Developing biomarkers of immune dysfunction to predict post-transplant AML relapse
To prevent relapse, at-risk patients must first be identified. Early recognition is essential, because established relapse compromises graft function and likely forecloses the possibility of influencing donor immune responses. Existing methods, such as minimal residual disease detection, can only be applied to a minority of patients and conveys no information regarding relapse mechanism. We hypothesise that biomarkers of immune dysfunction could predict disease recurrence in the majority of patients and guide manipulation of the donor immune response to avert relapse.
Recent studies have found that exhausted T cells (TEX) are detectable in the blood and bone marrow of patients who go on to experience relapse. We have established a study to collect peripheral blood samples at 8 timepoints from 300 transplant recipients. We are applying mass cytometry to confirm detection of TEX as a relapse biomarker and determine the time point(s) and cell surface markers that allow optimum prediction. Recent studies have also identified plasma proteomic signatures of anti-leukaemic T-cell activity. We are applying SWATH-MS to serial blood samples to discover novel protein biomarkers of relapse. We use topological data analysis and machine learning to identify distinct trajectories of relapse (Figure 2), because biomarkers of specific disease trajectories are likely to outperform those based on aggregated data.
Figure 2. Diagram depicting integrated analysis of SWATH-MS, mass cytometry and clinical data to discover signatures that predict AML relapse.
Identifying drivers of post-transplant T-cell exhaustion
Exhaustion is a distinct state of T-cell differentiation characterised by impaired effector function. Immune checkpoint inhibitors can reinvigorate or prevent the exhaustion of T cells and have revolutionised the management of several solid tumours. Evidence now implicates T-cell exhaustion as a mechanism of post-transplant AML relapse. Leukaemia-reactive exhausted T cells are present at relapse, typically expressing multiple inhibitory receptors, whose cognate ligands are expressed by AML cells (Figure 3A). Checkpoint inhibitors can induce post-transplant AML remissions, suggesting that T-cell exhaustion is a modifiable mechanism of relapse, but the increased risk of graft-versus-host disease has limited use. It is therefore necessary to identify context-specific drivers of T-cell exhaustion to inform treatments that re-establish anti-leukaemic T-cell responses without causing graft-versus-host disease.
Exhausted T cells are highly diverse, both gene expression and immunophenotype vary with clinical context. There are multiple potential drivers of exhaustion, including inhibitory cell signalling, suppressive cytokines and hypoxia. DNA sequences termed ‘enhancers’ are critical regulators of cell lineage specification, which integrate cell signals and environmental cues to determine context-specific gene expression. We are using patient samples to map enhancer activity and transcription factor binding at exhaustion-associated genes to identify the drivers of exhaustion most relevant to post-transplant AML relapse (Figure 3B).
Figure 3. (A) Mean ±SEM percentage of bone marrow CD8+ T cells expressing the indicated inhibitory receptors for cases of AML relapse (n=11) and time-matched remission (n=14) following transplant. Comparisons by unpaired t-test. TEX typically express multiple inhibitory checkpoint receptors, such as PD-1, TIM-3 & 2B4. Following transplant, isolated expression of PD-1 typically represents T-cell activation, whereas co-expression of PD-1 and TIM-3 identifies TEX with functional impairment and leukaemia-reactivity. (B) Diagram depicting the hypothesis that context-specific drivers of T-cell exhaustion can be identified by mapping enhancer activity and transcription factor binding at exhaustion-associated genes. There are multiple potential causes of T-cell exhaustion, but ultimately each must interact with the core TEX transcriptional programme to influence TEX differentiation.
Enhancers determine cell lineage specification by integrating signals from multiple pathways through the binding of different combinations of transcription factors. The pattern of enhancer activity and transcription factor binding therefore reflects the pathways that are most responsible for driving gene expression in a given context.
Inducing leukaemic differentiation to augment donor T-cell responses
In addition to T-cell exhaustion, murine studies identify poor antigen presentation as detrimental to anti-leukaemic T-cell responses. Professional antigen-presenting cells, such as macrophages, activate CD4+ T cells by displaying antigen on major histocompatibility complex class II (MHCII) together with co-stimulatory molecules. AML often expresses MHCII and transcriptional downregulation is common at post-transplant relapse, where leukaemic cells lacking MHCII elicit weaker donor T-cell responses, suggesting a mechanism of immune evasion. We are investigating the potential of compounds that induce monocyte-macrophage differentiation to drive leukaemic expression of MHCII and co-stimulatory molecules, to enable robust CD4+ T-cell activation, enhance CD8+ T-cell effector function and promote successful disease clearance.
Figure 4. Macrophage-like cells derived from patient AML. Stained for MHCII (green) and the co-stimulatory molecule CD86 (red).