that azacitidine with ipilimumab showed interesting activity, with response rates of about 30% to 40%, as well as a signal for improved survival. We believe there may be a role for ipilimumab in MDS, and potentially in AML, since ipilimumab, but not nivolumab, seems to have both single-agent and combination activity with azacitidine.

Learning from this experience, we are also evaluating ipilimumab in AML. We have an ongoing study that is combining azacitidine with nivolumab and now adding ipilimumab at a low dose of 1 mg/kg, which is one-third of the standard dose and given every 6 weeks; an update on this triplet therapy was also presented at ASH.[6] Just as has been seen in solid tumors, in which the combination of these two immune checkpoint inhibitors usually doubled or tripled response rates, we are hoping that this combination will further improve response rates and may even work in patients who have lower CD3 infiltration in the bone marrow, thereby overcoming the impact of low CD3 as a negative predictive biomarker. There are also a number of other company-sponsored and institutional phase I/II studies with other immune checkpoint inhibitors, such as pembrolizumab, pidilizumab, and durvalumab, most in combination with azacitidine.

One approach that I think will be very interesting, and first shown by our colleagues at the Dana-Farber Cancer Institute, is using an immune checkpoint inhibitor after a stem cell transplant. It was shown that this is quite an efficacious approach in patients who have relapsed AML post-transplant: single-agent ipilimumab showed complete remission in 5 of 12 patients, including skin and lymph node clearance of disease. We have an ongoing trial looking at the safety and efficacy of combining ipilimumab and nivolumab in patients who have either relapsed disease or have high-risk features before or after transplant, including adverse cytogenetics, TP53 mutation, and minimal residual disease positivity ( identifier: NCT03600155). We believe that in the setting of a transplant, enhancing immune surveillance with these drugs could potentially improve relapse-free and overall survival in high-risk patients who often do poorly even with transplant.

Q: Are there other important immunotherapy modalities that are now being tested for AML patients that we haven't discussed yet?

DR. DAVER:There are a number of other immunotherapy modalities that have been gaining traction in the last 3 to 4 years. In fact, at the 2018 ASH meeting, a number of studies were presented on immunotherapies such as bispecific antibodies, CAR T cells, monoclonal antibodies, and, of course, immune checkpoint inhibitor antibodies. I think we are seeing what was seen with solid tumors, such as lung cancer, renal cancer, and melanoma a few years ago-initially the targeted approaches took precedence and these were developed and became mainstream, but now there are a large onslaught of immunotherapy approaches that are becoming a major part of therapy.

One approach that is going to be important in hematologic malignancies is treatment with monoclonal antibodies. There will be updates on trials of monoclonal antibodies targeting CD33 (IMGN779; identifier: NCT02674763), as well as a CD123 antibody (IMGN632; NCT03386513), that we are working on at MD Anderson. Both of these are antibody-drug conjugates that carry a potent toxin on the antibody, and both have shown single-agent complete responses in patients with AML who are relapsed after failing standard therapy. IMGN632 appears especially active, with a CR/CRi rate of 26% as a single agent in all evaluable relapsed AML patients in the phase I study. We believe that when we combine these antibodies with azacitidine, venetoclax, or other therapies, we could enhance their activity even more.

The other group of drugs to look out for are the bispecific antibodies. Blinatumomab was approved by the FDA for acute lymphocytic leukemia (ALL). Blinatumomab links to CD3 on T cells and to CD19 antigen on the blast cell, and by doing that, brings the T cell in close proximity to the ALL blast cell, resulting in T-cell mediated death of the blasts. Based on this concept, many companies have started developing bispecific antibodies for AML, including Amgen's AMG 330, Xencor's XmAb CD3-CD133 antibody, and MacroGenics' CD3-CD133 antibody. We are seeing that they have single-agent activity in the relapsed AML setting. However, thus far the response rates have been lower than what we have seen with blinatumomab in ALL patients. The responses seem to be best in patients with low-burden disease, so we believe they may actually play a role in this setting, which is also where they happen to work best in ALL patients.

Lastly, there is early development of CAR T cells in AML, although the datasets are not large or quite yet clear. In ALL and lymphoma, we have data on CAR T cells that show high response rates. There are some early signals in AML, including a City of Hope study that showed responses with CD123 CAR T-cell therapy in multiply relapsed AML.[7] However, I think CAR T-cell therapy is about 4 to 5 years away from being mainstream in this setting. It is more likely that immune checkpoint inhibitors, bispecific antibodies, and monoclonal antibodies will be commercialized in the near future for the treatment of AML.

Financial Disclosure:Dr. Daver has received research funding from AbbVie, Bristol-Myers Squibb, Daiichi Sankyo, Genentech, ImmunoGen, Incyte, Nohla Therapeutics, Pfizer, and Servier. He has served as a consultant to AbbVie, Agios, Astellas, Bristol-Myers Squibb, Daiichi Sankyo, Genentech, ImmunoGen, Incyte, Jazz Pharmaceuticals, Novartis, Pfizer, and Servier.


1. Williams P, Basu S, Garcia-Manero G, et al. The distribution of T-cell subsets and the expression of immune checkpoint receptors and ligands in patients with newly diagnosed and relapsed acute myeloid leukemia. Cancer. 2018 Nov 30. [Epub ahead of print]

2. Daver N, Garcia-Manero G, Basu S, et al. Efficacy, safety, and biomarkers of response to azacitidine and nivolumab in relapsed/refractory acute myeloid leukemia: a non-randomized, open-label, phase 2 study. Cancer Discov. 2018 Nov 8. [Epub ahead of print]

3. Yang H, Bueso-Ramos C, DiNardo C, et al. Expression of PD-L1, PD-L2, PD-1 and CTLA4 in myelodysplastic syndromes is enhanced by treatment with hypomethylating agents. Leukemia. 2014;28:1280-8.

4. Garcia-Manero G, Daver NG, Montalban-Bravo G, et al. A phase II study evaluating the combination of nivolumab (Nivo) or ipilimumab (Ipi) with azacitidine in Pts with previously treated or untreated myelodysplastic syndromes (MDS). Blood. 2016;128(suppl):abstr 344.

5. Garcia-Manero G, Sasaki K, Montalban-Bravo G, et al. A phase II study of nivolumab or ipilimumab with or without azacitidine for patients with myelodysplastic syndrome (MDS). Presented at the 2018 American Society of Hematology Annual Meeting & Exposition; Dec 1-4, 2018; San Diego, CA. Abstract 465.

6. Daver NG, Garcia-Manero G, Basu S, et al. Safety, efficacy, and biomarkers of response to azacitidine (AZA) with nivolumab (Nivo) and AZA with Nivo and ipilimumab (Ipi) in relapsed/refractory acute myeloid leukemia: a non-randomized, phase 2 study. Presented at the 2018 American Society of Hematology Annual Meeting & Exposition; Dec 1-4, 2018; San Diego, CA. Abstract 906.

7. Budde L, Song JY, Kim Y, et al. Remissions of acute myeloid leukemia and blastic plasmacytoid dendritic cell neoplasm following treatment with CD123-specific CAR T cells: a first-in-human clinical trial. Blood. 2017;130:811.