High-Dimensional Analysis of Acute Myeloid Leukemia Reveals Phenotypic Changes in Persistent Cells during Induction Therapy

Abstract

The plasticity of AML drives poor clinical outcomes and confounds its longitudinal detection. However, the immediate impact of treatment on the leukemic and non-leukemic cells of the bone marrow and blood remains relatively understudied. Here, we conducted a pilot study of high dimensional longitudinal monitoring of immunophenotype in AML. To characterize changes in cell phenotype before, during, and immediately after induction treatment, we developed a 27-antibody panel for mass cytometry focused on surface diagnostic markers and applied it to 46 samples of blood or bone marrow tissue collected over time from 5 AML patients. Central goals were to determine whether changes in AML phenotype would be captured effectively by cytomic tools and to implement methods for describing the evolving phenotypes of AML cell subsets. Mass cytometry data were analyzed using established computational techniques. Within this pilot study, longitudinal immune monitoring with mass cytometry revealed fundamental changes in leukemia phenotypes that occurred over time during and after induction in the refractory disease setting. Persisting AML blasts became more phenotypically distinct from stem and progenitor cells due to expression of novel marker patterns that differed from pre-treatment AML cells and from all cell types observed in healthy bone marrow. This pilot study of single cell immune monitoring in AML represents a powerful tool for precision characterization and targeting of resistant disease.

Fig 1. Overview of mass cytometry phenotyping in early AML therapy.

(A) A timeline of AML induction shows blood and bone marrow collection goals for phenotypic comparisons as in Fig 2. Samples collected and analyzed for individual patients are listed in Fig 3 and Figure A in S1 File. (B) viSNE analysis of all live cells from the diagnosis marrow of one AML patient F004 is shown. Cells were arranged on the viSNE map along unitless x and y viSNE axes according to 27-dimensional phenotype (Table A in S1 File) so that phenotypically similar cells were placed near each other. Cellular abundance is indicated with a shaded contour plot where outliers start at 10% and each 2% contour is shaded a lighter color from purple to yellow. (C) On the same viSNE axes as in (B), diagnostic bone marrow cells from patient F004 were graphed and shaded according to identity determined by immunophenotype. AML blast cells were shaded red and non-blast cells were shaded grey. (D) On the same viSNE axes as in (B), diagnostic bone marrow cells from patient F004 were graphed and shaded according to expression of CD45 on a rainbow heatmap (log-like arcsinh₁₅ scale).

Fig 2. Phenotypic distance from hematopoietic stem cells distinguishes healthy cell populations and AML blasts from different individuals.

(A) A 27-dimensional viSNE analysis compares an equivalent number of live cells from normal bone marrow and from each of four AML patient bone marrow samples obtained at diagnosis prior to treatment. (B) 27-dimensional phenotypic distance of normal bone marrow mononuclear populations from healthy hematopoietic stem cells (HSCs) was measured in the viSNE analysis from (A) and is shown in blue. (C) As in (B), the HSC distance for the blast populations from four AML patients was measured and is shown in blue.

Fig 3. Computational analysis of samples throughout induction allows visualization of both remission and persistent AML.

27-dimensional viSNE analysis of all live cells from all sample collection time points for two AML patients is shown (left). Clinical response is indicated for each patient on the left. Leukemic blast areas were determined by location of cells at diagnosis and analysis of marker expression, as in Fig 1. At right, cells taken from the time point shown (red) were compared to all cells (grey). Differences in the location of cells within the viSNE map resulted from changes in protein expression.

Fig 4. Analysis of immunophenotypic change throughout induction—F001 and F003.

At top, mass cytometry quantification of blast percentage for each sample from two patients is shown. Below, clinical cytometry and microscopy data of each blood (clinical samples collected at different time on same day) and bone marrow (clinical and research samples collected at same time) sample throughout induction is shown. The patient whose samples are displayed on the left (F001) achieved remission (defined as <5% marrow blasts at recovery). On the right, a refractory patient’s samples are shown, in whom a very high blast percent were seen at Day 14.

Fig 5. Rare subsets at diagnosis become prominent in persistent leukemia in a patient with refractory AML.

(A) Blasts from Day 0 and Day 14 were gated out from prior viSNE maps for patient F003 (shown in Fig 3) and remapped in viSNE together. Gates were drawn around subpopulations and based on relative cell abundance, as in Fig 1B. (B) Percent of total blasts for each gate at both time points. (C) A heat map of median marker expression for each gate at both time points is shown.

Fig 6. Single cell analysis of immunophenotype in AML subpopulations at diagnosis and mid-induction persistence.

Biaxial density plots show the markers with the highest standard deviation across subsets compared with CD34 on individual cells in live bone marrow AML blasts from patient F003. Columns show AML blast populations identified (Fig 5) from Day 0 or Day 14, whichever was more abundant. Red boxes highlight key changes in protein expression discussed in the text. Standard deviation (SD) of each row’s marker across all populations is indicated to the right of the plots. Stem index (Fig 2) of each population is indicated above each column. CD34 and the top 11 most variable markers were graphed. CD4 was included as an example of a low expression marker that did not change.

Fig 7. Blast populations become increasingly phenotypically different from both differentiated cells and non-malignant hematopoietic stem cells.

27-dimensional viSNE analysis compared normal bone marrow cells, AML blasts from Day 0 from patient F003 (subpopulations 1–6 only), and AML blasts from Day 14 from patient F003 (subpopulation 7–13 only). (A) Normal bone marrow mononuclear cells are shown (left). A cartoon outline of the major healthy subpopulations was identified and HSC distance measured (right). (B) Pre-treatment (Day 0) and post-treatment (Day 14) samples are compared same map as in (A) and HSC distance of each subpopulation is shown.