Does protocol heterogeneity in active surveillance influence clinical outcomes? Insights from a multicenter prostate cancer cohort.

INTRODUCTION
Active surveillance (AS) is now the standard management for men with low-risk prostate cancer, designed to reduce overtreatment while maintaining oncologic safety. Landmark randomized and prospective cohort studies, such as ProtecT and PRIAS, have consistently demonstrated the long-term safety of AS, reporting very low prostate cancer-specific mortality despite variations in progression and treatment rates [12]. Consequently, AS is endorsed as the standard of care in major international guidelines [34].
However, despite widespread adoption, substantial heterogeneity persists in how AS is implemented across institutions. Variability is particularly evident in eligibility criteria, the role and timing of confirmatory biopsies, and the intensity of surveillance schedules. This raises critical concerns: do protocol differences alter when patients transition to treatment, and more importantly, do they influence oncologic outcomes?
Most prior studies have reported outcomes from single-institution series, registries, or systematic reviews [56], limiting direct comparison across centers. Meanwhile, advances such as multiparametric magnetic resonance imaging (mpMRI) and molecular biomarkers have refined risk stratification and surveillance decisions, but the fundamental question remains: does protocol variation materially alter patient outcomes [78]?
Therefore, this study was designed to directly compare AS protocols across three hospitals with distinct eligibility and monitoring policies. We specifically assessed their impact on (i) the duration of surveillance, (ii) the rate of transition to active treatment, and (iii) the pathological outcomes in men undergoing radical prostatectomy (RP). This multi-institutional analysis aims to clarify whether divergent AS strategies lead to meaningful differences in oncologic safety.

MATERIALS AND METHODS
This retrospective multicenter cohort included consecutive patients who entered AS from the inception of each institution’s AS program through the data lock, specifically: Hospital A (from 2016 onward), Hospital B (from 2015 onward), and Hospital C (from 2014 onward). Institutional Review Board of Veterans Health Service Medical Center approval was obtained (approval number: BOHUN 2025-08-022), and the requirement for informed consent was waived owing to the retrospective design.

1. Patient selection and exclusion
Eligible patients were those diagnosed with clinically localized prostate cancer (cT1–T2) and initially managed with AS. Each institution applied distinct inclusion criteria regarding Gleason Grade Group (GGG), prostate-specific antigen (PSA) level, and number of positive biopsy cores, as summarized in Table 1. Patients with incomplete clinical data, prior hormonal therapy (HT), or follow-up shorter than 12 months were excluded. For secondary analyses of surgical pathology, only men with available prostatectomy specimens were included.

2. Clinical evaluation and surveillance protocols
Baseline assessments included age, PSA, and biopsy findings. mpMRI was performed at the discretion of each institution for clinical staging or follow-up. Surveillance protocols differed by center (Table 1), but generally consisted of serial PSA every 3–6 months, MRI every 6–12 months, and repeat biopsy according to institutional policy. Specifically, Hospital A performed confirmatory biopsy only if clinical changes occurred (elevated PSA, MRI progression, or patient anxiety), Hospital B mandated biennial biopsies, and Hospital C recommended one confirmatory biopsy within 1–2 years followed by biopsy only if clinically indicated.

3. Endpoints
The primary endpoint was transition from AS to active treatment, defined as RP, radiotherapy (RT), or HT. Secondary endpoints included AS duration, defined as the time from diagnostic biopsy to treatment initiation or last follow-up, and pathological outcomes among patients who underwent RP, including final GGG and pathologic stage.

4. Statistical analysis
Categorical variables were compared using the chi-square or Fisher’s exact test. Continuous variables were assessed with the Kruskal–Wallis test when not normally distributed. Time-to-event outcomes (AS continuation and time to treatment) were analyzed with Kaplan–Meier curves and compared using log-rank tests. Cox proportional hazards models were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs), adjusting for baseline covariates (age, PSA, and initial GGG). All analyses were performed using R software, version 4.2.2 (R Foundation for Statistical Computing). A two-sided p-value <0.05 was considered statistically significant.

RESULTS

1. Patient characteristics and institutional differences
Among 232 men managed with AS (Hospital A, n=99; Hospital B, n=83; Hospital C, n=50), the median follow-up was 38.5 months. Baseline characteristics differed significantly across institutions (Table 2). Patients at Hospital A were younger (mean age, 69.0 years) than those at Hospital B (mean age, 74.2 years) and Hospital C (mean age, 74.3 years) (p<0.001). PSA at diagnosis was similar among groups (p=0.818). The distribution of GGG also varied: Hospital C had the highest proportion of GGG 1 (96.0%), whereas Hospital B had the lowest (69.9%) (p<0.001).

2. Repeat biopsy practices
The frequency of repeat biopsy varied markedly (Table 2). A second biopsy was performed in 20.2% of patients at Hospital A, compared with 65.1% at Hospital B and 66.0% at Hospital C (p<0.001). Third biopsies were conducted only at Hospital B (19.3%).

3. Duration of AS
The duration of AS also varied significantly (Fig. 1A). Median AS duration was longest at Hospital B (41.6 months), followed by Hospital C (27.6 months) and Hospital A (21.1 months) (p<0.001). Among patients who eventually underwent treatment, the median time from diagnosis to therapy was longest in Hospital B (30.8 months), compared with Hospital A (25.9 months) and Hospital C (15.4 months) (p=0.030) (Fig. 1B). Kaplan–Meier analysis confirmed these differences (Figs. 2, 3). Five-year retention rates were 53.2% in Hospital A, 79.8% in Hospital B, and 59.1% in Hospital C. Log-rank tests showed that Hospital B differed significantly from both A (p=0.008) and C (p=0.007), whereas A and C did not differ (p=0.358).

4. Transition to active treatment
Overall, 60 patients (25.9%) discontinued AS (Table 2). Transition rates were 23.2% at Hospital A, 18.1% at Hospital B, and 44.0% at Hospital C (p=0.003). Treatment modalities also differed (p<0.001): RP was most common in Hospital C (95.5%), while HT was used only in Hospital A (21.7%) and Hospital B (46.7%). Fig. 4 illustrates these institutional differences, presenting percentages calculated from the entire AS cohort including patients who remained on surveillance. In multivariable Cox models (Fig. 5), Hospital B showed the lowest hazard of transition compared with Hospital A (HR 0.49, 95% CI 0.25–0.95). For RP specifically, hazards were lowest in Hospital B (HR 0.20, 95% CI 0.07–0.62) and higher in Hospital C (HR 1.87, 95% CI 0.94–3.72), relative to Hospital A, which did not reach statistical significance.

5. Pathologic outcomes after RP
Thirty-eight patients underwent RP during AS (14 at Hospital A, 4 at Hospital B, 20 at Hospital C). Final pathological outcomes were comparable across institutions (Table 3). Distribution of GGG (p=0.899) and pathologic stage (p=0.203) did not differ significantly. Although extraprostatic extension appeared more frequent in Hospital C (45.0%), this was not statistically significant.

DISCUSSION
AS is widely adopted for men with low-risk (and/or favorable intermediate-risk in selected circumstances) in prostate cancer to avoid overtreatment while maintaining oncologic safety. In this multicenter cohort, Hospital C managed only low-risk patients and performed one routine confirmatory biopsy, whereas Hospitals A and B also included some favorable intermediate-risk cases. Hospital A did not perform routine confirmatory biopsy, while Hospital B implemented biennial scheduled re-biopsies (Table 1). The substantial inter-institutional differences in protocol elements were associated with divergent AS duration and treatment transition patterns. However, pathologic grade and stage among men who ultimately underwent RP did not differ significantly across hospitals. These findings suggest that protocol heterogeneity influences when patients discontinue AS rather than what is found at surgery.
Prior literature underscores how protocol components shape the clinical trajectory of AS. Confirmatory and surveillance biopsy strategies remain central: omission or delay risks missing higher-grade disease, whereas structured rebiopsy improves selection fidelity [59]. Large programs such as PRIAS demonstrated the importance of confirmatory biopsy for durable AS eligibility [2], while consensus reviews recommend tailoring monitoring intensity to patient risk rather than applying uniform schedules [5]. At the same time, mpMRI has reduced reliance on systematic repeat biopsy in select scenarios, but its negative predictive value is imperfect; MRI-guided approaches are best considered complementary, not substitutive, to histologic confirmation [101112]. Consistent with this, it has been demonstrated that AS candidates harboring the Prostate Imaging Reporting and Data System (PI-RADS) 5 lesions were diagnosed with unfavorable disease in more than 80% of prostatectomy specimens, whereas the vast majority (>90%) presented with PI-RADS≤4 lesions, highlighting the limited discriminative value of MRI alone and the need for systematic or targeted biopsy [13].
Several studies have extended the application of prostate MRI by incorporating additional clinical or imaging-derived parameters to improve risk prediction. An Italian multi-institutional cohort showed that combining MRI findings with PSA density and baseline core number improved prediction of reclassification risk compared with MRI alone [14]. Furthermore, a prospective Korean cohort adopting strict AS criteria—including exclusion of PI-RADS 5 lesions and incorporation of PSA density and core involvement—achieved significantly better progression-free and treatment-free survival than non-strict protocols, reinforcing that MRI must be integrated with clinical and pathological variables to ensure oncologic safety [15]. Recent imaging research has introduced lesion density—the ratio of lesion length to prostate volume—as a novel MRI-derived parameter that improves the prediction of clinically significant prostate cancer in targeted biopsies [16]. Combining MRI with biomarkers (e.g., Prostate Health Index [PHI], 4Kscore) may further refine rebiopsy decisions and reduce procedural burden without sacrificing safety [8]. For instance, recent data underscores the diagnostic value of PHI even in AS candidates with non-index lesions on mpMRI or following negative combined biopsies, supporting its role in enhanced risk stratification for the decision to proceed with or discontinue surveillance [17]. In parallel, emerging molecular classifiers (e.g., Decipher, Oncotype DX GPS) show promise in predicting upgrading during AS and may enable more individualized rebiopsy schedules and earlier intervention for those who need it, while reassuring the majority who do not [1819].
Why did more intensive protocols in our cohort not translate into better pathology at RP? One explanation is that tumor biology may dominate long-term outcomes once patients are reasonably selected, outweighing differences in surveillance intensity. This interpretation aligns with previous reviews showing earlier intervention under intensive surveillance but not necessarily superior surgical pathology [620], and with ProtecT’s long-term data reporting similar prostate cancer-specific mortality across monitoring, surgery, and RT despite differing treatment rates [1].
Still, not all studies agree. Some analyses caution that prolonged surveillance without confirmatory biopsy increases upgrading at prostatectomy [21], and prospective cohorts have linked omission of confirmatory biopsy with more adverse pathology at treatment [22]. These discrepancies likely reflect how surveillance is conducted. Centers that integrate consistent MRI review with at least one early confirmatory biopsy appear able to mitigate risks while avoiding the burden of frequent rebiopsies.
Clinical implications follow. First, protocol flexibility is defensible if grounded in common denominators: (i) at least one confirmatory biopsy (within 12–24 months) or MRI-triggered histologic confirmation as a safeguard against misclassification; (ii) risk-adapted surveillance cadence for PSA/MRI; and (iii) transparent criteria for leaving AS (grade or volume progression, or patient preference). Second, institutional patterns matter: in our study, the hospital with biennial biopsies showed the lowest hazard of treatment transition, while the center favoring rebiopsy and surgery showed higher RP transition rates—yet post-RP pathology remained comparable. For patients, these differences translate into distinct experiences on AS with potentially similar oncologic outcomes.
Institutional heterogeneity likely shaped the observed transition patterns. Hospital C’s higher conversion rate and predominance of RP may reflect its strict criteria for continuing AS and a proactive preference for surgical treatment, whereas Hospital B’s greater use of HT may relate to older age, comorbidities, or patient preference. Differences in monitoring intensity—particularly rebiopsy frequency—likely influenced detection of grade progression. Accordingly, the lower transition rate at Hospital A may reflect underdetection rather than superior disease stability. These differences highlight how enrollment practices and institutional culture can influence outcomes and emphasize the need for consensus-based standardization of AS criteria and monitoring intensity.
This study has several limitations. First, baseline differences, particularly in age and GGG distribution across institutions, may have influenced transition patterns. Although multivariable analyses adjusted for these factors, residual confounding cannot be excluded, and internal validity may be threatened by baseline imbalances that cannot be fully addressed with regression. As noted in the preceding paragraph, observed differences may in part reflect enrollment practices and institutional culture rather than protocol effects per se. Second, interpretation of surgical pathology outcomes is constrained by the relatively small number of men undergoing RP (n=38). The study was not powered to evaluate equivalence of adverse pathology across institutions; non-significant findings in the RP subset should be considered inconclusive. These findings should therefore not be overgeneralized to the entire AS cohort. Third, MRI utilization varied substantially across institutions, potentially introducing heterogeneity in detection and monitoring. This inconsistency may have influenced both biopsy decisions and timing of treatment transition, limiting the comparability of outcomes.

CONCLUSIONS
Institutional heterogeneity in AS protocols significantly influenced the duration of surveillance and the timing of treatment transition but did not alter adverse pathology at surgery. These findings suggest that flexibility in protocol design may be acceptable if anchored by common safeguards such as an early confirmatory biopsy and risk-adapted monitoring; however, inferences about oncologic safety should be interpreted cautiously given baseline imbalances and the limited RP subgroup. Our results, together with other multi-institutional experiences, support the need for evidence-based standardization to optimize patient outcomes while preserving the safety and practicality of AS.