Neuroendocrine differentiation in adenocarcinoma should be detected in a timely manner. Both the elevation of serum NSE levels and the presence of visceral metastases indicate the development of neuroendocrine differentiation in prostate cancer. Despite their high sensitivity, the CgA and NSE serum neuroendocrine markers lack specificity for SCC/NEPC (2, 4, 5). Pathological diagnosis is the most reliable diagnosis, and the acquisition of reliable biopsy specimens usually depends on good compliance of the patient and the puncture technique of the doctor (6). We obtained four pre- and posttreatment biopsy specimens, which revealed the histological progression of tumors ( Table 1 ) ( Supplementary Figure 2 ). PSMA, P504s, and NKX3.1 were expressed in nearly all prostatic adenocarcinomas. Only these markers were expressed at initial diagnosis, indicating the adenocarcinoma origin of the tumor. The CgA, Syn, and CD56 markers are expressed in nearly all prostate cancers with neuroendocrine differentiation (7). Following its progression to mCRPC, IHC assessment of the thoracic vertebral biopsy specimen revealed positive signals for PSMA, P504s, NKX3.1, Syn, CD56, and CgA, which indicated adenocarcinoma admixed with NEPC after intense therapeutic pressures of ADT therapy. Likewise, IHC assessment of the seminal vesicle biopsy specimen showed positive signals for CAM5.2, Syn, CD56, and CgA, but negative for NKX3.1 and P501s, indicating the predominance of SCC/NEPC in the patient during this period. Under the pressure of treatment, NEPC transformed into a poorly differentiated small cell phenotype, losing the expression of adenocarcinoma origin-specific markers, such as PSMA and NKX3.1.

As the special genetic pattern of NEPC reflects its cellular origin, the genetic alterations in two samples collected at different stages were investigated using whole-exome sequencing. The first was a mixed adenocarcinoma and neuroendocrine cancer sample (M-sample) at CRPC, whereas the second was a pure SCC/NEPC (S-sample) at the last stage. We described the mutational landscape and applied phylogenetic reconstruction to analyze the dynamic clonal progression. We estimated and clustered the cancer cell fraction (CCF) to track and visualize the clonal evolution of each tumor using PyClone (8).

We observed that both samples shared the most recent common ancestor (MRCA) encompassing 23 mutations and an exon1-exon13 truncation of the neurofibromin 2 (NF2) gene, indicating their common origin ( Figure 3 ). Notably, a double deletion of TP53 and RB1 was identified at the initiation of neuroendocrine stages, while a truncation of NF2 resulted in sensitivity to platinum chemotherapy (9). A specific subclone harboring 37 mutated genes and an AR amplification was observed in the M-sample. We integrated the mutation frequency of all mutated genes in different types of prostate cancer in this patient, according to the existing data using “cBioPortal” (http://www.cbioportal.org) as previously described (10) ( Figure 3A ). We observed that the CDC27 oncogene was frequently mutated in NEPC samples (7.4%, 4/54 public samples). CDC27 has been recognized as either a tumor suppressor gene or an oncogene in different neoplasms. Furthermore, CDC27 increases the stemness of cancer stem cells in colorectal cancer (11). A private subclone harboring three copy number alterations and five mutations was detected in the S-sample. Among them, the RUNX1 gene, a transcription factor playing key roles in the regulation of stem cell fate, might be driving the neuroendocrine progression of tumor (12). Although Beltran et al. observed that the RUNX1 copy number deletion was meaningful in NEPC, a RUNX1 amplification could also occur, further supporting the key role of RUNX1 in neuroendocrine-differentiated tumors (13). Consequently, the two samples obtained individual subclones after MRCA in the clonal evolutionary tree ( Figures 3C, D ). Some driving events in these subclones might have played a key role in tumor progression or even in promoting neuroendocrine differentiation.

Precision oncology is based on tumor biopsies and sequencing to identify therapeutic targets. Here, two posttreatment biopsy specimens with histologic progression were analyzed using whole-exome sequencing. An AR amplification was detected in the first mixed tumor sample but not in the second pure SCC/NEPC sample, indicating the downregulation of AR in NEPC compared with that in adenocarcinoma (5, 13). In addition, the inactivation of RB1 and TP53 tumor suppressor genes is frequently associated with small cell cancer (14, 15), and is a key mechanism of resistance to antiandrogen therapy and lineage plasticity (16, 17). Identical mutations in the DNA-binding domains of TP53, RB1, and MYC have also been observed in both NGS profiles, indicating the possible derivation of NEPC from adenocarcinoma (18–20). We attempted to identify therapeutic targets, such as DNA damage repair (DDR) pathway genes (BRCA1, BRCA2, ATM, and CDK12), which could be paired with poly ADP-ribose polymerase (PARP) inhibitors (6, 21). However, we did not detect any alterations in DDR pathway genes in the NGS profiling. Similarly, other studies have also shown the rare occurrence of alterations in DDR pathway genes in SCC/NEPC (2). We observed alterations in PIK3CA and NF2, which could be associated with the phosphatidylinositol-3-kinase/mammalian target of rapamycin (PI3K/mTOR) pathway. However, the curative effect of mTOR inhibitors in prostate cancer remains uncertain (22, 23). Considering the uncertain curative effect and the poor physical condition of the patient, we elected the regimen recommended by the guidelines (6, 24).

Prostate cancer with neuroendocrine differentiation shows remarkable clinical heterogeneity, as revealed by the constantly changing clinical symptoms, such as bone pain, liver metastases, and meningeal metastases. We deduced that adenocarcinoma and SCC/NEPC alternately dominated the shift in the process of disease. We elected ADT combined with docetaxel as the initial treatment, but changed to etoposide combined with cisplatin and pembrolizumab after the detection of neuroendocrine differentiation. In this case, the key advantage of the approach used is the timely correction of treatment strategy according to the pathological type and the construction of tumor subclone structures using sequential sampling. In addition, we advanced chemotherapy at the initial diagnosis based on the NCCN guidelines (6), and timely chose the regimen with etoposide and platinum (EP) + PD1 when the tumor was revealed to be neuroendocrine differentiation and mixed adenocarcinoma. Despite accomplishing a temporary response, regrettably, we could not find a regimen that could ideally control both adenocarcinoma and SCC/NEPC to improve the prognosis. We have also supplemented our report with the research on the origin of SCC/NEPC from the perspective of clonal evolution. Previous studies have reported the transdifferentiation phenomenon of NEPC and its molecular changes, that is, prostate adenocarcinoma cells could transform into a neuroendocrine phenotype (25, 26). We described here the clonal evolution from mixed adenocarcinoma and neuroendocrine cancer to pure SCC/NEPC. Meanwhile, some molecular alterations have also been provided, which could hopefully be used to monitor the progress of SCC/NEPC in the future.

Notably, this case study has some limitations. While longitudinal sampling combined with bulk sequencing could distinguish tumor subclones and monitor tumor progression, for the special tumor type observed in this case—with higher heterogeneity and more complex cellular components—single-cell sequencing may be a better and more accurate solution (27). Alternatively, in a previous report, transcription factors have been utilized for the molecular subtyping of small cell lung cancer; however, this has not been applied in clinical practice (28). Accordingly, further research is required to explore the value of transcription factor typing to guide clinical treatment.

Despite a periodic clinical response, the patient died from respiratory failure, severe anemia, infection, and brain edema in December 2020. The last MRI showed metastases of the meninges, clivus, skull, and sphenoid bone ( Figure 2D ).

If the IHC assessment of the initial puncture biopsy tissue detects potential neuroendocrine differentiation, doctors should offer patients more frequent imaging examination and analyses of serum markers.

Tumor biopsy and IHC assessment must be repeated at different stages of the disease to judge the progress of neuroendocrine differentiation and adjust the treatment regimen.

Tumor sequencing remains necessary even though changes might not be necessarily detected, as it could help clinicians better understand the genomic changes during neuroendocrine differentiation.