MSACL 2026 Abstract

Spatially Informed DBSI-MALDI Analysis Reveals Reproducible Microstructure-Molecular Associations in Prostate Cancer

Keervani Kandala (1,2,§), Kainen Utt (1,§), Cody Weimholt (3), Sheng-Kwei Song (1), Peggi M. Angel (4), Richard R. Drake (4), Joseph E. Ippolito (1,5,6), José Marcio Luna (1,2,6)
(1) Mallinckrodt Institute of Radiology, Washington University in St. Louis, St. Louis, MO, USA (2) Division of Biological and Biomedical Sciences, Washington University in St. Louis, St. Louis, MO, USA (3) Department of Pathology and Immunology, Washington University in St. Louis, St. Louis, MO, USA, (4) Department of Pharmacology and Immunology, Medical University of South Carolina, Charleston, SC, USA, (5) Department of Biochemistry and Molecular Biophysics, Washington University in St. Louis, St. Louis, MO, USA (6) Alvin J. Siteman Cancer Center, St. Louis, MO, USA (§) Co-first Authors

Jose Marcio Luna, MS, PhD (Presenter) Washington University in St. Louis

Presenter Bio: Dr. José Marcio Luna is an Assistant Professor at the Mallinckrodt Institute of Radiology at Washington University in St. Louis. He leads a research program focused on advancing cancer diagnosis and prognostication through innovative data integration methods, bringing expertise in cancer research, machine learning, data sciences, and computational medical imaging. Dr. Luna earned his M.S. and Ph.D. (with a Ph.D. minor in Applied Mathematics) in Engineering from the University of New Mexico in 2010 and 2014, respectively, where he focused on modeling intrinsically random nonlinear systems. In 2021, he completed postdoctoral training in Radiology at the University of Pennsylvania, developing AI-based methods to predict treatment effectiveness in non-small cell lung cancer (NSCLC) and contributing to interpretable machine learning methods for medical applications. Dr. Luna received the 2021 Marlene Shlomchik Fellowship and an Emerson Collective grant during his postdoc. With 36 peer-reviewed publications, including 12 in prestigious journals such as PNAS and Nature Scientific Reports, Dr. Luna is now leading a research initiative funded by an NCI Transition Career Development Award. His work focuses on identifying prognostic biomarkers from radiological and histopathological imaging to enhance treatment guidance and prognostication for various cancers.

Relevant Financial Disclosures (within past 24 months, reported on Apr 21, 2026) No relevant financial relationship(s) to disclose.

INTRODUCTION:
Non-invasive characterization of tumor molecular composition remains a major challenge in prostate cancer (PCa), as clinical imaging primarily reflects tissue structure rather than underlying biochemical processes. Diffusion Basis Spectrum Imaging (DBSI) [1,2], an advanced diffusion imaging technique, provides quantitative measures of tissue microstructure, while matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) [3-5] reports spatially resolved molecular profiling. Establishing robust associations between these modalities could enable imaging-based surrogates of molecular composition.

OBJECTIVES:
To quantify spatially robust associations between DBSI-derived microstructural features and MALDI-MSI-derived molecular signatures in PCa, and to evaluate their potential for non-invasive characterization of tumor composition.

METHODS:
We analyzed 8 PCa specimens with matched DBSI and MALDI-MSI data, selected to represent a range of structural collagen content. Principal component analysis (PCA) was performed independently for each modality using a cohort-wide pooled model constructed from background-masked (tissue-only) pixel or voxel values. Each specimen was subsequently projected into this shared PCA space to ensure consistent feature representation and enable direct inter-specimen comparison.

For MALDI-MSI, data were normalized by total ion current prior to analysis, and PCA was performed using all 504 m/z channels spanning approximately 600–2500 Da. For DBSI, PCA included quantitative microstructural metrics and b0 intensity (including fiber fraction, directional diffusivities λ1–λ3, isotropic ADC, and restricted, hindered, and free water fractions), with highly collinear variables excluded. This resulted in 9 DBSI principal components and 10 MALDI principal components.

Spearman rank correlations were computed within each specimen across all principal component pairs across modalities. To account for spatial autocorrelation, spatial block permutation testing (1,000 permutations per specimen) was implemented using a Queen’s adjacency framework to preserve local spatial structure during resampling. Statistical significance was determined from permutation-derived null distributions generated separately for each specimen.

RESULTS:
The fifth DBSI principal component, associated with fiber density, demonstrated a reproducible negative correlation with the first MALDI principal component, representing dominant molecular variance, in 7/8 specimens (87.5%) with permutation-adjusted significance (p < 0.001). Effect sizes were consistent across specimens (ρ range: -0.18 to -0.29; mean ρ ≈ -0.19). In contrast, naïve correlation testing produced substantially inflated significance (p ≈ 10⁻¹⁵), demonstrating that spatial autocorrelation can markedly bias statistical inference and that spatial permutation testing is required for valid analysis. This pattern suggests a reproducible coupling between tissue microstructure and molecular composition across independent specimens. Additional reproducible associations were observed between the fourth DBSI principal component and MALDI components in up to 75% of specimens, though with smaller effect sizes (|ρ| ≈ 0.08–0.14). One specimen demonstrated increased coupling (47 significant component pairs versus 12 to 26 in others), potentially reflecting biological heterogeneity or increased statistical power. Across all specimens, 181 out of 720 correlations (25.2%) were significant at p < 0.05 after spatial correction, and 47 out of 720 (6.5%) at p < 0.001.

CONCLUSIONS:
We identify reproducible, spatially robust associations between DBSI-derived microstructural features and MALDI-MSI-derived molecular signatures in prostate cancer. These relationships persist after correcting for spatial autocorrelation, supporting that diffusion imaging captures reproducible features associated with underlying tissue composition. From a clinical perspective, this work supports the potential of advanced diffusion imaging as a non-invasive surrogate for molecular characterization of the tumor microenvironment. While effect sizes are modest, their consistency across independent specimens supports a biologically meaningful multimodal relationship. As these findings are derived from a small cohort, ongoing work in larger cohorts is underway to further evaluate robustness and generalizability.

REFERENCES:
1. Wang, X. et al. Diffusion basis spectrum imaging detects and distinguishes coexisting subclinical inflammation, demyelination and axonal injury in experimental autoimmune encephalomyelitis mice. NMR in Biomedicine 27, 843-852 (2014).
2. Wang, Y. et al. Differentiation and quantification of inflammation, demyelination and axon injury or loss in multiple sclerosis. Brain 138, 1223-1238 (2015).
3. Wallace, E. N. et al. An N-glycome tissue atlas of 15 human normal and cancer tissue types determined by MALDI-imaging mass spectrometry. Scientific Reports 14, 489 (2024).
4. Angel, P. M. & Zambrzycki, S. C. Predictive value of collagen in cancer. Advances in cancer research 154, 15-45 (2022).
5. Scott, D. A. et al. Specific N-linked glycosylation patterns in areas of necrosis in tumor tissues. International journal of mass spectrometry 437, 69-76 (2019).