Blast vacuolization in AML patients indicates adverse-risk AML and is associated with impaired survival after intensive induction chemotherapy

Deep learning predicts therapy-relevant genetics in acute myeloid leukemia from Pappenheim-stained bone marrow smears

ABSTRACT

The detection of genetic aberrations is crucial for early therapy decisions in acute myeloid leukemia (AML) and recommended for all patients. Because genetic testing is expensive and time consuming, a need remains for cost-effective, fast, and broadly accessible tests to predict these aberrations in this aggressive malignancy. Here, we developed a novel fully automated end-to-end deep learning pipeline to predict genetic aberrations directly from single-cell images from scans of conventionally stained bone marrow smears already on the day of diagnosis. We used this pipeline to compile a multiterabyte data set of >2 000 000 single-cell images from diagnostic samples of 408 patients with AML. These images were then used to train convolutional neural networks for the prediction of various therapy-relevant genetic alterations. Moreover, we created a temporal test cohort data set of >444 000 single-cell images from further 71 patients with AML. We show that the models from our pipeline can significantly predict these subgroups with high areas under the curve of the receiver operating characteristic. Potential genotype-phenotype links were visualized with 2 different strategies. Our pipeline holds the potential to be used as a fast and inexpensive automated tool to screen patients with AML for therapy-relevant genetic aberrations directly from routine, conventionally stained bone marrow smears already on the day of diagnosis. It also creates a foundation to develop similar approaches for other bone marrow disorders in the future.

Label-Free Intracellular Multi-Specificity in Yeast Cells by Phase-Contrast Tomographic Flow Cytometry

In-flow phase-contrast tomography provides a 3D refractive index of label-free cells in cytometry systems. Its major limitation, as with any quantitative phase imaging approach, is the lack of specificity compared to fluorescence microscopy, thus restraining its huge potentialities in single-cell analysis and diagnostics. Remarkable results in introducing specificity are obtained through artificial intelligence (AI), but only for adherent cells. However, accessing the 3D fluorescence ground truth and obtaining accurate voxel-level co-registration of image pairs for AI training is not viable for high-throughput cytometry. The recent statistical inference approach is a significant step forward for label-free specificity but remains limited to cells' nuclei. Here, a generalized computational strategy based on a self-consistent statistical inference to achieve intracellular multi-specificity is shown. Various subcellular compartments (i.e., nuclei, cytoplasmic vacuoles, the peri-vacuolar membrane area, cytoplasm, vacuole-nucleus contact site) can be identified and characterized quantitatively at different phases of the cells life cycle by using yeast cells as a biological model. Moreover, for the first time, virtual reality is introduced for handling the information content of multi-specificity in single cells. Full fruition is proofed for exploring and interacting with 3D quantitative biophysical parameters of the identified compartments on demand, thus opening the route to a metaverse for 3D microscopy.

VEXAS syndrome: A review of bone marrow aspirate and biopsies reporting myeloid and erythroid precursor vacuolation

Myeloid and erythroid precursor vacuolation is a common dysplastic finding associated with myeloid malignancies, toxins, drug, and nutritional deficiencies. It has been described as a core morphologic feature in VEXAS (vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) syndrome. We sought to determine the number of cases attributable to VEXAS syndrome in bone marrow biopsies and aspirates (BAMB) reporting myeloid precursor vacuolation. We reviewed 1318 individual BAMB reports from January 2020 to July 2021 where "vacuole(s)," "vacuolation," or "vacuolated" was reported. Bone marrow biopsies with vacuolation confined to blasts or those completed as routine workup prior to stem cell transplant or post induction chemotherapy for AML (acute myeloid leukemia) were excluded. Myeloid and erythroid precursor vacuolation was noted in 219 reports representing 210 patients. The most common etiology was myelodysplastic syndrome (MDS) (38.6%), AML (16.7%), lymphoproliferative disorders and multiple myeloma (7.6%), drug or toxin exposure (5.2%) myeloproliferative neoplasm (MPN) or MPN/MDS overlap syndrome (4.3%). VEXAS syndrome was determined to be the etiology in 2.9% of patients. Two additional cases of VEXAS syndrome with bone marrow biopsies reported in the specified time frame did not explicitly report myeloid or erythroid precursor vacuolation but were identified based on clinical suspicion and repeat BAMB review. Myeloid and erythroid precursor vacuolation is a dysplastic feature attributable to VEXAS syndrome in at least 2.9% of cases. Standardized reporting of vacuolization, triaging of molecular sequencing and optimal treatment of this disorder are critical issues facing those seeing patients with suspected VEXAS syndrome.

A higher percentage of leukemic blasts with vacuoles predicts unfavorable outcomes in patients with acute myeloid leukemia

Cytoplasmic vacuoles, which are a morphological feature of dysplasia, can be observed under a microscope at initial diagnosis. Recently, this typical morphological feature has been found to be associated with impaired survival. To investigate the clinical significance of the grading of blasts with vacuoles in acute myeloid leukemia (AML), we retrospectively studied 152 patients newly diagnosed with non-M3 AML. The patients were categorized into three groups according to the percentage of blasts with vacuoles (>20 %, 11-20 %, 0-10 %). A high percentage of blasts with vacuoles (>20 %) was positively associated with the European Leukemia Net (2017-ELN) high-risk AML, a complex karyotype, TP53 and IDH1/2 mutations, and CD71 expression and negatively associated with the ELN low-risk category. Importantly, patients who had a higher percentage of blasts with vacuoles had a lower complete remission rate in response to first-cycle induction chemotherapy. The overall survival and event-free survival of patients who had a higher percentage of blasts with vacuoles were significantly shorter. Moreover, multivariate analysis showed that blast vacuolization was an independent high prognostic factor for AML. In conclusion, a higher percentage of leukemic blasts with vacuoles predicts worse outcomes in AML and may have potential as a prognostic marker.