Integrative Molecular Analyses of the MD Anderson Prostate Cancer Patient-derived Xenograft (MDA PCa PDX) Series

Authors

Anselmino N

Labanca E

Shepherd PDA

Dong J

Yang J

Song X

Nandakumar S

Kundra R

Lee C

Schultz N

Zhang J

Araujo JC

Aparicio AM

Subudhi SK

Corn PG

Pisters LL

Ward JF

Davis JW

Vazquez ES

Gueron G

Logothetis CJ

Futreal A

Troncoso P

Chen Y

Navone NM

Doi

10.1158/1078-0432.CCR-23-2438

PMID: 38488813 · DOI: 10.1158/1078-0432.CCR-23-2438 · Journal: Clinical Cancer Research (2024)

TL;DR

This study performed comprehensive molecular characterization (WGS, targeted sequencing with the T200.1 panel, and RNA-seq) of 44 patient-derived xenograft (PDX) models from 38 prostate cancer patients at MD Anderson Cancer Center. The PDX cohort spans adenocarcinoma and neuroendocrine prostate cancer morphologies, multiple disease states, and metastatic sites. The authors found that 91% of PDXs harbored oncogenic alterations in AR, RB1, TP53, or PTEN, and that DNA damage response (DDR) pathways were differentially upregulated in neuroendocrine compared to adenocarcinoma PDXs. The study also identified an FGFR1 transcriptomic signature associated with bone metastasis. All data are available in cBioPortal as prad_msk_mdanderson_2023.

Cohort & data

  • 44 PDX models derived from 38 patients with prostate cancer (PRAD), including adenocarcinoma, neuroendocrine prostate cancer (PRNE), poorly differentiated carcinomas, and sarcomatoid carcinomas.
  • Samples from primary sites and metastases (bone, lymph node, circulating tumor cells), both therapy-naive and castration-resistant.
  • Sequencing: WGS (30X), targeted sequencing (T200.1 panel, 263 genes, 400X), and RNA-seq.
  • Aligned to hg19 reference genome; mouse reads (mm10) filtered out.
  • Cross-referenced with TCGA-PRAD and SU2C patient datasets.
  • Data deposited in cBioPortal as prad_msk_mdanderson_2023 and dbGaP (phs003420.v1.p1).

Key findings

  • Driver gene prevalence: 91% of PDXs carried oncogenic alterations in at least one of AR, RB1, TP53, or PTEN (PMID:38488813).
  • TMPRSS2-ERG fusion was detected in 13 PDX models; 10 of these correlated with increased ERG expression. Lack of correlation in the remaining 3 was consistent with low/null AR expression.
  • TMPRSS2-ETV4 fusion was found in 2 PDXs with increased ETV4 expression. ETV1 fusions (ETV1-FOXA1, ACSL3-ETV1) were detected but did not correlate with expression changes.
  • CDK12 alterations were identified in 4 PDXs through integrative analysis, including biallelic mutations (174-6, 322-2) and structural/splicing-mediated inactivation (117-9, 328-5) that would have been missed without multi-platform integration.
  • RB1 alterations were significantly more frequent in NEPC than adenocarcinoma (P = 0.0002 for combined CNV + mutation), and RB1 and AR expression were significantly lower in NEPC (P = 9 x 10^-5 for RB1; P = 4 x 10^-9 for AR mRNA).
  • DDR pathway upregulation in NEPC: Unsupervised clustering of DDR-associated gene expression segregated models concordant with morphologic classification. NEPC samples exhibited elevated MYCN and/or AURKA expression without gene amplification. The same DDR pattern was confirmed in the SU2C dataset.
  • FGFR1 signature: Promoter CpG methylation was inversely correlated with FGFR1 expression in both PDXs and TCGA-PRAD. Three downstream genes (NRP2, LRP4, TGFBI) segregated PDXs by FGFR1 level and were preferentially expressed in bone metastases in SU2C.
  • Heterogeneity pairs derived from different areas of the same tumor shared core driver alterations but exhibited unique mutations (e.g., NRAS Q61K in 144-4; SPEN mutations in 146 pair; KMT2D in 316-1).
  • Longitudinal pairs showed conserved alteration profiles, suggesting treatment does not select for particular genomic alterations at the same tumor site.

Genes & alterations

  • AR – Oncogenic mutations (H875Y, T878A), amplification, enhancer region amplification, and ARv7 splice variant expression in multiple PDXs. AR expression was significantly reduced in NEPC (P = 4 x 10^-9).
  • PTEN – Deep deletions prevalent across the cohort; some models showed expression loss without deep deletion or mutation, implicating other regulatory mechanisms.
  • RB1 – Deep deletions and mutations, significantly enriched in NEPC. Expression loss without genomic alteration observed in some models.
  • TP53 – Mutations frequent across the cohort; higher frequency of TP53 mutations noted in NEPC.
  • CDK12 – Biallelic loss via dual mutations or structural variation in 4 PDXs; CDK12-loss-associated focal copy-number gains observed.
  • TMPRSS2 – Fusion partner for ERG (13 models) and ETV4 (2 models).
  • ERG – Overexpressed when fused with TMPRSS2 in AR-expressing models.
  • FGFR1 – Expression regulated by promoter methylation; elevated in bone metastases; novel downstream signature (NRP2, LRP4, TGFBI) defined.
  • SPOP – Mutations detected in heterogeneity pair 316.
  • FOXA1 – Involved in ETV1-FOXA1 gene fusions.
  • NRAS – Oncogenic Q61K mutation in PDX 144-4 (NEPC), seldom implicated in prostate cancer.
  • SPEN – Mutations in heterogeneity pairs (146, 316), implicated as a hormone-inducible transcription repressor in prostate cancer.
  • RNF43 – Mutations in pair 316, a negative regulator of the Wnt pathway.
  • KMT2D – Mutation in PDX 316-1.
  • KDM6A – Deep deletion in NEPC pair 144.
  • AURKA, MYCN – Elevated expression in NEPC without gene amplification; linked to DDR and neuroendocrine differentiation via the MYCN-CDK5-RB1-E2F1 axis.
  • ATM, BRCA1, BRCA2, CHEK2 – DDR pathway genes with driver genomic alterations predominantly in adenocarcinoma PDXs, while NEPC showed transcriptomic upregulation.

Clinical implications

  • The PDX platform enables informed selection of clinically annotated models for drug testing (organoid-based assays demonstrated for cisplatin, paclitaxel, cabazitaxel, bicalutamide, enzalutamide, and niraparib).
  • The FGFR1 downstream signature (NRP2, LRP4, TGFBI) could serve as an indicator of FGFR pathway activation, potentially stratifying patients for FGFR-targeted therapy (e.g., dovitinib) in bone-metastatic CRPC.
  • DDR upregulation in NEPC, combined with evidence that PARP inhibition impairs MYCN-CDK5-RB1-E2F1-mediated neuroendocrine differentiation, supports investigation of PARP inhibitors in NEPC.
  • Conserved driver alteration profiles in longitudinal pairs suggest that treatment resistance may be driven by non-genomic (e.g., epigenomic, microenvironmental) mechanisms rather than selection for new driver mutations.

Limitations & open questions

  • Cohort size (44 PDXs from 38 patients) limits statistical power for detecting differences in alteration frequency between morphologic groups.
  • PDXs are grown subcutaneously in untreated mice, so effects of treatment pressure and organ-specific niches on phenotype are not fully captured.
  • Some models showed expression loss of PTEN or RB1 without detectable genomic alteration, suggesting epigenomic or posttranscriptional regulation that was not profiled in this study.
  • The biological consequences of several detected gene fusions (e.g., ETV1-FOXA1) remain to be elucidated.
  • Whether ARv7 can independently drive prostate cancer (versus being a passenger) remains unresolved.
  • The FGFR1 signature (NRP2, LRP4, TGFBI) requires prospective clinical validation as a biomarker for FGFR-targeted therapy response.
  • Alignment was performed to hg19 rather than the current GRCh38 reference.

Citations from this paper used in the wiki

  • “91% of PDXs presented oncogenic alterations in AR, RB1, TP53, or PTEN” (Results, Prostate cancer drivers section).
  • “RB1-altered models were more frequent in NEPC than in Ad (P = 0.0002)” (Results, Molecular comparison section).
  • “DDR-associated mechanisms emerged as differentially regulated between adenocarcinoma and neuroendocrine prostate cancer” (Abstract).
  • “CpG methylation at the FGFR1 promoter was associated with low expression” and the NRP2/LRP4/TGFBI signature (Results, Exploiting the applicability section).
  • “TMPRSS2-ERG fusion in 13 models, but only 10 correlated with increased ERG expression” (Results, Gene fusions section).
  • “Four PDXs with CDK12 alterations in the cohort” including integrative detection of biallelic inactivation (Results, Prostate cancer drivers section).

This page was processed by crosslinker on 2026-05-04.