Label-Space Mismatch Distorts Cross-Cohort Evaluation in Breast Cancer Molecular Subtyping
Abstract
Cross-cohort evaluation of breast cancer subtype classifiers is unreliable when source and target cohorts use different annotation standards: METABRIC labels six classes, PAM50 defines five, and a claudin-low prediction cannot be correct under a five-class standard. A classifier transferred from 1,369 METABRIC patients (train n=958) to TCGA-BRCA (n=987) and GSE21653 (n=266) separates three evaluation operations routinely conflated in cross-site comparisons. (1) Label-space restriction - an evaluation-rule change, not a retraining gain - is the dominant correction (73.76%->79.74% on TCGA-BRCA; 63.16%->68.42% on GSE21653); (2) EM prior correction on the reconciled space is an optional calibration step (Brier fidelity up, macro F1 down; to 80.65% and 74.06%); and (3) the target label space is not recoverable from unlabelled predictions (EM retains prior 0.0751 on the absent class; calibrated detectors fail), so compliance must come from cohort documentation. The protocol exposes why internal validation overstates cross-cohort value: multimodal fusion gains do not transfer (no configuration beats mRNA alone, n=694; McNemar p=0.13-1.00), and the genomic panel adds nothing to pathological complete response prediction over clinical variables (DeltaAUC = +0.0045). Clinically, without label-space compliance, cross-site accuracy comparisons misattribute annotation-standard differences to model skill; Luminal A/B and claudin-low boundary errors map to distinct treatment decisions; and subtype accuracy is a weaker portable endpoint than survival discrimination (external C-index 0.577-0.584). Reporting the cohort's allowed label set and applying label-space compliance are therefore prerequisites before any cross-cohort performance claim.
Related
- VARION: A Network Propagation Framework for Individual Patient Somatic Mutation Interpretation in Cancer Molecular Subtyping
- Crowding distorts features, not objects: structured target confusions in a 4AFC conjunction task
- HLA-DQ molecular mismatch predicts acute rejection and donor-specific antibody development after pediatric liver transplantation
- Uncorrected publication bias distorts the spindle-memory link: Comment on Chen et al (2024) “Sleep spindles consolidate declarative memory with tags: A meta-analysis of adult data”
- Hidden but Expanding Sampling Bias Distorts Prokaryotic Genome Collections
BibTeX
@misc{liu2026labelspace,
title = {{Label-Space Mismatch Distorts Cross-Cohort Evaluation in Breast Cancer Molecular Subtyping}},
author = {Suxing Liu and Byungwon Min and Zhijian Zheng and Jiahao Wang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.24962},
eprint = {2610.24962},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.24962},
}
RIS
TY - UNPB
TI - Label-Space Mismatch Distorts Cross-Cohort Evaluation in Breast Cancer Molecular Subtyping
AU - Suxing Liu
AU - Byungwon Min
AU - Zhijian Zheng
AU - Jiahao Wang
PY - 2026
DA - 2026/10/18
PB - arXiv
DO - 10.48550/arxiv.2610.24962
UR - https://arxiv.org/abs/2610.24962
AB - Cross-cohort evaluation of breast cancer subtype classifiers is unreliable when source and target cohorts use different annotation standards: METABRIC labels six classes, PAM50 defines five, and a claudin-low prediction cannot be correct under a five-class standard. A classifier transferred from 1,369 METABRIC patients (train n=958) to TCGA-BRCA (n=987) and GSE21653 (n=266) separates three evaluation operations routinely conflated in cross-site comparisons. (1) Label-space restriction - an evaluation-rule change, not a retraining gain - is the dominant correction (73.76%->79.74% on TCGA-BRCA; 63.16%->68.42% on GSE21653); (2) EM prior correction on the reconciled space is an optional calibration step (Brier fidelity up, macro F1 down; to 80.65% and 74.06%); and (3) the target label space is not recoverable from unlabelled predictions (EM retains prior 0.0751 on the absent class; calibrated detectors fail), so compliance must come from cohort documentation. The protocol exposes why internal validation overstates cross-cohort value: multimodal fusion gains do not transfer (no configuration beats mRNA alone, n=694; McNemar p=0.13-1.00), and the genomic panel adds nothing to pathological complete response prediction over clinical variables (DeltaAUC = +0.0045). Clinically, without label-space compliance, cross-site accuracy comparisons misattribute annotation-standard differences to model skill; Luminal A/B and claudin-low boundary errors map to distinct treatment decisions; and subtype accuracy is a weaker portable endpoint than survival discrimination (external C-index 0.577-0.584). Reporting the cohort's allowed label set and applying label-space compliance are therefore prerequisites before any cross-cohort performance claim.
ER -