Antibiotics and observation have a similar impact on asymptomatic patients with a raised PSA

What's known on the subject? and What does the study add? As PSA exhibits suboptimal specificity, different strategies have been proposed in order to decrease the number of negative biopsies. An empirical course of antibiotics has been proposed as a cost-saving strategy to differentiate patients with benign cancer as it could potentially avoid unnecessary biopsies. Despite being limited by its non-randomized open-label design, our data suggest that no specific PSA reduction threshold can accurately discriminate prostate cancer. The empirical use of antibiotics for asymptomatic patients with elevated PSA levels should be discouraged. OBJECTIVES To compare the influence of a 4-week course of empirical antimicrobial therapy or observation on the prostate-specific antigen (PSA) levels of asymptomatic patients with a raised baseline PSA. To identify whether a decrease in PSA can predict the risk of prostate cancer (PCa) detection on prostate biopsy. PATIENTS AND METHODS Patients were referred to our ambulatory centre because of a raised PSA level (> 2.5 ng/mL) with a normal digital rectal examination. A 12-core prostate biopsy was indicated in these patients and they were offered antibiotic treatment with levofloxacin 500 mg daily for 30 days. Patients who did not agree to use antibiotics but who still showed interest in participating underwent simple observation, serving as controls. Total and free PSA levels at baseline and after 45 days were measured. Variation in PSA level was calculated. All patients underwent a 12-core prostate biopsy 6 weeks after the initial visit. RESULTS In all, 245 men were enrolled, but 43 were lost due to follow-up. A total of 145 patients who used antibiotics and 57 controls were included in the analysis. The median baseline PSA levels were 7.6 and 7.7 ng/mL in the antibiotic and control groups, respectively, with median follow-up levels of 6.8 and 7.0 ng/mL. The follow-up PSA level was significantly lower than the initial PSA level (P = 0.009). Mean absolute and percentage variation in PSA level were similar in both groups (P = 0.828 and 0.128, respectively). The overall PCa detection rate was 15.8%, and did not differ among the groups (P = 0.203). Regarding the percentage variation in PSA level, patients diagnosed with PCa tended to have their PSA level increased (22.4 vs -5.3%; P = 0.001). Indeed, a decrease of 20% in PSA was not predictive of a negative prostate biopsy (P = 0.41). The area under the receiver operating characteristic curve for percentage PSA variation as a predictor of PCa was 0.660. CONCLUSIONS PSA levels tend to fall when repeated after 45 days, regardless of antibiotic use. Despite being associated with the chance of PCa, no percentage PSA variation threshold value exhibits satisfactory discriminatory properties.