The limited sensitivity of the historical treatment response criteria for acute myeloid leukemia (AML) has resulted in a different paradigm for treatment compared with most other cancers presenting with widely disseminated disease. Initial cytotoxic induction chemotherapy is often able to reduce tumor burden to a level sufficient to meet the current criteria for “complete” remission. Nevertheless, most AML patients ultimately die from their disease, most commonly as clinically evident relapsed AML. Despite a variety of available salvage therapy options, prognosis in patients with relapsed or refractory AML is generally poor. In this review, we outline the commonly utilized salvage cytotoxic therapy interventions and then highlight novel investigational efforts currently in clinical trials using both pathway-targeted agents and immunotherapy based approaches. We conclude that there is no current standard of care for adult relapsed or refractory AML other than offering referral to an appropriate clinical trial.

Measurable residual disease (MRD) testing after initial chemotherapy treatment can predict relapse and survival in acute myeloid leukemia (AML). However, it has not been established if repeat molecular or genetic testing during chemotherapy can offer information regarding the chemotherapy sensitivity of the leukemic clone. Blood from 45 adult AML patients at day 1 and 4 of induction (n = 35) or salvage (n = 10) cytotoxic chemotherapy was collected for both quantitative real-time PCR (qPCR) assessment (WT1) and next generation sequencing (>500 × depth) of 49 gene regions recurrently mutated in MDS/AML. The median age of subjects was 62 (23–78); 42% achieved a complete response. WT1 was overexpressed in most patients tested but was uninformative for very early MRD assessment. A median of 4 non-synonymous variants (range 0–7) were detected by DNA sequencing of blood on day 1 of therapy [median variant allele frequency (VAF): 29%]. Only two patients had no variants detectable. All mutations remained detectable in blood on day 4 of intensive chemotherapy and remarkably the ratio of mutated to wild-type sequence was often maintained. This phenomenon was not limited to variants in DNMT3A, TET2, and ASXL1. The kinetics of NPM1 and TP53 variant burden early during chemotherapy appeared to be exceptions and exhibited consistent trends in this cohort. In summary, molecular testing of blood on day 4 of chemotherapy is not predictive of clinical response to cytotoxic induction therapy in AML. The observed stability in variant allele frequency suggests that cytotoxic therapy may have a limited therapeutic index for clones circulating in blood containing these mutations. Further validation is required to confirm the utility of monitoring NPM1 and TP53 kinetics in blood during cytotoxic therapy.

We report here the largest study to date of adult AML patients tested for measurable residual disease (MRD) at the time of autologous hematopoietic cell transplantation (auto-HCT). Seventy-two adult patients transplanted between 2004–2013 at a single academic medical center (UCSF) were eligible for this retrospective study based on availability of cryopreserved GCSF mobilized autologous peripheral blood progenitor cell (PBPC) leukapheresis specimens (“autografts”). Autograft MRD was assessed by molecular methods (RQ-PCR for WT1 alone or a multigene panel) and by multi-parameter flow cytometry (MPFC). WT1 RQ-PCR testing of the autograft had low sensitivity for relapse prediction (14%) and a negative predictive value of 51%. MPFC failed to identify MRD in any of 34 autografts tested. Combinations of molecular MRD assays however improved prediction of post auto-HCT relapse. In multivariate analysis of clinical variables, including age, gender, race, cytogenetic risk category, and CD34+ cell dose, only autograft multigene MRD as assessed by RQ-PCR was statistically significantly associated with relapse. One year after transplantation only 28% patients with detectable autograft MRD were relapse free, compared with 67% in the MRD negative cohort. Multigene MRD, while an improvement on other methods tested, was however suboptimal for relapse prediction in unselected patients with specificity of 83% and sensitivity of 46%. In patients with known chromosomal abnormalities or mutations, however, better predictive value was observed with no relapses observed in MRD negative patients in the first year after auto-HCT compared with 83% incidence of relapse in the MRD positive patients (HR 12.45, p=0.0016). In summary, increased personalization of MRD monitoring by use of a multigene panel improved the ability to risk stratify patients for post auto-HCT relapse. WT1 RQ-PCR and flow cytometric assessment for AML MRD in autograft samples had limited value for predicting relapse following auto-HCT. We demonstrate that cryopreserved autograft material presents unique challenges for AML MRD testing due to masking effects of previous GCSF exposure on gene expression and flow cytometry signatures. In the absence of information regarding diagnostic characteristics, sources other than GCSF-stimulated PBSC leukapheresis specimens should be considered as alternatives for MRD testing in AML patients undergoing auto-HCT.