Stockholm3 Versus Prostate-specific Antigen in Prostate Cancer Screening: 9-year Outcomes Demonstrating Improved Detection of Aggressive Cancers and Reduced Overdiagnosis from the STHLM3 Trial.

[BACKGROUND AND OBJECTIVE] Prostate-specific antigen (PSA) thresholds (≥3 or 4 ng/ml) are used to balance prostate cancer (PCa) detection against false positives; yet, aggressive PCa can occur at a low PSA and indolent tumors at a high PSA level. Long-term data clarifying aggressiveness across PSA thresholds are limited.

[METHODS] The STHLM3 screening trial enrolled 59 088 men. All received PSA testing; those with PSA ≥1 ng/ml underwent the multianalyte Stockholm3 blood test. We analyzed men treated with radical prostatectomy or radiotherapy (n = 968). Outcomes were any biochemical recurrence (BCR), high-risk BCR, and PCa-specific mortality. Incidence across four baseline groups-(1) elevated PSA (≥3 ng/ml) and Stockholm3 (≥11), (2) elevated Stockholm3 alone (≥11), (3) elevated PSA alone (≥3 ng/ml), and (4) neither elevated PSA (<3 ng/ml) nor elevated Stockholm3 (<11)-was compared using Gray's test and competing-risk regression.

[KEY FINDINGS AND LIMITATIONS] Follow-up was 8.9 yr. The 5-yr cumulative rates of any/high-risk BCR were as follows: 13%/9.0% for both elevated Stockholm3 (≥11) and elevated PSA (≥3 ng/ml), 9.4%/5.3% for elevated Stockholm3 alone, 1.5%/0% for elevated PSA alone, and 0%/0% for nonelevated results on both tests (p < 0.001). Compared with PSA-only elevation, Stockholm3-only elevation showed a hazard ratio (HR) of 1.8 (95% confidence interval 0.8-3.9; p = 0.2) for any BCR and an HR of 8.8 (1.06-72; p = 0.044) for high-risk BCR.

[CONCLUSIONS AND CLINICAL IMPLICATIONS] Some men with PSA <3 ng/ml harbor aggressive PCa with a substantial risk of recurrence after upfront curative treatment. Risk predictive blood tests, such as Stockholm3, used at lower PSA thresholds, can identify these men, while few clinically important cancers are missed when biopsy is deferred with PSA ≥3 ng/ml but low Stockholm3 scores.