Discussion

In this multicenter retrospective cohort of patients with PDA, our findings are consistent with the aggressive clinical behavior previously attributed to this histologic subtype.

In our cohort, the median PSA at diagnosis was 9.7 ng/mL, reflecting PSA heterogeneity in PDA. This aligns with studies disputing comparable PSA levels between ductal and acinar tumors [11–13]. Packiam et al. showed PSA in localized pure PDA is higher than ISUP 1–2 acinar cancers, yet lower than ISUP 4–5 disease, suggesting PSA reflects tumor differentiation rather than histology alone [13]. That study included only pure ductal carcinoma; our cohort included mixed variants, explaining broader PSA distribution. Mixed histology may lead to higher PSA levels due to a larger acinar component, thereby masking the typically modest PSA secretion associated with pure ductal growth patterns. PSA > 20 ng/mL did not predict earlier systemic therapy (HR 2.01, 95% CI 0.77–5.25), supporting limited prognostic value. This hypothesis is supported by PSA levels in synchronous metastatic patients, with five of eight patients showing PSA levels below 20ng/mL despite metastatic disease. This may contribute to delayed diagnosis and underestimation of disease extent, emphasizing the need for imaging-based staging and careful histopathologic assessment rather than PSA-driven risk stratification.

In our cohort, nearly 10% of patients presented with synchronous metastatic disease at diagnosis, reflecting the well-described propensity of PDA to manifest at an advanced stage. This rate is higher than in typical acinar adenocarcinoma and aligns with reports of early extraprostatic spread. Also, the rate of metastatic disease at diagnosis aligns with the literature, which reports rates ranging from 10% to over 25%, especially in pure ductal cases [14, 15].

Similar to Ranasinghe et al., our cohort showed high BCR (29.3%) and short median time to systemic therapy (23 months), emphasizing aggressive PDA biology [5, 16]. However, the relatively short interval to systemic therapy initiation must also be interpreted in the context of evolving imaging standards during the study period. As PSMA-PET/CT was not uniformly available across all participating centers and years of inclusion, occult micrometastatic disease at baseline may have remained undetected in a subset of patients initially classified as non-metastatic. This potential understaging may have contributed to the early need for systemic treatment observed in our cohort. These findings support that PDA biology is prone to early dissemination, driven by extraprostatic extension or occult micrometastases. Notably, apart from ISUP ≥ 4, neither PSA level nor other conventional clinicopathological parameters reliably predicted the need for systemic therapy, highlighting the limitations of standard risk-stratification tools in this histologic subtype [18]. Early systemic treatment raises the question of earlier multimodal approaches, similar to high-risk PCA. Ranasinghe et al. reported poorer ADT responses in PDA, but our cohort demonstrated measurable ADT activity, suggesting variable endocrine sensitivity [5].

Guidelines and APCCC 2022 suggest aggressive histologic variants may benefit from earlier ADT, potentially neoadjuvant or perioperative [17]. From a clinical perspective, our findings support the consideration of PDA as a clinically high-risk entity, regardless of the percentage of ductal histology, and our findings are consistent with current pathological consensus recommendations that generally discourage active surveillance in the presence of ductal differentiation. Given the tendency toward early systemic progression, upfront consideration of treatment intensification appears biologically plausible. However, PDA-specific trials examining multimodal strategies are clearly needed. Such an approach would require prospective evaluation. Overall, these findings underscore the need for PDA-specific prognostic markers and tailored therapeutic strategies that extend beyond conventional PSA-driven risk stratification.

Notably, ADT monotherapy had longer median duration than ARPI or chemotherapy. At first glance, this may suggest that a subset of PDA retains clinically relevant androgen sensitivity. However, this finding must be interpreted in the context of the disease state and biological selection. Patients historically received ADT monotherapy in routine practice, typically in the hormone-sensitive prostate cancer (HSPC) setting, where androgen receptor (AR) signaling remains the predominant driver of tumor growth and responses to ADT are common [18]. In our cohort, most first-line non-curative patients received ADT (n = 14), fewer ARPI (n = 7), few chemotherapy (n = 3).

ARPI and chemotherapy were used in later stages (CRPC), likely explaining shorter durations rather than inferior efficacy [19]. Stepwise escalation likely explains shorter ARPI/chemotherapy duration versus ADT, not inferior activity. Because dedicated evidence for PDA is limited, therapeutic strategies are frequently extrapolated from PAC. In PAC, adjuvant or salvage ADT, with or without RTx, has been shown to improve disease-free survival in patients with adverse pathological features, including nodal involvement or postoperative recurrence [20]. Consequently, ADT-based approaches have also been widely adopted for PDA, supported by consistent AR expression in most PDA tumors and reports of long-term ADT responses in small RTx cohorts and metastatic series [6, 21–26]. However, PDA does not mirror PAC biology uniformly. Prior work has shown that intrinsic pathway upregulation in PDA can attenuate ADT efficacy, indicating early resistance mechanisms in a subset of tumors [5]. In this context, our finding of longer ADT duration likely reflects the subgroup of PDA that retains AR-driven biology.

In contrast, the shorter treatment windows observed with ARPIs and chemotherapy plausibly represent patients whose disease had already transitioned toward AR-independent or more aggressive phenotypes. These observations are consistent with the concept that PDA comprises both androgen-sensitive and androgen-resistant biological subtypes. Second, the observed differences in treatment duration may also reflect a selection bias, as patients who could remain on ADT alone might have had lower disease burden or slower biological progression, as ADT monotherapy was an established therapeutic concept before dual therapies. Conversely, those escalated early to ARPIs or chemotherapy often represent inherently aggressive disease phenotypes or have already transitioned to CRPC.

Another important finding in our cohort is the considerable discrepancy between diagnostic prostate biopsy and final prostatectomy histology: a substantial proportion of PDA cases were not identified on initial biopsy but only on the surgical specimen. A small subset of patients classified as low-risk according to biopsy-based d’Amico classification showed ductal differentiation only in the radical prostatectomy specimen. This has important clinical implications. If the probability of ductal differentiation is similar across treatment groups in PAC, many patients managed non-surgically are unlikely ever to receive the complete histological assessment that would reveal a ductal component. Consequently, prognostic information may remain unknown, and PDA prevalence underestimated. As contemporary consensus statements increasingly support radiotherapy-based approaches for specific high-risk presentations, the shift toward non-surgical management may amplify this “hidden” burden of PDA [17, 27]. In this context, primary surgical management may offer a unique diagnostic advantage by enabling comprehensive histopathological assessment, which in turn can directly influence postoperative risk stratification and adjuvant treatment decisions. Patients with ductal differentiation might therefore particularly benefit from an initial surgical approach, as radiotherapy-based strategies may preclude definitive histological characterization and thereby limit access to histology-driven therapeutic escalation. In practice, this argues for heightened pathological scrutiny of biopsy specimens, with targeted sampling where feasible, and for careful consideration of surgical management when ductal differentiation is suspected preoperatively.

Furthermore, routine reporting of any ductal features, even if focal, should be established and incorporated into the guidelines. It should be considered if patients with high-risk features despite a negative biopsy for ductal patterns should receive an intensive staging and closer follow-up. Together, these measures could reduce the risk that ductal histology remains occult and ensure that patients receive risk-adapted counseling and treatment planning.

Several limitations of our study must be acknowledged. First, as a retrospective multicenter analysis, treatment selection bias and heterogeneity in management approaches are unavoidable. Moreover, the primary endpoint of systemic therapy-free survival represents a pragmatic clinical endpoint rather than a standardized oncologic surrogate such as PSA-based biochemical recurrence-free survival or metastasis-free survival. As initiation of systemic therapy varied across institutions and clinical contexts, the analyzed endpoint encompassed heterogeneous treatment situations, including biochemical recurrence and initiation of treatment for overt metastatic disease. However, due to incomplete and non-standardized longitudinal PSA follow-up data across centers, established PSA-based recurrence definitions could not be consistently applied in this retrospective cohort. We were therefore unable to establish uniform criteria for biochemical recurrence or harmonize the clinical triggers for initiating systemic therapy across participating centers. In addition, detailed clinical information regarding the specific indication for systemic therapy initiation (adjuvant, salvage, or metastatic setting), as well as comprehensive data on metastatic burden, anatomical distribution, and precise timing relative to disease recurrence, were not consistently available across all participating centers due to the retrospective and multicenter nature of the study. To avoid including patients undergoing immediate adjuvant radiotherapy or early salvage therapy for PSA persistence, only patients undergoing treatment 6 months after initial local therapy were included in the final analysis. Second, the number of patients receiving systemic therapy and the lack of data granularity in this setting limit our analyses to univariable models. Despite the multicenter design, the cohort remains small, highlighting the rarity of this subtype. Given the limited follow-up duration, survival outcomes should therefore be interpreted as exploratory and hypothesis-generating rather than definitive. Second, detailed data on the percentage of ductal growth infiltration at the time of biopsy were unavailable in the majority of cases, limiting biological correlative analyses. Furthermore, baseline staging was not standardized across centers and evolved substantially during the inclusion period, particularly as the implementation of PSMA-PET/CT increased. Consequently, some patients classified as non-metastatic at diagnosis may have harbored occult metastatic disease not detectable by conventional imaging. Detailed pathological characterization of positive surgical margins, including anatomical location and margin extent, was not consistently available across participating centers and therefore could not be systematically analyzed. These features may provide additional prognostic information in PDA and should be addressed in future prospective studies. Finally, follow-up remains relatively short for a disease with a potentially long natural history, and specific outcome measures (e.g., cancer-specific survival stratified by systemic therapy subtype) could not be assessed in a standardized way. Future studies incorporating extended follow-up as well as molecular and genomic characterization may further refine the biological understanding and therapeutic stratification of PDA.