Introduction

Precision oncology heralds an era in which tumors are biopsied and profiled in the metastatic setting with the goal of identifying therapeutic targets. Although next generation sequencing (NGS) of “N of 1” cases have identified mechanisms of exceptional response to investigational therapies [1–4], such approaches have largely not been applied to exceptional non-responders. Likewise, NGS profiling of pre-/post-treatment samples in cases with marked histologic progression, which enables assessment of progression mechanisms, is challenging due to difficulties in obtaining and assessing routine diagnostic biopsy samples. Here, we describe NGS assessment of routine clinical samples from a patient diagnosed with metastatic PCa at a young age who rapidly progressed and died from disease approximately 1 year from diagnosis. Importantly, while his primary tumor exclusively contained conventional prostatic adenocarcinoma, a post-treatment liver metastasis biopsy exclusively contained prostatic small cell carcinoma/NePC. Hence, this case provided a unique opportunity to assess the utility of NGS-based profiling of serial routine biopsy specimens from an “exceptional non-responder” who showed rapid histologic progression during treatment.

Methods

Study oversight

The patient signed a consent form to participate in an IRB-approved research study to sequence tumor and germline DNA from men presenting with metastatic PCa before age 60 years.

Tumor sequencing and analysis

Post-mortem, we performed targeted next generation sequencing (NGS) on DNA and RNA co-isolated from macrodissected formalin-fixed paraffin-embedded (FFPE) tissue sections from the original diagnostic prostate biopsy specimen (PR-259) and the post-treatment liver biopsy specimen (PR-258). Multiplexed PCR-based NGS (Ampliseq) was performed using 40-ng DNA and the Ion Torrent Comprehensive Cancer Panel (CCP), which targets the coding region of 409 cancer-related genes with 15,992 amplicons (1,688,650 targeted bases) [5, 6]. Multiplexed PCR-based NGS (Ampliseq) was also performed using 20 ng RNA and the RNA component of the Oncomine Comprhensive Panel (OCP), which uses a total of 154 primer pairs to target known gene fusion isoforms, including those involving recurrent 5’ (TMPRSS2, SLC45A3, C15ORF21) and 3’ (ERG, ETV1, ETV4, ETV5, and BRAF) fusion partners in PCa [7]. Detailed description of sequencing, data analysis using validated pipelines, sequencing statistics, and all identified high-confidence somatic variants are given in the Supplement (Additional file 1 and Tables 1 and 2).

Table 1 Sequencing statistics for the diagnostic prostate biopsy sample containing conventional adenocarcinoma (PR-259) and subsequent liver metastasis with small cell/neuroendocrine carcinoma (PR-258)
Full size table
Table 2 High confidence, non-synonymous variants identified in the diagnostic prostate biopsy sample containing conventional adenocarcinoma (Dx [PR-259]) and subsequent liver metastasis with small cell/neuroendocrine carcinoma (NePC [PR-258])
Full size table

Case presentation

A 47-year-old male participated in PCa screening due to positive family history of PCa (father). Prostate-specific antigen (PSA) was initially elevated at 13.3 ng/mL, however, prostate biopsy was negative. His PSA rose one year later to 170 ng/mL, and repeat prostate biopsy revealed Gleason 4 + 5 = 9 prostate adenocarcinoma involving all 12 cores (Fig. 1, top and middle panels). Computed tomography (CT) of the abdomen and pelvis showed an enlarged left iliac lymph node but no other metastases. Bone-scan showed metastases involving the lumbar vertebrae. He enrolled in a clinical trial and was treated with an oral anti-androgen along with leuprolide. On this regimen, PSA decreased to a nadir of <4.0 ng/mL over 6 months, and CT scan showed reduction in the size of the enlarged iliac node and normal liver parenchyma.

Fig. 1
figure 1

Histology of diagnostic (Dx) prostate biopsy (PR-259) with Gleason score 4 + 5 = 9 adenocarcinoma and subsequent post-treatment (Tx) liver biopsy (PR-258) containing small cell/neuroendocrine prostate carcinoma (NePC). Hematoxylin and eosin-stained diagnostic biopsy cores (top and middle panels) and liver biopsy (bottom panel) are shown. Original magnification 10× (insets indicated by green boxes 40×)

Full size image

Two months later, the patient presented with jaundice and was found to have elevated transaminases. MRI showed a pseudocirrhotic appearance of the liver (Fig. 2). Liver biopsy revealed small cell/NePC (Fig. 1, bottom panel). Despite poor performance status and after discussion of risks and benefits, he was initiated on dose-reduced oral etoposide along with carboplatin and continued leuprolide. Although the patient demonstrated an initial clinical response, he elected to end treatment. He died under hospice care 3 months following NePC diagnosis.

Fig. 2
figure 2

MRI Liver at the time of clinical progression. Axial T2-weighted MRI shows interval development of innumerable solid lesions in the liver, replacing a majority of the parenchyma in both hepatic lobes with development of a pseudocirrhotic appearance of the liver with a nodular surface contour

Full size image

Results

NGS of DNA isolated from the routine FFPE diagnostic prostate biopsy specimen (PR-259) and the post-treatment liver biopsy specimen (PR-258) identified a total of two prioritized high-confidence somatic variants. As shown in Fig. 3, a SMAD4 c1605delC p.L535fs frameshifting variant was present in both PR-259 (36 % variant allele frequency) and PR-258 (67 % variant allele frequency). In contrast, a TP53 c.C844T p.R282W non-synonymous variant was exclusively called in the NePC specimen (PR-258; 69 % variant allele frequency). This variant was markedly enriched in PR-258, and was only present at a variant allele frequency of 1.6 % (12/755 reads) in the diagnostic pre-treatment specimen (PR-259). These results are consistent with clonal origin and marked enrichment of the TP53 R282W variant exclusively in the post-treatment NePC specimen. Exome sequencing of germ line DNA isolated from white blood cells confirmed the TP53 and SMAD4 variants as somatic (see Table 2).

Fig. 3
figure 3

Next generation sequencing (NGS) genomic profiles support transdifferentiaton from prostatic adenocarcinoma (PR-259) to small cell/neuroendocrine prostate carcinoma (NePC, PR-258). Copy number plots and prioritized somatic mutations are shown. Points represent the log2 copy number ratio for all targeted genes (shown in genome order). Clonal gains and losses are shown in red and blue, respectively. Prioritized high-level copy number alterations (CNAs) alterations and somatic mutations (with variant allele frequency [%] and coverage depth [x]) are indicated. Clonal prioritized SMAD4 mutation and SCC enriched TP53 mutation and MYCL, PTEN, RB1, and MAP2K4 copy number alterations are indicated

Full size image

We next compared NGS-derived copy number profiles between PR-259 and PR-258 using our well-validated approach [5–7]. Copy number profiling revealed broad one copy loss of 10q (containing PTEN), 18q, and a complex alteration on chromosome 19 in both specimens, whereas the post-treatment liver biopsy (PR-258) exclusively demonstrated focal, high-level MYCL amplification, and focal homozygous PTEN, RB1, and MAP2K4 deletions. Lastly, no gene fusions were identified in either PR-259 or PR-258 from targeted multiplexed PCR-based RNAseq on co-isolated RNA (see Additional file 1). Taken together with the somatic variant analysis, copy number profiling supported the clonal relationship between PR-259 and PR-285, and identified highly enriched, focal, high-level copy number alterations in the post-therapy NePC specimen.

Discussion

Small cell carcinoma/(NePC) is a rare PCa variant with an aggressive phenotype. Although de novo NePC constitutes <1 % of all PCa, autopsy series of castration-resistant prostate cancer (CRPCa) suggest the presence of NePC in 10–25 % of cases [8]. Despite high initial overall response rates (75–85 %) to platinum combinations, relapse to a chemo-refractory state is nearly universal with a median survival of less than 18 months [9].

Although initially posited as due to clonal selection of malignant neuroendocrine cells [10, 11], recent genetic evidence supports a model of NePC development due to transformation of prostate adenocarcinoma cells to a neuroendocrine phenotype, termed transdifferentiation (see review [12]). Consistent with the concept of a common clonal origin, recurrent prostate adenocarcinoma-specific alterations, such as recurrent ETS gene rearrangements, show concordant status in PCa admixed with NePC, and ETS rearrangement frequency is similar in conventional PCa and NePC [13–15]. Additionally, identical mutations in the DNA-binding domains of TP53 have been observed in paired prostate adenocarcinoma and NePC [16]. The molecular mechanism of NePC development via transdifferentiation is also supported by a recent report showing gene amplification of AURKA and MYCN present in 65 % of adenocarcinomas that develop into NePC following ADT whereas only 5 % of unselected adenocarcinomas showing similar amplifications [17]. Lastly, RNAseq profiling in matched NePC and prostate adenocarcinomas showed downregulation of the transcriptional complex REST, which is integral to the repression of neuronal differentiation [18].

Despite the growing evidence in support of transdifferentiation, limited data has been published comparing comprehensive molecular profiling of a primary prostate adenocarcinoma and the subsequent metastatic NePC. Our current study reflects, to our knowledge, the first comprehensive profiling of paired diagnostic FFPE biopsy and subsequent NePC specimens. The presence of a somatic, deleterious SMAD4 variant in both the diagnostic and NePC specimens is consistent with clonal origin and the transdifferentiation model. Among 816 sequenced PCas in the cBioPortal database [19], six (0.7 %) harbor somatic SMAD4 mutations, including one sample with two SMAD4 mutations (Table 3). Importantly, six of the seven reported mutations impact the MH2 domain (including known inactivating and truncating mutations [20,21]) and are recurrent in the COSMIC database [22]. Of note, detailed mechanistic studies demonstrate that SMAD4 loss leads to an aggressive PCa phenotype in mouse models [23], providing a likely candidate driver of the aggressive phenotype in this exceptional “non-responder.”

Table 3 SMAD4 mutation and MYCL1 amplification frequency in prostate cancer NGS and copy number profiling studies available in cBioPortal
Full size table

The TP53 p.R282W mutation enrichment and homozygous RB1 loss in the NePC sample herein supports single gene studies and our recent targeted NGS profiling of eight NePC that show frequent inactivation of these genes in NePC [7, 24, 25]. Likewise, in a recent study, we used qRT-PCR and a combination of exome/targeted NGS to profile distinct conventional PCa and NePC components from an FFPE transurethral resection specimen, which demonstrated enrichment of a TP53 p.N151fs mutation exclusively in the NePC component [26]. Although both oncogenic and metastasis suppressive roles for MAP2K4 have been reported in PCa [27–29], its role in NePC has not been described and will require additional investigation.

As described above, recurrent MYCN amplifications have been well-described in NePC [17]. Although a recent report identified recurrent MYCL amplifications in ~25 % of untreated Gleason score 7 PCa (>2 copies in 8–20 % of malignant glands) [30], clonal, high-level MYCL amplifications have not been observed in 1166 prior SNP-, aCGH-, or NGS-based copy number profiled untreated PCa or CRPC in cBioPortal (Table 3). However, in our previous NGS-based profiling of 116 aggressive PCas, we identified a single NePC (of 8 profiled) that harbored a high-level MCYL amplification (Additional file 2: Figure S1) [7]. Likewise, copy number profiling of mouse NePC resulting from prostate-specific p53 and RB inactivation identified recurrent MYCL gains [31]. MYCL amplifications and gene fusions have also been identified and shown to drive proliferation in small cell lung carcinomas [32–34]. Taken together with our previous NGS profiling study, herein we identify recurrent MYCL1 amplifications in NePC, which will need to be confirmed in additional NePC cohorts.

Alternate mechanisms for the development and maintenance of NE transdifferentiation have been described. The process of “epithelial plasticity” provides evidence for the diverse phenotype of NE-like tumor cells, such as the variable expression of epithelial and NE markers following androgen deprivation [12, 35–37]. This plasticity, which can occur via epithelial-to-mesenchymal transition (EMT) or mesenchymal-to-epithelial transition (MET), is regulated by a complex system of transcriptional networks and signaling pathways. The TMPRSS2-ERG fusion gene and certain microRNAs (i.e., miR-200 family) appear to promote the EMT phenotype leading, in part, to the castrate-resistant state [38, 39].

Importantly, neuroendocrine cancers involving other organ sites appear to have distinct molecular aberrations and highlight the need for individualized therapies [40]. For example, telomerase reverse transcriptase (TERT) promoter mutations are observed in many human epithelial cancers as well as the vast majority of urothelial neuroendocrine carcinomas, however, they are rarely found in NE-prostate or -lung cancers [41, 42]. Given the molecular heterogeneity of neuroendocrine carcinomas, targeted approaches guided by appropriate biomarker identification, rather than or in addition to cytotoxic therapies, are paramount to improve outcomes [43].

A variety of novel therapeutics targeting receptor tyrosine kinases, mammalian target of rapamycin (mTOR), angiogenesis, cell cycle, epigenetics, and immunotherapy have been tested, largely in small cell lung cancer, with limited success [44, 45]. Specific to the mutational landscape of NEPC, a number of targeted therapies have been investigated in in vitro and in murine models with varied success (see review [46]). Targeting tumor suppressor loss (TP53, RB1, and PTEN) is particularly relevant to NEPC given the high frequency of these alterations. For example, SAR405838, a novel small molecule inhibitor of the oncoprotein murine double minute 2 (MDM2)-TP53 protein-protein interaction, showed significant activity in wild-type TP53 murine models, including LNCaP prostate cancer lines [47]. Multiple small molecules that can activate TP53 are in early phase clinical trials, however, none at this time are recruiting patients with NEPC [46,48]. A phase II study of MLN8237, a small molecule inhibitor of Aurora Kinase A, is currently the only molecularly targeted trial enrolling men with CRPC with neuroendocrine features (NCT01799278).

A limitation of the current report is that it is based on NGS from one patient. Future case studies should consider application of immunohistochemical and morphoproteomic analyses, which might elucidate alternative mechanisms of resistance. The application of these tools has previously revealed means of response and resistance in two cases of refractory Ewing sarcoma that responded to combination therapy with insulin-like growth factor 1 receptor and mTOR inhibition [49].

“N of 1” cases provide unique hypothesis-generating opportunities with the potential to provide new information about pathogenic mechanisms and/or therapeutic response [2, 4]. We suggest that profiling of “exceptional non-responders” and temporally/histologically distinct tumor components [26, 50], as shown herein, may be as informative as “exceptional responder” studies and can exploit the wealth of archived diagnostic tissue specimens. Such studies may be particularly important for identifying the prognostic and predictive associations of rare alterations, such as SMAD4 mutations in prostate cancer, as well as identifying adaptive alterations associated with treatment resistance/progression such as MCYL amplifications.

Conclusions

Through comprehensive profiling of archived diagnostic and liver biopsy specimens from a single patient with an aggressive clinical course, we identify molecular alterations associated with rapid progression from prostatic adenocarcinoma to NePC, and more broadly identify MYCL as a recurrently amplified gene specifically in NePC.