The clinical and genomic landscape of patients with DDX41 variants identified during diagnostic sequencing.

Deleterious germ line variants in DDX41 are a common cause of genetic predisposition to hematologic malignancies, particularly myelodysplastic neoplasms (MDS) and acute myeloid leukemia (AML). Targeted next-generation sequencing was performed in a large cohort of sequentially recruited patients with myeloid malignancy, covering DDX41 as well as 30 other genes frequently mutated in myeloid malignancy. Whole genome transcriptome sequencing data was analyzed on a separate cohort of patients with a range of hematologic malignancies to investigate the spectrum of cancer predisposition. Altogether, 5737 patients with myeloid malignancies were studied, with 152 different DDX41 variants detected. Multiple novel variants were detected, including synonymous variants affecting splicing as demonstrated by RNA-sequencing. The presence of a somatic DDX41 variant was highly associated with DDX41 germ line variants in patients with MDS and AML, and we developed a statistical approach to incorporate the co-occurrence of a somatic DDX41 variant into germ line variant classification at a very strong level (as per the American College of Medical Genetics and Genomics/Association for Molecular Pathology guidelines). Using this approach, the MDS cohort contained 108 of 2865 (3.8%) patients with germ line likely pathogenic/pathogenic (LP/P) variants, and the AML cohort 106 of 2157 (4.9%). DDX41 LP/P variants were markedly enriched in patients with AML and MDS compared with those in patients with myeloproliferative neoplasms, B-cell neoplasm, and T- or B-cell acute lymphoblastic leukemia. In summary, we have developed a framework to enhance DDX41 variant curation as well as highlighted the importance of assessment of all types of genomic variants (including synonymous and multiexon deletions) to fully detect the landscape of possible clinically relevant DDX41 variants.

AML classification in the year 2023: How to avoid a Babylonian confusion of languages.

In parallel to the 5th edition of the World Health Organization Classification of Haematolymphoid Tumours (WHO 2022), an alternative International Consensus Classification (ICC) has been proposed. To evaluate the impact of the new classifications on AML diagnoses and ELN-based risk classification, we analyzed 717 MDS and 734 AML non-therapy-related patients diagnosed according to the revised 4th WHO edition (WHO 2017) by whole genome and transcriptome sequencing. In both new classifications, the purely morphologically defined AML entities decreased from 13% to 5%. Myelodysplasia-related (MR) AML increased from 22% to 28% (WHO 2022) and 26% (ICC). Other genetically-defined AML remained the largest group, and the abandoned AML-RUNX1 was mainly reclassified as AML-MR (WHO 2022: 77%; ICC: 96%). Different inclusion criteria of AML-CEBPA and AML-MR (i.a. exclusion of TP53 mutated cases according to ICC) were associated with differences in overall survival. In conclusion, both classifications focus on more genetics-based definitions with similar basic concepts and a large degree of agreement. The remaining non-comparability (e.g., TP53 mutated AML) needs additional studies to definitely answer open questions on disease categorization in an unbiased way.

Explainable AI identifies diagnostic cells of genetic AML subtypes.

Explainable AI is deemed essential for clinical applications as it allows rationalizing model predictions, helping to build trust between clinicians and automated decision support tools. We developed an inherently explainable AI model for the classification of acute myeloid leukemia subtypes from blood smears and found that high-attention cells identified by the model coincide with those labeled as diagnostically relevant by human experts. Based on over 80,000 single white blood cell images from digitized blood smears of 129 patients diagnosed with one of four WHO-defined genetic AML subtypes and 60 healthy controls, we trained SCEMILA, a single-cell based explainable multiple instance learning algorithm. SCEMILA could perfectly discriminate between AML patients and healthy controls and detected the APL subtype with an F1 score of 0.86±0.05 (mean±s.d., 5-fold cross-validation). Analyzing a novel multi-attention module, we confirmed that our algorithm focused with high concordance on the same AML-specific cells as human experts do. Applied to classify single cells, it is able to highlight subtype specific cells and deconvolve the composition of a patient's blood smear without the need of single-cell annotation of the training data. Our large AML genetic subtype dataset is publicly available, and an interactive online tool facilitates the exploration of data and predictions. SCEMILA enables a comparison of algorithmic and expert decision criteria and can present a detailed analysis of individual patient data, paving the way to deploy AI in the routine diagnostics for identifying hematopoietic neoplasms.

Artificial intelligence in hematological diagnostics: Game changer or gadget?

The future of clinical diagnosis and treatment of hematologic diseases will inevitably involve the integration of artificial intelligence (AI)-based systems into routine practice to support the hematologists' decision making. Several studies have shown that AI-based models can already be used to automatically differentiate cells, reliably detect malignant cell populations, support chromosome banding analysis, and interpret clinical variants, contributing to early disease detection and prognosis. However, even the best tool can become useless if it is misapplied or the results are misinterpreted. Therefore, in order to comprehensively judge and correctly apply newly developed AI-based systems, the hematologist must have a basic understanding of the general concepts of machine learning. In this review, we provide the hematologist with a comprehensive overview of various machine learning techniques, their current implementations and approaches in different diagnostic subfields (e.g., cytogenetics, molecular genetics), and the limitations and unresolved challenges of the systems.

Improved survival in metastatic breast cancer: results from a 20-year study involving 1033 women treated at a single comprehensive cancer center.

PURPOSE: Diagnosis and treatment of breast cancer have changed profoundly over the past 25 years. The outcome improved dramatically and was well quantified for early stage breast cancer (EBC). However, progress in the treatment of metastatic disease has been less convincingly demonstrated. We have studied survival data of patients with metastatic breast cancer (MBC) from a large academic cancer center over a period of 20 years. METHODS: Data from 1033 consecutive MBC patients who were treated at the Department of Medical Oncology of the West German Cancer Center from January 1990 to December 2009 were retrospectively analyzed for overall survival (OS) and risk factors. Patients were grouped in 5-year cohorts, and survival parameters of each cohort were compared before and after adjustment for risk factors. RESULTS: Overall survival of patients with MBC treated at specialized center has significantly improved from 1990 to 2010 (hazard ratio 0.7, 95%CI 0.58-0.84). The increments in OS have become less profound over time (median OS 1990-1994: 24.2 months, 1995-1999: 29.6 months, 2000-2004: 36.5 months, 2005-2009: 37.8 months). CONCLUSION: Survival of patients with MBC has improved between 1990 and 2004, but less from 2005 to 2009. Either this suggests an unnoticed shift in the patient population, or a lesser impact of therapeutic innovations introduced in the most recent period.