Research Article: Cross-modal fusion of cytomorphology and 18 F-FDG PET/CT for non-invasive bone marrow immune microenvironment decoding in multiple myeloma
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 ? 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.
Introduction:
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.
Read more