Search bioRxiv⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/2023.01.31.525753

Robust biomarker discovery through multiplatform multiplex image analysis of breast cancer clinical cohorts

Eng, J. R.·Bucher, E.·Hu, Z.·Sanders, M.·Chakravarthy, B.·Pietenpol, J. A.·Gibbs, S. L.·Chin, K.
View original source ↗Export metadata ↓

Abstract

AO_SCPLOWBSTRACTC_SCPLOWSpatial profiling of tissues promises to elucidate tumor-microenvironment interactions and enable development of spatial biomarkers to predict patient response to immunotherapy and other therapeutics. However, spatial biomarker discovery is often carried out on a single patient cohort or imaging technology, limiting statistical power and increasing the likelihood of technical artifacts. In order to analyze multiple patient cohorts profiled on different platforms, we developed methods for comparative data analysis from three disparate multiplex imaging technologies: 1) cyclic immunofluorescence data we generated from 102 breast cancer patients with clinical follow-up, in addition to publicly available 2) imaging mass cytometry and 3) multiplex ion-beam imaging data. We demonstrate similar single-cell phenotyping results across breast cancer patient cohorts imaged with these three technologies and identify cellular abundance and proximity-based biomarkers with prognostic value across platforms. In multiple platforms, we identified lymphocyte infiltration as independently associated with longer survival in triple negative and high-proliferation breast tumors. Then, a comparison of nine spatial analysis methods revealed robust spatial biomarkers. In estrogen receptor-positive disease, quiescent stromal cells close to tumor were more abundant in good prognosis tumors while tumor neighborhoods of mixed fibroblast phenotypes were enriched in poor prognosis tumors. In triple-negative breast cancer (TNBC), macrophage proximity to tumor and B cell proximity to T cells were greater in good prognosis tumors, while tumor neighborhoods of vimentin-positive fibroblasts were enriched in poor prognosis tumors. We also tested previously published spatial biomarkers in our ensemble cohort, reproducing the positive prognostic value of isolated lymphocytes and lymphocyte occupancy and failing to reproduce the prognostic value of tumor-immune mixing score in TNBC. In conclusion, we demonstrate assembly of larger clinical cohorts from diverse platforms to aid in prognostic spatial biomarker identification and validation. SO_SCPLOWTATEMENTC_SCPLOWO_SCPCAP C_SCPCAPO_SCPLOWOFC_SCPLOWO_SCPCAP C_SCPCAPO_SCPLOWSIGNIFICANCEC_SCPLOWOur single-cell spatial analysis of multiple clinical cohorts uncovered novel biomarkers of patient outcome in breast cancer. Additionally, our data, software, and methods will help advance spatial characterization of the tumor microenvironment.

Explore related subjects

cancer biologyPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Eng, J. R., Bucher, E., Hu, Z., Sanders, M., Chakravarthy, B., Pietenpol, J. A., Gibbs, S. L., Chin, K.. 2023-02-03. Robust biomarker discovery through multiplatform multiplex image analysis of breast cancer clinical cohorts. https://doi.org/10.1101/2023.01.31.525753

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Glycolysis and interferon transcription in melanoma immunotherapy

Glycolysis and interferon signalling have both been associated with responses to PD-1 blockade, but their joint transcriptional relationship with treatment outcome remains uncertain. We examined whether a state with high glycolysis and low interferon- response was associated with reduced disease control in melanoma. Fixed Hallmark gene sets were scored by single-sample gene set enrichment analysis in GSE91061, GSE78220 and an additional Liu cohort. Disease control analyses included 47, 26 and 100 patients. Conditional odds ratios for the joint state were 6.06 (exact 95% confidence interval, 1.04 to 65.61), 0.35 (0.026 to 3.11) and 1.03 (0.39 to 2.73), respectively. Holm-adjusted P values for the two initial comparisons were 0.073 and 0.365. The additional exploratory P value was 1.000. Continuous models and analyses that omitted each patient in turn did not provide consistent evidence of an adverse association. IFN scores were negatively correlated with DNA-derived tumour purity in 102 patients (Spearman {rho} = -0.671). Among 40 patients with paired biopsies, disease control was associated with higher on-treatment IFN scores after baseline adjustment, although the paired subset had lower baseline glycolysis than patients without paired profiles. In patient-level single-cell analyses, the baseline CD8 T cell correlation weakened when an alternative rank score used a common gene background. The defined joint state was not consistently associated with poorer disease control across cohorts. Bulk IFN scores were related to tumour purity, and cellular correlations were sensitive to scoring choices. These transcriptomic results do not measure pyruvate concentration, glycolytic flux or STAT1 pyruvylation.

cancer biology↗
bioRxiv2026

Psychoneuroimmunology of Viral Hepatitis and Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis

Hepatocellular carcinoma (HCC) and viral hepatitis (VH) are recognized primarily as hepatic diseases, yet they impose a substantial neuropsychiatric disorders (NPDs) encompassing depression, anxiety, cognitive impairment, and fatigue. Whether these manifestations reflect psychological reactions or biologically grounded psychoneuroimmune dysfunction remains poorly understood. Here, we present a pre-registered systematic review and meta-analysis (PROSPERO 2026 CRD420251081757) synthesizing evidence from 24 studies combining psychosocial outcomes with immune dysregulation markers across HCC and VH populations. Bibliometric and network analyses identified a growing psychoneuroimmunology research hub linking cytokine signaling, microglial activation, blood-brain barrier (BBB) disruption, and gut-liver-brain axis dysregulation to the neuropsychiatric phenotype of these diseases. Meta-analysis revealed a pooled hazard ratio (HR) of 1.92 (95% CI: 1.09-3.39) for adverse outcomes in HCC and a pooled HR of 2.28 (95% CI: 1.48-3.52) for depression associated to genetic variants in VH. TNF-a, but not IL-6, showed a tendency toward lower levels in VH patients with versus without NPDs, suggesting differential inflammatory roles. The literature, however, remains fragmented by heterogeneous designs and a paucity of longitudinal data. A psychoneuroimmunological approach, integrating immune, endocrine, neural, and psychosocial dimensions, is essential for capturing the full systemic impact of HCC and VH and should inform future mechanistic research and multidisciplinary clinical strategies.

cancer biology↗
bioRxiv2026

Nutrient Availability as a Driver of Transcriptome Reprogramming in Osteosarcoma

Background: Gene expression in mammalian cells is intricately regulated by environmental and metabolic cues. Among these, nutrient availability represents a key feature of the tumor microenvironment and has recently emerged as a potent modulator of oncogenic signaling networks. In this study, we employed a cell-line model of osteosarcoma, a rare but highly aggressive bone malignancy affecting both humans and dogs. We demonstrate that altering the concentration of a single nutrient can profoundly reprogram gene expression in osteosarcoma cells, revealing a previously underexplored phenomenon: nutrient-dependent gene regulation. This finding highlights potential opportunities for therapeutic intervention and underscores the broader implications of metabolic regulation for personalized medicine and nutrition. Methods/Results: We performed bulk RNA sequencing of an established osteosarcoma cell line cultured under varying concentrations of two key nutrients, glucose and glutamine. Differential gene expression analysis revealed a pronounced nutrient-dependent transcriptional response. The identified genes were functionally associated with pathways involved in cancer, development, nervous system regulation, metabolism, cellular signaling, and immune responses. We identified seven distinct expression patterns, one of which exhibited a striking correlation with cell growth rate and resembled a FOS-like transcriptional signature. Notably, the proto-oncogene FOS was among the identified genes and displayed a strong growth-rate-dependent expression pattern. Conclusion/Impact: Our findings demonstrate that modulation of nutrient availability profoundly reshapes the gene expression landscape of osteosarcoma cells, revealing nutrient-responsive targets with potential therapeutic relevance. This work expands our understanding of metabolic regulation in cancer and suggests that nutrient-based interventions may provide a controllable means of influencing tumor progression. Beyond oncology, such approaches may also have broader implications for developmental biology, autoimmune disorders, and nervous system function. More broadly, this study highlights the potential of metabolic interventions as a strategy for targeted modulation of gene expression programs.

cancer biology↗