Deep Learning-Powered Spatial Transcriptomics Analysis to Predict Immune Checkpoint Inhibitor Response from Pre-Treatment Digital Pathology Slides

Logeshwaran Rajasekaran ORCID ,  Basheer Mohamed ORCID ,  Somnath Balasaheb Thigale ORCID ,  Balasubramaniyan Ramalingam ,  Venkatesan Kuppan ,  Mahalakshmi
    Received: 18 July 2026; Revised: 10 August 2026; Accepted: 17 September 2026; Published: 20 October 2026

    Abstract

    Response to immune checkpoint inhibitors (ICIs) is governed in part by the spatial organization of the tumour microenvironment, yet most deep learning models applied to digital pathology treat tissue regions independently and therefore do not represent the neighbourhood interactions that shape immune activation. This study develops and retrospectively evaluates a Neighborhood-aware Masked Graph Spiking Neural Network (NAMG-SNN) that classifies ICI responders and non-responders from pre-treatment haematoxylin and eosin whole-slide images. Whole-slide images are stain-normalized, tiled and encoded into patch-level embeddings that combine statistical, textural, morphological and deep semantic descriptors. A spatial graph is then constructed over the retained patches and pruned by an adaptive similarity mask that suppresses weak or artefactual neighbourhood edges. The masked graph is propagated through a leaky integrate-and-fire spiking network whose accumulated firing rates form a spatial transcriptomic representation, which is finally classified into responder and non-responder categories. The model was evaluated on an integrated cohort assembled from The Cancer Genome Atlas (TCGA), the Panoptic segmentation of nUclei and tissue in Melanoma pathology Challenge (PUMA challenge) and the Phoenix virtual spatial transcriptomics resource, using patient-level splitting. On the integrated test data the framework reached 97.31% accuracy, 96.94% precision, 97.82% recall, 97.38% F1-score and 98.87% area under the receiver operating characteristic curve (AUC), with a Matthews correlation coefficient of 0.947 and Cohen’s kappa of 0.946, exceeding convolutional, transformer and graph baselines trained under identical conditions. Transcriptomic reconstruction reached a Pearson correlation of 0.962, mean absolute error of 0.051, root mean square error of 0.081 and structural similarity index (SSIM) of 0.966. Ablation confirmed that neighbourhood-aware graph construction, adaptive masking and spike-based learning each contributed to performance. These results are retrospective and computational.

    Keywords

    References

    1. 1. Schuiveling, M.; Liu, H.; Eek, D.; et al. Associations between spatial distribution of immune cell subsets and clinical outcomes in patients with advanced melanoma treated with immune checkpoint inhibitors: Results from the PUMA challenge. Med. Image Anal. 2026, 115, 104334.
    2. 2. Williams, H.L.; Frei, A.L.; Koessler, T.; et al. The current landscape of spatial biomarkers for prediction of response to immune checkpoint inhibition. Npj Precis. Oncol. 2024, 8, 178.
    3. 3. Nosa-Ihaza, E.A.; Edeh, E.C.; Geng, W.B.; et al. Artificial intelligence in cancer immunotherapy: Current trends in predicting response and personalizing treatment. J. Egypt. Natl. Cancer Inst. 2026, 38, 30.
    4. 4. Liu, W.; Puri, A.; Fu, D.; et al. Dissecting the tumor microenvironment in response to immune checkpoint inhibitors via single-cell and spatial transcriptomics. Clin. Exp. Metastasis 2024, 41, 313–332.
    5. 5. Zhou, X.M.M.; D’Amiano, A.J.; Lu, C.; et al. Spatial proteomics for investigating solid tumor resistance mechanisms. Cancer Metastasis Rev. 2025, 44, 76.
    6. 6. Greenwald, N.F.; Nederlof, I.; Sowers, C.; et al. Temporal and spatial composition of the tumor microenvironment predicts response to immune checkpoint inhibition in metastatic TNBC. Nat. Cancer 2026, 7, 435–450.
    7. 7. Peyraud, F.; Guegan, J.P.; Rey, C.; et al. Spatially resolved transcriptomics reveal the determinants of primary resistance to immunotherapy in NSCLC with mature tertiary lymphoid structures. Cell Rep. Med. 2025, 6, 101915.
    8. 8. Karlsen, V.T.; Majid, U.; Lindholm, H.T.; et al. Single cell spatial transcriptomics identifies coordinated cellular programs that are associated with good prognosis in microsatellite stable colorectal cancer. bioRxiv 2025. DOI: https://doi.org/10.1101/2025.03.31.646286
    9. 9. Xin, Y.; Yang, L.; Liu, C.; et al. Spatial transcriptomics of immune ecotypes for predicting immunotherapy outcomes in head and neck squamous cell carcinoma. Oral Oncol. 2026, 175, 107887.
    10. 10. Cui, X.; Liu, S.; Song, H.; et al. Single-cell and spatial transcriptomic analyses revealing tumor microenvironment remodeling after neoadjuvant chemoimmunotherapy in non-small cell lung cancer. Mol. Cancer 2025, 24, 111.
    11. 11. Han, Y.B.; Lee, S.; Lee, J.O.; et al. Spatial transcriptomics reveal high T cell and monocyte status as predictive and prognostic markers in pancreatic cancer. J. Transl. Med. 2025, 23, 576.
    12. 12. Tran, M.; Gindra, R.H.; Putze, P.; et al. Pan-cancer virtual spatial transcriptomics from routine histology with Phoenix. bioRxiv 2026. DOI: https://doi.org/10.64898/2026.04.25.720812
    13. 13. Mukherjee, S.; Patiyal, S.; Pal, L.R.; et al. Pan-cancer prediction of tumor immune activation and response to immune checkpoint blockade from tumor transcriptomics and histopathology. bioRxiv 2025. DOI: https://doi.org/10.1101/2025.06.27.661875
    14. 14. Shen, Z.; Brand, D.; Simard, M.; et al. AI-powered immune profiling from histopathology slides for chemo-radiotherapy outcome prediction in rectal cancer: A study using clinical trial and real-world cohorts. EBioMedicine 2025, 122, 105993.
    15. 15. Howard, F.M.; Dolezal, J.; Hieromnimon, H.; et al. Integration of gene expression and digital histology to predict treatment-specific responses in breast cancer. medRxiv 2025. DOI: https://doi.org/10.1101/2025.08.25.25334393
    16. 16. Belete, M.; Thakkar, N.; Khatri, I.; et al. Integrated histopathology-transcriptomic biomarker enhances survival prediction in HNSCC patients treated with immunotherapy. Transl. Oncol. 2026, 70, 102822.
    17. 17. Al-Mansour, F.S.H.; Almasoudi, H.H.; Albarrati, A. Mapping molecular landscapes in triple-negative breast cancer: Insights from spatial transcriptomics. Naunyn-Schmiedeberg’s Arch. Pharmacol. 2025, 398, 11125–11143.
    18. 18. Lubo, I.; Hernandez, S.; Wistuba, I.I.; et al. Novel spatial approaches to dissect the lung cancer immune microenvironment. Cancers 2024, 16, 4145.
    19. 19. Ma, A.; Xiang, L.; Yuan, J.; et al. Spatial transcriptomics decodes breast cancer microenvironment heterogeneity: From multidimensional dynamic profiling to precision therapy blueprint construction. Biomolecules 2025, 15, 1067.
    20. 20. Manogaran, S.; Ramadoss, R.; Selvam, S.P.; et al. Artificial intelligence-driven spatial transcriptomics in OSCC: Mapping the tumor microenvironment and personalizing therapy. J. Oral Biol. Craniofac. Res. 2025, 15, 1862–1873.
    21. 21. Voglis, S.; Schulz, D.; Eling, N.; et al. Spatially-resolved single-cell imaging of melanoma brain metastases identifies localized immune patterns predictive of immune checkpoint blockade response. Neuro-Oncol. 2026, 8, 2211–2223.
    22. 22. Ran, L.; Hu, G.; Fan, C.; et al. Deciphering the mechanistic landscape of immune checkpoint blockade in ccRCC: From molecular drivers to therapeutic responses. Front. Immunol. 2026, 17, 1774959.
    23. 23. Di Mauro, A.; De Cecio, R.; Simonelli, S.; et al. Modern pathology-driven strategies in neoadjuvant immunotherapy for head and neck squamous cell carcinoma: From residual tumor quantification to spatial and AI-based biomarkers. Cancers 2026, 18, 1020.
    24. 24. Choo, J.; Zhao, J.J.; Lau, M.C.; et al. Spatial predictors of response to chemo-immunotherapy in microsatellite stable metastatic colorectal cancer. Nat. Commun. 2026, 17.
    25. 25. Nameki, R.; Kinong, J.; Huang, C.H.; et al. Spatial remodeling of the urothelial carcinoma tumor microenvironment shapes response to neoadjuvant atezolizumab. medRxiv 2026. DOI: https://doi.org/10.64898/2026.04.15.26350980
    26. 26. Shamis, S.A.; Savioli, F.; Ammar, A.; et al. Spatial transcriptomic analysis of tumour with high and low CAIX expression in TNBC tissue samples using GeoMxTM RNA assay. Histol. Histopathol. 2024, 39, 177–200.
    27. 27. Yu, C.; Shi, C.; Zhou, H.; et al. Spatially resolved whole-transcriptomic profiling of EGFR-mutated lung adenocarcinomas stratified by PD-L1 expression. Cell. Oncol. 2026, 49, 2.
    28. 28. Zheng, Y.; Sadee, C.; Ozawa, M.; et al. Single-cell multimodal analysis reveals tumor microenvironment predictive of treatment response in non-small cell lung cancer. Sci. Adv. 2025, 11, eadu2151.
    29. 29. Pybus, A.; Kirchgaessner, R.; Nguyen, J.; et al. Predicting anti-PD-1 immune checkpoint blockade response in melanoma patients with spatially aware machine learning models. Npj Precis. Oncol. 2026, 10.
    30. 30. Baena, J.C.; Victoria, J.S.; Toro-Pedroza, A.; et al. Smart CAR-T nanosymbionts: Archetypes and proto-models. Front. Immunol. 2025, 16, 1635159.
    ×