MOSAIC: An Artificial Intelligence-Based Framework for Multimodal Analysis, Classification, and Personalized Prognostic Assessment in Rare Cancers.

PURPOSE: Rare cancers constitute over 20% of human neoplasms, often affecting patients with unmet medical needs. The development of effective classification and prognostication systems is crucial to improve the decision-making process and drive innovative treatment strategies. We have created and implemented MOSAIC, an artificial intelligence (AI)-based framework designed for multimodal analysis, classification, and personalized prognostic assessment in rare cancers. Clinical validation was performed on myelodysplastic syndrome (MDS), a rare hematologic cancer with clinical and genomic heterogeneities. METHODS: We analyzed 4,427 patients with MDS divided into training and validation cohorts. Deep learning methods were applied to integrate and impute clinical/genomic features. Clustering was performed by combining Uniform Manifold Approximation and Projection for Dimension Reduction + Hierarchical Density-Based Spatial Clustering of Applications with Noise (UMAP + HDBSCAN) methods, compared with the conventional Hierarchical Dirichlet Process (HDP). Linear and AI-based nonlinear approaches were compared for survival prediction. Explainable AI (Shapley Additive Explanations approach [SHAP]) and federated learning were used to improve the interpretation and the performance of the clinical models, integrating them into distributed infrastructure. RESULTS: UMAP + HDBSCAN clustering obtained a more granular patient stratification, achieving a higher average silhouette coefficient (0.16) with respect to HDP (0.01) and higher balanced accuracy in cluster classification by Random Forest (92.7% ± 1.3% and 85.8% ± 0.8%). AI methods for survival prediction outperform conventional statistical techniques and the reference prognostic tool for MDS. Nonlinear Gradient Boosting Survival stands in the internal (Concordance-Index [C-Index], 0.77; SD, 0.01) and external validation (C-Index, 0.74; SD, 0.02). SHAP analysis revealed that similar features drove patients' subgroups and outcomes in both training and validation cohorts. Federated implementation improved the accuracy of developed models. CONCLUSION: MOSAIC provides an explainable and robust framework to optimize classification and prognostic assessment of rare cancers. AI-based approaches demonstrated superior accuracy in capturing genomic similarities and providing individual prognostic information compared with conventional statistical methods. Its federated implementation ensures broad clinical application, guaranteeing high performance and data protection.

Early transfusion patterns improve the Molecular International Prognostic Scoring System (IPSS-M) prediction in myelodysplastic syndromes.

BACKGROUND: The Molecular International Prognostic Scoring System (IPSS-M) is the new gold standard for diagnostic outcome prediction in patients with myelodysplastic syndromes (MDS). This study was designed to assess the additive prognostic impact of dynamic transfusion parameters during early follow-up. METHODS: We retrieved complete transfusion data from 677 adult Swedish MDS patients included in the IPSS-M cohort. Time-dependent erythrocyte transfusion dependency (E-TD) was added to IPSS-M features and analyzed regarding overall survival and leukemic transformation (acute myeloid leukemia). A multistate Markov model was applied to assess the prognostic value of early changes in transfusion patterns. RESULTS: Specific clinical and genetic features were predicted for diagnostic and time-dependent transfusion patterns. Importantly, transfusion state both at diagnosis and within the first year strongly predicts outcomes in both lower (LR) and higher-risk (HR) MDSs. In multivariable analysis, 8-month landmark E-TD predicted shorter survival independently of IPSS-M (p < 0.001). A predictive model based on IPSS-M and 8-month landmark E-TD performed significantly better than a model including only IPSS-M. Similar trends were observed in an independent validation cohort (n = 218). Early transfusion patterns impacted both future transfusion requirements and outcomes in a multistate Markov model. CONCLUSION: The transfusion requirement is a robust and available clinical parameter incorporating the effects of first-line management. In MDS, it provides dynamic risk information independently of diagnostic IPSS-M and, in particular, clinical guidance to LR MDS patients eligible for potentially curative therapeutic intervention.

How to classify risk based on clinical and molecular modeling: integrating molecular markers in the risk assessment of myelodysplastic syndrome.

Myelodysplastic syndrome (MDS), also known as "myelodysplastic neoplasm," is a heterogeneous group of clonal myeloid neoplasms that typically affects older adults. The clinical phenotype, symptoms, and complications relate to the depth of cytopenia and progression to acute myeloid leukemia (AML). The diagnosis of MDS relies on morphologic criteria, such as evidence of dysplasia, disordered maturation, and increasing blast counts, which separate the disease into histologic subtypes with different probabilities for progression to AML. The treatment of MDS is often risk-adapted depending on the prognostic profile of each patient's disease. There has been a coevolution of diagnostic and prognostic systems for MDS developed over the past 40 years, both of which have now incorporated molecular markers. The new International Prognostic Scoring System-Molecular (IPSS-M) improves partitioning of patients compared to prior versions with resultant upgrading of 34% of patients into higher-risk groups due to the presence of mutations. The new IPSS-M also more accurately distinguishes intermediate-risk patients separating them into two tiers. The two new diagnostic classifications include MDS defined by mutations in SF3B1 and TP53, though there are differences in diagnostic criteria. Future efforts to refine MDS prognostication could investigate the interface between MDS and clonal cytopenia of undetermined significance, expand access to genomic testing, obtain results in a less invasive manner, and develop treatment-response predictors and dynamic risk models.

Flow Cytometric Assessment of Myelodysplastic Syndromes/Neoplasms.

Myelodysplastic syndromes/neoplasms (MDS) are a heterogeneous class of hematopoietic stem cell neoplasms characterized by ineffective hematopoiesis leading to peripheral cytopenias. This group of diseases is typically diagnosed using a combination of clinical, morphologic, and genetic criteria. Many studies have described the value of multiparametric flow cytometry (MFC) in the diagnosis, classification, and prognostication of MDS. This review summarizes the approach to MDS diagnosis and immunophenotypic characterization using MFC and describes the current state while highlighting future opportunities and potential pitfalls.

Classification, risk stratification and response assessment in myelodysplastic syndromes/neoplasms (MDS): A state-of-the-art report on behalf of the International Consortium for MDS (icMDS).

The guidelines for classification, prognostication, and response assessment of myelodysplastic syndromes/neoplasms (MDS) have all recently been updated. In this report on behalf of the International Consortium for MDS (icMDS) we summarize these developments. We first critically examine the updated World Health Organization (WHO) classification and the International Consensus Classification (ICC) of MDS. We then compare traditional and molecularly based risk MDS risk assessment tools. Lastly, we discuss limitations of criteria in measuring therapeutic benefit and highlight how the International Working Group (IWG) 2018 and 2023 response criteria addressed these deficiencies and are endorsed by the icMDS. We also address the importance of patient centered care by discussing the value of quality-of-life assessment. We hope that the reader of this review will have a better understanding of how to classify MDS, predict clinical outcomes and evaluate therapeutic outcomes.

Assessment and validation of the molecular international prognostic scoring system for myelodysplastic syndromes.

The Molecular International Prognostic Scoring System (IPSS-M) is a novel risk stratification model for myelodysplastic syndromes (MDS) that builds on the IPSS and IPSS-R by incorporating mutational data. The model showed improved prognostic accuracy over the IPSS-R across three endpoints: overall survival (OS), leukemia-free survival (LFS) and leukemic transformation. This study aimed to validate the findings of the original in a large cohort of MDS patients, as well as assess its validity in therapy-related and hypoplastic MDS. We retrospectively reviewed clinical, cytogenetic and molecular data for 2355 MDS patients treated at the Moffitt Cancer Center. Correlative analysis between IPSS-R and mean IPSS-M scores and outcome predictions was performed on LFS, OS and leukemic transformation. Using the IPSS-M, patients were classified as Very Low (4%), Low (24%), Moderate-Low (14%), Moderate-High (11%), High (19%) and Very-High risk (28%). Median OS was 11.7, 7.1, 4.4, 3.1, 2.3, and 1.3 years from VL to VH risk subgroups. Median LFS was 12.3, 6.9, 3.6, 2.2, 1.4, and 0.5 years respectively. For patients with t-MDS and h-MDS the model retained its prognostic accuracy. Generalized use of this tool will likely result in more accurate prognostic assessment and optimize therapeutic decision-making in MDS.

Assessment of Factors Associated With the Evaluation of Children for Leukemia Predisposition Syndromes: A Retrospective Single-center Study.

Five to 10% of children with cancer are thought to have a cancer predisposition syndrome (CPS). Referral guidelines for leukemia predisposition syndromes are limited and vague, requiring the treating provider to determine whether patients should have a genetics evaluation. We evaluated referrals to the pediatric cancer predisposition clinic (CPP), the prevalence of CPS in those who elected to pursue germline genetic testing, and assessed for associations between a patient's medical history and the diagnosis of a CPS. Data were obtained via chart review of children diagnosed with leukemia or myelodysplastic syndrome between November 1, 2017, and November 30, 2021. A total of 22.7% of pediatric leukemia patients were referred for evaluation in the CPP. Of the participants evaluated with germline genetic testing, the prevalence of a CPS was 25%. Our study was able to find a CPS in different malignancies, including acute lymphoblastic leukemia, acute myeloid leukemia, and myelodysplastic syndrome. We did not find associations between a participant with an abnormal CBC before diagnosis or hematology visit and the diagnosis of a CPS. Our study supports that a genetic evaluation should be available to all children with leukemia as medical and family history alone is not predictors of a CPS.

Wilms' tumor 1 expression combined with genetic mutations for prognostic assessment in MDS.

Overexpression of Wilms' tumor (WT1) is frequently observed in myelodysplastic syndrome (MDS), which has been proposed as a prognostic marker. However, the prognostic role of WT1 expression in different contexts remains to be fully elucidated. We retrospectively assessed the relationships between WT1 levels and preexisting prognostic factors to further investigate its prognostic role under different contexts. In our study, WT1 expression was positively correlated with WHO 2016 classification and IPSS-R stratification. Lower WT1 expression was found in relation to TET2, TP53, CD101, or SRSF2 mutations, while mutant NPM1 patients possessed higher level. Notably, WT1 overexpression maintained its inferior prognostic effect on overall survival (OS) in TP53-wild patients but not in TP53-mutated group. In multivariate analysis, higher WT1 expression was a risk factors for OS in EB patients without TP53 mutations. Overall, WT1 expression was useful to predict prognosis for MDS and its prognostic role was impacted by some gene mutations.

How can we improve response assessments in MDS? Strategies to improve response assessment in MDS treatment paradigms.

Evaluating response to treatment in MDS represents a major challenge due to its associated complexity and heterogeneity. Although response criteria have been proposed by the IWG and revised on several occasions, these criteria have limitations. This review has outlined some refinements that can be used to improve response assessment and to ensure the identification of clinically meaningful endpoints.

Impact of Hypomethylating Agent Use on Hospital and Emergency Room Visits, and Predictors of Early Discontinuation in Patients With Higher-Risk Myelodysplastic Syndromes.

BACKGROUND: Previous analyses using the SEER-Medicare database have reported substantial underutilization of hypomethylating agents (HMAs) among patients with higher-risk myelodysplastic syndromes (MDS), and an association between poor HMA persistence and high economic burden. We aimed to compare rates of hospitalizations and emergency room (ER) visits among patients with higher-risk MDS according to use or non-use of HMA therapy, and to explore factors associated with early discontinuation of HMA therapy. PATIENTS AND METHODS: We used the 2010-2016 SEER-Medicare database to identify patients aged ≥66 years with a new diagnosis of refractory anemia with excess blasts (RAEB; a surrogate for higher-risk MDS) between 2011 and 2015. New hospitalizations and ER visits during the 12 months following MDS diagnosis were determined. Treatment discontinuation was defined as stopping HMA therapy before 4 cycles. RESULTS: Overall, 664 (55.8%) patients were HMA users and 526 (44.2%) non-users. Non-users had more hospitalizations (mean 0.47 vs. 0.30, P < .001) and ER visits (mean 0.69 vs. 0.41, P = .005) per month than HMA users. Among HMA users, 193 (29.1%) discontinued HMA therapy before 4 cycles, and 91 (47.2%) of these after 1 cycle. Older age and poor performance status were associated with higher risk of HMA discontinuation. CONCLUSION: An increased rate of hospitalizations and ER visits occurred in HMA non-users vs. HMA users. Approximately one-third of patients discontinued HMA therapy early. Predictors of discontinuation included older age and poor performance status. Novel approaches are needed to improve utilization and persistence with HMA therapy and associated outcomes, particularly among these higher-risk groups.

Integrating germline variant assessment into routine clinical practice for myelodysplastic syndrome and acute myeloid leukaemia: current strategies and challenges.

Over the last decade, the field of hereditary haematological malignancy syndromes (HHMSs) has gained increasing recognition among clinicians and scientists worldwide. Germline mutations now account for almost 10% of adult and paediatric myelodysplasia/acute myeloid leukaemia (MDS/AML). As our ability to diagnose HHMSs has improved, we are now faced with the challenges of integrating these advances into routine clinical practice for patients with MDS/AML and how to optimise management and surveillance of patients and asymptomatic carriers. Discoveries of novel syndromes combined with clinical, genetic and epigenetic profiling of tumour samples, have highlighted unique patterns of disease evolution across HHMSs. Despite these advances, causative lesions are detected in less than half of familial cases and evidence-based guidelines are often lacking, suggesting there is much still to learn. Future research efforts are needed to sustain current momentum within the field, led not only by advancing genetic technology but essential collaboration between clinical and academic communities.

A geno-clinical decision model for the diagnosis of myelodysplastic syndromes.

The differential diagnosis of myeloid malignancies is challenging and subject to interobserver variability. We used clinical and next-generation sequencing (NGS) data to develop a machine learning model for the diagnosis of myeloid malignancies independent of bone marrow biopsy data based on a 3-institution, international cohort of patients. The model achieves high performance, with model interpretations indicating that it relies on factors similar to those used by clinicians. In addition, we describe associations between NGS findings and clinically important phenotypes and introduce the use of machine learning algorithms to elucidate clinicogenomic relationships.

Persistence to hypomethylating agents and clinical and economic outcomes among patients with myelodysplastic syndromes.

OBJECTIVES: To evaluate hypomethylating agent (HMA) persistence in patients with myelodysplastic syndromes (MDS), and examine its association with healthcare resource utilization (HRU) and progression to acute myeloid leukemia (AML). METHODS: A total of 2,400 adults diagnosed with MDS initiating HMAs were included from IBM MarketScan databases during 1/1/2011-3/31/2018. The index date was HMA initiation following MDS diagnosis. Patients were classified according to their persistence status by the end of a fixed 'landmark period' of 4 months post-index. RESULTS: Median persistence to HMAs was 5.6 months (95% CI: 5.2, 6.1); HMA non-persistence increased with time. Non-persistent patients had a significantly higher non-HMA-related HRU burden than persistent patients [adjusted incidence rate ratios, outpatient visits: 1.12 (95% CI: 1.10, 1.14); inpatient visits: 1.48 (95% CI: 1.30, 1.69); emergency department visits 1.30 (95% CI: 1.12, 1.50); all p-values < 0.001]. All-cause and HMA-related outpatient visits were lower among non-persistent patients, likely because of fewer HMA administration-related visits. The incidence rate of AML was numerically, although not significantly, higher in non-persistent patients, when starting follow-up at the end of the landmark period. When follow-up began at the index date, non-persistent patients had a significantly higher rate of AML [adjusted hazard ratio, 1.88 (95% CI: 1.53, 2.32); p-value < 0.001]. CONCLUSIONS: HMA non-persistence, which increased over time, was associated with significantly higher non-HMA-related HRU, and numerically higher AML progression in MDS patients initiating HMAs. Future studies should evaluate predictors of HMA non-persistence in this patient population.

Direct Medical Costs Associated With Treatment Nonpersistence in Patients With Higher-Risk Myelodysplastic Syndromes Receiving Hypomethylating Agents: A Large Retrospective Cohort Analysis.

BACKGROUND: Suboptimal use of hypomethylating agents (HMAs) among higher-risk myelodysplastic syndrome (HR-MDS) patients can translate into worse health outcomes and economic burden. We estimated the direct medical costs associated with HMA treatment nonpersistence among HR-MDS patients. PATIENTS AND METHODS: Using the Surveillance, Epidemiology, and End Results-Medicare linked database, a retrospective cohort of patients diagnosed with refractory anemia with excess blasts (RAEB), a diagnosis that substantially overlaps with HR-MDS, between January 2011 and December 2015 was analyzed. Patients who had ≥ 1 year of continuous Medicare enrollment before diagnosis and who did not receive stem cell transplant or lenalidomide in the follow-up period were included. Patients receiving HMAs were stratified into HMA persistent (≥4 HMA cycles) and HMA nonpersistent (<4 cycles or a gap of ≥ 90 days between cycles) groups. Healthcare resource use and costs during the follow-up period were reported descriptively as total and per patient per month (PPPM). Weighted generalized linear models (GLM) were used to compare estimated healthcare resource use and costs between HMA groups. RESULTS: Among the 664 patients with RAEB, 295 (44.4%) were HMA nonpersistent and 369 (55.6%) HMA persistent. On the basis of weighted GLM analysis, the HMA nonpersistent group incurred significantly (P < .05) higher total PPPM costs compared to the HMA persistent group ($18,039 vs. $13,893), particularly for hospitalization ($3,375 vs. $2,131), and emergency room ($5,517 vs. $2,867) costs. CONCLUSION: There is a substantial economic burden associated with early discontinuation of guideline-recommended HMA therapy in RAEB patients. The study findings necessitate closer care management in this population in order to improve outcomes and reduce healthcare spending.

Real-world diagnostic testing patterns for assessment of ring sideroblasts and SF3B1 mutations in patients with newly diagnosed lower-risk myelodysplastic syndromes.

INTRODUCTION: The presence of ring sideroblasts (RS) and mutation of the SF3B1 gene are diagnostic of lower-risk (LR) myelodysplastic syndromes (MDS) and are correlated with favorable outcomes. However, information on testing and reporting in community-based clinical settings is scarce. This study from the Connect® MDS/AML Disease Registry aimed to compare the frequency of RS and SF3B1 reporting for patients with LR-MDS, before and after publication of the 2016 World Health Organization (WHO) MDS classification criteria. METHODS: Ring sideroblasts assessment and molecular testing data were collected from patients with LR-MDS at enrollment in the Registry. Patients enrolled between December 2013 and the data cutoff of March 2020 were included in this analysis. RESULTS: Among 489 patients with LR-MDS, 434 (88.8%) underwent RS assessment; 190 were assessed prior to the 2016 WHO guidelines (Cohort A), and 244 after (Cohort B). In Cohort A, 87 (45.8%) patients had RS identified; 29 (33.3%) patients had RS < 15%, none of whom underwent molecular testing for SF3B1. In Cohort B, 96 (39.3%) patients had RS identified; 31 (32.3%) patients had < 15% RS, with 13 undergoing molecular testing of which 10 were assessed for SF3B1. CONCLUSIONS: In the Connect® MDS/AML Registry, only 32% of patients with <15% RS underwent SF3B1 testing after the publication of the WHO 2016 classification criteria. There was no change in RS assessment frequency before and after publication, despite the potential impact on diagnostic subtyping and therapy selection, suggesting an unmet need for education to increase testing rates for SF3B1 mutations.

Exploring blast composition in myelodysplastic syndromes and myelodysplastic/myeloproliferative neoplasms: CD45RA and CD371 improve diagnostic value of flow cytometry through assessment of myeloblast heterogeneity and stem cell aberrancy.

BACKGROUND: Flow cytometry immunophenotyping (FCIP) can improve diagnosis of myelodysplastic syndromes (MDS) and myelodysplastic/myeloproliferative neoplasms (MDS/MPN), although its application is challenging due to difficulties in standardization, complexity of antibody panels and subjective interpretation of data. Since blasts are invariably affected in these disorders, we developed a FCIP approach for detailed and objective analysis of the blast population. METHODS: FCIP using a one-tube 10-color (13-marker) antibody panel was performed on bone marrow samples from 23 MDS and 8 MDS/MPN patients, 21 cytopenic patients non-diagnostic for MDS (Non-MDS), and 16 Control samples. RESULTS: MDS and MDS/MPN cases demonstrated one to several immunophenotypic abnormalities including: increased myeloblasts, decreased stage-1 hematogones, aberrant stem cells, abnormal myeloblast heterogeneity/divergence from normal, increased or decreased CD45 intensity, increased CD117 or CD123 intensity, decreased CD38 intensity, and aberrant expression of lineage markers (CD5, CD19, CD56). A Blast score was developed that showed sensitivity of 80.6% and specificity of 90.5% for immunophenotypic diagnosis of MDS and MDS/MPN. Expression levels of CD45RA and CD371 were used to evaluate abnormal myeloblast heterogeneity and stem cell aberrancy. Both these features were, for the first time, incorporated into a scoring system and resulted in 19% increase in the sensitivity of the assay for lower-risk MDS. CONCLUSION: Deep immunophenotypic analysis of the blast population is valuable for diagnosis of MDS and MDS/MPN and can potentially provide sensitivity and specificity figures comparable to those previously described using more comprehensive panels that assess maturing myelomonocytic and erythroid elements in addition to progenitor cells.

Diagnosis and prognosis are supported by integrated assessment of next-generation sequencing in chronic myeloid malignancies. A real-life study.

Next-generation sequencing (NGS) is used to investigate the presence of somatic mutations. The utility of incorporating routine sequencing to guide diagnosis and therapeutic decisions remains unclear. We report the findings of an observational, multicenter study that aimed to assess the impact of somatic mutation testing by NGS in a reallife setting of chronic myeloid malignancies. A total of 177 patients were enrolled, partitioned into two overlapping groups. In group A (n=94), the indication was to search for clonal hematopoiesis, in a context of suspected myelodysplastic syndrome or myeloproliferative neoplasia. In group B (n=95), the theranostic impact of somatic mutations was studied. A panel of 34 genes was used on DNA extracted from blood or bone marrow samples. Within group A, the detection of clonal hematopoiesis supported the diagnosis of chronic myeloid malignancies for 31 patients while the absence of clonal hematopoiesis ruled out the suspected diagnosis in 47 patients. Within group B, NGS identified prognostically relevant somatic mutations in 32 patients, which had a therapeutic impact in 18 cases. By determining the presence or absence of somatic mutations, the application of NGS in daily practice was found to be useful for an integrated final diagnosis in 83% of the patients. Moreover, the search for somatic mutations had a prognostic impact that led to treatment modification in 19% of the cases. This study outlines the fact that adequate implementation of new investigations may have a significant positive medico-economic impact by enabling appropriate management of patients.

Integrating patient-centered factors in the risk assessment of MDS.

Myelodysplastic syndromes are clonal myeloid neoplasms that primarily present in older adults. Although leukemia develops in approximately 25% to 30% of individuals, the significantly shortened survival in this population is attributed more commonly to nonleukemic causes. The current prognostic scoring systems for leukemia and overall survival based on disease characteristics are becoming increasingly sophisticated and accurate with the incorporation of molecular data. The addition of patient-related factors such as comorbidity, disability, frailty, and fatigue to these new models may improve their predictive power for overall survival, treatment toxicity, and health care costs. To improve the generalizability of clinical trial results to the real world, geriatric assessment testing should become a standard of care in MDS clinical trials.