Clinicopathological and Genetic Analyses of Small Cell Neuroendocrine Carcinoma of the Prostate: Histological Features for Accurate Diagnosis and Toward Future Novel Therapies Short Running Title: Characteristics of Prostatic Small Cell Neuroendocrine Carcinoma

Arika Ida Kanagawa Cancer Center Yoichiro Okubo (  yoichiro0207@hotmail.com ) Kanagawa Cancer Center Rika Kasajima Kanagawa Cancer Center Research Institute Kota Washimi Kanagawa Cancer Center Shinya Sato Kanagawa Cancer Center Emi Yoshioka Kanagawa Cancer Center Kimito Osaka Kanagawa Cancer Center Takahisa Suzuki Kanagawa Cancer Center Yayoi Yamamoto Kanagawa Cancer Center Tomoyuki Yokose Kanagawa Cancer Center Takeshi Kishida Kanagawa Cancer Center Yohei Miyagi Kanagawa Cancer Center


Introduction
Adenocarcinoma constitutes the majority of histological types of prostate cancer (1).Primary neuroendocrine (NE) tumors of the prostate have also been identi ed and can be divided into four types based on the World Health Organization classi cation: adenocarcinoma with NE differentiation, welldifferentiated NE tumor, small cell NE carcinoma (SCNC), and large cell NE carcinoma (2).Of these, adenocarcinoma with NE differentiation and SCNC occur at a certain frequency (2).Approximately 0.5% to 2% of cases of primary SCNC of the prostate have been reported among prostate malignancies (3), in which morphological features are essential for the diagnosis of this tumor (4).However, diagnosis based on morphological features alone is sometimes challenging.Moreover, positive immunoreactivity for NE markers does not always provide a de nitive diagnosis of NE carcinoma of the prostate due to the concept of prostate adenocarcinoma with NE differentiation (5).Given that therapeutic strategies differ depending on histological type, diagnoses are critical (4,6).
An aggressive variant of SCNC occurs in castration-resistant prostate cancer (CRPC) after hormonal therapy and taxane chemotherapy (7)(8)(9), referred to as therapy-related NE prostate cancer (t-NEPC) (8).Genomic and transcriptional alterations are known to occur in CRPC cases, resulting in SCNC (10).De novo type SCNC may occur, and there are several reports on the frequency of de novo SCNC and t-NEPC cases.A high proportion of t-NEPC cases among patients with SCNC of the prostate has been reported (11), whereas other studies have indicated that SCNC cases constitute approximately equal proportions of t-NEPC and de novo types (8).In this regard, various mechanisms, including the presence or absence of a history of therapy, may be associated with SCNC of the prostate (12).
Recently, the concept of prostate cancer with diffuse NE differentiation has been reported, raising concerns regarding its differentiation from SCNC (5).Here, we rst aimed to clarify the characteristics of prostate NE tumors diagnosed by biopsy and determine factors contributing to an accurate diagnosis.Moreover, the presence of ALK mutation (ALK F1174C) in SCNC of the prostate was reported, highlighting its potential use as a novel therapeutic target (13).Therefore, we further explored the possibility of future novel therapeutic strategies.

Study design
This study included cases biopsied after January 2017 due to the shift from systemic prostate needle biopsies to the addition of extra biopsies based on prebiopsy magnetic resonance imaging ndings at our institute.As per the previous methodology of our institution, Primary pathological ndings were rst recorded by YO and reviewed by a second pathologist (YM) according to the 2014 International Society of Urological Pathology system (14-16).
In the event of disagreement on diagnostic ndings, another trained pathologist was recruited for discussion until a nal consensus was reached.From January 2017, clinical information, including age, body mass index, and pre-biopsy serum prostate-speci c antigen (PSA) levels, and pathological ndings, such as the presence of tumor lesions, histological type (including suspected NE tumors), number of positive scores, highest Gleason grade (only for adenocarcinoma cases), lesion length, and immunohistochemical con rmation of NE marker expression, were recorded by YO for database generation.Cases were extracted from the database if NE tumors were suspected based on morphological ndings and immunohistochemical con rmation of NE markers, such as synaptophysin and chromogranin A. Subsequently, cases were re-reviewed using actual specimens.Serum neuron-speci c enolase (NSE) and pro-gastrinreleasing peptide (ProGRP) levels were con rmed in the most recent values at the time of biopsy.Metastasis to other organs and prognosis were con rmed using electronic medical records.The diagnostic criteria for small cell neuroendocrine carcinoma (SCNC) of the prostate were based on the World Health Organization classi cation (2), with emphasis on morphology and immunoreactivity of NE markers.Given that prostate cancer can exhibit partial NE differentiation, adenocarcinoma with NE in this study was de ned as cases with diffuse immunoreactivity for NE markers (prostatic carcinoma with diffuse NE differentiation), based on a previous report (5).The de nition of "diffuse" is controversial, but in this study, we included cases in which at least 50% were positive.In each of the cases in this study, only a prostate biopsy was conducted, and no additional surgically resected specimens were obtained.

Statistical analyses
Statistical analyses were performed using International Business Machines (IBM) Statistical Package for the Social Sciences Statistics version 25 (IBM Corp., Armonk, NY, USA).Serum NSE, ProGRP, and PSA levels and the H-score and Ki-67 labeling index were subjected to statistical analysis.Student's t-test was used to compare two groups.One-way analysis of variance (one-way ANOVA) and the Bonferroni post hoc test were used to compare three groups.A p-value <0.05 was considered statistically signi cant.

Clinicopathological results
Between January 2017 and May 2021, 1014 prostate biopsies were conducted at our hospital, among which 654 (64.5%) prostate cancer cases were con rmed.After primary evaluation of hematoxylin and eosin-stained specimens, immunohistochemistry of synaptophysin and chromogranin A was conducted in 32 cases (32/654, 4.9%) because there were cases suspected of neuroendocrine tumors but could not be determined by morphological images alone.Components positive for both synaptophysin and chromogranin A were identi ed in 11 cases.Seven cases (7/654, 1.1%) contained SCNC components, two of which were pure SCNC and ve were mixed SCNC-adenocarcinoma.Four of the 11 cases (4/654, 0.6%) were prostatic carcinoma with diffuse NE differentiation (Figure 1).Although t-NEPC accounts for a subset of SCNC of the prostate (10,11), cases in this study were treatment-naive, with the exception of a patient in one case who had been treated with hormonal therapy (Case 6) for 18 months before the biopsy.
All prostatic carcinomas with diffuse NE differentiation were Gleason Grade 5 and were di cult to differentiate from SCNC morphologically.In contrast to SCNC, these cases exhibited the following features: nuclear structures with densely stained nuclei and inconspicuous nucleoli observed at high magni cation, lower nuclear-to-cytoplasmic ratio, and foamy cytoplasm, consistent with previous reports (2,5).Age, histological type, serum NSE, and ProGRP and PSA levels; organs with metastasis; and outcomes are summarized in Table 1.The mean (± standard deviation) serum NSE, ProGRP, and PSA values in SCNC were 51.9±78.7 ng/mL, 2381.3±5086.2pg/mL, and 31.3±56.7 ng/mL, respectively, whereas these values in prostate cancer with diffuse NE differentiation were 10.8±5.1 ng/mL, 50.5±35.3pg/mL, and 1603.7±2396.5 ng/mL, respectively (PSA values in the case with hormonal therapy and unmeasured values were excluded).There were no signi cant differences in serum NSE, Pro GRP, and PSA levels between SCNC and prostate cancer with diffuse NE differentiation (Student's t-test, p=0.257, p=0.313, and p=0.281, respectively).

Immunohistochemical results
One case with a history of hormonal therapy before the biopsy was excluded.Immunohistochemical results of the remaining 10 cases are summarized in Tables 2 and 3 and Figures 2 and 3. Compared with adenocarcinoma components, SCNC exhibited a signi cantly higher Ki-67 labeling index in mixed SCNCadenocarcinoma (one-way ANOVA, p<0.01) and prostatic carcinoma with diffuse NE differentiation (one-way ANOVA, p<0.01).Immunohistochemical PSA positivity in SCNC demonstrated a trend of lower H-scores, although this did not reach statistical signi cance (p=0.09 and p=0.1 for adenocarcinoma components in mixed SCNC-adenocarcinoma and prostatic carcinoma with diffuse NE differentiation, respectively).Cases with extremely high positivity (overexpression) or complete negativity (null-type) for p53 were identi ed in both SCNC and prostatic carcinoma with diffuse NE differentiation (Figures 4 and  5).No signi cant differences were noted in any immunohistochemical nding pertaining to adenocarcinoma components between mixed SCNCadenocarcinoma and prostatic carcinoma with diffuse NE differentiation.

Hotspot driver mutations
Hotspot driver mutations were evaluated in available cases, including one of the two pure SCNCs, three of the four prostatic carcinomas with diffuse NE differentiation, and four of the ve mixed SCNC-adenocarcinoma cases.Both SCNC and adenocarcinoma components were examined in mixed SCNCadenocarcinoma case.The one pure SCNC was subjected to the FoundationOne CDx assay (Foundation Medicine Inc., Cambridge, MA).In summary, ALK mutation was not detected, whereas TP53 mutation was the most frequently identi ed.In one case of SCNC, a mutation in PIK3CA was also detected.Results are summarized in Table 4.

Discussion
SCNC of the prostate is a rare but highly aggressive cancer (8).In this study, seven cases of SCNC (including two pure cases and ve mixed SCNCadenocarcinoma cases) were identi ed with a frequency of approximately 1%.Further, four cases of prostatic carcinoma with diffuse NE differentiation were observed.Generally, prostatic carcinoma with diffuse NE differentiation does not differ from the conventional type of prostate adenocarcinoma with similarly high GG in prognosis (17), although this remains controversial (2).Further, the difference in prognosis between SCNC and prostatic carcinoma with diffuse NE differentiation remains controversial (2,4,6).In this study, two of the four patients with prostatic carcinoma with diffuse NE differentiation died (although one death was due to a complicated gastric cancer), and three of the seven patients with SCNC died.However, cases of prostate cancer with diffuse NE differentiation in this study were limited to those with suspected SCNC morphologically and con rmed immunoreactivity for NE markers, all of which were aggressive cases corresponding to GG5.Thus, assessing the true difference in prognosis between SCNC and prostatic carcinoma with diffuse NE differentiation remains challenging.Lymph nodes and bones are well-established metastatic sites in prostate cancer (14,(25)(26)(27).Apart from these organs, one case of liver metastasis was identi ed among the four cases of adenocarcinoma (Case 1), whereas distant metastases were identi ed in three of the seven cases of SCNC (Cases 5-7, brain, liver, lung, and skin).Although the number of cases should be considered, SCNC exhibited a tendency of distant metastases to sites other than lymph nodes or bones, consistent with previous reports (8, 28, 29).Clinical data revealed a trend of low serum PSA levels and extremely high serum ProGRP levels in SCNC cases.However, one case of SCNC with high PSA levels was identi ed, and four of the seven SCNC cases did not exhibit high ProGRP and NSE levels.Furthermore, the mean serum NSE, ProGRP, and PSA values were different between SCNC and prostatic carcinoma with diffuse NE differentiation, although the differences were not statistically signi cant.Although the small number of cases should be considered, the heterogeneity of serum NSE and ProGRP levels in SCNC cases may also be relevant.Although extremely high serum NSE and ProGRP values suggest the presence of SCNC, it may be di cult to completely distinguish between SCNC and diffusely NE-differentiated prostate cancer based on serum PSA and NE marker values alone.ALK mutation was not detected in this study.Including the FoundationOne CDx assay results of one case, only ve SCNC cases were genetically analyzable (Cases 4-7 and Case 11); therefore, further investigations are warranted to elucidate the pro le of targetable mutations, such as ALK F1174C in SCNC of the prostate.Comparison of gene mutations between the SCNC and adenocarcinoma components from the same mixed SCNC-adenocarcinoma was possible for two cases.No mutations were identi ed in the adenocarcinoma component, and ERBB2 mutation in the SCNC component was noted in one case.
The other case revealed truncating loss-of-function mutations in the tumor suppressor genes PTEN and TP53 in the adenocarcinoma component.Notably, the SCNC component of this case shared the same TP53 mutation at the nucleotide alteration level (c.820_821delGT; p.Val274fs), implying the existence of a common precursor lesion committed to tumorigenesis (30,31).The SCNC component did not contain PTEN mutations.PTEN mutations appeared at or after the splitting of the two histologic types.Two cases of TP53 truncating mutations exhibited a null-type pattern on immunostaining for p53.Consistent with reports on other cancers (32), nonsynonymous amino acid-replacing TP53 mutations were identi ed in two cases of prostate cancer with diffuse NE differentiation and one case of pure SCNC with high expression of p53.Moreover, a PIK3CA mutation (p.His1047Arg) was detected in one case of SCNC (Case 7).PIK3CA mutations have been identi ed in various types of cancers (33) and have also been reported to be a genetic driver of prostate cancer (34).However, to the best of our knowledge, this mutation has not been reported in prostate SCNC.A combination of PIK3CA inhibitors, alpelisib and fulvestrant, has been proposed as a novel therapy for breast cancer (35)(36)(37).Our genetic ndings suggest that the presence of this PIK3CA mutation in SCNC may offer a novel therapeutic target.
Histopathologically, the basic morphological features of SCNC are relatively small tumor cells with inconspicuous nucleoli, densely stained nuclei with increased chromatin, a high nuclear-to-cytoplasmic ratio, and stretchability often resulting in nuclear lines (38).The signi cance of these morphological features has increased due to the establishment of the concept of prostatic carcinoma with diffuse NE differentiation (2).However, distinguishing SCNC from prostatic carcinoma with diffuse NE differentiation based on morphology and immunohistochemistry for NE markers alone can be challenging.The H-score for chromogranin A tended to be higher for SCNC than for prostatic carcinoma with diffuse NE differentiation, although this difference was not statistically signi cant.Further, no signi cant difference in immunoreactivity for synaptophysin was noted between the two tumor types, highlighting the di culty in differentiating SCNC from prostatic carcinoma with diffuse NE differentiation based solely on immunoreactivity to NE markers.Although high p53 immunoreactivity has been reported in SCNC of other organs (39)(40)(41), both SCNC and adenocarcinoma with diffuse NE differentiation of the prostate frequently exhibited a mutated TP53 pattern, that is, overexpression or null-type, which precluded these tumor types from being distinguished.In particular, both prostate cancer with diffuse NE differentiation and SCNC may exhibit overexpression or null-type immunoreactivity for p53.Pathologists, urologists, and oncologists should therefore be aware that p53 immunoreactivity is insu cient to distinguish between these two histological types.In contrast, the Ki-67 labeling index, a well-known indicator of malignant potential (42)(43)(44)(45), was signi cantly different between the two histologic types.The average labeling index for SCNC was 70%, which was approximately three times signi cantly higher than that for prostatic carcinoma with diffuse NE differentiation.With regard to prostatic epithelial cell markers, a degree of AR and prostein immunoreactivity was observed in SCNC, whereas less PSA immunoreactivity was noted, although this did not reach statistical signi cance.In SCNC of the prostate, AR and prostein are not useful in differentiating prostate cancer, as they maintain immunoreactivity.However, PSA, in addition to the Ki-67 labeling index, might be an auxiliary marker in the diagnosis of SCNC, because a considerable, although not statistically signi cant, reduction in PSA immunoreactivity was observed in SCNC.
Further studies are warranted, our ndings suggest that a high Ki-67 labeling index and low PSA immunoreactivity may contribute to an accurate diagnosis of SCNC.Although other novel markers for SCNCs have been reported (46,47), an advantage of our approach is that it can be applied using antibodies commonly employed in the diagnostic pathology department.In addition, immunohistochemistry for prostate markers other than PSA, such as AR and prostein (48-50), may be suitable for excluding metastatic SCNC from other organs.Nevertheless, we once more wish to state the limitations of this study.The study includes seven cases of SCNC and 4 cases of prostatic carcinoma with diffuse NE differentiation.Therefore, we acknowledge that the small number of cases is one of the limitations of this study.

Conclusion
Our study provides insight into histological characteristics that may facilitate the diagnosis of SCNC of the prostate.Moreover, genetic analysis ndings suggest that the presence of PIK3CA mutation in this tumor may offer a novel therapeutic target.
Ki-67 labeling index for each group.Ki-67 labeling index for the small cell neuroendocrine carcinoma (SCNC) group (n=6), adenocarcinoma area in mixed SCNC-adenocarcinoma group (n=4), and prostatic carcinoma with diffuse neuroendocrine (NE) differentiation group (n=4) (excluding one case with hormonal therapy).The Ki-67 labeling index was higher in the SCNC group than in the other two groups (one-way analysis of variance and the Bonferroni post hoc test. P-values were <0.01, respectively).
adenocarcinoma area in mixed SCNC-adenocarcinoma (n=4), and prostatic carcinoma with diffuse neuroendocrine (NE) differentiation (n=4) (excluding one case with hormonal therapy).No signi cant differences were observed between the three groups (one-way analysis of variance).Abbreviations: AR, androgen receptor; PSA, prostate-speci c antigen.

Table 1 .
Clinicopathological data for each caseAbbreviations: NSE, neuron-speci c enolase; NE, neuroendocrine; ProGRP, pro-gastrin-releasing peptide; PSA, prostate-speci c antigen; GG, Gleason Grade; SCNC, small cell neuroendocrine carcinoma; DOD, died of disease; AWD, alive with disease; DOC, died of other causes.The means±standard deviations of serum NSE, ProGRP, and PSA values in SCNC were 51.9±78.7 ng/mL, 2381.3±5086.2pg/mL, and 31.3±56.7 ng/mL, respectively, whereas these values in prostate cancer with diffuse NE differentiation were 10.8±5.1 ng/mL, 50.5±35.3pg/mL, and 1603.7±2396.5 ng/mL, respectively (PSA values in the one case with hormonal therapy and unmeasured values were excluded).No signi cant differences (Student's t-test) in serum NSE, ProGRP, and PSA levels were observed between small cell carcinoma and prostatic carcinoma with diffuse NE differentiation (excluding cases treated with hormonal therapy).All of the adenocarcinomas con rmed in this study were classi ed as Gleason Grade 5.
Abbreviations: AR, androgen receptor; NE, neuroendocrine; PSA, prostate-speci c antigen; SCNC, small cell neuroendocrine carcinoma; SD, standard deviation Immunohistochemical results for small cell neuroendocrine carcinoma (SCNC, n=6, comprising two cases of pure SCNC and four cases of mixed SCNCadenocarcinoma), adenocarcinoma areas in cases of mixed SCNC-adenocarcinoma (n=4), and prostatic carcinoma with diffuse NE differentiation (n=4), excluding one case in which hormonal therapy was performed.In principle, H-scores were employed in the assessments.Only the Ki-67 labeling index was based on the percentage of hotspots.

Table 4 .
Results of hotspot driver mutations Abbreviations: NE, neuroendocrine; SCNC, small cell neuroendocrine carcinoma Results of hotspot driver mutations for each case.Case 7 data were based on a FoundationOne CDx assay report.* Since SGK mutations are unavailable in the Ion AmpliSeq Cancer Hotspot Panel v2, they were analyzed only in Case 7, in which the FoundationOne CDx assay was applied.SGK mutations were not analyzed in other cases.