Pheno-MYCN maps the morphological footprint of MYCN amplification in paediatric neuroblastoma

Abstract

MYCN amplification and segmental chromosomal abnormalities have long been prognostic markers in paediatric neuroblastoma, and are often accompanied by morphological changes that portend high-risk disease. Yet quantitative morphological assessment is limited to countable features, such as the mitosis-karyorrhexis index. Other high-risk features, such as nuclear size, pleomorphism and anaplasia, lack formal criteria. This leaves a gap: current simple pathology metrics cannot capture, or localise, the full morphological landscape of high-risk biology, such as disease driven by MYCN. Seeking high-risk cases that standard criteria may miss, we developed Pheno-MYCN, a weakly supervised framework linking slide-level MYCN prediction on 189 H&E whole-slide images to interpretable phenotypic sub-populations that expert review mapped to neuroblastoma morphologies. The aim is not a stronger classifier, but to reveal the morphological basis of MYCN prediction. Cell-level profiling found MYCN-associated differences across sub-populations. Amplified tumour was densely cellular but disorganised, with less connected networks, lower cellular diversity and greater abundance of haemorrhagic and acellular necrotic regions. MYCN-amplified-like tissue was identifiable per slide from these features alone (AUC 0.93-1.00, leave-one-slide-out) and traced as a continuous gradient within tumours. Thus MYCN amplification leaves a concrete, interpretable footprint, readable and localisable on routine H&E.