Clinical and genomic features of SPOP-mutant prostate cancer

Abstract

Background

Inactivating missense mutations in the SPOP gene, encoding speckle-type poxvirus and zinc-finger protein, are one of the most common genetic alterations in prostate cancer.

Methods

We retrospectively identified 72 consecutive prostate cancer patients with somatic SPOP mutations, through next-generation sequencing analysis, who were treated at the Johns Hopkins Hospital. We evaluated clinical and genomic characteristics of this SPOP-mutant subset.

Results

SPOP alterations were clustered in the MATH domain, with hotspot mutations involving the F133 and F102 residues. The most frequent concurrent genetic alterations were in APC (16/72 [22%]), PTEN (13/72 [18%]), and TP53 (11/72 [15%]). Our SPOP-mutant cancers appeared to be mutually exclusive with tumors harboring the TMPRSS2-ERG fusion, and were significantly enriched for Wnt pathway (APC, CTNNB1) mutations and de-enriched for TP53/PTEN/RB1 alterations. Patients with mtSPOP overall had durable responses to androgen deprivation therapy (ADT) with a median time-to-castration-resistance of 42.0 (95% CI, 25.7–60.8) months. However, time-to-castration-resistance was significantly shorter in SPOP-mutant patients with concurrent TP53 mutations (HR 4.53; p=0.002), HRD pathway (ATM, BRCA1/2, CHEK2) mutations (HR 3.19; p=0.003), and PI3K pathway (PTEN, PIK3CA, AKT1) alterations (HR 2.69; p=0.004). In the castration-resistant prostate cancer setting, median progression-free survival was 8.9 (95% CI, 6.7-NR) months on abiraterone and 7.3 (95% CI, 3.2-NR) months on enzalutamide. There were no responses to PARP inhibitor treatment.

Conclusions

SPOP-mutant prostate cancers represent a unique subset with absent ERG fusions and frequent Wnt pathway alterations, with potentially greater dependency on androgen signaling and enhanced responsiveness to ADT. Outcomes are best for SPOP-altered patients without other concurrent mutations.

Background

Inactivating point mutations in the gene encoding speckle-type poxvirus and zinc-finger protein (SPOP) are one of the most common alterations in prostate cancer, occurring at a frequency of 6–15% across localized and metastatic cancers (1). These alterations are clustered in the substrate-binding MATH domain, appear to be an early event in prostate tumorigenesis (2), and may be enriched in localized prostate cancers compared to metastatic cancers (3). SPOP inactivation results in androgen receptor (AR) overexpression at the protein level and enhanced AR-driven cell proliferation (4). Together, these findings have implicated SPOP alterations as defining a novel subclass of prostate cancers (5).

The SPOP protein is an E3 ubiquitin ligase adaptor (6), which interacts with steroid receptor coactivator 3 (SRC-3). SRC-3 appears to be critical in AR-dependent transcriptional activity and promotes cellular proliferation in prostate cancer (7). Mutant SPOP (mtSPOP) is unable to engage with SRC-3 to promote proteasome-mediated degradation of SRC-3, suggesting that wild-type SPOP (wtSPOP) acts as a tumor suppressor (6). Studies show that patients with mtSPOP may have superior responses to androgen deprivation therapy (ADT) (8) and may also exhibit more favorable responses to abiraterone (9) compared to wtSPOP patients, likely attributable to enhanced addiction to the AR pathway in patients with mtSPOP. Other studies have implicated the SPOP protein in DNA double-strand break repair, suggesting that mtSPOP prostate cancers are characterized by genomic instability and may be sensitive to PARP inhibition and other synthetic lethal therapeutic approaches (3).

Additional studies assessing mtSPOP prostate cancers are needed to further examine the impact of SPOP mutations and their relevance to clinical outcomes in prostate cancer. This study aimed to characterize concurrent genetic alterations in patients with SPOP mutations, as well as their clinical outcomes to systemic therapies for hormone-sensitive prostate cancer (HSPC) and castration-resistant prostate cancer (CRPC).

Methods

We retrospectively identified 72 consecutive prostate cancer patients at a single academic institution (Johns Hopkins Hospital) with somatic SPOP mutations from genomic testing of primary tumors (n=57), metastatic lesions (n=13), or liquid biopsies (n=2). SPOP mutations were detected using tissue-based targeted next-generation sequencing (NGS) performed in a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory: Foundation Medicine (Cambridge, MA). The two liquid biopsies were performed on circulating tumor DNA as part of a separate study on genetic alterations identified from cell-free DNA (10).

Records of these 72 patients were reviewed to determine baseline characteristics, therapies received, and clinical outcomes. The primary endpoints of interest included time-to-castration-resistance following initiation of ADT in the HSPC setting, and PSA progression-free survival (PSA-PFS) on abiraterone and/or enzalutamide in the CRPC setting. PSA-PFS was determined using the PCWG3 criteria (11), and time-to-castration-resistance was calculated based on PSA progression following initiation of continuous ADT for recurrent or metastatic disease. We also constructed Cox proportional hazards survival models to assess the effect sizes of SPOP and other associated gene mutations with outcomes. Both adjusted (by race, age at diagnosis, and Gleason score) and unadjusted models were fitted using individual mutational frequencies and aggregated mutational frequencies, which included Wnt pathway activating mutations, HRD pathway mutations, and PI3K pathway mutations. Wnt pathway mutations were defined as mutations in APC or CTNNB1; homologous repair deficiency (HRD) as mutations in ATM, BRCA1/2 or CHEK2; PI3K pathway mutations as alterations in AKT1, PIK3CA or PTEN. Any individual mutational frequency with less than five occurrences was not assessed by modeling. Kaplan-Meier survival curves were constructed for the mutations and mutational cohorts found to have statistical significance.

We next assessed the prevalence of concurrent mutations compared to other data sets. Publicly available datasets (n=22) from prostate cancer studies uploaded to cBioPortal (12, 13) were analyzed for clinically relevant pathogenic mutations in localized (i.e. primary tumor specimens) and metastatic prostate cancers. Studies were pooled to interrogate mutations including copy number alterations, point mutations, and structural variants. Clinically relevant mutations selected for comparison included AKT1, APC, AR, ATM, BRCA1, BRCA2, CHEK2, CTNNB1, MYC, PIK3CA, PTEN, RB1, TMPRSS2-ERG, and TP53. Only prostate adenocarcinoma specimens with “sample type” (i.e. primary vs metastatic) information and non-overlapping data were included, with a final study count of 12 (14–25). To generate this analysis, the cgdsr API in R (26) was utilized to query for mutation data. The studies utilized to generate this analysis, as well as the localized and metastatic specimens identified from each study is shown in Supplemental Table 3. Analysis of Individual patient level data for the purposes of baseline patient characteristics and clinical outcomes was not performed due to the lack of standardized data input from each individual study. The proportion of mutation occurrence in the aggregated studies were compared to those of our cohort using Fisher’s exact tests, and significance was set at p<0.01 to account for multiple comparisons.

Results

Seventy-two (72) men with prostate cancer and pathogenic SPOP mutations were included in the final analysis, out of a total of 586 men with available NGS data (prevalence: 12.3%). The median age at diagnosis was 64.2 years (range, 46 to 85). Eighteen of 72 (25.0%) were African American or Black, representing an enrichment compared to an overall prevalence of 13% African American or Black patients in our whole NGS database. Fifty of 72 patients (69.4%) were diagnosed with Gleason grade group 4–5 disease. Twenty-three of 72 men (31.9%) were diagnosed with de novo metastatic disease. Baseline patient characteristics are further summarized in Table 1. Comparison of the baseline characteristics of our cohort to patients who underwent NGS at our institution without SPOP mutations (ie. wtSPOP), and to patients enrolled in the standard-of-care (SOC) alone or SOC plus docetaxel (27) arms of the STAMPEDE trial is available in Supplemental Tables 1 and 2, respectively. The STAMPEDE trial was selected as a point of comparison due to the enrollment of a mixed proportion of metastatic and non-metastatic HSPC.

Table 1:

Baseline clinical characteristics of patients with mtSPOP prostate cancer (n=72) at Johns Hopkins Hospital, as well as systemic treatments received in the hormone-sensitive prostate cancer (HSPC) and castration-resistant prostate cancer (CRPC) settings.

Characteristic	mtSPOP patients (n=72)
Race
African American	18 (25.0%)
Asian	2 (2.8%)
White	52 (72.2%)
Age at diagnosis (Years)
Mean (SD)	64.2 (7.39)
Median [Min, Max]	64 [46, 85]
Tumor mutation burden (Mutations/Mb)
Mean (SD)	3.31 (4.14)
Median [Min, Max]	3.00 [0, 29.0]
Prostatectomy
No	35 (48.6%)
Yes	37 (51.4%)
Primary or salvage radiotherapy
No	30 (41.7%)
Yes	42 (58.3%)
T Stage at diagnosis
1/2	29 (40.3%)
3/4	29 (40.3%)
N Stage at diagnosis
0	54 (75.0%)
1	18 (25.0%)
M Stage at diagnosis
0	49 (68.1%)
1	23 (31.9%)
Gleason grade group
1	1 (1.4%)
2	3 (4.2%)
3	11 (15.3%)
4	23 (31.9%)
5	27 (37.5%)
PSA at Diagnosis (ng/mL)
Mean (SD)	53.8 (135)
Median [Min, Max]	9.90 [1.85, 802]
Treatments received for HSPC
ADT alone	38 (52.8%)
ADT + Abiraterone	6 (8.3%)
ADT + Enzalutamide	3 (4.2%)
ADT + Docetaxel	11 (15.2%)
PARP inhibitor	3 (8.3%)
Treatments received for CRPC	n=31
Abiraterone	21 (67.7%)
Enzalutamide	16 (51.6%)
Docetaxel	6 (19.4%)
Cabazitaxel	6 (19.4%)
PARP inhibitor	4 (12.9%)

F133 was the most frequently mutated SPOP residue, occurring in 39 (54.2%) of instances, of which F133L was the most commonly observed alteration (20 counts). There were four patients who harbored two mutations each in the SPOP gene. Figure 1 depicts the distribution of mutations within the SPOP gene, the vast majority of which are in the MATH domain, with one mutation (M254T) located in the BTB domain. The most frequently co-occurring pathogenic mutations were in APC (16/72 [22.2%] patients), PTEN (13/72 [18.1%] patients), and TP53 (11/72 [15.3%] patients). Of note, there were patients who harbored more than one pathogenic mutation per gene. Interestingly, 26 men (36.1%) harbored only an SPOP mutation without any other concurrent genetic driver event. Concurrent TMPRSS2-ERG fusions were virtually never observed within this cohort (1/72 [1.4%] patients), suggesting mutual exclusivity. Figure 2 shows the most frequent clinically relevant co-occurring pathogenic alterations (AKT1, APC, AR, ATM, BRCA1/2, CHEK2, CTNNB1, MYC, PIK3CA, PTEN, RB1, TMPRSS2-ERG, and TP53) in our cohort of patients with SPOP-altered prostate cancer.

Figure 1.

Lollipop plot depicting locations of individual SPOP mutations in cohort of mtSPOP patients, by frequency of mutation observed.

Figure 2.

Oncoprint figure showing concurrent genetic alterations in our cohort of mtSPOP prostate cancer patients. Of the 72 mtSPOP cases, 26 tumors (36%) had no other genetic driver mutations.

Data from pooled studies using cBioPortal were analyzed to determine whether the frequencies of clinically relevant mutations seen in our cohort of SPOP-mutated patients were different from those in unselected patients with localized and metastatic prostate cancers. These results are summarized in Table 2. SPOP mutations occurred at a frequency of 12.7% in localized and 16.8% in metastatic prostate cancers in cBioPortal. Our cohort of mtSPOP patients was enriched for mutations in APC (22.2% vs 10.5% in localized specimens; p=0.003). Conversely, TMPRSS2-ERG fusions were rarely observed in our cohort (1.4%) compared to the public datasets (33.2% in localized and 39.6% in metastatic specimens, both with p<0.001), suggesting mutual exclusivity. Indeed, co-occurrence of SPOP and TMPRSS2-ERG fusions were rare in the pooled cBioportal cohorts as well, identified in 2.7% and 6.5% of primary and metastatic specimens, respectively, neither of which were statistically significant compared to the occurrence in our mtSPOP cohort. Our SPOP-altered patients also had lower frequencies of alterations involving TP53, PTEN and RB1. When considering biological pathways, our SPOP-mutated patients had significantly lower frequencies of HRD pathway and PIK3 pathway mutations compared to metastatic specimens, and more Wnt pathway alterations compared to unselected primary prostate cancers. Overall, our cohort of mtSPOP patients had mutational frequencies in other key genes (AKT1, AR, BRCA2, CHEK2, CTNNB1, MYC, PTEN, TP53) that were more akin to localized as opposed to metastatic prostate cancer specimens.

Table 2:

We performed iterative Fisher’s Exact Tests comparing the frequency of localized and metastatic mutations found in aggregated cBioPortal prostate cancer datasets with respect to the frequencies found in our study. Panel A depicts frequencies of individual mutations across our cohort, compared to localized prostate cancer specimens, and metastatic prostate cancer specimens. Numerical values within parentheses () in the row for TMPRSS2-ERG depicts the number of specimens with co-occurrence of TMPRSS2-ERG fusions and SPOP mutations. Panel B depicts the frequencies of pooled mutations (Wnt activating mutations, HRD pathway, and PIK3 pathway) across our cohort, compared to localized prostate cancer specimens, and metastatic prostate cancer specimens. Due to the multiple comparisons, only p values of <0.01 were considered significant.

A)
	mtSPOP cohort n=72		Localized specimens n=2,903		Metastatic specimens n=1,562
	Frequency	Percentage	Frequency	Percentage	Frequency	Percentage
SPOP	72 (of 586)	12.3%	370	12.7%	262	16.8%
AKT1	3	4.2%	111	3.8%	238	15.2% *
APC	16	22.2%	304	10.5% *	373	23.9%
AR	3	4.2%	99	3.4%	749	48.0% **
ATM	5	6.9%	256	8.8%	300	9.2%
BRCA1	1	1.4%	206	7.1%	135	8.6%
BRCA2	5	6.9%	428	14.7%	408	26.1% **
CHEK2	3	4.2%	195	6.7%	307	19.7% **
CTNNB1	4	5.6%	166	5.7%	283	18.1% *
MYC	4	5.6%	462	15.9%	584	37.4% **
PIK3CA	5	6.9%	282	9.7%	290	18.6%
PTEN	13	18.1%	591	20.4%	685	43.9% **
RB1	2	2.8%	670	23.1% **	527	33.7% **
TMPRSS2-ERG	1	1.4%	964 (79)	33.2% ** (2.7%)	618 (101)	39.6% ** (6.5%)
TP53	11	15.3%	762	26.2%	807	51.7% **

B)
	mtSPOP cohort n=72		Localized specimens n=2,903		Metastatic specimens n=1,562
	Frequency	Percentage	Frequency	Percentage	Frequency	Percentage
Wnt activating	20	27.8%	414	14.3% *	548	35.1%
HRD pathway	21	29.2%	762	26.2%	673	43.1% **
PIK3 pathway	12	16.7%	799	27.5%	831	53.2% **

^*p <0.01,

^**p <0.001

Fifty-eight of 72 patients (80.6%) received ADT for recurrent or metastatic disease. About half of these men also received concurrent treatment with an anti-androgen or a taxane (see Table 1). Median time-to-castration-resistance was 42.0 (95% CI, 25.7–60.8) months in these SPOP-mutant patients, which is longer than expected in unselected patients (28). Co-occurring pathogenic mutations in TP53 resulted in shorter time-to-castration-resistance (adjusted HR 4.53; p=0.002). Mutations were also grouped by the molecular pathways affected and subsequently analyzed. Here, the presence of HRD pathway and PI3K activating mutations also conferred a more rapid time-to-castration-resistance (adjusted HR 3.19; p=0.03 and HR 2.69; p=0.04, respectively) as illustrated in Table 3. The Kaplan-Meier survival curve for TP53-mutated patients, as well as for those with pooled (i.e. Wnt pathway, HRD pathway, and PI3K pathway) mutations are depicted in Figure 3. Outcomes for patients with the most commonly mutated SPOP residues, F133 and F102, were not significantly different compared to the rest of the cohort (Supplemental Table 4).

Table 3:

A Cox proportional-hazards model was fitted to assess the presence of mutations and their influence on time to castration resistance on first-line ADT. The table shows the hazard ratios (HR), the respective confidence intervals, and the significance of mutational occurrence of specified Wnt activating, HRD pathway, PIK3 pathway, and TP53 mutations. Indicators (race, age at diagnosis, and Gleason grade) were also included in the adjusted model to account for potential confounding effects.

	HR	Lower 95%CI	Upper 95%CI	p Value
Mutations (Unadjusted)
Wnt Activating	0.92	0.42	2.03	0.85
HRD Pathway	2.43	1.07	5.54	0.03*
PIK3 Pathway	2.64	1.22	5.69	0.01*
TP53	3.19	1.45	6.98	0.003*
Mutations (Adjusted † )
Wnt Activating	1.26	0.47	3.33	0.65
HRD Pathway	3.19	1.13	9.01	0.03*
PIK3 Pathway	2.69	1.04	7.00	0.04*
TP53	4.53	1.77	11.6	0.002*

^†The Cox proportional-hazards model for adjusted estimates included race, age at diagnosis, and Gleason grade as variables within the model.

Figure 3.

Kaplan-Meier curves and Cox proportional-hazards models adjusted for race, age at diagnosis, and Gleason grade fitted with the presence and absence (black and colored curves, respectively) of a particular class of mutations: Wnt activating, HRD pathway, PIK3 pathway, and TP53 mutations. Baseline features for race (white, majority), age at diagnosis (64, median), and Gleason grade (4, mean) were used across all models. Median PFS estimates are shown corresponding to dotted lines along with each hazard ratio.

Of the 31/72 patients (43.1%) who progressed to CRPC, 21/31 (67.7%) received abiraterone and 16/31 (51.6%) received enzalutamide; 9/31 patients (29.0%) received both agents sequentially in the CRPC setting. A large proportion of enzalutamide-treated patients had previously received abiraterone first. Median PSA-PFS was 8.9 (95% CI, 6.7-NR) months on abiraterone, and 7.3 (95% CI, 3.2-NR) months on enzalutamide. In a pooled analysis of mtSPOP patients who received abiraterone or enzalutamide, co-existing Wnt pathway, HRD pathway, and PI3K pathway mutations did not affect PSA-PFS, as shown in Table 4. Six of 31 patients (19.4%) received docetaxel, and 6/31 (19.4%) received cabazitaxel for CRPC. Survival analysis on chemotherapies was not performed due to the insufficient sample size.

Table 4:

A Cox proportional-hazards model was fitted to assess the presence of mutations and their influence on pooled PSA-PFS on abiraterone and enzalutamide. The table shows the hazard ratios (HR), the respective confidence intervals, and the significance of mutational occurrence of specified Wnt activating, HRD pathway, PIK3 pathway, and TP53 mutations. Indicators (race, age at diagnosis, and Gleason grade) were also included in the adjusted model to account for potential confounding effects.

	HR	Lower 95%CI	Upper 95%CI	p Value
Mutations (Unadjusted)
Wnt Activating	0.78	0.33	1.85	0.57
HRD Pathway	0.92	0.35	2.47	0.87
PIK3 Pathway	1.10	0.50	2.38	0.82
TP53	1.72	0.68	4.34	0.25
Mutations (Adjusted † )
Wnt Activating	0.80	0.20	3.23	0.76
HRD Pathway	0.77	0.23	2.55	0.67
PIK3 Pathway	1.01	0.38	2.63	0.99
TP53	1.75	0.54	5.56	0.35

^†The Cox proportional-hazards model for adjusted estimates included race, age at diagnosis, and Gleason grade as variables within the model.

In this SPOP-mutated cohort, 7/72 patients (9.7%) of received a PARP inhibitor (olaparib or rucaparib): three in the HSPC setting (with two of them receiving concurrent enzalutamide), and four in the CRPC setting. None of the patients on PARP inhibitor monotherapy had co-occurring mutations in BRCA1/2, and none demonstrated radiographic or PSA responses to PARP inhibitor treatment.

Discussion

Our efforts to understand the clinical heterogeneity of prostate cancer have led to the genomic profiling of localized and metastatic prostate cancers over the last decade (20, 22, 25, 29). Recent evidence suggests that evolution from SPOP-mutated to CHD1-deleted prostate cancer defines a unique molecular subtype of prostate cancer distinct from ERG-overexpressing tumors, the latter of which is frequently paired with frequent PTEN deletion (30). The two events – mtSPOP and ERG fusion – together are considered to be synthetic lethal and their downstream effects on AR appear to be critically distinct (31). Previous in vitro studies have shown that SPOP mutants (particularly Y87, W131, and F133) disrupt substrate binding, and furthermore that cells transfected with the very commonly mutant (F133V) demonstrated increased invasion (1). Induction of SPOP mutations in normal prostate cells was recently shown to be sufficient in promoting oncogenic chromatin remodeling, increased DNA accessibility, and AR binding (32). Together, these findings point towards a mechanistic foundation for the clinical significance of SPOP mutations, conferring dependence on AR-mediated signaling and responsiveness to hormonal therapies.

Our cohort was enriched for African American or Black patients, who in previous studies were shown to have SPOP mutations at equal or higher frequencies than their Caucasian counterparts as well as fewer ERG fusions (33–35). Our mtSPOP patients appeared to have comparable mutational landscapes to unselected localized prostate cancer specimens, with the exception of over-representation of APC mutations and under-representation of TP53/PTEN/RB1 alterations and TMPRSS2-ERG fusions. The latter finding is consistent with prior studies (1, 36) and our understanding of molecular progression in early prostate cancer (30). CHD1 deletion status was unable to be assessed, as this gene was not available within the Foundation Medicine panel. APC was the most frequently co-mutated gene, resulting in Wnt pathway activation. β-catenin, the effector molecule of the Wnt pathway, has previously been shown to act as an AR coactivator in prostate cancer (37, 38). This activation may further reflect dependence on the AR signaling axis induced by mtSPOP, which may explain why the presence of APC mutations or Wnt activating mutations as a class did not appear to confer worse prognosis in first-line ADT. Prior work has shown that Wnt activating mutations are associated with resistance to first-line abiraterone and enzalutamide in the CRPC setting (39), but in our cohort of mtSPOP patients the co-existence of Wnt activating mutations did not influence PSA-PFS on these agents in CRPC (possibly due to small patient numbers).

Patients with defects in HRD genes are known to benefit from PARP inhibition (40), though among the SPOP-altered patients who received PARP inhibitors in our cohort, none showed objective responses (but none had concurrent HRD mutations). However, the presence of HRD pathway mutations was found to confer shorter time to castration resistance on first-line ADT. Indeed, prior studies have implicated worse outcomes of patients with germline BRCA1/2 mutations in terms of metastasis-free survival after diagnosis of localized prostate cancer (41). There were no major differences in outcomes of mtSPOP patients co-harboring HRD mutations in response to abiraterone or enzalutamide in CRPC, a finding that has been shown previously for unselected patients with HRD mutations (42). Alterations in PTEN, a tumor suppressor and negative regulator of the PI3K pathway, are found to accumulate with disease progression (43) and appears to play a significant role in development of castration resistance (44). TP53 is frequently mutated in prostate cancer (22) and is associated with worse time to progression in localized prostate cancer (45). While presence of PTEN loss/inactivation itself was not prognostically significant, alterations in the PI3K pathway as well as TP53 mutations in our cohort conferred worse outcomes on first-line ADT but not to abiraterone or enzalutamide in CRPC. These data suggest that the best outcomes in SPOP-mutated patients are achieved in those without other concurrent genetic driver alterations.

With the rise of new systemic therapies and earlier adoption of therapies previously reserved for the mCRPC setting, the utilization of genomic profiling and biomarker-directed therapy selection will be critical in the coming years to strategically select patients who are most likely to derive benefit. Our results show that, overall, patients harboring mtSPOP had durable responses to first-line ADT, but this cohort notably includes a heterogenous group of patients with metastatic and nonmetastatic hormone-sensitive disease, as well as patients receiving concurrent chemotherapy (i.e., docetaxel) and/or androgen-directed therapy (i.e. abiraterone, enzalutamide). Preliminarily, this study suggests that patients with mtSPOP have roughly equivalent durations of response to abiraterone and enzalutamide in CRPC, although many enzalutamide-treated patients previously received abiraterone. These findings require further confirmatory studies but have the potential to lead to utilizing SPOP as a biomarker for treatment selection.

To our knowledge, this is the largest cohort of SPOP-mutated patients from a single institution with comprehensive characterization of co-occurring pathogenic mutations and clinical outcomes. The goal of this paper is primarily descriptive and hypothesis-generating. This dataset is limited to patients who underwent NGS at the treating oncologist’s discretion, which tends to be biased towards those who had recurrent or metastatic tumors. The primary limitation of this study is the lack of clinical outcomes for wtSPOP patients to serve as a control, however, baseline characteristics of wtSPOP patients appear skewed towards more White patients and those with higher Gleason scores. When compared to the STAMPEDE cohorts, our cohort of mtSPOP patients had fewer patients with metastatic disease at diagnosis (and therefore were treated with ADT alone in the HSPC setting), however, with similar distributions of Gleason ≤7 and 8–10 disease. When comparing time to castration resistance to failure-free survival, our cohort appears to fare better than the STAMPEDE SOC plus docetaxel cohort (42.0 vs 37.0 months), though this difference may be attributable to the greater proportion of patients with metastatic disease in the STAMPEDE cohort. We were not able to describe clinical outcomes to taxane chemotherapies due to very small patient numbers. Furthermore, we used a heterogeneous tissue source that comprised both metastatic and primary tumor biopsies as well as two ctDNA specimens. Finally, we were unable to confirm the association with concurrent CHD1 deletions since this gene was not included on our NGS platform. Thus, we were not able to explore differences in clinical outcomes between SPOP-mutated patients with intact versus deleted CHD1 status.

However, our results affirm that somatic mutations in SPOP are clinically significant and define a distinct molecular subtype of prostate cancer worthy of further investigation. Ongoing studies assessing the impact of mtSPOP on immunotherapy response are also of particular interest since mtSPOP is known to stabilize PD-L1 at the protein level (46), suggesting a mechanistic basis for improved outcomes for mtSPOP on PD-1 inhibitors and other immunotherapies. In addition, SPOP has been shown to participate in DNA repair of double-strand breaks and appears to have a similar effect as BRCA2 inactivation in vivo and may confer sensitivity to PARP inhibitor (3), which warrants further investigation on the impact of SPOP mutations and response to PARP inhibition, even though this was not observed in our study.

Conclusion

SPOP mutations define a unique subset of prostate cancers that are enriched for high Gleason scores, African American race, co-existence of Wnt pathway activating mutations, and virtual absence of TMPRSS2-ERG fusions. Time-to-castration-resistance on first-line ADT appears longer than expected compared to genomically-unselected patients; while those with concurrent TP53 mutations, HRD pathway mutations, and PI3K pathway mutations may have inferior outcomes. Further studies are necessary to characterize the impact of SPOP alterations on outcomes to chemotherapies, PARP inhibitors, and immunotherapies.