Chronic Use of NSAIDs and/or Statins Does Not Affect PSA or PSA Velocity in Men at High Risk for Prostate Cancer

NSAIDs and statins are of interest in cancer prevention. However, data thus far have been conflicting and inconclusive (1). The 2003 positive study on celecoxib showing an effect in familial adenomatous polyposis [28% (P = 0.003) polyp reduction in 88 patients] and the subsequent FDA approval of celecoxib for chemoprevention in this high-risk population (2) intensified the focus for examining the effects of NSAIDs in cancer chemoprevention, including prostate cancer. However, the positive effects have not translated into other diseases. Similarly, there are compelling data about chemopreventive effects of statins; however, these results are controversial (3, 4).

Many men at high risk for prostate cancer have associated comorbidities requiring long-term use of NSAIDs and statins, which presents a challenge in studying these agents in a randomized trial. Furthermore, an independent effect of these drugs on PSA and PSA velocity (rate of PSA change over time) in those at high-risk could interfere with appropriate diagnosis and monitoring of patients with a positive prostate biopsy. Thus, our goal was to determine whether these medications elicited an independent effect of PSA in high-risk men.

Data were obtained from the negative biopsy trial, a randomized, double-blind, placebo-controlled, phase III chemoprevention trial designed to investigate the effect of selenium supplementation on prevention of prostate cancer in high-risk men, details of which are available in our earlier publication (5). In the current study, medication use data were obtained by questionnaire at baseline and at twice-yearly follow-up visits. Baseline association was investigated using multiple linear regression analysis. PSA values were transformed via the logarithmic function to correct for skewed distribution. Statistical significance of an interaction term between medication use and time-on-study was used to assess the impact of the association between medication use and change in PSA levels over time through mixed-effects regression models. We project to have 96% power using a two-sample t test to detect a difference between PSA levels at an α of 0.05. Using PSA velocity of 0.51 and 0.95 ng/mL/year for aspirin users and nonusers, respectively (6) and assuming an average of 5 repeated measures per subject, we project power of 99% for PSA velocity for the sample size.

Table 1 displays the mean and SDs for subjects using aspirin, other NSAIDs, and statins at baseline. Using multiple linear regression analysis, the coefficient of association between baseline PSA and aspirin, other NSAID, or statins is −0.17, 0.66, and −0.24, with respective P values as 0.75, 0.36, and 0.62. Table 2 shows the PSA values and SD of participants for each visit number, with N denoting the number of subjects in each group at each visit. Results of mixed-effect regression models indicate that medication use is not associated with statistically significant change in PSA velocity (P values 0.76, 0.40, and 0.98, respectively).

This is the first study to use a longitudinal study design and mixed-model analysis to determine the effect of medication use on PSA and PSA velocity. These results indicated that aspirin, other NSAIDs, and statins may not interfere with the diagnosis and prognosis of prostate cancer and hence may be safely used in men at high risk for prostate cancer. The Irish Tumor Registry showed that aspirin did not significantly affect prostate cancer risk [HR, 0.88; 95% confidence interval, CI, 0.67–1.15], and a modest survival effect was observed (HR, 0.61; 95% CI, 0.37–0.99) in the group defined as high-dose (75 mg daily; ref. 7). Although other studies suggest that there is an effect (8). Similarly, the data thus far on statins on prostate cancer risk are compelling but inconclusive. A meta-analysis of clinical data by Zhang and colleagues (4) suggests a protective effect (OR, 1.195; 95% CI, 1.018–1.403, P = 0.029); however, study limitations including lack of prospective trials confound conclusions. Another review by Moon and colleagues discussed the weaknesses in current data (3), which further demonstrates that more information is needed.

Strengths of the study include longitudinal study design that uses PSA data measured at multiple time points to increase the power of the study to provide a more reliable estimate of the patient's PSA velocity. Limitations include recall bias that may occur due to the medication data being obtained through questionnaires. However, attempts were made to minimize this bias by standardized administration of these questionnaires by trained study personnel. In addition, if recall bias does play a role, its effect would be reduced due to it affecting the medication users and nonusers equally. Average duration of time subjects were on the study was 3.5 years. This may not have been enough time for medication use to affect PSA. Role of medication compliance and dosage could not be addressed in this study due to small numbers.

No potential conflicts of interest were disclosed.

Conception and design: A.M. Algotar, M.S. Stratton, R. Behnejad, S.P. Stratton

Development of methodology: A.M. Algotar, R. Behnejad, S.P. Stratton

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.M. Algotar, M.S. Stratton, R. Behnejad, S.P. Stratton

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.M. Algotar, M.S. Stratton, R. Behnejad

Writing, review, and/or revision of the manuscript: A.M. Algotar, M.S. Stratton, R. Behnejad

Study supervision: A.M. Algotar, M.S. Stratton, S.P. Stratton

The authors thank the study participants and families for their contribution to this work.

This study was supported by PHS grants CA079080 and CA023074.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.