Purpose:

Advances in prostate cancer lag behind other tumor types partly due to the paucity of models reflecting key milestones in prostate cancer progression. Therefore, we develop clinically relevant prostate cancer models.

Experimental Design:

Since 1996, we have generated clinically annotated patient-derived xenografts (PDXs; the MDA PCa PDX series) linked to specific phenotypes reflecting all aspects of clinical prostate cancer.

Results:

We studied two cell line–derived xenografts and the first 80 PDXs derived from 47 human prostate cancer donors. Of these, 47 PDXs derived from 22 donors are working models and can be expanded either as cell lines (MDA PCa 2a and 2b) or PDXs. The histopathologic, genomic, and molecular characteristics (androgen receptor, ERG, and PTEN loss) maintain fidelity with the human tumor and correlate with published findings. PDX growth response to mouse castration and targeted therapy illustrate their clinical utility. Comparative genomic hybridization and sequencing show significant differences in oncogenic pathways in pairs of PDXs derived from different areas of the same tumor. We also identified a recurrent focal deletion in an area that includes the speckle-type POZ protein-like (SPOPL) gene in PDXs derived from seven human donors of 28 studied (25%). SPOPL is a SPOP paralog, and SPOP mutations define a molecular subclass of prostate cancer. SPOPLdeletions are found in 7% of The Cancer Genome Atlas prostate cancers, which suggests that our cohort is a reliable platform for targeted drug development.

Conclusions:

The MDA PCa PDX series is a dynamic resource that captures the molecular landscape of prostate cancers progressing under novel treatments and enables optimization of prostate cancer–specific, marker-driven therapy.

Translational Relevance

We report the generation of a large number of patient-derived xenografts (PDXs; MDA PCa PDX series) representative of the clinical spectrum of prostate cancer (PCa). The MDA PCa PDX series is a dynamic resource that captures the molecular landscape of prostate cancers over time. It provides insight into the biological basis that accounts for heterogeneity and serves as an invaluable resource for discovery, therapy development, and optimization of personalized therapy targeting prostate cancer–specific molecular markers. The models developed thus far have led to the identification of clinically relevant therapy targets and have proven valuable for drug testing.

Metastatic prostate cancer (PCa) that progresses after androgen ablation therapy [i.e., castration-resistant prostate cancer (CRPC)] remains incurable. Prostate cancer consist of clinical subsets including the typical prostate adenocarcinoma with a predictable pattern of progression, as well as other subsets with atypical clinical behavior similar to that of prostatic small-cell neuroendocrine carcinoma [aggressive variant prostate cancer (AVPC)], which accounts for about 40% of lethal prostate cancer (1, 2). Prostate cancer is also heterogeneous at the morphologic and molecular level (3). Thus, responses to targeted therapies differ between subpopulations of patients with prostate cancer, and it is essential to have a spectrum of biological models that reflect the diverse clinical, morphologic, and molecular prostate cancer phenotypes for further research and therapy development.

Prostate cancer research has historically suffered from a lack of patient-derived models. Efforts to develop prostate cancer patient-derived xenografts (PDXs) have been undertaken in several institutions (4). However, few prostate cancer PDX banks with substantial number of models have been established. Some contain PDXs derived from autopsy samples (the LuCaP series) with no information on tumor progression after the samples have been acquired. Others are mainly derived from primary tumors (the Living Tumor Laboratory series and the PC series developed in Erasmus Medical Center, Rotterdam, the Netherlands; ref. 4). Therefore, despite these efforts, we still lack a collection of clinically annotated prostate cancer PDXs that reflects the full spectrum of potentially lethal disease, namely therapy-naïve and therapy-resistant prostate cancers reflecting clinical and morphologic variants derived from primary sites as well as metastases, and can be related to the tumor donor's progression to therapy, which is monitored by expert genitourinary oncologists. We have approached this challenge by establishing PDXs using tissue specimens taken from patients with potentially lethal prostate cancer undergoing surgical resection or biopsy of primary tumors and/or metastatic sites demonstrating clinical progression. This is an ongoing program that provides a diverse repository of well-annotated tissue samples and PDXs that can be linked prospectively with specific stages and states of prostate cancer progression and also reflects the clinical and molecular evolution of therapy-resistant prostate cancer.

Here, we report the morphologic and molecular characterization of two cell line–derived xenografts and the first 80 PDXs derived from 47 human prostate cancer donors developed in our program. Of these, 47 PDXs derived from 22 donors are working models and can be expanded either as cell lines (MDA PCa 2a and 2b) or PDXs. This study constitutes the foundation for the development and use of PDXs for precision oncology. Furthermore, as this is a dynamic repository, we also established PDXs from human prostate cancer that are not described in this report because they are currently being characterized and will be the subject of a follow-up publication. These new PDXs include those derived from prostate cancers progressing on new hormonal agents, enzalutamide and abiraterone. Enquiries about their availability to the scientific community can be sent to the corresponding author of this article.

The prostate cancer PDX program at MD Anderson Cancer Center

To efficiently process all aspects of PDX development, the MDA prostate cancer PDX program operates within a highly integrated network of physicians, scientists, laboratory staff, and resources within the Tissue Biospecimen and Pathology Resource at MD Anderson Cancer Center (Houston, TX). These individuals include urologists, oncologists, interventional radiologists, and pathologists, as well as staff who provide regulatory compliance support for archived biospecimen requests, consent validation, material transfer agreements, data management, and sample distribution.

PDX development

In our program, prostate cancer tissue samples used for PDX development were derived from therapeutic or diagnostic procedures, namely, radical prostatectomies, orthopedic, and neurosurgical procedures to palliate complications, and biopsies of metastatic lesions. Written informed consent was obtained from patients before sample acquisition, and all samples were processed according to a protocol approved by the Institutional Review Board of the University of Texas MD Anderson Cancer Center (Houston, TX). The studies were conducted in accordance with the Belmont Report and the U.S. Common Rule. On collection, the specimens were placed in cold (4°C), sterile α-MEM (Gibco; Invitrogen), and small pieces were then implanted into subcutaneous pockets of 6- to 8-week-old male CB17 SCID mice (Charles River Laboratories; Fig. 1A). Tissue adjacent to the implanted samples (mirror image) were formalin-fixed, paraffin-embedded (FFPE), and tissue sections were used for quality control. Mice were monitored weekly for tumor growth. Once the initial implanted tumor grew in the mouse (passage 0), it was harvested and sequentially passaged to five mice (Fig. 1A). At each mouse-to-mouse passage, representative samples of each PDX were frozen in a DMSO solution for later implantation and PDX expansion, and FFPE, flash-frozen, and frozen in optimal cutting temperature (OCT) medium for histopathologic and molecular studies (Fig. 1A).

To achieve a reliable and reproducible method for PDX development, we established a standard operating procedure with an optimized process for tissue samples, including shortening the time from sample acquisition to implantation in mice and selecting tissue samples with the highest percentage of viable cells. We do not maintain PDXs in mice unless they are being utilized by investigators to perform experiments. Historically, once initial growth was observed and the PDX underwent five or more serial passages in mice, the likelihood that the model would continue to propagate in mice is high (80%–90%). Also, in our experience most PDXs will regrow after cryopreservation either in the first or second attempt, provided that the tissue used for cryopreservation was obtained from a PDX that reached passage 5. Every time we harvested tissue for cryopreservation, a mirror tissue sample for quality control was obtained. The viability of tissue at the time of cryopreservation is essential for regrowth. One caveat is that we have not tested the viability of cryopreserved tissue for more than 1 year as we have consistently generated new vials at shorter intervals.

All animal experiments were conducted in accordance with accepted standards of animal care and were approved by the Institutional Animal Care and Use Committee at The University of Texas MD Anderson Cancer Center (Houston, TX).

PDX growth in intact and castrated mice

All studies were performed using 6- to 8-week-old male CB17 SCID mice. Twenty-three mice were subcutaneously implanted with MDA PCa 183-A. Nine of these mice were monitored for tumor growth over time (intact mice). Seven were castrated when tumor volume was equal or higher than 500 mm3 (castrated mice). The remaining seven mice were monitored for tumor volume before and 10 days after castration (tumor volume before and after castration).

Twenty-one mice were subcutaneously implanted with MDA PCa 180-30. Seven of these mice were monitored for tumor growth (intact mice), and 14 were castrated when tumor volume was equal or higher than 500 mm3. Values were used for the plots of castrated mice and tumor volume before and after castration.

PDX growth intrabone in mice treated with vehicle or FGFR inhibitor

Twenty-three male CB17 SCID mice were injected with cells derived from MDA PCa 118b into the distal ends of the femurs according to published protocols (5). Twelve days after cell injection, 10 mice were treated with vehicle and 13 with erdafitinib (Janssen Pharmaceuticals) by oral gavage. Using the same protocol, 20 male CB17 SCID mice were injected with cells derived from MDA PCa 183-A into the distal ends of their femurs. Twenty-three days after cell injection, 10 mice were treated with vehicle and 10 with erdafitinib by oral gavage. After 3 weeks of treatment of MDA PCa 118b–bearing mice and 5 weeks of treatment of MDA PCa 183-A–bearing mice, tumor volume was measured from serial sagittal magnetic resonance (MR) images in vehicle- and erdafitinib-treated mice as reported previously (6).

Statistical analysis

For the studies on tumor size response to mice castration, the change of tumor volume over time was explored using spaghetti plots. Linear mixed models were fit to assess the change of tumor volume over time in all groups, while taking into account the intramouse correlations. All statistical analyses were performed using SAS and Splus.

Two-sample t tests were used to study tumor volume response to erdafitinib treatment.

The prostate cancer PDX program at MD Anderson Cancer Center

In 1996, we initiated efforts to develop patient-derived models of prostate cancer and established two cell lines (MDA PCa 2a and MDA PCa 2b; ref. 7). Subsequent attempts (n = 25) at establishing in vitro models were unsuccessful. These attempts include prostate cancer samples obtained at radical prostatectomy and samples from different metastatic sites (bone, lymph node, skin, and ascites). The methods used during these attempts have been published previously (8). In the cases derived from primary prostate cancer, we saw initial cell growth in most cases, but normal prostatic epithelial cell overgrowth was the most common outcome. In the cases derived from metastases, we frequently obtained short-term cultures, but the cells eventually stopped growing or underwent cell death. Methods and alternatives used during these attempts are discussed in a previous publication (8).

From 1996 to April 2010, we processed prostate cancer tissue samples from 163 patients for PDX development, including samples from different areas of the same tumor to understand tumor heterogeneity. In total, during that period, we established 80 PDXs derived from 47 human prostate cancer donors; of these, 47 PDXs derived from 22 donors are working models and can be expanded either as cell lines or PDXs (Table 1; Supplementary Table S1; Supplementary Results). These PDXs are named: MDA PCa followed by a number unique to the donor tumor, tumor site, and procedure date. Currently, the success rate for PDX development is 30%–40%, irrespective of the site of origin of the tissue. The variability probably depends on the amount of viable tissue submitted for PDX development. It takes between 3 months and 3 years (depending on the rate of growth) for a tissue sample implanted in mice to be passed through five mice sequentially. The PDXs developed are derived from primary prostate cancers or areas of direct extension to adjacent organs (bladder and rectum) and from metastases, and include therapy-naïve and -resistant adenocarcinomas, as well as clinical and histopathologic subtypes (Fig. 1A; Table 1; Supplementary Table S1). For brevity, from now on the term “PDXs” will be inclusive of MDA PCa2a-T and 2b-T xenografts (T indicates a cell line–derived xenograft).

As previously mentioned, this is a dynamic repository and, to date, in addition to the PDXs described in this report, we have also established PDXs from 52 patients with prostate cancer, which are currently being characterized. These PDXs derived from primary sites, bone, or soft-tissue metastases include those derived from treatment-naïve prostate cancers and prostate cancers progressing on first- and second-generation androgen deprivation therapy, chemotherapy, and other therapies.

Morphologic and IHC features of human donor tumors and PDXs

Early- and late-passage PDXs (passage 1–3 and 5–6, respectively) retain the morphology of the human prostate cancer donor (Table 1). PDXs also have the same IHC profile [androgen receptor (AR), synaptophysin (SNP), chromogranin A (CGA), or CD56 (NCAM)] as their human prostate cancer donor and are either adenocarcinomas (Fig. 1B) or neuroendocrine carcinomas, which includes small-cell morphology (Fig. 1C). PDXs established from mixed adenocarcinoma and neuroendocrine carcinoma, reflect one or both morphologic components present in the human prostate cancer (Fig. 1D; Supplementary Fig. S1; Supplementary Results). Neuroendocrine carcinoma PDXs are all AR negative and positive for one or more neuroendocrine markers (SNP, CGA, or CD56; ref. 9; Fig. 1C; Table 1; Supplementary Results). Adenocarcinomas with negative or low expression of AR and negative expression of SNP and CGA (e.g., MDA PCa 83 and MDA PCa 118b) fit the recently published definition of double-negative prostate cancers (10).

Of the 47 donor tumors used to establish the PDXs, 32 were adenocarcinomas (68%), and 11 were neuroendocrine carcinoma or mixed adenocarcinoma and neuroendocrine carcinoma (23%; Fig. 1E; Table 1). This distribution recapitulates the morphologic landscape of potentially lethal human prostate cancer (1, 2). In the other four cases (9%), either there was no material available for review (e.g., serous fluids), there were no malignant cells based on the pathology report, or they were unusual prostate cancer morphologies (Table 1; Supplementary Table S1; Supplementary Results).

AR, ERG, and PTEN status in PDXs and a subgroup of 15 PDX–human donor tumor pairs

Previous reports indicate that 50%–60% of human prostate cancers have recurrent rearrangements involving ERG, ETV1, ETV4, or ETV5. AR signaling status, a critical determinant of prostate cancer behavior, is a therapeutic target for prostate cancer, and aberrant ERG expression cooperates with PTEN deletions to promote prostate cancer progression (11). IHC and FISH analyses of these genes demonstrated that adenocarcinoma PDXs derived from 27 of the 29 human donors available for study were AR positive (93%), and ERG expression was positive in 12 of 26 of the AR-positive samples available for study (46%; Fig. 1F). MDA PCa 2a-T and 2b-T and MDA PCa 177 have ETV1 rearrangement as confirmed by FISH and RNA-ISH (ref. 12; Supplementary Fig. S2). PDXs derived from neuroendocrine carcinomas, which do not express AR, also did not express ERG in all but one case (Supplementary Table S2). PDXs derived from 18 of 26 adenocarcinomas (69%) available for evaluation had homo- or heterozygous PTEN deletion (Fig. 1F; Supplementary Table S2; Supplementary Results). PDXs derived from six of seven neuroendocrine carcinomas (86%) had homo- or heterozygous PTEN deletion (Fig. 1F). PTEN status was variable in PDXs derived from mixed adenocarcinoma and neuroendocrine carcinoma (Supplementary Table S1). In summary, the morphologic distribution and molecular features of the PDXs established were consistent with the reports on human prostate cancer in the general population (13).

Finally, to further understand how AR, ERG, and PTEN status in prostate cancer PDXs recapitulate human prostate cancer donors, we studied 15 PDX–human donor pairs and found concordance in most samples. Figure 2 shows results of the 11 pairs that were not archived, more details are in Supplementary Results.

MDA PCa PDX growth in intact and castrated male mice

Androgen ablation is the standard first-line treatment for metastatic prostate cancer. The response duration is highly variable and second-generation androgen ablation therapeutics (abiraterone and enzalutamide) have significant activity in patients with CRPC. Nevertheless, the emergence of resistance remains inevitable in most cases. Thus, to develop effective treatment strategies aimed at improving patient management, there is an urgent need to further understand the mechanisms that account for treatment response and resistance.

To assess the potential utility of our models in furthering our understanding of the mechanisms underlying CRPC progression and to develop effective therapies, we studied the effect of surgical castration of PDX-bearing mice on the growth of the tumors. Testosterone levels in intact male mice were shown to mimic standard androgen ablation, and castrated mice mimic abiraterone-treated patients (14). Castration was performed as published previously (7). For these studies we selected two models, MDA PCa 183-A, derived from a treatment-naïve prostate cancer, and MDA PCa 180-30, derived from a prostate cancer that progressed on androgen deprivation and chemotherapy.

Change of tumor volume over time in the intact and castrated groups was assessed by a linear mixed model, while considering the intramouse correlations. We found a statistically significant difference between the slopes of the tumor volumes in subcutaneous MDA PCa 183-A growing in intact and castrated mice (P < 0.0001; Fig. 3A). Accordingly, when we analyzed the growth slope of MDA PCa 183-A before and after castration in the same mouse, we found a statistically significant increase in tumor volume before castration (slope = 26.8; P < 0.0001) and a statistically significant decrease in tumor volume after castration (slope = −18.0; P < 0.0001). Of interest, in the castrated group, tumors in three of seven mice grew fast and reached a volume above what is allowed by our animal facilities guidelines and had to be killed. The tumor volume in the other four mice either plateaued and/or showed a subsequent decline (Fig. 3A). Although, the small number of mice does not have enough power for a statistically significant conclusion, these results suggest that two different populations of cells exist in MDA PCa 183-A PDXs, which is in alignment with the fact that the cells were derived from an untreated human tumor, and therefore, there was no prior treatment selection. In contrast, although the growth rate of MDA PCa 180-30 was slowed down after castration (P = 0.008), there was a more uniform response to castration in this tumor with a clear relapse occurring in all tumors over time (Fig. 3A). This is in alignment with the fact that MDA PCa 180-30 was derived from a human prostate cancer that underwent multiple therapies, and therefore, there was a selection. These results indicate that the MDA PCa PDXs are useful models to study CRPC progression.

Although, further analysis of these tumors is beyond the scope of this report, in a technical note, when we find a statistically significant difference in tumor volume between drug- versus vehicle-treated mice, we perform morphologic, IHC, and molecular analysis of tumors. Briefly, we cut a longitudinal section in the middle of the harvested tumor and prepare FFPE blocks for morphologic and IHC analysis. Adjacent tissue pieces are cut and flash frozen for molecular studies. Alternatively, fresh tissue can be used to perform single-cell sequencing. One benefit of performing preclinical studies using PDXs in mice is that certain IHC and molecular studies can be species specific.

Effect of a specific pan-FGFR inhibitor on two bone-derived PDXs

Bone-forming metastases dominate the clinical picture of men with CRPC (15, 16). Studies by our group and others have implicated the FGF axis in the pathogenesis of prostate cancer bone progression (5, 17), and we showed that blockade of FGFRs with the receptor tyrosine kinase inhibitor, dovitinib, has clinical activity in a subset of men with CRPC and bone metastases (6). More recent studies implicate the FGF axis in the progression to androgen ablation and other therapies (10, 18). We, therefore, assessed the antitumor activity of erdafitinib, a novel and selective pan-FGFR tyrosine kinase inhibitor, preclinically. For these studies, we selected two bone-derived PDXs that recapitulate the bone-forming phenotype observed in human prostate cancer, MDA PCa 118b and MDA PCa 183-A (Fig. 3B and C). RT-PCR analysis indicated that FGFR1 was high in MDA PCa 118b and about 20-fold lower in MDA PCa 183-A. All other receptors were expressed at very low levels (Fig. 3B and C). Accordingly, FGFR1 expression was high in MDA PCa 118b and not detectable in MDA PCa 183-A by the IHC analysis (Fig. 3B and C). Cells derived from these PDXs were injected into the distal ends of the femurs of intact male SCID mice according to published protocols (5) and mice were subjected to erdafitinib (12.5 mg/kg body weight twice a day) or vehicle administration by oral gavage. Potent antitumor effect of erdafitinib against prostate cancer cells derived from MDA PCa 118b (P < 0.007), but not MDA PCa 183-A, was observed by MR analysis (Fig. 3). These results indicate that erdafitinib is active in controlling the growth of FGFR1-expressing prostate cancer cells in bone and that our PDXs are informative in the preclinical setting.

The same technical note that was outlined at the end of the previous section applies here, only that in tumor-bearing bones it is challenging to cut each tumor in equally representative pieces. Therefore, we use a set of bone tumor (one per each mouse) for morphologic, IHC, and molecular studies. Of course, the flash frozen material will have a significant contribution of mouse (bone) cells. Alternatively, the femoral shafts of the tumors bearing femur can be flashed to obtain a product enriched for tumor cells.

Copy-number alteration in PDXs derived from different areas of the same human prostate cancer

We performed array comparative genomic hybridization (aCGH) to assess copy-number changes in different PDXs derived from the same human prostate cancer donor.

Neuroendocrine carcinoma PDXs MDA PCa 144-13 and 144-4 lost cyclin-dependent kinase inhibitor 2A (CDKN2C) (1p32.3) and PTEN (10q23.31), and had a partial loss of RB (13q14) (Supplementary Fig. S3). However, notable differences in copy-number changes in specific regions were also found (Supplementary Fig. S4, Supplementary Results).

Common copy-number alterations among MDA PCa 146-10, 146-12, and 146-20 were found despite morphologic differences, including loss of PTEN and loss of MAP3K1 via a focal homozygous deletion in chromosome 5 (Supplementary Fig. S5), which increases sensitivity to MEK inhibition (19). However, many marked differences in specific regions were identified (Supplementary Fig. S5).

MDA PCa 153 PDXs are adenocarcinomas that share many gains and losses, including PTEN loss (Fig. 4AC). However, MDA PCa 153-14 has AR amplification and a deletion in the area encompassing the CDKN2A (Fig. 4B), suggesting alternative mechanisms of progression within cells derived from the same tumor. In this context, it is noticeable that ERG fusion by deletion is very prominent in MDA PCa 153-7 but not in 153-14, showing further evidence of the intratumor heterogeneity typical of prostate cancer (Fig. 4D). Other cases are presented in Supplementary Results (Supplementary Figs. S6 and S7).

These results illustrate the high degree of heterogeneity in prostate cancer, which may underlie the diverse mechanisms of progression to targeted therapies.

Gene expression analysis by next-generation RNA-sequencing

We subsequently focused on MDA PCa 144 and MDA PCa 146 pairs to better understand human prostate cancer heterogeneity at the gene expression level. We found a strong correlation in gene expression between MDA PCa 144-4 and 144-13 PDXs (r = 0.89 to 0.90; Fig. 4E, top). However, 3,351 genes were differentially regulated between these PDXs (Supplementary Table S3), and gene set enrichment analysis (GSEA) identified hallmark MYC, E2F, and G2–M-phase checkpoints as the most significantly enriched target gene sets in MDA PCa 144-13 (Fig. 4E, bottom). This is in alignment with the 47-fold higher expression of MYC in MDA PCa 144-13 compared with 144-4 (Supplementary Table S4). New therapeutic approaches are emerging to target this oncogene (20).

We also found a strong correlation in gene expression between MDA PCa 146-10 and 146-12 (Fig. 4F, top). However, 3,022 genes were differentially regulated between these PDXs (Supplementary Table S5), and GSEA showed that hallmark androgen response genes and fatty acid metabolism target genes are among the most significantly enriched in MDA PCa 146-12 compared with 146-10 (Fig. 4F, bottom; Supplementary Table S6). This is in alignment with our previous finding that MDA PCa 146-12, but not 146-10, expresses AR (Fig. 1D). Recent studies demonstrate that AR is tightly linked with lipogenesis in prostate cancer and suggest that targeting fatty acid metabolism will inhibit AR signaling (21). Our studies emphasize the clinical relevance of MDA PCa PDX models in reflecting the heterogeneity of prostate cancer and identifying different pathways (e.g., MYC and AR signals) that drive the growth of heterogeneous cell populations within a single cancer.

Copy-number alteration and SPOPL mutational analyses

Results of aCGH analysis of 37 PDXs derived from 28 human tissues identified gains and losses previously reported for prostate cancer (Supplementary Fig. S8; ref. 13). Supplementary Table S7 outlines specific rearrangements (TMPRRS2-ERG/DSCAM/PRDM15/ETV1), amplifications (AR and MYC), and deletions (PTEN, SPOP, SPOPL, and TP53) identified in these PDXs.

Importantly, by aCGH, we observed that nine PDXs derived from seven human prostate cancer donors displayed a focal deletion in the area encompassing speckle-type POZ protein-like (SPOPL) gene (7/28, 25%; Fig. 5A; Supplementary Table S7). Deletions in SPOPL have been previously reported in prostate cancer (22) and comprise about 7% of the prostate cancers in The Cancer Genome Atlas (TCGA) dataset (Fig. 5B; ref. 22). SPOPL is a MATH-BTB protein that shares an overall 85% sequence identity with SPOP (a SPOPL paralog). SPOP was recently reported to be mutated in about 8% of prostate cancers and defines a molecular subclass (23). No mutations were found in SPOP in our cohort, but we found a heterozygous deletion in a region including SPOP (17q21.33) in MDA PCa 101 (Fig. 5C). Other focal deletions identified in our PDXs include those spanning tumor suppressors TP53 (17p13.1), CDKN1B (12p13.1), MAP3K1 (5q11.2), and FANCD2 (3p26) (MDA PCa 79; 101; 115-C1; 146-10, -12, -20; 160-29; and 183-A). These genomic alterations have been reported in human prostate cancer (22, 23), which further supports the clinical relevance of our models.

The MDA PCa PDX series includes models derived from prostate cancers encompassing the entire clinical spectrum, including hormone-naïve and CRPC, primary tumors and metastases, and typical and AVPC. It also includes prostate cancer morphologic variants (e.g., adenocarcinoma and neuroendocrine carcinoma). As the MDA PCa PDX program is constantly accruing samples for PDX development, it captures the evolving molecular landscape of prostate cancer progressing under novel therapies. To date, we have 154 PDXs derived from 99 patients with prostate cancer, including the nonarchived models from the 47 human donor tumors described here that can be expanded either as cell lines (MDA PCa 2a and 2b) or PDXs. The racial distribution of patient donors of the established prostate cancer models reflects the patient population treated at our institution (88 Caucasian, six African American, and five Hispanic).

Previous sequencing studies of multiple metastatic sites at different times provided unique insights into tumor evolution (24, 25). However, tissue samples obtained at certain timepoints during therapy are snapshots of progression. The MDA PCa PDX series provides a biological tool to study the role of these alterations in the progression of the disease experimentally. Furthermore, current clinical trials in metastatic CPRC focus on targeting pathways that are altered in this advanced disease state. Clinical trials are the benchmark for establishing the therapeutic activity of drugs, and the value of biomarkers of response and/or resistance to drugs should be established in prospective studies. But studies involving patients and patients' biological material are limited owing to feasibility, cost, and ethical constraints. Therefore, PDXs are particularly important for establishing the preclinical antitumor activity and tolerability of new drugs or drug combinations prior to clinical studies. The use of PDXs is also important for determining the contribution of the tumor microenvironment (e.g., bone) to prostate cancer progression and resistance to therapy, the antitumor effect of drugs, the identification of predictors of treatment response, and the emergence of drug-resistant clones (biomarker identification). This evidence underscores the importance of the availability of PDXs in the preclinical setting for drug selection, therapy development, selection of markers of treatment response, and identification of promising combination treatment strategies. Furthermore, advances in therapy development can be achieved by integrated analysis of clinical and coclinical studies with PDXs.

One of the most important subjects of prostate cancer research is the identification of mediators of progression and resistance to therapy (relapse), which can be accelerated by using preclinical approaches. A deep understanding of the molecular mechanisms underlying the results of these preclinical studies requires the ability to introduce genetic alterations into prostate cancer cells. However, as it happens with human prostate cancer cells, prostate cancer cells derived from PDXs do not establish as cell lines, which is a significant obstacle to genetically edit these cells. The recent establishment of organoid technology (26, 27) provides a biological platform for prostate cancer cell propagation in vitro, which can be manipulated (e.g., genetic manipulation and in vitro selection). This new methodology complements PDXs as preclinical models of prostate cancer. However, 3D organoid growth conditions do not recapitulate the tumor microenvironment. Thus, findings using 3D organoids need to be complemented by studies using PDXs to translate those findings to the clinic (28). One final point that has not been successfully addressed by PDXs is the study of the immune system in cancer (28).

SPOPL is a component of a cullin-based ubiquitin ligase complex and its clinical significance is reported for medulloblastoma (29). In the TCGA provisional dataset, SPOP mutation was reported in 11% and SPOPL was deleted in 7% of primary prostate cancer with a tendency to cooccur (Fig. 5B). Interestingly, in the SU2C/PCF dream team dataset (23), this cooccurrence is absent. There is an increase in SPOP amplification and a reduction in the incidence of both SPOP mutation and SPOPL deletion, but the combination of these alterations totaled 8% (Fig. 5D). The enriched incidence of SPOP mutations in earlier disease relative to metastatic CRPC has been reported by others (22, 30). Together, these results suggest that there is a selection for mutually exclusive SPOP mutation or SPOPL loss in metastatic CRPC. Our cohort is enriched for SPOPL deletions, and therefore, constitutes a unique resource to study the role of SPOPL deletion in CRPC.

Finally, we show here that the response of the PDX to castration is different depending on the source of the PDX (e.g., tumor donor prior therapy) and as expected, PDXs derived from a treatment-naïve tumor have a better response to castration than those derived from prostate cancers that progress after therapy. Furthermore, our models also reflect the bone phenotype typical of prostate cancer bone metastases and predict tumor response to targeted therapies.

In summary, The MDA PCa PDX collection presented here include adenocarcinoma and neuroendocrine carcinoma PDXs derived from primary prostate cancers and bone metastases, prostate and bone being the two most frequent sites of progression. These were obtained from therapy naïve and prostate cancers progressing on androgen deprivation and chemotherapy, therefore, constitutes a unique, clinically relevant, resource for preclinical studies to understand mechanisms of treatment response and resistant to standard as well as less common therapies (e.g., gefitinib). In this context, of particular relevance is the availability of PDXs derived from different areas of the same tumor that would enable the identification of divergent mechanism of progression to therapy in the same tumor due to molecular heterogeneity. Also, the MDA PCa PDXs cohort presented here is unique in that is enriched for SPOPL deletions.

Finally, our studies are in-line with those of others (31, 32), demonstrating that MDA PCa PDXs reflect the human donor tumor and are useful for drug testing (33). Furthermore, MDA PCa PDXs have provided unique insights into the biology of prostate cancer. For example, a study of MDA PCa 118b implicated the FGF axis in the pathogenesis of prostate cancer bone metastasis (5), which led to the initiation of a study that demonstrated clinical activity (6). Subsequent studies using our MDA PCa PDXs implicated the FGF axis in progression to therapy (18). A recent study confirmed our reports of the role of the FGF axis in the pathogenesis of advanced prostate cancer (10) and reported a new subgroup that utilizes the FGF axis in progression. MDA PCa 118b fell within this subgroup, further attesting the clinical relevance of our models. Furthermore, the MDA PCa PDXs contributed to the discovery of distinct classes of chromosomal rearrangements in prostate cancer cells (12), the identification of new therapeutic approaches for combination therapy targeting DNA damage response genes (34, 35), the elucidation of new biological roles of genes in prostate cancer (36), and the identification of new mechanisms underlying neuroendocrine differentiation (37). A list of studies that have used the MDA PCa PDXs developed in our program is outlined in Supplementary Table S1-Addendum.

Conclusion

The MDA PCa PDX series provides insight into the biological basis that accounts for prostate cancer heterogeneity and serves as an invaluable resource for discovery, therapy development, and optimization of personalized therapy targeting prostate cancer–specific drivers of progression. Prostate cancer lag behind other tumor types in the marker-informed classification for treatments. The transition to such classification can be achieved by linking detailed characterization of human cancers to “driver mechanisms” in model systems.

E. Efstathiou reports grants, personal fees, and non-financial support from Sanofi, Janssen, Astellas, Tolmar, and Bayer outside the submitted work. S.A. Tomlins reports grants, and personal fees from Astellas/Medivation [consultant for (with honoraria) and sponsored research agreement), personal fees from Janssen, and personal fees from Strata Oncology (co-founder of, equity holder in, prior consultant for, and current employee) outside the submitted work, and reports that the University of Michigan has been issued patents on ETS gene fusions in prostate cancer, on which S.A. Tomlins is a co-inventor and included in the royalty streams issued, licensed, and with royalties paid (The diagnostic field of use was licensed to Hologic/Gen-Probe Inc., which has sublicensed rights to Roche/Ventana Medical Systems). B.M. Broom reports grants from NIH during the conduct of the study. N.M. Navone reports grants from Prostate Cancer Foundation, The University of Texas MD Anderson Moon Shot Program, NCI Cancer Center Support Grant (P30CA16672), Cancer Center Prostate Cancer SPORE (NIH/NCI P50 CA140388-08), David H. Koch Center for Applied Research in Genitourinary Cancers at MD Anderson (Houston, TX), Janssen Research and Development, and NIH/NCI U01 CA224044 during the conduct of the study. No potential conflicts of interest were disclosed by the other authors.

N. Palanisamy: Resources, data curation, formal analysis, investigation, methodology. J. Yang: Resources, formal analysis, investigation, methodology. P.D.A. Shepherd: Resources, data curation, investigation, methodology. E.M. Li-Ning-Tapia: Data curation, formal analysis. E. Labanca: Data curation, investigation, writing-review and editing. G.C. Manyam: Software, formal analysis. M.K. Ravoori: Data curation, formal analysis. V. Kundra: Data curation, formal analysis, validation, investigation. J.C. Araujo: Resources, investigation, methodology. E. Efstathiou: Resources, formal analysis, investigation. L.L. Pisters: Resources, validation. X. Wan: Resources, data curation, investigation. X. Wang: Formal analysis, visualization. E.S. Vazquez: Visualization, writing-review and editing. A.M. Aparicio: Resources, investigation. S.L. Carskadon: Resources, data curation. S.A. Tomlins: Data curation, formal analysis. L.P. Kunju: Data curation, investigation. A.M. Chinnaiyan: Resources, formal analysis, validation. B.M. Broom: Data curation, formal analysis. C.J. Logothetis: Conceptualization, resources, funding acquisition. P. Troncoso: Conceptualization, resources, data curation. N.M. Navone: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

We thank Sarah E. Townsend for editing the article. We thank Agilent Technologies for the “Agilent University Relations Grant Award to N. Palanisamy.” This work was supported by the Prostate Cancer Foundation, The University of Texas MD Anderson Moon Shot Program, NCI Cancer Center Support Grant (P30CA16672), Cancer Center Prostate Cancer SPORE (NIH/NCI P50 CA140388), David H. Koch Center for Applied Research in Genitourinary Cancers at MD Anderson (Houston, TX), Janssen Research and Development, and NIH/NCI U01 CA224044.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Beltran
H
,
Tomlins
S
,
Aparicio
A
,
Arora
V
,
Rickman
D
,
Ayala
G
, et al
Aggressive variants of castration-resistant prostate cancer
.
Clin Cancer Res
2014
;
20
:
2846
50
.
2.
Logothetis
CJ
,
Gallick
GE
,
Maity
SN
,
Kim
J
,
Aparicio
A
,
Efstathiou
E
, et al
Molecular classification of prostate cancer progression: foundation for marker-driven treatment of prostate cancer
.
Cancer Discov
2013
;
3
:
849
61
.
3.
Vlachostergios
PJ
,
Puca
L
,
Beltran
H
. 
Emerging variants of castration-resistant prostate cancer
.
Curr Oncol Rep
2017
;
19
:
32
.
4.
Navone
NM
,
van Weerden
WM
,
Vessella
RL
,
Williams
ED
,
Wang
Y
,
Isaacs
JT
, et al
Movember GAP1 PDX project: an international collection of serially transplantable prostate cancer patient-derived xenograft (PDX) models
.
Prostate
2018
;
78
:
1262
82
.
5.
Li
ZG
,
Mathew
P
,
Yang
J
,
Starbuck
MW
,
Zurita
AJ
,
Liu
J
, et al
Androgen receptor-negative human prostate cancer cells induce osteogenesis in mice through FGF9-mediated mechanisms
.
J Clin Invest
2008
;
118
:
2697
710
.
6.
Wan
X
,
Corn
PG
,
Yang
J
,
Palanisamy
N
,
Starbuck
MW
,
Efstathiou
E
, et al
Prostate cancer cell-stromal cell crosstalk via FGFR1 mediates antitumor activity of dovitinib in bone metastases
.
Sci Transl Med
2014
;
6
:
252ra122
.
7.
Navone
NM
,
Olive
M
,
Ozen
M
,
Davis
R
,
Troncoso
P
,
Tu
SM
, et al
Establishment of two human prostate cancer cell lines derived from a single bone metastasis
.
Clin Cancer Res
1997
;
3
:
2493
500
.
8.
Navone
NM
,
Olive
M
,
Troncoso
P
. 
Isolation and culture of prostate cancer cell lines
.
Methods Mol Med
2004
;
88
:
121
32
.
9.
Epstein
JI
,
Amin
MB
,
Beltran
H
,
Lotan
TL
,
Mosquera
JM
,
Reuter
VE
, et al
Proposed morphologic classification of prostate cancer with neuroendocrine differentiation
.
Am J Surg Pathol
2014
;
38
:
756
67
.
10.
Bluemn
EG
,
Coleman
IM
,
Lucas
JM
,
Coleman
RT
,
Hernandez-Lopez
S
,
Tharakan
R
, et al
Androgen receptor pathway-independent prostate cancer is sustained through FGF signaling
.
Cancer Cell
2017
;
32
:
474
89
.
11.
Choucair
K
,
Ejdelman
J
,
Brimo
F
,
Aprikian
A
,
Chevalier
S
,
Lapointe
J
. 
PTEN genomic deletion predicts prostate cancer recurrence and is associated with low AR expression and transcriptional activity
.
BMC Cancer
2012
;
12
:
543
.
12.
Tomlins
SA
,
Laxman
B
,
Dhanasekaran
SM
,
Helgeson
BE
,
Cao
X
,
Morris
DS
, et al
Distinct classes of chromosomal rearrangements create oncogenic ETS gene fusions in prostate cancer
.
Nature
2007
;
448
:
595
9
.
13.
Grasso
CS
,
Wu
YM
,
Robinson
DR
,
Cao
X
,
Dhanasekaran
SM
,
Khan
AP
, et al
The mutational landscape of lethal castration-resistant prostate cancer
.
Nature
2012
;
487
:
239
43
.
14.
Michiel Sedelaar
JP
,
Dalrymple
SS
,
Isaacs
JT
. 
Of mice and men–warning: intact versus castrated adult male mice as xenograft hosts are equivalent to hypogonadal versus abiraterone treated aging human males, respectively
.
Prostate
2013
;
73
:
1316
25
.
15.
Carlin
BI
,
Andriole
GL
. 
The natural history, skeletal complications, and management of bone metastases in patients with prostate carcinoma
.
Cancer
2000
;
88
:
2989
94
.
16.
Loberg
RD
,
Logothetis
CJ
,
Keller
ET
,
Pienta
KJ
. 
Pathogenesis and treatment of prostate cancer bone metastases: targeting the lethal phenotype
.
J Clin Oncol
2005
;
23
:
8232
41
.
17.
Valta
MP
,
Tuomela
J
,
Bjartell
A
,
Valve
E
,
Vaananen
HK
,
Harkonen
P
. 
FGF-8 is involved in bone metastasis of prostate cancer
.
Int J Cancer
2008
;
123
:
22
31
.
18.
Varkaris
A
,
Corn
PG
,
Parikh
NU
,
Efstathiou
E
,
Song
JH
,
Lee
YC
, et al
Integrating murine and clinical trials with cabozantinib to understand roles of MET and VEGFR2 as targets for growth inhibition of prostate cancer
.
Clin Cancer Res
2016
;
22
:
107
21
.
19.
Xue
Z
,
Vis
DJ
,
Bruna
A
,
Sustic
T
,
van Wageningen
S
,
Batra
AS
, et al
MAP3K1 and MAP2K4 mutations are associated with sensitivity to MEK inhibitors in multiple cancer models
.
Cell Res
2018
;
28
:
719
29
.
20.
Nowak
DG
,
Katsenelson
KC
,
Watrud
KE
,
Chen
M
,
Mathew
G
,
D'Andrea
VD
, et al
The PHLPP2 phosphatase is a druggable driver of prostate cancer progression
.
J Cell Biol
2019
;
218
:
1943
57
.
21.
Zadra
G
,
Loda
M
. 
When fat goes down, prostate cancer is on the ropes
.
Mol Cell Oncol
2019
;
6
:
1595308
.
22.
Cancer Genome Atlas Research Network
. 
The molecular taxonomy of primary prostate cancer
.
Cell
2015
;
163
:
1011
25
.
23.
Abida
W
,
Cyrta
J
,
Heller
G
,
Prandi
D
,
Armenia
J
,
Coleman
I
, et al
Genomic correlates of clinical outcome in advanced prostate cancer
.
Proc Natl Acad Sci U S A
2019
;
116
:
11428
36
.
24.
Gundem
G
,
Van Loo
P
,
Kremeyer
B
,
Alexandrov
LB
,
Tubio
JM
,
Papaemmanuil
E
, et al
The evolutionary history of lethal metastatic prostate cancer
.
Nature
2015
;
520
:
353
7
.
25.
Beltran
H
,
Prandi
D
,
Mosquera
JM
,
Benelli
M
,
Puca
L
,
Cyrta
J
, et al
Divergent clonal evolution of castration-resistant neuroendocrine prostate cancer
.
Nat Med
2016
;
22
:
298
305
.
26.
Gao
D
,
Vela
I
,
Sboner
A
,
Iaquinta
PJ
,
Karthaus
WR
,
Gopalan
A
, et al
Organoid cultures derived from patients with advanced prostate cancer
.
Cell
2014
;
159
:
176
87
.
27.
Beshiri
ML
,
Tice
CM
,
Tran
C
,
Nguyen
HM
,
Sowalsky
AG
,
Agarwal
S
, et al
A PDX/organoid biobank of advanced prostate cancers captures genomic and phenotypic heterogeneity for disease modeling and therapeutic screening
.
Clin Cancer Res
2018
;
24
:
4332
45
.
28.
Risbridger
GP
,
Toivanen
R
,
Taylor
RA
. 
Preclinical models of prostate cancer: patient-derived xenografts, organoids, and other explant models
.
Cold Spring Harb Perspect Med
2018
;
8
:
a030536
.
29.
Hu
Y
,
Yang
L
,
Zhang
M
,
Huang
Z
,
Lin
J
,
Zhang
N
. 
Expression and clinical relevance of SPOPL in medulloblastoma
.
Oncol Lett
2017
;
14
:
3051
6
.
30.
Abida
W
,
Armenia
J
,
Gopalan
A
,
Brennan
R
,
Walsh
M
,
Barron
D
, et al
Prospective genomic profiling of prostate cancer across disease states reveals germline and somatic alterations that may affect clinical decision making
.
JCO Precis Oncol
2017
;
2017
.
doi:10.1200/PO.17.00029
.
31.
Tzelepi
V
,
Zhang
J
,
Lu
JF
,
Kleb
B
,
Wu
G
,
Wan
X
, et al
Modeling a lethal prostate cancer variant with small-cell carcinoma features
.
Clin Cancer Res
2012
;
18
:
666
77
.
32.
Aparicio
A
,
Tzelepi
V
,
Araujo
JC
,
Guo
CC
,
Liang
S
,
Troncoso
P
, et al
Neuroendocrine prostate cancer xenografts with large-cell and small-cell features derived from a single patient's tumor: morphological, immunohistochemical, and gene expression profiles
.
Prostate
2011
;
71
:
846
56
.
33.
Brenner
JC
,
Ateeq
B
,
Li
Y
,
Yocum
AK
,
Cao
Q
,
Asangani
IA
, et al
Mechanistic rationale for inhibition of poly(ADP-ribose) polymerase in ETS gene fusion-positive prostate cancer
.
Cancer Cell
2011
;
19
:
664
78
.
34.
Li
L
,
Chang
W
,
Yang
G
,
Ren
C
,
Park
S
,
Karantanos
T
, et al
Targeting poly(ADP-ribose) polymerase and the c-Myb-regulated DNA damage response pathway in castration-resistant prostate cancer
.
Sci Signal
2014
;
7
:
ra47
.
35.
Zhang
W
,
Liu
B
,
Wu
W
,
Li
L
,
Broom
BM
,
Basourakos
SP
, et al
Targeting the MYCN-PARP-DNA damage response pathway in neuroendocrine prostate cancer
.
Clin Cancer Res
2018
;
24
:
696
707
.
36.
Salameh
A
,
Lee
AK
,
Cardo-Vila
M
,
Nunes
DN
,
Efstathiou
E
,
Staquicini
FI
, et al
PRUNE2 is a human prostate cancer suppressor regulated by the intronic long noncoding RNA PCA3
.
Proc Natl Acad Sci U S A
2015
;
112
:
8403
8
.
37.
Zhang
Y
,
Zheng
D
,
Zhou
T
,
Song
H
,
Hulsurkar
M
,
Su
N
, et al
Androgen deprivation promotes neuroendocrine differentiation and angiogenesis through CREB-EZH2-TSP1 pathway in prostate cancers
.
Nat Commun
2018
;
9
:
4080
.
38.
Cerami
E
,
Gao
J
,
Dogrusoz
U
,
Gross
BE
,
Sumer
SO
,
Aksoy
BA
, et al
The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data
.
Cancer Discov
2012
;
2
:
401
4
.
39.
Gao
J
,
Aksoy
BA
,
Dogrusoz
U
,
Dresdner
G
,
Gross
B
,
Sumer
SO
, et al
Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal
.
Sci Signal
2013
;
6
:
pl1
.

Supplementary data