From established biology to candidate archetypes in high-grade serous ovarian carcinoma: the ARCHETYPE-PX evidence framework and consensus method

Background. High-grade serous ovarian carcinoma (HGSOC) causes most deaths from ovarian cancer, and treatment remains largely empirical outside the homologous-recombination-deficient subgroup. Genomic, immune, stromal and treatment-induced states with different therapeutic vulnerabilities are increasingly well described, but rarely inform routine clinical decisions. Treatment choices are discrete and routine inputs are limited. We therefore define a compact set of measurable, mechanistically interpretable and treatment-relevant candidate states to provide the biological prior for ARCHETYPE-PX, a Virtual Human Twin (VHT) intended to model individual disease trajectories and treatment response. Methods. We developed a literature-grounded approach to define candidate archetypes and evaluate them through expert consensus. A targeted evidence map covered three domains: genome and DNA-repair biology; the tumour and immune microenvironment; and the malignant compartment, including plasticity, spatial organisation and treatment-induced change. We organised the evidence into working definitions and partially populated six-field dossiers covering mechanism, molecular features, prevalence, assays, therapeutic hypothesis and behaviour under treatment. Candidates represent potentially overlapping biological layers. An offline browser instrument supports independent assessment of biological validity, detectability from routine and emerging clinical data, and clinical or therapeutic relevance. The consensus process combines a modified Delphi with RAND/UCLA-informed decision principles and prespecified rules for agreement, disagreement, minimum information and robustness. Initial methodological outputs. Version 2.3 contains 80 phenotype–evidence entries: 27 genome/repair, 28 immune/microenvironment and 25 malignant-compartment entries. Thirty-three record an association with progression-free survival, overall survival or treatment response. Sixteen working candidates were specified for appraisal: six genomic or repair states, six immune or microenvironmental states and four malignant-compartment states. The rating pathway and structured export are implemented. Expert evaluation is ongoing; no final membership, tier or consensus result is inferred from interim ratings. Interpretation. ARCHETYPE-PX provides a transparent, versioned framework linking established HGSOC biology to structured expert appraisal. The evidence profiles and reference mapping make the basis for candidate evaluation explicit, without establishing clinically validated archetypes or predictive biomarkers. Consensus completion, modelling, functional testing and clinical validation will determine whether these states can support treatment stratification.