Related Experiment Video
Updated: Oct 24, 2025
08:50
Predictive Immune Modeling of Solid Tumors
Published on: February 25, 2020
7.2K
How the variability between computer-assisted analysis procedures evaluating immune markers can influence patients'
Marylène Lejeune1,2,3, Benoît Plancoulaine4,5,6, Nicolas Elie6
1Department of Pathology, Oncological Pathology and Bioinformatics Research Group, Hospital de Tortosa Verge de la Cinta, Carrer de les Esplanetes, 14, 43500, Tortosa, Spain. mlejeune.ebre.ics@gencat.cat.
Histochemistry and Cell Biology
|August 12, 2021
Summary
Computer-assisted image analysis (CAI) algorithm differences impact immune biomarker identification in breast cancer biopsies. Developing improved CAI strategies is crucial for consistent, reliable analysis and predicting patient relapse.
Area of Science:
- Oncology
- Biomedical Imaging
- Computational Pathology
Background:
- Computer-assisted image analysis (CAI) algorithms can yield discrepancies in identifying immune biomarkers from immunohistochemically stained breast cancer biopsies.
- These variations may affect the accuracy of predicting disease outcomes and patient relapse.
- Standardization of CAI procedures is needed for reliable biomarker assessment.
Purpose of the Study:
- To compare three distinct CAI procedures (A, B, and C) for quantifying positive marker areas in post-neoadjuvant chemotherapy biopsies from triple-negative breast cancer (TNBC) patients.
- To evaluate how differences in CAI procedures influence the assessment of associations between immune markers and patient relapse risk.
- To investigate the consistency and reliability of CAI in biomarker analysis for TNBC.
Main Methods:
- Analysis of 3304 digital images from 118 TNBC patients' post-neoadjuvant chemotherapy biopsies.
- Immunohistochemical staining for seven immune markers (CD4, CD8, FOXP3, CD21, CD1a, CD83, HLA-DR).
- Comparison of three CAI procedures (A, B, C) measuring positive pixel areas, utilizing principal component analysis (PCA) and Cox multivariate regression.
Main Results:
- Paired CAI procedures showed good agreement for most markers at low concentrations, but procedures B/C and B/A exhibited higher probability of differences than C/A.
- PCA identified two patient groups with significantly different relapse probabilities, primarily based on CD8, CD1a, and HLA-DR data.
- Multivariate regression revealed similarities in relapse-associated factors for procedures A and C, contrasting with procedure B.
Conclusions:
- Discrepancies among CAI procedures can lead to inconsistent identification of predictive breast cancer biomarkers.
- General agreement between CAI methods does not ensure consistent identification of the same predictive markers.
- Further development of CAI strategies is essential to enhance sensitivity and improve the reliability of immune biomarker analysis in TNBC.
Keywords:
Computer-aided image analysisImmune responseImmunohistochemistryPredictive markersPrincipal component analysisRelapseTriple-negative breast cancer
