Exploiting a living biobank to delineate mechanisms underlying disease-specific chromosome instability

Abstract

Chromosome instability (CIN) is a cancer hallmark that drives tumour heterogeneity, phenotypic adaptation, drug resistance and poor prognosis. High-grade serous ovarian cancer (HGSOC), one of the most chromosomally unstable tumour types, has a 5-year survival rate of only ~30% - largely due to late diagnosis and rapid development of drug resistance, e.g., via CIN-driven ABCB1 translocations. However, CIN is also a cell cycle vulnerability that can be exploited to specifically target tumour cells, illustrated by the success of PARP inhibitors to target homologous recombination deficiency (HRD). However, a lack of appropriate models with ongoing CIN has been a barrier to fully exploiting disease-specific CIN mechanisms. This barrier is now being overcome with the development of patient-derived cell cultures and organoids. In this review, we describe our progress building a Living Biobank of over 120 patient-derived ovarian cancer models (OCMs), predominantly from HGSOC. OCMs are highly purified tumour fractions with extensive proliferative potential that can be analysed at early passage. OCMs have diverse karyotypes, display intra- and inter-patient heterogeneity and mitotic abnormality rates far higher than established cell lines. OCMs encompass a broad-spectrum of HGSOC hallmarks, including a range of p53 alterations and BRCA1/2 mutations, and display drug resistance mechanisms seen in the clinic, e.g., ABCB1 translocations and BRCA2 reversion. OCMs are amenable to functional analysis, drug-sensitivity profiling, and multi-omics, including single-cell next-generation sequencing, and thus represent a platform for delineating HGSOC-specific CIN mechanisms. In turn, our vision is that this understanding will inform the design of new therapeutic strategies.

Keywords: Chromosome instability; cancer genomics; ovarian cancer; tumour heterogeneity.

Fig. 1

Treatment timelines of patients with HGSOC. Line graphs showing CA125 levels, measured via blood sampling, for patients 74 and 110 over time following diagnosis. Graphs are annotated to show surgery (blue up arrows), when ascites were collected (orange down arrows), which ascites generated OCMs (orange stars) and when the patient died (black star). Vertical coloured bars along the top of the plot area show the timing of the indicated chemotherapy treatments

Fig. 2

HGSOC is characterised by extensive chromosome instability. Genome-wide chromosome copy number profiles determined by shallow single-cell whole-genome sequencing (scWGS) of OCMs derived from patients 38, 64, 152 and 246. For each OCM, each row represents a single cell, with chromosomes plotted as columns and the copy number indicated by the colour. The four OCMs shown represent examples whereby genomes are marked by whole-chromosome aneuploidies, rearranged chromosomes, tetrasomies or monosomies. Karyotypes previously shown in Nelson et al. and Coulson-Gilmer et al. (Licenses at https://creativecommons.org/licenses/by/4.0/)

Fig. 3

Living Biobank ascites pipeline. A Bar graph summarising the biopsy pipeline, showing that from June 2016 to March 2023, 715 biobank alerts yielded 454 ascites samples from 209 patients, in turn yielding 127 OCMs. B–D Summary of OCM collection with pie charts showing breakdown of subtypes based on pathology assessment (B); chemo-naïve (CN) vs. post-treatment (C); and longitudinal subsets (D). In (D), while 61 patients are represented by a single OCM (blue), 9 patients have 2 longitudinal OCMs (grey), and 3 patient subsets have 3 longitudinal OCMs (orange), etc.

Fig. 4

Chromosome instability generates highly divergent subclones. A scWGS-derived karyotypes of EpCAM-positive and EpCAM-negative subpopulations present in the OCM generated from the 3^rd ascites sample collected form patient 64. B Table summarising characteristics of OCMs 64-1 and the two 64-3 subpopulations. C Speculative ploidy reversal event to explain how the two 64-3 subpopulations might have been generated. Karyotypes in A adapted from Nelson et al. (License at https://creativecommons.org/licenses/by/4.0/)

Fig. 5

Primary HGSOC can display very different histopathologies. Representative 20× immunohistochemistry images of the primary tumours from patients 92 and 109, stained to detect p53, PAX8, WT1 and Cytokeratin 7. Patient 92 images adapted from Coulson-Gilmer et al. (License at https://creativecommons.org/licenses/by/4.0/). Scale bar, 100 μm. Panels are representative images from single experiment

Fig. 6

OCM gene expression analysis. Principal component (PC) analysis of RNAseq-derived global gene expression profiles, distinguishing stromal and tumour clades, and showing the close relationship of longitudinal OCMs samples from patients 64, 66, 74, 110, 118 and 124, with numbers inside the symbol indicating the ascites number. 69* is a stromal culture. Published data collated from Nelson et al. , Barnes et al. , Coulson-Gilmer et al.

Fig. 7

TP53 mutation profile. A Pie charts showing the number of different TP53 mutation subtypes in the TCGA dataset compared with the subset of OCMs for which TP53 data is currently available. B Comparison of missense TP53 mutations in the TCGA dataset (grey) versus the OCM subset (purple). OCM data collated from Nelson et al. , Coulson-Gilmer et al. ; TCGA data from cBioPortal (Cerami et al. ; Gao et al. 2013)