Venetoclax-based regimens for newly diagnosed patients with acute myeloid leukemia (AML) who are not suitable candidates for intensive induction chemotherapy have had rapid and widespread uptake. There are at least two reasons for this: (i) there was previously no consensus on or enthusiasm for a standard-of-care therapy in this population, and (ii) outcomes from the venetoclax treatment arms were regarded as clinically impactful.1,2 As we settle in to the post-venetoclax AML era, one thing is clear: those of us who work in the AML field are greedy. We have quickly become accustomed to a well-tolerated therapy with the potential for rapid and deep remissions, and we are done with marveling at response rates in the 60-70% range. Attention has turned to the 30-40% who do not respond to this regimen. We look forward to a future in which we develop interventions to augment or replace venetoclax-based regimens, but to reach this promised land, we must be able to reliably recognize, a priori, those patients least likely to respond. Once upon a time, when intensive induction chemotherapy was the only reasonable intervention for a patient with newly diagnosed AML, rules were written regarding who was and who was not likely to respond to these regimens. After decades of experience using induction chemotherapy, those rules were codified into prognostic systems that judgmentally labeled AML: the hoped-for “favorable” strain, the much-feared “adverse” flavor, and the murky “intermediate” group. Of course, these characteristics were never inherent to AML, but were instead a reflection of response to a particular treatment. In a world with one treatment, however, this subtlety was lost, and these categories came to define the disease subtypes themselves, not describe their response to induction chemotherapy. When another effective treatment arrived, this one with a wholly different mechanism, we had to be reminded that the traditional labels, defined by response to intensive chemotherapy, could not be extrapolated without rigorous study and testing. Indeed, as we have gained experience with venetoclax, we have learned that some traditional risk factors, such as adverse cytogenetic profiles, do not carry adverse implications.3 Others, such as TP53 mutations, still do,3 and still others that had previously been prognostically neutral, such as IDH mutations, are associated with better responses.4 But we cannot limit our analyses to traditional risk factors; biases such as biases such as these, when attempting to uncover predictors for a novel therapy, have the potential to prevent the discovery of new and important factors that may not involve chromosomal abnormalities or gene mutations.
In this issue of Haematologica, Kuusanmaki et al. and the Finnish group make further progress in advancing the field of venetoclax response prediction in AML.5 They have been leaders in this movement; 3 years ago, in this Journal, this team made the novel observation that venetoclax response might vary by the degree of maturation of AML, with more primitive disease having higher sensitivity and more mature forms having greater resistance.6 This unexpected observation of stage of differentiation as a predictive marker has since been validated, by our group and others, in retrospective studies of patients receiving treatment and with further mechanistic work.7, 8
They have now made the logical next step: seeking to predict, prospectively, whether an individual patient might respond to venetoclax with ex vivo testing. The authors designed a pilot study for newly diagnosed or relapsed/refractory AML patients who at baseline had bone marrow or peripheral blood sampled, to which multiple measures of ex vivo sensitivity testing were applied using multiple culture conditions and measures of efficacy. All patients (N=39) then received a standard venetoclax+azacitidine regimen, regardless of their sensitivity testing results, which were not communicated to the clinicians treating the patients. Comparison of the predicted versus actual response yielded an encouraging positive predictive value of 88%, and the ex vivo test was able to predict a cohort with superior overall survival.5
The group showed that not accounting for heterogeneity inherent to this disease led to false predictions of resistance. Interestingly, ex vivo efficacy was affected by culture conditions, with the strongest correlations occurring with the use of conditioned media. Furthermore, measurement by flow cytometry had the highest correlation with in vivo efficacy.5 Ultimately, this method largely recapitulates previous preclinical and clinical findings regarding the heterogeneity of response in subsets of cells with some minor exceptions that are likely due to limited representation.
Previous groups have attempted similar measures of predicting drug sensitivity ex vivo to guide therapy.9,10 Importantly, these have largely concentrated on response to conventional chemotherapy agents. Furthermore, accounting for disease heterogeneity, and utilization of multiple media conditions in an iterative fashion, makes the report by Kuusanmaki et al. distinctive and particularly exciting.
The authors highlight many of hurdles to developing their assay as a fully-realized clinical test. These include logistical and quality issues around the samples, false predictions, inability to identify small subclones, and scalability issues for its use in multiple laboratories. Addressing these challenges will not be trivial, but this process will be crucial to bringing this type of assay to the clinic.
The manuscript by Kuusanmaki et al. is an admirable first step to guiding venetoclax-based therapy prospectively by a response prediction assay that is rapid and accurate. Indeed, the authors report that they are currently using results of this assay to decide whether or not to administer venetoclax+azacitidine to relapsed/refractory AML patients in an ongoing follow-up study. If successful, one can envision a near-future clinical trial design landscape in which patients, after screening, are assigned to venetoclax with a single backbone therapy if they are predicted to respond well, a “triplet” if they might encounter resistance that the third agent could overcome, or a non-venetoclax regimen if they are likely to be refractory (Figure 1). We eagerly anticipate the next phase of their study, and hope we can continue to rely on the Finnish to diminish uncertainty in predicting venetoclax responders in AML patients.
- Received January 20, 2023
- Accepted February 7, 2023
DAP receives research funding from Abbvie and serves as a consultant for Abbvie and Genentech.
BS and DAP wrote the manuscript.
- DiNardo CD, Jonas BA, Pullarkat V. Azacitidine and venetoclax in previously untreated acute myeloid leukemia. N Engl J Med. 2020; 383(7):617-629. https://doi.org/10.1056/NEJMoa2012971PubMedGoogle Scholar
- Wei AH, Montesinos P, Ivanov V. Venetoclax plus LDAC for newly diagnosed AML ineligible for intensive chemotherapy: a phase 3 randomized placebo-controlled trial. Blood. 2020; 135(24):2137-2145. https://doi.org/10.1182/blood.2020004856PubMedPubMed CentralGoogle Scholar
- Pollyea DA, Pratz KW, Wei AH. Outcomes in patients with poor-risk cytogenetics with or without TP53 mutations treated with venetoclax and azacitidine. Clin Cancer Res. 2022; 28(24):5272-5279. https://doi.org/10.1158/1078-0432.CCR-22-1183PubMedPubMed CentralGoogle Scholar
- Pollyea DA, DiNardo CD, Arellano ML. Impact of venetoclax and azacitidine in treatment-naive patients with acute myeloid leukemia and IDH1/2 mutations. Clin Cancer Res. 2022; 28(13):2753-2761. https://doi.org/10.1158/1078-0432.CCR-21-3467PubMedPubMed CentralGoogle Scholar
- Kuusanmaki H, Kytola S, Vanttinen I. Ex vivo venetoclax sensitivity testing predicts treatment response in acute myeloid leukemia. Haematologica. 2023; 108(7):1768-1781. Google Scholar
- Kuusanmaki H, Leppa AM, Polonen P. Phenotype-based drug screening reveals association between venetoclax response and differentiation stage in acute myeloid leukemia. Haematologica. 2020; 105(3):708-720. https://doi.org/10.3324/haematol.2018.214882PubMedPubMed CentralGoogle Scholar
- Pei S, Pollyea DA, Gustafson A. Monocytic subclones confer resistance to venetoclax-based therapy in patients with acute myeloid leukemia. Cancer Discov. 2020; 10(4):536-551. https://doi.org/10.1158/2159-8290.CD-19-0710PubMedPubMed CentralGoogle Scholar
- White BS, Khan SA, Mason MJ. Bayesian multi-source regression and monocyte-associated gene expression predict BCL-2 inhibitor resistance in acute myeloid leukemia. NPJ Precis Oncol. 2021; 5(1):71. https://doi.org/10.1038/s41698-021-00209-9PubMedPubMed CentralGoogle Scholar
- Frismantas V, Dobay MP, Rinaldi A. Ex vivo drug response profiling detects recurrent sensitivity patterns in drug-resistant acute lymphoblastic leukemia. Blood. 2017; 129(11):e26-e37. https://doi.org/10.1182/blood-2016-09-738070PubMedPubMed CentralGoogle Scholar
- Snijder B, Vladimer GI, Krall N. Image-based ex-vivo drug screening for patients with aggressive haematological malignancies: interim results from a single-arm, open-label, pilot study. Lancet Haematol. 2017; 4(12):e595-e606. https://doi.org/10.1016/S2352-3026(17)30208-9PubMedPubMed CentralGoogle Scholar
Figures & Tables
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.