AI and Cancer Research ── Myeloma ── 2026-11-05
Multiple myeloma: treatment, trials and AI
- Abstract
To determine intermodal agreement of AI-derived volumetric body composition metrics between [18 F]FDG-PET/CT-derived low-dose CT and whole-body MRI (wb-MRI). We retrospectively analyzed same-day whole-body low-dose CT, obtained from [ 18 F]FDG-PET/CT, and wb-MRI from consecutive multiple myeloma patients from the prospective GMMG-HD7- and GMMG-HD8-/DSMM-XIX-trials. An automated deep learning pipeline based on the TotalSegmentator model, and its MRI variant were used to segment skeletal muscle (SM), subcutaneous- (SAT) and abdominal visceral adipose tissue (VAT) from CT and MRI, respectively. Volumetric measures and single-slice estimates were derived. Agreement between CT- and MRI-based metrics was assessed using Pearson r, intraclass correlation coefficients (ICC), and Bland-Altman analyses. Forty-two multiple myeloma patients with same-day PET/CT and wb-MRI were included. For AI-based volumetric body composition metrics, CT and MRI demonstrated high agreement (ICC > 0.95) across all SM, SAT and VAT. However, Bland-Altmann plots revealed systematic modality-related differences between MRI and CT: both SM (bias: -2582 ml, p < 0.0001) and VAT (bias: -211 ml, p < 0.0001) were significantly lower when extracted from MRI compared to CT, while for SAT, the difference was not statistically significant (bias: -263 ml, p = 0.1629). For AI-based single slice analysis, agreement remained high for SAT and VAT (ICC > 0.97), while being lower for SM (ICC 0.87). Bland-Altman plots showed that SM was systematically and significantly lower when obtained from MRI than from CT (bias: -18.4cm 2 , p < 0.0001). Automated body composition measures from (PET-)CT and wb-MRI are highly concordant but show consistent intermodal biases that should be considered when pooling multimodal data or applying modality-agnostic thresholds. Not applicable.
Journal IF-equivalent: 5.7 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Holzschuh JC, Rotkopf LT, Neelsen C, Bauer F, Zhang KS, Menis E, et al. AI-based automated body composition analysis from whole-body MRI and (PET-) CT: an intermodal comparison from same-day examinations. Cancer Imaging. 2026 Oct 24;26(1):149. doi:10.1186/s40644-026-01134-5. PMID: 42928255.Checked: Abstract only - Abstract
Patients with multiple myeloma accumulate hundreds of clinical documents over years of sequential therapy, and every treatment decision depends on synthesis across that history. Whether large language model-based systems can match expert agreement on this task, and where they fail, has not been established. On longitudinal records of 811 patients treated over 25 years (44,962 documents and 1,334,677 laboratory values), an agentic reasoning system guided by an indexed clinical skill library was compared against single-pass retrieval-augmented generation (RAG), iterative RAG, iterative RAG augmented with the same skill library, and full-context input on 469 expert-adjudicated patient-question pairs, with an exploratory cross-dataset evaluation on MIMIC-IV. The skill-guided agent exceeded the ceiling at which iterative RAG (75.7%) and full-context input (75.8%) converged, reaching 80.1% concordance, with its advantage widening on the most complex tasks (+11.2 percentage points) and longest records. Providing the same skill library to iterative RAG matched the agent in aggregate (79.0%) and on shorter records, whereas the agent retained a significant advantage on the longest records (+10.3 percentage points in the top decile, p = 0.017) and a numerical, non-significant advantage on the most complex questions (+5.4 percentage points). The aggregate error rate (12.2%) was comparable to expert disagreement (13.6%), but a higher proportion of system errors were clinically significant (57.8 versus 18.8%).
Journal IF-equivalent: 9.8 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Moll J, Lübberstedt J, Nuernbergk C, Stroh J, Mertens L, Purcarea A, et al. Skill-guided agentic reasoning enables longitudinal synthesis of myeloma records at expert-level concordance. NPJ Precis Oncol. 2026 Oct 14;10(1):389. doi:10.1038/s41698-026-01733-2. PMID: 42875467; PMCID: PMC13669396.Checked: Full text checked - Abstract
Although effective, current chimeric antigen receptor T-cell (CAR-T) production methods-centralized, manual, and complex-are cost-intensive, time-consuming, and prone to variability. AIDPATH (AI-Powered Decentralized Production for Advanced Therapies in the Hospital) proposes a decentralized, automated alternative that integrates patient-specific data; optimizes resource use; and potentially improves cell viability, manufacturing efficiency, and patient outcomes. The aim of this study was to compare CAR-T therapy produced by the AIDPATH platform to both ciltacabtagene autoleucel (cilta-cel) and standard of care (SoC) for patients with triple-class refractory multiple myeloma (MM) over a 40-year time horizon from the hospital perspective in Germany. A partitioned survival model reflecting 3 health states (progression-free disease, progressed disease, and death) was used. The analysis used clinical trial data for cilta-cel, real-world data for SoC, and estimated parameters for AIDPATH due to the developmental status of the platform. The primary outcome was the incremental cost-effectiveness ratio; secondary outcomes included sensitivity and scenario analyses. On the basis of the model used in this analysis, AIDPATH was dominant compared to both cilta-cel and SoC. Most costs for CAR-T therapies were driven by acquisition and adverse events. Sensitivity analyses showed that the results were most influenced by discount rates and assumptions about progression-free survival. Scenario analyses, including reduced adverse events and shorter vein-to-vein times for AIDPATH, further supported its cost-effectiveness. This is the first study to assess the cost-effectiveness of a CAR-T product generated with AI support in Germany from the hospital perspective. AIDPATH was found to be a cost-effective alternative to both cilta-cel and SoC, making it a promising option for future implementation. Although further data are needed, this study provides valuable guidance for health care stakeholders, reimbursement discussions, and future research.
Journal IF-equivalent: 4.2 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Boot IWA, Weltin AL, de Graaf Y, Vrijhoef HJM. Cost-Effectiveness of AI-Powered Decentralized Production for Advanced Therapies in the Hospital–Generated Chimeric Antigen Receptor T-Cell Therapy for Multiple Myeloma From a German Hospital Perspective: Early Health Technology Assessment Study. JMIR Cancer. 2026 Oct 13;12:e92253. doi:10.2196/92253. PMID: 42868529; PMCID: PMC13718277.Checked: Full text checked - Abstract
Multiple myeloma (MM) is a plasma cell malignancy characterized by complex cytogenetic abnormalities and a supportive bone marrow microenvironment that promotes tumor growth and drug resistance. Despite the lack of curative options, novel targeted therapies show promise in overcoming resistance and improving patient outcome. Therefore, this study aimed to identify and validate prognostic biomarkers and underlying mechanisms in MM using integrated bioinformatics, machine learning, and experimental approaches. This study used bioinformatics analyses to identify key candidate genes and molecular pathways in MM. Differentially expressed genes (DEGs) were identified from datasets (GSE5900, GSE7116, GSE146649), followed by functional enrichment, protein-protein interaction network, and survival analyses. GAS2L3 was identified as a key gene in MM, strongly associated with prognosis. A GAS2L3 -based risk model predicted patient survival. Functional assays, including Cell Counting Kit-8, Transwell, and terminal deoxynucleotidyl transferase dUTP nick end labeling, showed that GAS2L3 silencing suppressed cell proliferation, migration, and invasion, while promoting apoptosis in RPMI-8226 cells. Immune infiltration analysis revealed significant associations between GAS2L3 and immune cell populations, including memory B cells and natural killer cells. Single-cell transcriptomic analysis identified the HLA-E-KLRK1 signaling axis as a key regulator of MM progression. This study uncovered novel molecular markers and immune regulatory pathways, providing insight into MM prognosis and possible therapeutic strategies.
Journal IF-equivalent: 1.8 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Li W, Li L, Sun Y, Zhao T. Machine learning-based prognostic model and immune landscape analysis in multiple myeloma. Am J Transl Res. 2026;18(9):8501-21. doi:10.62347/eoza1443. PMID: 42923674.Checked: Abstract only - Abstract
Acute kidney injury (AKI) frequently complicates sepsis in patients with multiple myeloma (MM), yet no validated predictive tool integrates the inflammatory and nutritional disturbances specific to this dual-disease population. We aimed to develop and externally validate an interpretable machine learning (ML) model-based on composite inflammatory-nutritional indices-to predict in-hospital AKI in septic patients with MM, and to deploy it as a web-based point-of-care tool. In this retrospective cohort study, 362 septic MM patients (development cohort: 253 training, 109 internal test) and 71 temporal validation patients were enrolled from Shaanxi Provincial People's Hospital (2020-2025). Moreover, an independent external validation cohort from a separate tertiary hospital ( n = 118, Xi'an No. 3 Hospital). Candidate predictors included 8 composite indices (e.g., NLR, SII, LAR, AAPR) alongside routine clinical and laboratory variables. Boruta and recursive feature elimination (RFE) identified a parsimonious predictor set. Six ML algorithms (logistic regression, SVM, MLP, LightGBM, XGBoost, Random Forest) were compared using AUC, calibration, and decision curve analysis. The final model was interpreted with SHAP and implemented as a public Streamlit application. AKI occurred in 206/362 (56.9%) patients. Eight predictors were selected: NLR, LAR, creatinine, SII, PTA, cystatin C, CRP, and AAPR. Random Forest demonstrated the most robust and generalizable performance, achieving a temporal validation AUC of 0.8574 (accuracy 0.7465, sensitivity 0.8205, specificity 0.6562). Moreover, Random Forest also achieved an external validation AUC of 0.9650 (accuracy 0.8983, sensitivity 0.9000, specificity 0.8971), with favorable calibration and consistent net clinical benefit. SHAP analysis identified higher NLR, LAR, SII, cystatin C, creatinine, and CRP, together with lower PTA and AAPR, as the most influential predictors in the model. The model is freely available at (https://gftcs94ast7omb44cg2izq.streamlit.app/). An interpretable Random Forest model employing eight routinely available indicators accurately predicts AKI in septic MM patients, bridging algorithmic complexity and bedside applicability. This tool may facilitate early risk stratification and personalized renal protection in this high-risk group.
Journal IF-equivalent: 5.1 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Zhang Q, Xin Q, Guo H. A machine-learning-derived online prediction model based on inflammatory and nutritional composite indicators for acute kidney injury in sepsis patients with multiple myeloma. Front Nutr. 2026 Sep 7;13:1874801. doi:10.3389/fnut.2026.1874801. PMID: 42769257; PMCID: PMC13590401.Checked: Full text checked - Abstract
Prognostic stratification in multiple myeloma (MM) relies on staging systems fixed at diagnosis, discarding temporal information accumulated during treatment. We developed a dynamic multimodal framework that predicts residual overall survival from observation windows of 1-18 months post-diagnosis. The model integrates DeepInsight-transformed gene expression, longitudinal trajectories of 10 laboratory analytes, and treatment history through missingness-aware gated fusion. On the Multiple Myeloma Research Foundation (MMRF) cohort from the Relating Clinical Outcomes in Multiple Myeloma to Personal Assessment of Genetic Profile (CoMMpass) study (n = 752), five-fold-specific models trained on the development dataset achieved a mean concordance index (C-index) of 0.773 ± 0.024 and 1-year time-dependent area under the receiver operating characteristic curve (AUC) of 0.789 ± 0.021 on a common held-out CoMMpass validation split, outperforming the evaluated survival-learning baselines including DeepSurv, a Cox proportional hazards neural network, and random survival forests. Kaplan-Meier stratification showed significant separation at all primary landmarks (log-rank $P<.001$, hazard ratios 3.46-3.93). A distilled student model retaining only the DeepInsight gene expression representation and five baseline clinical features transferred to an independent microarray cohort (GSE24080, n = 507) without retraining, achieving a C-index of 0.672 and a time-dependent AUC at 1-year of 0.740, supporting cross-cohort transferability in a reduced-input setting. Interpretability analyses recovered ubiquitin-proteasome, endoplasmic reticulum (ER) stress, and Interferon Alpha Response signals consistent with established myeloma biology. These findings support the potential of dynamic multimodal modeling for longitudinal prognostic assessment in MM.
Journal IF-equivalent: 7.9 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Jia S, Lysenko A, Boroevich KA, Sharma A, Tsunoda T. Dynamic multimodal survival prediction in multiple myeloma integrating gene expression, longitudinal laboratory measurements, and treatment history. Brief Bioinform. 2026 Sep 7;27(5):bbag475. doi:10.1093/bib/bbag475. PMID: 42704271; PMCID: PMC13548322.Checked: Full text checked - Abstract
The immune microenvironment contributes substantially to the biological and clinical heterogeneity of multiple myeloma (MM), yet immune-related molecular biomarkers with reproducible prognostic value remain limited. Here, we developed a 12-gene immune-related gene signature (IRGS) using an integrative machine-learning framework and evaluated its prognostic performance across multiple MM cohorts. The IRGS consistently stratified overall survival and remained independently associated with outcome after adjustment for established clinical covariates. Its prognostic discrimination was comparable to that of IFM15 and generally exceeded that of MRCIX6 and a mitophagy-related signature across the evaluated validation datasets. Single-cell RNA sequencing further revealed marked cell type-dependent variation in the activity of the 12-gene module, with comparatively low activity in plasma cells and higher activity in several non-plasma compartments, indicating that the bulk-derived IRGS reflects a multicellular bone marrow transcriptional context rather than an exclusively malignant plasma cell intrinsic program. Somatic mutation analysis identified distinct mutational patterns between IRGS-defined groups, including relative enrichment of DIS3 mutations in the low-IRGS group and MUC16 mutations in the high-IRGS group, together with a modestly higher tumor mutational burden in low-IRGS patients. Transcriptome-based drug-response prediction further suggested differential therapeutic vulnerabilities, with high- and low-IRGS groups showing distinct predicted sensitivity patterns across apoptosis-, DNA damage-, BET-, checkpoint-, and replication-stress-related agents. Collectively, these findings define the IRGS as a complementary immune-associated molecular biomarker for prognostic stratification in MM and provide a framework linking prognosis with multicellular transcriptional context, somatic mutational characteristics, and candidate therapeutic vulnerabilities.
Journal IF-equivalent: 4.4 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Wang K, Zhao C, Shi J, Liu S, Chen Y, Wang J, et al. Machine Learning‐Derived Immune Gene Signature Predicts Prognosis and Therapeutic Vulnerabilities in Multiple Myeloma. Cancer Sci. 2026 Sep 6 [Epub ahead of print]. doi:10.1111/cas.70523. PMID: 42702361; PMCID: PMC13546800.Checked: Full text checked - Abstract
To develop a deep learning (DL) approach for automatic segmentation and accurate risk stratification in multiple myeloma (MM) using whole-body [ 18 F]FDG PET/CT. This retrospective study included MM patients who underwent [ 18 F]FDG PET/CT between August 2013 and December 2023. To automatically segment focal lesions, an nnU-Net architecture was trained using dual-channel PET/CT inputs. For diffuse/mixed patterns, the tumor was automatically identified via bone segmentation using liver SUVmedian thresholds. Subsequently, DL and radiomics features were extracted using Pyradiomics and the STU-Net encoder. Then a deep learning radiomics nomogram (DLRN) was constructed using the Cox proportional hazards model and evaluated using the calibration curve, the receiver operating characteristic (ROC) curve, the Kaplan-Meier curve, and decision curve analysis. The study included 345 patients (median age, 59 years [IQR, 35-67 years], 198 male). The nnU-Net achieved a median DSC of 0.64-0.77 for focal lesions segmentation across cohorts. The DLRN was constructed by integrating deep learning radiomics score (DLRS), lactate dehydrogenase (LDH), and β2-microglobulin (β2-MG). The DLRN achieved an area under ROC curve (AUC) of 0.87 (95% confidence interval [CI]: 0.82-0.93), 0.84 (95% CI: 0.73-0.96), and 0.88 (95% CI: 0.76-0.99) for 3-year overall survival (OS) status prediction in the training, internal and external testing cohorts, which outperformed the International Staging System (ISS) (all P < 0.05). Furthermore, the DLRN can effectively identify high-risk individuals (all P < 0.05), demonstrated good agreement between predicted and observed survival probabilities, and provided clinical net benefit. The pattern-specific DL approach achieved automated whole-body tumor segmentation in MM, and the established DLRN demonstrated improved risk stratification capability.
Journal IF-equivalent: 5.5 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Xiao M, Zhong Y, Hao H, Yu X, Hu D, Wang J, et al. The deep learning radiomics nomogram for risk stratification in multiple myeloma using automatic whole-body [18F]FDG PET/CT segmentation approach. Eur J Nucl Med Mol Imaging. 2026 Jul 27;53(12):7274-87. doi:10.1007/s00259-026-08095-8. PMID: 42503527.Checked: Abstract only - Abstract
Multiple myeloma (MM) is a hematopoietic system malignancy characterized by clonal proliferation of abnormal plasma cells, commonly presenting with renal impairment (RI) that significantly impacts patient's quality of life. The objective of this study was to develop a predictive model for assessing the risk of MM with RI. This retrospective study included 1029 patients with MM, who were categorized into two groups: MM without RI group and MM with RI group. Specifically, the modeling dataset (Cohort 1) consisted of 792 patients (401 without RI and 391 with RI), and the validation dataset (Cohort 2) consisted of 237 patients (119 without RI and 118 with RI). Eight machine learning (ML) algorithms were assessed using the modeling dataset (n = 792), and model validation was conducted on the validation dataset (n = 237). The predictive performance of ML models was evaluated through the area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis. The optimal model among the eight ML algorithms was XGBoost. The predictive model constructed based on the XGBoost algorithm, RI-Lab6, incorporated six features: NLR, LDH, Hb, ACa, AGR and β2 microglobulin. In the modeling dataset (cohort 1), RI-Lab6 demonstrated an AUC of 0.99 (95% CI, 0.98-0.99), while in the validation dataset (cohort 2), it achieved an AUC value of 0.98 (95% CI, 0.96-0.99). These results indicated that the RI-Lab6 model possessed reliable predictive value for RI risk in MM patients and facilitated early identification of RI in these individuals.
Journal IF-equivalent: 7.6 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Zhou M, Feng S. Development and temporal validation of a machine learning-based model for predicting renal impairment in multiple myeloma: a single-center retrospective cohort study. BMC Med Inform Decis Mak. 2026 Jul 24;26(1):377. doi:10.1186/s12911-026-03689-6. PMID: 42816840; PMCID: PMC13625446.Checked: Full text checked - Abstract
We aimed to develop a survival prediction system integrating radiomics and clinical features for newly diagnosed multiple myeloma (MM) and to compare the performance of different feature sets and algorithms in the early prediction of progression-free survival (PFS). This study retrospectively included 300 MM patients between June 2022 and June 2024, with their baseline positron emission tomography, computed tomography, and magnetic resonance imaging radiomics features and clinical variables collected. Following construction of a radiomics-based risk score (Rad-score), seven machine learning-based survival models were established using the clinical feature set, the image feature set, and the integrated feature set. The fusion feature set-based gradient boosting model (GBM) showed numerically favorable overall performance in predicting 12-month PFS, suggesting that the integration of radiomics and clinical variables may provide complementary predictive information for early risk stratification. The model effectively distinguished high-, intermediate-, and low-risk patients (log-rank p<0.001), and calibration curve analysis and decision curve analysis revealed favorable calibration and high clinical net benefits. Shapley additive explanations analysis showed that the Rad-score was one of the most important features in the model, indicating that radiomics information may contribute prognostic value within the integrated feature space. β2-microglobulin, age, lactate dehydrogenase, blood calcium, platelet count, and hemoglobin were also identified as key contributing features. The integration of radiomics and clinical variables may improve early PFS prediction in MM patients. The GBM using fused features showed relatively good discriminative ability, potential clinical applicability, and favorable interpretability. Bu çalışmada, yeni tanı almış multipl miyelom (MM) hastalarında radyomik ve klinik özellikleri bütünleştiren bir sağkalım tahmin sistemi geliştirilmesi ve progresyonsuz sağkalımın (PFS) erken tahmininde farklı özellik setleri ile algoritmaların performanslarının karşılaştırılması amaçlandı. Bu retrospektif çalışmaya Haziran 2022 ile Haziran 2024 tarihleri arasında 300 MM hastası dahil edildi. Hastaların başlangıç pozitron emisyon tomografisi, bilgisayarlı tomografi ve manyetik rezonans görüntüleme radyomik özellikleri ile klinik değişkenleri toplandı. Radyomik tabanlı bir risk skoru (Rad-skoru) oluşturulduktan sonra klinik özellik seti, görüntüleme özellik seti ve bütünleşik özellik seti kullanılarak makine öğrenmesine dayalı yedi sağkalım modeli geliştirildi. Birleşik özellik setine dayalı gradyan artırma modeli (GBM), 12 aylık PFS’nin tahmininde sayısal olarak daha iyi bir genel performans gösterdi. Bu bulgu, radyomik ve klinik değişkenlerin bütünleştirilmesinin erken risk sınıflandırması açısından tamamlayıcı öngörü bilgileri sağlayabileceğini düşündürdü. Model; yüksek, orta ve düşük riskli hastaları etkili biçimde ayırt etti (log-rank p<0,001). Kalibrasyon eğrisi ve karar eğrisi analizleri, modelin uygun kalibrasyona ve yüksek klinik net faydaya sahip olduğunu gösterdi. Shapley katkı açıklamaları analizi, Rad-skorunun modeldeki en önemli özelliklerden biri olduğunu ortaya koyarak radyomik bilgilerin bütünleşik özellik alanında prognostik değer sağlayabileceğini gösterdi. β2-mikroglobulin, yaş, laktat dehidrogenaz, kan kalsiyum düzeyi, trombosit sayısı ve hemoglobin de modele önemli katkı sağlayan özellikler olarak belirlendi. Radyomik ve klinik değişkenlerin bütünleştirilmesi, MM hastalarında erken PFS tahminini iyileştirebilir. Birleşik özellikleri kullanan GBM, görece iyi ayırt edici performans, potansiyel klinik uygulanabilirlik ve uygun yorumlanabilirlik gösterdi.
Journal IF-equivalent: 0.8 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Sun S, Huang Z, Tang Y, Gu W, Wu L, Shen F. Establishment of a Multimodal Prognostic Prediction Model for Multiple Myeloma Patients Based on Radiomics and Clinical Features: A Retrospective Cohort Study. Turk J Haematol. 2026 Sep 2 [Epub ahead of print]. doi:10.4274/tjh.galenos.2026.33678. PMID: 42460671; PMCID: PMC13540411.Checked: Full text checked - Abstract
Multiple myeloma (MM) displays significant genetic heterogeneity, making it challenging to distinguish malignant from non-malignant plasma cells in single-cell datasets. Existing marker-based and CNV detection methods require manual intervention or high computational resources, limiting their scope. We developed a supervised deep learning autoencoder to classify malignant cells across MM and its precursor stages and validated its performance and biological relevance. The model outperformed alternative approaches, achieving mean AUCs of 0.86 on internal and 0.80 on external datasets, with strong performance on unseen samples across single-cell platforms (mean AUC 0.92). Despite training solely on MM samples, our model distinguishes malignant cells in precursor stages, with predicted malignant cell proportions increasing from monoclonal gammopathy of undetermined significance (MGUS, 29.00%) to smoldering multiple myeloma (SMM, 77.83%) and MM (96.08%). In paired samples, the model also accurately captured a reduction in malignant plasma cell proportions following treatment, reflecting patient variability in treatment response. Differential expression analysis uncovered a 12-gene malignant plasma signature linked to poor survival (HR = 2.7, Log-rank P = 0.0023) and validated in the MMRF CoMMpass cohort. Overall, our resulting model enables accurate malignant cell identification and provides biologically and clinically relevant insights for MM research.
Journal IF-equivalent: 9.8 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Satpathy S, Michaud ME, Pilcher WC, Bhasin SS, Bhasin MK. Deep learning-based classifier for malignant plasma cell identification in myeloma. NPJ Precis Oncol. 2026 Jul 9;10(1):418. doi:10.1038/s41698-026-01589-6. PMID: 42426241; PMCID: PMC13728225.Checked: Full text checked - Abstract
Patients with multiple myeloma are frequently exposed to prolonged, multiline and multidrug treatments, factors that may substantially influence overall health, physical reserve, and physiological resilience. Imaging of organs that are captured by whole-body scans can be opportunistically interrogated to derive direct, quantitative measures of body composition. Advances and standardization of whole-body magnetic resonance imaging (WBMRI) together with recent improvements in artificial intelligence (AI) methodologies for automated organ and tissue segmentation presents an opportunity to complement accurate disease assessments with body composition metrics. We report, to our knowledge, for the first time, development of an AI-based pipeline to enable automated quantitative image-derived phenotypes in nondiseased tissue from WBMRI in patients with multiple myeloma. We have demonstrated that a deep-learning pipeline can derive body composition metrics from routinely acquired clinical WBMRI. Throughout treatment, significant longitudinal changes were observed (P< .001), characterized by a decrease in abdominal skeletal muscle (ASM) alongside transient increases in abdominal subcutaneous and visceral adipose tissue. We have also shown that greater reserves of ASM (hazard ratio [HR], 0.60; 95% confidence interval [CI], 0.40-0.89) and abdominal subcutaneous adipose tissue (HR, 0.67; 95% CI, 0.46-0.98) at baseline are associated with better progression-free survival. Conversely increases in visceral adipose tissue over time was associated with inferior progression-free survival (HR, 2.89; 95% CI, 1.65-5.09). These findings support opportunistic body composition phenotyping from diagnostic WBMRI as a scalable biomarker for risk stratification (concordance index, 0.725) and mechanistic study. This trial was registered at www.clinicaltrials.gov as NCT02403102.
Journal IF-equivalent: 6.2 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Basty N, Kaiser M, Pawlyn C, Boyd K, Smith K, Thomas K, et al. AI-derived whole-body MRI metrics in patients with multiple myeloma reveal unique insights into body composition and outcomes. Blood Adv. 2026 Sep 9;10(18):6103-11. doi:10.1182/bloodadvances.2026020852. PMID: 42418394; PMCID: PMC13578971.Checked: Full text checked - Abstract
The bone marrow immune microenvironment (BMME) shapes treatment response in multiple myeloma (MM), yet routine diagnostic workup primarily assesses tumor burden rather than immune competence. We developed ImmunoCast-MM, a cross-modal deep learning framework that extracts immunologically relevant signals from two examinations routinely performed at diagnosis: Wright-Giemsa-stained bone marrow aspirate smears and whole-body 18 F-FDG PET/CT. The cytomorphology branch used DinoBloom embeddings to classify individual cells across five hierarchical levels. The PET/CT branch generated a multi-organ inflammation fingerprint from tumor, spleen, lymph node, and diffuse bone marrow compartments. A contrastive fusion module aligned the two imaging modalities with a flow cytometry reference panel and generated an Immune Dysfunction Index (IDI) along a learned effector-suppressor axis. ImmunoCast-MM was evaluated retrospectively in 243 patients with newly diagnosed MM. Associations with flow cytometric measurements, progression-free survival, and daratumumab response were assessed, with adjustment for International Staging System stage, cytogenetic risk, and age. On a held-out validation subset, decoded cytomorphologic indices correlated with matched flow cytometric fractions, with Spearman ρ values of 0.68-0.81 across three decoded index-panel pairs; all Benjamini-Hochberg-adjusted (p) values were (<0.001). Unsupervised clustering of the fused embeddings identified immune-competent and immune-exhausted archetypes that differed in progression-free survival and response to daratumumab. Separately, adding the IDI to conventional risk markers increased the concordance index from 0.58 to 0.75 ((p<0.001)) and improved the area under the receiver operating characteristic curve for daratumumab response from 0.55 to 0.81 ((p<0.001)). ImmunoCast-MM reframes two standard diagnostic examinations as non-invasive profilers of the BMME. The framework may support risk assessment and immunotherapy stratification, particularly in centers without access to high-dimensional flow cytometry.
Journal IF-equivalent: 6.0 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Gao M, Ouyang G, Liu Y, Chen S, Xu K. Cross-modal fusion of cytomorphology and 18F-FDG PET/CT for non-invasive bone marrow immune microenvironment decoding in multiple myeloma. Front Immunol. 2026 Jun 24;17:1856163. doi:10.3389/fimmu.2026.1856163. PMID: 42421956; PMCID: PMC13342052.Checked: Full text checked - Abstract
Multiple Myeloma (MM) is a complex hematological malignancy characterized by the uncontrolled proliferation of malignant plasma cells. Neutrophil Extracellular Traps (NETs) have recently been implicated in cancer, but the role in MM remains unclear. This study aimed to identify NETs-related biomarkers in MM and explore the underlying mechanisms using machine learning and experimental validation. Methods and Methods: Three MM-related microarray datasets were downloaded from the Gene Expression Omnibus database. Differential expression and weighted gene co-- expression network analyses were applied to identify NETs-related hub genes for MM. Machine learning algorithms were used to refine the biomarkers. The core genes were validated using receiver operating characteristic analysis, nomogram construction, and an external dataset. Functional analyses were performed to explore the underlying mechanisms. Drug-gene interaction analysis was performed to predict potential therapeutic drugs. Finally, gene expression was validated by quantitative real-time PCR in peripheral blood mononuclear cells from three MM patients and three healthy controls. A total of 24 NETs-related hub genes were identified. Machine learning algorithms identified three NETs-related core biomarkers for MM: KIT proto-oncogene (KIT), Hepatocyte Growth Factor (HGF), and F2R-like trypsin receptor 1 (F2RL1). The predictive model showed good predictive performance in both the training and external cohorts. Functional analyses showed significant associations between the biomarker genes and immune-related biological processes. Additionally, potential therapeutic drugs targeting these biomarkers were identified, including diphenylpyraline, Crizotinib, 1,4-chrysenequinone, and histamine. The expression patterns of the biomarker genes were confirmed in clinical MM samples. This study combined multiple machine learning models to identify key NETs-related biomarkers for MM. Clinical validation confirmed the relevance of the biomarkers in clinical practice. These findings provide a foundation for further research into the immune microenvironment and therapeutic strategies for MM. This study successfully identified and validated NETs-related biomarkers for MM through machine learning and experimental validation. Understanding the role of NETs-related biomarkers in MM could not only deepen our understanding of the disease but also offer information for clinical applications.
Journal IF-equivalent: 2.9 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Wu H, Zhu Y, Wang B, Yang W. Identification of Neutrophil Extracellular Traps-Related Biomarkers in Multiple Myeloma Using Machine Learning and Experimental Validation. Curr Med Chem. 2026 Jun 17;33. doi:10.2174/0109298673438668260604035517. PMID: 42316496.Checked: Abstract only