henhaobio
Sign in

bioRxivCancer BiologyPreprint

Patient-Derived Liver Cancer Organoids Reflect Tumor Biology and Their Growth Phenotype Correlates with Clinical Outcomes

Yeon Su KimiD, Yoon-Ha Go, Hyung Sun KimiD, 서재휘, Dong oh Kim, Dong‐Youn HwangiD, Jongman Yoo, Wookyeom Yang, Jin Hong Lim

PDFDOI
In-depth full-text digest available · Jump to digest ↓

Abstract

Liver cancer remains a major global health burden with high mortality and limited treatment response prediction tools. Patient-derived cancer organoids have emerged as promising preclinical models that recapitulate tumor heterogeneity; however, the biological significance of morphological diversity within established organoids remains poorly characterized in hepatocellular carcinoma (HCC). In this exploratory study, we investigated whether distinct organoid growth phenotypes reflect underlying tumor biology and correlate with clinical outcomes. We established liver cancer organoids from resected tumor tissues of 27 patients and analyzed their clinical, histological, and genomic characteristics. Organoids were classified morphologically into cystic and solid types. Whole exome sequencing (WES) was conducted on six matched tumor-organoid pairs to assess genomic fidelity. Associations between organoid establishment, growth characteristics, and clinical parameters were statistically evaluated. Progression-free survival (PFS) was analyzed using the Kaplan-Meier method and univariate Cox proportional hazards regression. Organoids were successfully established in 13 of 27 cases (48.1%). Solid-type organoids were significantly associated with shorter PFS compared to cystic types (HR = 13.91; p = 0.0039, log-rank test). Organoid establishment was more frequent in older patients (>70 years), those with HBV infection, and tumors with positive beta-catenin expression. WES analysis demonstrated high concordance in somatic mutation profiles and variant allele frequency distributions between tissues and corresponding organoids. In univariate Cox regression, organoid growth pattern (solid vs. cystic) showed a significant association with PFS within this exploratory cohort (p = 0.0207). Patient-derived liver cancer organoids preserved the genomic and histopathological features of the original tumors. Notably, solid morphology was associated with shorter PFS, suggesting that organoid growth phenotype may serve as a supplementary indicator of tumor biological behavior in HCC. Given the modest cohort size and the absence of multivariate analysis, these preliminary findings should be interpreted with caution and warrant further large-scale validation.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 3 (6 Oct 2026) · CC BY 4.0

Organoids grew from 13 of 27 resected hepatocellular carcinomas. The sequenced lines kept parental mutations, and solid rather than cystic growth marked shorter progression-free survival (HR 13.91).

Hepatocellular carcinoma (HCC) prognosis still relies on TNM and BCLC staging, which poorly reflect tumor biology, and patient-derived xenografts are slow, costly and engraft in only 30–40% of HCC cases. This exploratory study asks whether organoid growth shape after establishment, and not only culture success, mirrors the parental tumor and carries prognostic information, a question poorly characterized in HCC.

Key results

  1. SuggestedPatients whose tumors gave solid-type organoids progressed much sooner than those whose tumors gave cystic-type organoids. Growth pattern stayed significant in univariate Cox regression, as did TNM stage.

    Solid n = 3 vs cystic n = 10. Kaplan-Meier: median PFS of 7.46 months in the solid group, with the cystic median not reached; HR = 13.91 (log-rank p = 0.0039). Univariate Cox: growth status HR = 14.90, p = 0.0207; TNM stage HR = 7.27, p = 0.0157. A multivariate model was attempted but judged not statistically valid. · Establishment and morphological classification…; Association between organoid phenotype and…; Prognostic value of clinical and organoid-based…

  2. SuggestedSolid morphology was linked to disease recurrence but not to TNM stage or to serum AFP or PIVKA levels. Tumor size and AFP were only borderline prognostic in univariate Cox analysis.

    Chi-square: solid type vs recurrence p = 0.0031; no significant association with TNM stage or AFP/PIVKA. Univariate Cox: tumor size ≥ 5 cm p = 0.0959; AFP ≥ 100 ng/mL p = 0.0983; gender, grade, neoadjuvant therapy and CK19/EpCAM expression not significant. · Clinicopathological factors associated with…; Prognostic value of clinical and organoid-based…

  3. SuggestedOrganoids were established from about half of the resections. Success was more frequent in older patients, HBV-positive patients and beta-catenin-expressing tumors, but establishment success itself did not predict progression-free survival.

    13 of 27 cases (48.1%), compared with the 30–40% engraftment cited for HCC PDX models. Chi-square: age 70 years or older p = 0.0313; HBV-positive p = 0.0217; beta-catenin-positive p = 0.0035. Kaplan-Meier: PFS did not differ significantly between establishment success and failure. · Establishment and morphological classification…; Clinicopathological factors associated with…; Association between organoid phenotype and…; Discussion

  4. ShownWhole exome sequencing of six tumor-organoid pairs showed mostly shared somatic variants, concordant variant allele frequencies (VAFs) and similar tumor mutational burden (TMB). In Patient_16, both the cystic and the solid line matched the parental tumor.

    Most frequently mutated genes were conserved within each pair; TMB difference not significant (Wilcoxon rank-sum). Patient_16 tissue-organoid VAF Pearson r = 0.69 (cystic) and r = 0.68 (solid). 34 genes carried at least three mutations across the six organoid samples, among them MUC16, DPP6, SYNE1, BCR and NELL2. · Whole exome sequencing and bioinformatic…; Genomic fidelity between organoids and parental…; Discussion

  5. SuggestedIn Patient_09, the only pair carrying TP53 and CTNNB1 mutations, the organoid retained both drivers at higher VAFs, which fits clonal enrichment in culture. HLA types matched between tumor and organoid in most cases.

    Patient_09 only: TP53 p.Y234C VAF 0.276 in tissue vs 1.000 in organoid; CTNNB1 p.K335I 0.217 vs 0.475. HLA class I and II types called by OptiType and HLA-LA were identical between tumor and organoid in the majority of cases. · Genomic fidelity between organoids and parental…; HLA genotype concordance

Methods, data and reproducibility

Resected HCC tissue was dissociated with enzymes and grown in Matrigel with a growth-factor expansion medium. Organoids were classed as cystic or solid, compared with parental tumors by H&E, immunohistochemistry, whole exome sequencing and HLA typing, then related to clinical variables (chi-square) and progression-free survival (Kaplan-Meier, univariate Cox).

  • Resected liver cancer (HCC) tumors from patients enrolled at Gangnam Severance Hospital, Seoul, between 2023 and 2025; organoid derivation attempted for each · human · n = 27 patients; 13 lines established (10 cystic, 3 solid)
  • Matched tumor tissue and organoid pairs profiled by whole exome sequencing (Twist exome kit, NovaSeq 6000, 151 bp paired-end); Patient_16 gave both cystic and solid lines · human · n = six pairs (oncoplot spans 12 samples)
  • Matched tumor and organoid samples HLA-typed for class I and class II alleles from WES data · human
  • Tumor tissues and matched organoids stained by H&E and IHC (CK19, AFP, EpCAM, HepPar-1) to compare histology and marker patterns · human
  • Clinicopathological records (stage, grade, HBV/HCV, AFP, PIVKA, IHC markers, recurrence) and progression-free survival follow-up · human · n = 27 patients
Methods & tools
Matrigel 3D organoid culture with HCCO expansion medium, H&E and immunohistochemistry (EpCAM, HepPar-1, AFP, CK19, N-cadherin, vimentin), Twist Human Core Exome 2.0 kit; Illumina NovaSeq 6000, BWA-MEM, Mutect2, vcf2maf with Ensembl VEP, maftools v2.24.0, OptiType, HLA-LA, R survival package (Kaplan-Meier, univariate Cox), GraphPad Prism 8.0, VennDiagram, ggcorrplot, ggplot2, ggpubr
Reagents & environment
R v4.5.1 with maftools v2.24.0, survival, dplyr, writexl, VennDiagram, ggcorrplot and ggpubr; GraphPad Prism 8.0; reads aligned to GRCh37 (human_g1k_v37_decoy). No versions given for the other packages.
Reproducibility gaps
Data available only on reasonable request from the corresponding authors. No sequencing accession or analysis code repository is reported. Only the Patient_09 mutation profile is supplied, as S1 Table.
Limitations (6)
  • The cohort was small, with few events and missing covariate data, so no valid multivariate Cox model could be built. Solid morphology therefore cannot yet be called an independent prognostic factor. · stated by the authors
  • Why organoids grow cystic or solid is unknown. Candidate pathways from colorectal organoid studies (IGF1R signaling, stemness) were not tested. · stated by the authors
  • There were too few samples to compare genomes between cystic and solid organoids, so morphology-linked driver alterations could not be assessed. TP53 and CTNNB1 mutations appeared in only one pair (Patient_09). · stated by the authors
  • Only early-passage organoids were studied (morphology tracked up to P5). Longer culture could select clones or add mutations that erode genomic and phenotypic fidelity. · stated by the authors
  • The solid group holds only 3 patients, and Patient_16 produced both cystic and solid lines. The paper does not say how this patient was grouped for progression-free survival, so the HR estimates (13.91, 14.90) rest on very few events. · noted in this summary
  • The Discussion says derivation introduced no clonal selection bias, yet the Patient_09 TP53 VAF rose from 0.276 to 1.000 in culture, which the Results attribute to clonal enrichment. Fidelity also rests on six pairs with mostly qualitative concordance. · noted in this summary

Read this if you build or use patient-derived HCC organoids and want early evidence linking organoid morphology, establishment success and genomic fidelity to patient progression and recurrence.

Citations
0

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISO'Sullivan '26Crignis '20Álvarez-Gon '26Malham '26Betge '22Singla '25
66% semantic matchCancer Biology
Eabha O'Sullivan, Christina Cahill, Rebecca M. O'Brien+12 more·bioRxiv·Apr 10, 2026