Graphical abstract graphic file with name fx1.jpg [48]Open in a new tab Highlights * • Neoadjuvant PARPi + ARSi shows efficacy with manageable toxicity in high-risk PCa * • Biallelic HRR/BRCA2 alterations correlate with faster PSA decline * • Post-treatment analysis reveals MYC suppression and reduced proliferation-related pathways * • Drug-tolerant persister cells exhibit enhanced EMT and AP-1 activation __________________________________________________________________ Zhang et al. report a phase 2 trial evaluating neoadjuvant fuzuloparib plus abiraterone in treatment-naive men with high-risk localized prostate cancer. The combined pCR/MRD rate is 46%, with a 53% 2-year biochemical progression-free survival. The trial demonstrates feasibility, preliminary efficacy, and potential biomarkers for future studies. Introduction Radical prostatectomy (RP) is a standard therapeutic approach for patients with clinically localized prostate cancer (PCa). However, individuals presenting with high-risk features, such as serum prostate-specific antigen (PSA) levels exceeding 20 ng/mL, advanced stage (≥cT3), and high-grade tumors (grade group 4 or 5), face an elevated risk of biochemical recurrence and metastasis post RP.[49]^1 These patients, particularly those with nodal involvement (cN1), are most likely to benefit from intensified treatment strategies,[50]^2 which has led to the integration of multimodal approaches to optimize disease control.[51]^3^,[52]^4 Neoadjuvant conventional androgen deprivation therapy (ADT) has been shown to reduce positive surgical margins and extraprostatic extension but does not significantly improve survival outcomes.[53]^5 Over the past decade, intensified androgen receptor signaling inhibitors (ARSis), including abiraterone, enzalutamide, and apalutamide, have been explored as preoperative treatments for localized PCa. Favorable pathological responses, such as pathological complete response (pCR, defined as the absence of morphologically identifiable carcinoma in the RP specimen), and minimal residual disease (MRD, defined as the maximum diameter of residual tumor ≤5 mm) have been observed in 6.4%–30% of cases ([54]Table S1). Poly(adenosine diphosphate-ribose) polymerase inhibitors (PARPis), such as olaparib, talazoparib, and niraparib, have received approval from the US Food and Drug Administration for use in combination with ARSis as first-line treatment for patients with metastatic castration-resistant prostate cancer (mCRPC) harboring homologous recombination repair (HRR) gene defects (e.g., talazoparib plus enzalutamide) or BRCA mutations (e.g., olaparib plus abiraterone, niraparib plus abiraterone). However, the majority of research on ARSi and PARPi combination therapy has concentrated on the mCRPC stage, with limited studies addressing early-stage PCa. The prevalence of HRR mutations in localized PCa ranges from 8% to 12% in predominantly Caucasian cohorts.[55]^6^,[56]^7 Given the association of BRCA1/2 mutations with more advanced disease,[57]^8 their prevalence is likely higher among high-risk patients. Recent advancements in understanding the biology of PCa suggest that combining PARPi and ARSi may exert synergistic effects.[58]^9 Evidence indicates that this combination significantly prolongs radiographic progression-free survival in patients with mCRPC, irrespective of genomic alterations, though the extent of benefit for individuals without BRCA mutations remains debated.[59]^10^,[60]^11^,[61]^12 In hormone-naive settings, the infrequent occurrence of androgen receptor (AR) pathway alterations implies that ARSis may induce a state of “BRCAness,” enhancing tumor susceptibility to PARPis.[62]^9^,[63]^13 Fuzuloparib, a PARPi,[64]^14 has been approved in China for the treatment of high-grade, platinum-sensitive, recurrent ovarian cancer based on the FZOCUS-2 trial.[65]^15 A double-blind, randomized phase 3 trial ([66]NCT04691804) is currently investigating fuzuloparib combined with abiraterone versus placebo with abiraterone as first-line treatment in unselected patients with mCRPC. The FAST-PC trial was conducted to assess the efficacy and safety of fuzuloparib combined with abiraterone as neoadjuvant therapy in patients with localized high-risk PCa. Additionally, this study aimed to identify potential predictive biomarkers of treatment efficacy and to explore the genomic characteristics of drug-tolerant persister cells, thereby providing comprehensive evidence to guide and optimize future clinical trials. Results Patients Between June 23, 2021, and November 4, 2022, 35 eligible patients were enrolled and received the study treatment. Three patients discontinued the assigned neoadjuvant therapy, and two did not undergo RP, resulting in 30 patients included in the per-protocol (PP) population ([67]Figure S1). Baseline characteristics are summarized in [68]Table 1. The median age was 72 years, with 19 patients (54%) presenting with clinical T stage ≥ T3 and 10 patients (29%) having clinical lymph node metastasis. 19 patients (54%) were classified as International Society of Urological Pathology grade group 5 (16 with Gleason sum 9; 3 with Gleason sum 10). Furthermore, 32 patients (91%) were categorized as having National Comprehensive Cancer Network (NCCN) very high-risk disease,[69]^3 and 34 patients (97%) met the high-risk criteria of the STAMPEDE trial.[70]^16 Table 1. Baseline characteristics Characteristics Patients (N = 35) Age __________________________________________________________________ Median (range), years 72 (52–82) <60, n (%) 4 (11.4%) 60–70, n (%) 7 (20.0%) ≥70, n (%) 24 (68.6%) __________________________________________________________________ ECOG PS, n (%) __________________________________________________________________ 0 31 (88.6%) 1 4 (11.4%) __________________________________________________________________ Clinical T stage, N (%) __________________________________________________________________ T2 16 (45.7%) T3a 4 (11.4%) T3b 5 (14.3%) T4 10 (28.6%) __________________________________________________________________ Clinical N stage, N (%) __________________________________________________________________ N0 25 (71.4%) N1 10 (28.6%) __________________________________________________________________ Gleason score, N (%) __________________________________________________________________ 7 (3 + 4)[71]^a 1 (2.9%) 7 (4 + 3) 3 (8.6%) 8 12 (34.3%) 9 16 (45.7%) 10 3 (8.6%) __________________________________________________________________ PSA (ng/mL) Median (range), ng/mL 46.9 (3.41–470) <20, N (%) 5 (14.3%) 20–100, N (%) 20 (57.1%) ≥100, N (%) 10 (28.6%) __________________________________________________________________ NCCN risk group, N (%) __________________________________________________________________ High risk 3 (8.6%) Very high risk 32 (91.4%) __________________________________________________________________ STAMPEDE high-risk criteria[72]^b __________________________________________________________________ Yes 34 (97.1%) No 1 (2.9%) [73]Open in a new tab ECOG PS, Eastern Cooperative Oncology Group performance status; PSA, prostate-specific antigen. ^a The patient was diagnosed in 2016 and subsequently underwent active surveillance. The Gleason score was assessed based on the 2016 biopsy specimens. ^b Defined as node positive or, if node negative, having at least two of the following: tumor stage T3 or T4, Gleason score of 8–10, and PSA ≥ 40 ng/mL.[74]^16 Efficacy In the intention-to-treat (ITT) population, three patients (9%, 95% confidence interval [CI]: 2%–23%) achieved pCR, and 13 patients (37%, 95% CI: 22%–55%) achieved MRD. The study met its primary endpoint with a pathological response rate of 46% (95% CI: 29%–63%) in the ITT population ([75]Figure 1A). The PP population exhibited similar results, with a pCR rate of 10% (95% CI: 2%–27%) and an MRD rate of 40% (95% CI: 23%–59%) ([76]Figure 1A). Consistent pathological response rates were observed across all subgroups, except for patients with clinical T4 stage (20%, 95% CI: 3%–56%) ([77]Figure 1B). Representative pathological responses after neoadjuvant treatment are shown in [78]Figure 1C. Of the 33 patients who underwent RP, only one (3%) had positive surgical margins. Figure 1. [79]Figure 1 [80]Open in a new tab Pathological response in the FAST-PC trial (A) Pathological responses. (B) Subgroup analyses of pathological responses in the intention-to-treat (ITT) population; data are presented as the pathological response rate (%) and 95% CI. pCR, pathological complete response; MRD, minimal residual disease. (C) Representative pathological response of patients who achieved pCR or MRD before and after treatment (hematoxylin and eosin staining, along with confirmatory IHC for AE1/AE3 and p63). A black scale bar in the image represents 1 mm. Among the 30 patients who underwent RP with available baseline and pre-RP MRI data, 87% were staged as cT2 pre-RP compared to 47% at baseline. Radiological assessments indicated lower rates of lymph node involvement at RP compared to baseline (13% vs. 33%, respectively) ([81]Figures 2A and 2B). Representative radiological responses, including a patient (P34) with a biallelic BRCA2 mutation, are depicted in [82]Figure 2C. Figure 2. [83]Figure 2 [84]Open in a new tab Radiologic and biochemical response in the FAST-PC trial (A and B) The impact of neoadjuvant treatment on clinical TN stage (based on MRI) and correlation with pathological TN stage at final pathology (n = 30); three patients who did not receive RP at our hospital and two patients with missing post-treatment MRI data were excluded. (C) Representative radiological response before and after neoadjuvant treatment using MRI (P34). (D) Boxplots of the PSA level of the intention-to-treat population at baseline, following three cycles of neoadjuvant treatment, pre-RP, 1 month post RP, and 3 months post RP (n = 32–35); one PSA value 1 month post RP was missing due to loss to follow-up; two patients received RP following one cycle treatment due to adverse events, and their PSA values at the end of cycle 1 were measured at C4D1 and before RP; the PSA value was calculated as 0.01 ng/mL if it was equal to or lower than 0.01 ng/mL. The boxplot displays data distribution, where the bottom and top lines represent the minimum and maximum values (excluding outliers), the lower and upper edges of the box indicate the first and third quartiles, and the middle line within the box represents the median. Outliers, if present, are shown as individual points beyond the whiskers. Statistical analysis was performed by the paired Wilcoxon test; ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns: non-significant, p ≥ 0.05. RP, radical prostatectomy; MRI, magnetic resonance imaging; PSA, prostate-specific antigen; C4D1, day 1 of cycle 4; C6D28, day 28 of cycle 6. The patients presented with a median baseline PSA level of 46.9 ng/mL (interquartile range [IQR]: 30.1–114.0 ng/mL). Following three and six cycles of neoadjuvant therapy, median PSA reductions of 99.3% and 99.9% were achieved, respectively, resulting in median PSA levels of 0.21 (IQR: 0.12–0.52 ng/mL) and 0.06 ng/mL (IQR: 0.02–0.23 ng/mL) ([85]Figure 2D). Postoperatively, eight patients were advised to receive adjuvant therapy due to the presence of ypT3/4 or ypN1 disease. Four patients opted for surveillance instead of adjuvant therapy, while the remaining four patients underwent continuous adjuvant medical castration. Adjuvant radiotherapy was not administered due to COVID-19 restrictions. Following the COVID-19 period, these patients subsequently received salvage radiotherapy. One patient demonstrated radiological progression characterized by extensive tumor invasion and involvement of multiple pelvic lymph nodes following six treatment cycles. A rebiopsy indicated an active tumor with DNA mismatch repair (dMMR) deficiency. The patient achieved disease remission after transitioning to immunotherapy (tislelizumab) and subsequently underwent RP, with postoperative pathology confirming a pCR (which was not included in the calculation of pCR rates in our study). Follow-up The swimmer plot ([86]Figure 3A) delineates the duration of treatment and outcomes throughout the study period. As of the data cutoff on May 31, 2024, with a median follow-up of 21.7 months from enrollment (IQR, 20.7 to 27.4 months) and a median post-RP follow-up of 16.2 months (IQR, 14.1 to 21.9 months), the 2-year biochemical progression-free survival (bPFS) was 53%, and the 2-year metastasis-free survival (MFS) was 94% ([87]Figures 3B and 3C). During this period, one patient succumbed, and 14 patients experienced biochemical recurrence (BCR). There was no significant difference in bPFS between patients who achieved pCR/MRD and those who did not (p = 0.45, [88]Figure S2A), which may be influenced by the fact that four of the 16 non-pCR/MRD patients received continuous adjuvant medical castration. The recurrence sites of the 14 BCR patients are depicted in [89]Figure 3D, with the majority of recurrences localized to the prostate bed (n = 6) and pelvic lymph nodes (n = 4) ([90]Figure 3D). These patients were managed with salvage radiation plus medical castration. One patient who developed supraclavicular lymph node metastasis was treated with medical castration plus abiraterone. All recurrent patients exhibited a favorable PSA response (>90%). Figure 3. [91]Figure 3 [92]Open in a new tab Follow-up of the FAST-PC trial (A) Swimmer plot of 35 patients involved in the FAST-PC trial. (B) Kaplan-Meier curves of bPFS for FAST-PC patients; one patient who withdrew consent and two patients who did not receive RP were excluded. (C) Kaplan-Meier curves of MFS for FAST-PC patients; one patient who withdrew consent and two patients who did not receive RP were excluded. (D) Recurrence site indicated by PSMA-PET/CT (n = 14). (E) Kaplan-Meier curves of bPFS since testosterone recovery for patients who have undergone testosterone recovery. pCR, pathological complete response; MRD, minimal residual disease; PD, progressive disease; BCR, biochemical recurrence; MDS, myelodysplastic syndromes; bPFS, biochemical progression-free survival; RP, radical prostatectomy; MFS, metastasis-free survival; NR, not reached. To account for the potential impact of continuous adjuvant medical castration on bPFS and MFS post RP, we analyzed patients who were untreated post RP and had recovered testosterone levels (>150 ng/dL). The median bPFS duration since testosterone recovery was 5.3 months (95% CI, 5.3 to not reached [NR] months, [93]Figure 3E). Additionally, patients who achieved pCR/MRD appeared to have a longer bPFS since testosterone recovery, although this difference was not statistically significant (11.6 months [95% CI, 4.6 to NR months] vs. 4.4 months [95% CI, 3.3 to NR months], p = 0.24, [94]Figure S2B). Safety All 35 patients received at least one dose of the study treatment. Among them, 33 patients (94%) experienced at least one treatment-related adverse event (TRAE) ([95]Figure 4A). The most common TRAEs were anemia (66%, n = 23), increased alanine aminotransferase (26%, n = 9), and decreased platelet count (20%, n = 7). Grade 3 TRAEs were reported in five patients (14%), which included anemia (6%, n = 2), abnormal hepatic function (6%, n = 2), increased alanine aminotransferase (3%, n = 1), hypokalemia (3%, n = 1), and increased aspartate aminotransferase (3%, n = 1). Patients with grade 2/3 anemia generally resolved to grade 1 within 1 month following a dose reduction of fuzuloparib, except for one patient who withdrew consent ([96]Figure 4B). No transfusions were required. One patient developed grade 4 drug-induced liver injury and recovered after discontinuing the medication. Another patient succumbed to myelodysplastic syndrome (MDS), diagnosed 7 months post RP, which led to death 3 months later. Whole-exome sequencing (WES) revealed that this patient had a germline BRCA2 mutation and an ASXL1 clonal hematopoiesis of indeterminate potential (CHIP) variant, while four other patients also carried variants in CHIP-associated genes ([97]Table S2). Figure 4. [98]Figure 4 [99]Open in a new tab Treatment-related adverse events (A) Treatment-related adverse events during medication. (B) Hemoglobin levels; hemoglobin levels at baseline, nadir, pre-RP, post-RP, 2 weeks post RP, 3 months post RP, 6 months post-RP, and 12 months post RP; the thicker lines represent hemoglobin levels in patients with anemia that reached grade 2 or worse. TRAEs, treatment-related adverse events; RP, radical prostatectomy; MDS, myelodysplastic syndrome. Two patients (6%) discontinued treatment before RP due to TRAEs, one due to grade 4 drug-induced liver injury and the other due to grade 3 hepatic function abnormality and insomnia. Dose reductions were necessary for four patients (11%). One patient required dose reductions of both fuzuloparib and abiraterone due to grade 3 hepatic function abnormalities. Two patients required a single-dose reduction of fuzuloparib for grade 2/3 anemia, and one patient required a dose reduction of abiraterone for grade 3 insomnia. Serious adverse events were reported in five patients (14%): two treatment-related (one drug-induced liver injury and one MDS) and three non-treatment-related (one diabetic nephropathy, one pneumonia, and one rib fracture). Perioperative characteristics and complications are detailed in [100]Table S3. Robotic-assisted RP was performed in 21 patients (66%), while laparoscopic RP was performed in 10 patients (31%). The most frequent perioperative complications were fever (56%, n = 18), urinary tract infection (6%, n = 2), and urine extravasation (6%, n = 2). There were no Clavien-Dindo grade ≥3 complications, and no patients required readmission or surgical reintervention within 90 days post surgery. Preoperative ultrasound examinations of patients’ lower extremity deep veins revealed no thrombus formation. Pre-treatment genomic characteristics and correlation with treatment response [101]Figure 5A illustrates the genomic landscape of the pre-treatment biopsy samples, matched with the baseline T stage and biochemical and pathological responses. Detected pathogenic variants within the FAST-PC cohort are detailed in [102]Table S4. Among the genes of interest, FOXA1 exhibited the highest frequency of genomic mutations (5/27, 19%), while MYC gain/amplification (14/27, 52%) and RB1 shallow deletion (16/27, 59%) exhibited high frequencies of copy-number variants. Deleterious genomic alterations (pathogenic mutations and deep deletion) in the HRR pathway were observed in 10/27 patients (37%), including three germline BRCA2 mutations and one BRCA2 deep deletion. TMPRSS2-ERG fusion was detected in two patients (7%), and PTEN loss was identified in four patients (15%) ([103]Figure S3). Figure 5. [104]Figure 5 [105]Open in a new tab Pre-treatment molecular characteristics and correlation with treatment response (A) Oncoplot showing mutations and copy-number variants as assessed by WES of baseline biopsy samples (n = 27). (B) Correlation between HRR gene alterations and BRCA2 alterations with pathological response, adjusted for baseline T stage. (C) Correlation between HRR gene alterations and BRCA2 alterations with time to achieve PSA level <0.1 ng/mL, adjusted for baseline PSA. HRR, homologous recombination repair; AR, androgen receptor; IHC, immunohistochemistry; PSA, prostate-specific antigen; pCR, pathological complete response; MRD, minimal residual disease; PD, progressive disease; RP, radical prostatectomy; ALT, alteration; OR, odds ratio; HR, hazard ratio. WES data were utilized to calculate the homologous recombination deficiency (HRD) score, which assesses HRR deficiency. There was no significant difference in HRD scores between the HRR gene alteration group and the HRR gene wild-type group (p = 0.55). However, patients with BRCA2 alterations had significantly higher HRD scores compared to those with wild-type BRCA2 (p = 0.037). Despite this, the HRD score did not show a significant correlation with treatment response ([106]Figure S4). We also calculated tumor mutational burden (TMB) and the fraction of genome altered (FGA). Patients with HRR gene alterations exhibited significantly higher TMB (p = 0.013) and a trend toward higher FGA (p = 0.26). Neither TMB nor FGA showed a significant correlation with treatment response ([107]Figure S5). Notably, the sole patient (P11) who experienced disease progression during neoadjuvant treatment had the highest TMB (19.8 mutations/Mb) and was confirmed to have a dMMR status ([108]Figure S6). In our cohort, we assessed the allele status of HRR alterations ([109]Table S5) and found that six out of 10 patients had biallelic HRR alterations, while three out of four patients had biallelic BRCA2 alterations ([110]Figure 5A). After adjusting for baseline T stage, neither HRR alterations nor BRCA2 alterations, whether monoallelic or biallelic, were significantly associated with pathological response ([111]Figure 5B). However, after adjusting for baseline PSA levels, patients with biallelic HRR alterations and BRCA2 alterations demonstrated a shorter time to achieve PSA <0.1 ng/mL (p = 0.001 and p = 0.006, respectively, [112]Figure 5C). Genomic changes before and after treatment indicate potential sensitivity/tolerance mechanisms [113]Figure 6A presents the genomic characteristics of 19 patients with paired tissue samples collected before and after treatment. Most pathogenic mutations detected pre-treatment were not detected post-treatment, which was supported by the significantly reduced TMB after treatment (p < 0.001, [114]Figure S7). Figure 6. [115]Figure 6 [116]Open in a new tab Molecular changes before and after treatment indicate potential sensitivity/tolerance mechanisms (A) Oncoplot showing mutations and copy-number variants of paired pre- and post-treatment samples (n = 19). (B) Expression of MYC before and after treatment and enriched pathways of MYC target between pre- and post-treatment samples. The Benjamin-Hochberg method was used to calculate FDR adjusted p value for multiple comparisons; the boxplot displays data distribution, where the lower and upper edges of the box represent the first and third quartiles, and the middle line within the box represents the median. The whiskers extend to the minimum and maximum values within 1.5 times the interquartile range (IQR), while outliers, if present, are shown as individual points beyond the whiskers. (C) Volcano plot showing differentially expressed genes between pre-treatment samples (n = 27) and post-treatment samples (n = 29). Color dots denote genes that passed the adjusted p value and fold change thresholds. (D) GSEA comparing non-pCR/MRD samples to pCR/MRD samples before treatment (blue) and post-treatment samples to pre-treatment samples (red). (E) Histopathological analysis of post-treatment tissues: H&E, AE1/AE3 + p63, and vimentin staining. A black scale bar in the image represents 1 mm. pCR, pathological complete response; MRD, minimal residual disease; NES, normalized enrichment score; FDR, false discovery rate; AR, androgen receptor; EMT, epithelial-to-mesenchymal transition; NS, not significant; H&E, hematoxylin and eosin. Interestingly, MYC gain/amplification, present in 42% of patients before treatment, was undetectable in paired post-treatment tissues ([117]Figure 6A). RNA sequencing (RNA-seq) analysis of pre- and post-treatment tissues also confirmed a significant downregulation of MYC expression post treatment (p < 0.001). Gene set enrichment analysis (GSEA) further demonstrated that MYC target gene pathways were significantly suppressed following treatment ([118]Figure 6B). Subsequently, we conducted a comprehensive analysis of RNA-seq data from pre- and post-treatment specimens. Principal component analysis revealed a clear separation and significant transcriptomic distances between pre-treatment and post-treatment samples, as well as a homogeneous cluster among pCR/MRD and non-pCR/MRD samples before treatment ([119]Figures S8A and S8B). Unsupervised hierarchical clustering corroborated these findings ([120]Figure S8C). The most differentially expressed genes are presented in [121]Figure 6C. As anticipated, there was a marked downregulation of AR-related gene expression in post-treatment specimens, which corresponds with the downregulation observed in the hallmark androgen response pathway (false discovery rate = 5.76E−08, [122]Figure S9A). Both ARG10 and the AR signature showed significant reductions post treatment and demonstrated a strong correlation (p = 1.26E−10, [123]Figures S9B–S9D). The expression of AR did not significantly change before and after treatment (p = 0.865, [124]Figure S9E), and no AR amplification was detected post treatment. The neuroendocrine score also showed no significant change pre- and post treatment (p = 0.442, [125]Figure S9F). However, an increased expression of certain epithelial-to-mesenchymal transition (EMT) genes (CCN1, VIM, SGK1, SNAI2, and AREG) and activator protein 1 (AP-1)-related genes (CCN1, DUSP1, EGR1, FOS, ATF3, and FOSB) was noted in post-treatment specimens. Next, we conducted GSEA comparing pCR/MRD samples to non-pCR/MRD samples before treatment, as well as pre- and post-treatment samples ([126]Figure 6D). The pre-treatment comparison (blue) revealed that pCR/MRD specimens were enriched in E2F targets, G2M checkpoint, and MYC target pathways, suggesting increased tumor cell proliferation, while non-pCR/MRD specimens were enriched in the EMT pathway. These findings align with the overall patient characteristics before treatment and the residual tumor characteristics post treatment (red), indicating that factors influencing treatment sensitivity/tolerance may be present prior to treatment, with drug treatment reinforcing the selection of these features. To confirm that the observed changes in post-treatment residual tumor cells reflect true EMT rather than an increased fraction of stromal cells, we performed vimentin immunohistochemical staining on post-treatment tissue samples. Our results demonstrate clear evidence of EMT within carcinoma cells, independent of the stromal component. Representative images, including H&E staining, AE1/AE3+p63 staining, and vimentin staining, are presented in [127]Figure 6E to provide morphological evidence supporting our findings. Discussion To our knowledge, the FAST-PC trial is the first prospective phase 2 trial to evaluate the efficacy, safety, and potential biomarkers of PARPi combined with ARSi in localized high-risk PCa. The trial met its primary endpoint, indicating promising antitumor activity with a pathological response rate of 46% and manageable toxicity associated with the neoadjuvant regimen of fuzuloparib, abiraterone, prednisone, and ADT in very high-risk patient population, for whom a multimodality treatment approach is strongly needed due to poor outcomes by conventional treatments. A pooled analysis summarizing data from 201 patients with localized high-risk PCa who received neoadjuvant treatment with intensified ARSis reported an overall pathological response rate of 22.4% (95% CI: 16.8%–28.8%).[128]^17 Comparing the baseline characteristics of our patients to those in these studies revealed a higher prevalence of high-risk factors in our cohort ([129]Figure S10). Although cross-trial comparisons should be approached with caution, the combination of fuzuloparib, abiraterone, prednisone, and ADT demonstrated an increased pathological response rate in our cohort, despite the very high-risk features present. Since intensified ARSis are not the standard of care in the presurgical setting, we could not quantify the benefit of the combinations. Our study should be considered as signal-finding, and these benefits should be confirmed by randomized studies with long-term outcomes as endpoints. With a 2-year bPFS of 53% and a 2-year MFS of 94%, while only four patients received continuous adjuvant medical castration, the survival outcomes were favorable compared with the survival results from the STAMPEDE trial, in which the patients received continuous abiraterone and prednisolone with or without enzalutamide.[130]^16 Patients in our study who achieved pCR/MRD showed numerically better testosterone-recovered bPFS (11.6 vs. 4.4 months), suggesting that sustained pathological response may be associated with better control of micrometastases. Given that 97% of our patients met the STAMPEDE high-risk criteria, abiraterone and ADT were considered the standard of care in addition to local treatment.[131]^16 The favorable pathological response observed in our trial suggests that adding PARPi to the treatment regimen for these high-risk patients may enhance antitumor effects and may be an attractive alternative to be tested in the future randomized trials.[132]^18 The neoadjuvant regimen comprising fuzuloparib, abiraterone, prednisone, and ADT demonstrated a relatively safe profile with manageable adverse events. The incidence of grade ≥3 TRAEs was relatively low and did not impede the surgical procedure. Hepatic toxicity and anemia were the most common grade ≥3 TRAEs. Specifically, grade ≥3 hepatic toxicity was observed in four out of 35 patients (11%), a rate comparable to that reported in patients receiving abiraterone alone for localized high-risk PCa.[133]^19 Longitudinal monitoring of hemoglobin levels indicated a rapid recovery from anemia following a reduction in the fuzuloparib dose, and no blood transfusions were required. The risk of grade ≥3 anemia was numerically lower compared to the PROpel trial[134]^10 in mCRPC (9% vs. 15%), despite the older age of our cohort, suggesting that patients without bone metastases and prior treatment may have an increased tolerance to PARPi. Notably, a 76-year-old man with a germline BRCA2 mutation and an ASXL1 CHIP variant was diagnosed with MDS 7 months post RP, a known potential risk associated with PARPi treatment. The presence of CHIP in patients with solid tumors has been linked to an increased risk of subsequently developing therapy-related myeloid neoplasms like MDS and acute myeloid leukemia (AML).[135]^20 Recent studies have shown an increase in CHIP following PARPi treatment in men with advanced PCa, which may underlie the association of PARPi treatment with the rare but serious side effects of MDS and AML.[136]^21 Additional follow-up of men exposed to PARPis, especially those carrying CHIP, is needed to fully understand the long-term risks and complications. Our translational analysis confirmed distinct genomic features of PCa among different races. Consistent with Li et al.,[137]^22 Chinese patients exhibited a significantly lower probability of PTEN loss and TMPRSS2-ERG gene fusion, while showing a higher frequency of HRR pathway alterations, FOXA1 mutations, MYC amplification, and RB1 deletion. We explored several candidate biomarkers for their correlations with biochemical and pathological response. Our studies demonstrated that biallelic HRR gene alterations and biallelic BRCA2 alterations were significantly associated with a shorter time to achieve a PSA nadir of <0.1 ng/mL. This finding is consistent with previous studies indicating that inactivation of the second BRCA allele (biallelic) is associated with a better response to platinum/PARPi therapy.[138]^23 Interestingly, three out of four BRCA2 alterations were biallelic, while three out of six non-BRCA2 HRR pathway alterations were biallelic. The differing allelic statuses may plausibly explain the varying benefits observed among patients with BRCA alterations and non-BRCA HRR pathway alterations.[139]^24 Since patients with metastatic castrate-sensitive prostate cancer (mCSPC) may have multiple biopsy cores available for genomic sequencing, which are more likely to capture biallelic alterations compared to circulating tumor DNA,[140]^25 ongoing phase 3 mCSPC trials (e.g., [141]NCT04821622, [142]NCT04497844, and [143]NCT06120491) may include molecular stratification to test the hypothesis of better biochemical responses in patients with biallelic HRR alterations. It is noteworthy that the rate of pCR/MRD in patients with HRR pathway alterations or BRCA2 alterations was comparable to those with wild-type tumors. This discordance may suggest that the combination therapy is more potent in eradicating micrometastases, while the primary tumor may develop drug-tolerant mechanisms to survive. In our study, we did not observe a significant correlation between baseline HRD scores in PCa tissue and treatment response to the combined therapy. Several factors could explain this observation. First, some patients with lower HRD scores still responded favorably to treatment, which may be due to their sensitivity to ARSi. Studies have suggested that ARSi might induce an HRD-like state, potentially enhancing sensitivity to PARPi in these patients. Second, certain patients with higher HRD scores had substantial residual tumor post treatment, indicating that additional mechanisms, such as EMT pathway upregulation, might be contributing to the persistence of tumor cells independently of the HRD score. Finally, the current HRD scoring systems were primarily developed in breast and ovarian cancers, where genomic landscapes differ from PCa. Limited research has validated HRD scoring in PCa, suggesting a need to develop PCa-specific HRD scoring models to better capture its unique genomic characteristics. Prostate-specific membrane antigen positron emission tomography/computed tomography (PSMA PET/CT) imaging at the time of recurrence also indicated that only one patient had distant metastases, while most had regional or local disease. Thus, it is intriguing to further investigate whether adjuvant radiotherapy, following neoadjuvant treatment and surgery, could improve the oncological outcomes. Despite achieving a favorable biochemical response in the trial, with a median baseline PSA of 46.9 ng/mL, 54% of patients still presented with significant residual tumors (>5 mm). Paired analysis of pre- and post-treatment specimens provided insights into the dynamics of sensitive and tolerant clones. Notably, we observed a high rate of MYC clearance post treatment, consistent with findings in breast cancer studies.[144]^26 Multiple studies have suggested that MYC enhances PARPi sensitivity by inducing DNA damage through alternative non-homologous end joining[145]^27 or by impairing homologous recombination.[146]^28 In line with MYC amplification, gene expression analysis revealed that pre-treatment samples from patients who achieved pCR/MRD were associated with the activation of E2F, G2M, and MYC pathways. These pathways were significantly downregulated in residual tumor samples post treatment, indicating either more proliferative tumors or tumors experiencing premature entry into the S phase and replication stress, both associated with PARPi[147]^29^,[148]^30 and ARSi[149]^31 responsiveness. Consistent with previous results from neoadjuvant ADT treatment results, AR pathway activity was significantly downregulated post treatment.[150]^32^,[151]^33 At the same time, there is no evidence of AR pathway resistance or neuroendocrine transformation following 6 months of ARSi combined with PARPi therapy. Our analysis of pre- and post-treatment specimens indicated that tolerance to ARSi combined with PARPi therapy is characterized by an adaptive drug-tolerant persister intermediate EMT state in residual tumor cells. This phenotypic change mirrors those observed in vitro in human-derived PCa cells[152]^34^,[153]^35 and clinical trial patients.[154]^29^,[155]^31 Recent findings support that whereas EMT induction directly contributes to partial resistance, it more generally facilitates phenotypic plasticity over the long term, enabling the emergence of additional mechanisms that increase drug resistance.[156]^36 Importantly, EMT is not a one-way road but a highly dynamic and reversible process,[157]^37 which also explains why patients remain sensitive to medical castration after recurrence in our trial. Our study is the first to clinically validate EMT as an intermediate state of resistance in trial specimens. Currently, numerous clinical trials are underway exploring therapies that inhibit EMT (e.g., [158]NCT05546879, [159]NCT05445791, [160]NCT06203821, and [161]NCT05550415), offering hope for further optimization of combined treatment strategies. We observed an upregulation of AP-1-related genes post treatment. The members of AP-1 transcription factor family are key players in phenotypic plasticity as they coordinate global epigenetic and transcriptional changes, for instance during the acquisition of a partial EMT phenotype.[162]^38 AP-1 is a druggable target for which multiple classes of small-molecule inhibitors have been developed.[163]^39 Therefore, it is possible to consider therapeutically targeting AP-1 in the context of ARSi combined with PARPi therapy for the purpose of overcoming resistance. In conclusion, the neoadjuvant regimen of fuzuloparib, abiraterone, prednisone, and ADT demonstrated favorable efficacy and manageable toxicity in patients with localized high-risk PCa. Our study identified that biallelic HRR pathway alterations and biallelic BRCA2 alterations were associated with a shorter time to PSA nadir. There was a notable decrease in MYC gain/amplification and cell proliferation characteristics (E2F targets, G2M checkpoint, and MYC targets) following ARSi combined with PARPi therapy. Conversely, the activation of EMT and AP-1 was reinforced in drug-tolerant persister cells. Future trials are necessary to validate the clinical benefits of this combination and to explore interventions aimed at overcoming drug tolerance mechanisms following ARSi and PARPi combinations. Limitations of the study This study has several limitations. First, it employs a single-arm design with a relatively small sample size, necessitating cautious interpretation when comparing results with other neoadjuvant studies. Second, the molecular investigations represent post hoc analyses that require further validation in ongoing neoadjuvant PARPi trials (e.g., [164]NCT05873192, [165]NCT05498272, [166]NCT04030559, and [167]NCT04812366). Third, this study did not assess health-related quality of life, an important aspect for understanding the overall impact of the treatment. Finally, as our study primarily included patients from an Asian population, the findings may have limited generalizability to non-Asian populations. Resource availability Lead contact Further information and requests for resources should be directed to the lead contact, Yao Zhu (zhuyao@fudan.edu.cn). Materials availability This study did not generate new, unique reagents. Data and code availability The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2021) in National Genomic Data Center (Nucleic Acids Res 2022), China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academic Sciences (GSA-Human:HRA008404). Data access ([168]http://ngdc.cncb.ac.cn/gsa-human) is subjected to the regulations by the Ministry of Science and Technology of the People’s Republic of China. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the [169]lead contact upon request. Acknowledgments