- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
The use of 3-hydroxy-3-methyl-glutaryl coenzyme A (HMG CoA) inhibitors, or statins, has steadily increased in recent years . From 2003 to 2004, approximately 24 million Americans were on a statin drug, almost double the number of statin users in 1999–2000 . Recent studies suggest that statins may reduce the risk of prostate cancer . However, these findings are controversial as other studies have failed to confirm this association and, in addition, have reported an increased risk of prostate cancer for obese statin users . With regard to treatment, a significant percentage of patients will undergo radical prostatectomy (RP) for organ-confined prostate cancer and most will be cured based on recent data demonstrating a 15-year prostate-cancer-specific mortality (PCSM) of 12% after treatment . The benchmark for disease eradication and post-prostatectomy surveillance is the ability to maintain an undetectable PSA level. In most contemporary studies, biochemical recurrence (BCR) is defined as a PSA level >0.2 ng/mL after RP . In a large cohort of 3478 men undergoing RP, the 5- and 10-year BCR-free survival rates were 80% and 68% respectively . Multiple studies have shown that tumour stage, Gleason sum (GS) and PSA reliably predict the likelihood of BCR after RP . The biological effects of statins on prostate cancer are unclear but studies have shown that they may lower PSA levels in men [7,8]. The potential decrease in PSA for statin users may have an impact on the rates of BCR after RP but there are limited data to support this hypothesis.
Few data suggest that statins have a durable effect on treatment-related outcomes, specifically BCR and PCSM, following definitive local therapy. Given the natural history of prostate cancer, patient age at presentation and the prevalence of hypercholesterolaemia and cardiac disease in this population, a significant number of patients who present with the disease are likely to be taking a statin. Furthermore, with some evidence suggesting that statins may lower PSA and that PSA plays a role in determining the risk of BCR as well as PCSM it would be of interest to determine whether an association exists between use of these medications and oncological outcome after surgery. The goal of the present study was to determine the effect of statin use on BCR following RP in a contemporary cohort of men.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Patients from the Columbia University Comprehensive Urologic Oncology Database, approved by the institutional review board, were used for analysis. The medical records of 3198 men who underwent RP between 1990 and 2008 were retrospectively reviewed. Data on statin use were extracted from the admission or discharge records of patients at the time of RP. Patients were excluded from the analysis if they had (1) <2 years of adequate follow-up (with at least three consecutive PSA values), (2) neo-adjuvant or adjuvant therapy in the form of hormones, radiation and/or chemotherapy and (3) insufficient pathological data (PSA, GS, TNM stage, lymph node status, margin status). Based on the aforementioned criteria, a total of 1261 patients were included in the final analysis and, of these, 281 (22%) were taking a statin at the time of surgery (Fig. 1).
Figure 1. Algorithm for patient analysis. Numbers in parentheses represent percentages of total. The first number is the percentage of the total group of RP patients ( n= 3198); the second number is the percentage of the total for the respective group of statin users or non-users ( n= 2526 non-users, n= 672 statin users).
Download figure to PowerPoint
RP was performed by multiple surgeons via either a retropubic or a robotic approach and pelvic lymph node dissection was performed at the discretion of the attending surgeon. BCR was defined as PSA ≥0.2 ng/mL after a previously undetectable PSA three months postoperatively. Descriptive statistics (mean, median) for preoperative and postoperative variables were compared between statin users and non-users using Student’s t test and the chi-squared test where appropriate. Preoperative variables included age, race, PSA, clinical T stage, and GS. Postoperative variables included GS, pathological T and N stage and margin status. Given that statin medications were introduced in the early 1990s and that their frequency of use has steadily risen with time, we incorporated the decade of surgery (1990–2000 vs 2001–present) as a variable in the statistical model. Univariate and multivariate Cox proportional hazards models were performed to determine the association between preoperative variables, postoperative variables, decade of surgery, statin use and risk of BCR.
Kaplan–Meier survival analysis was performed to compare the 5-year BCR-free survival rates for statin users vs non-users. Patients were also stratified based on risk for disease progression as per D’Amico: low risk (PSA <10, GS ≤ 6 and stage cT1 or T2a), medium risk (PSA = 10–20 or GS = 7, or stage cT2b or T2c) and high risk (PSA > 20 or GS ≥ 8 or stage ≥ cT3) . Kaplan–Meier survival analysis was then performed within each risk category (low, medium, high) to compare the 5-year BCR-free survival rates for statin users vs non-users. All data were analysed using Stata SE version 9.0 (StatCorp LE, College Station, TX, USA). A P value ≤0.05 was considered statistically significant for all analyses.
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
A total of 1261 patients were included in the final analysis (281 statin users and 980 non-users) (Fig. 1). The mean age at the time of treatment was 60 years and the median follow-up was 36 months (Table 1). The majority of patients were white and there were no significant differences in the distribution of racial groups between statin users vs non-users. The median preoperative PSA for the entire group was 7 ng/mL and statin users had a median PSA of 6.4 ng/mL vs 7.1 ng/mL for non-users (P < 0.05) (Table 1). The median prostate biopsy GS was 6 for the entire group with statin users having a median GS of 7 vs 6 for non-users, but this difference was not statistically significant (P= 0.076) (Table 1). The majority of patients for both groups were clinical stage T1c, and there was a significantly higher proportion of T1 patients in the group of statin users vs non-users (Table 1). With respect to pathological data, the most common stage was T2c, N0 and the median pathological GS was 7 for the entire group as well as for both groups. However, statin users had a higher proportion of patients with pathological GS ≥ 7 (80%) compared with non-users (67%) and this was statistically significant (P= 0.002). Statin users had a slightly higher positive margin rate (29% vs 24%) than non-users, but this difference was not statistically significant (Table 1).
Table 1. Demographic and clinical data for entire study cohort
| ||All patients||Statin non-users||Statin users|| P |
|Total number (%)||1261 (100)||980 (77.7)||281 (22.3)|| |
|Median age||60||59||62|| P < 0.01|
|Race|| || || || |
| White||876 (69.4)||697 (71.1)||179 (63.7)||NS|
| Black||157 (12.5)|| 116 (11.8)||41 (14.6)|| |
| Latin|| 117 (9.3)||88 (9.0)||29 (10.3)|| |
| Other|| 111 (8.8)||79 (8.1)||32 (11.4)|| |
|Median follow-up (months)||36||35||33.5||NS|
|Median preoperative PSA||7||7.1||6.4||0.04|
|Median biopsy GS||6||6||7||NS*|
|Clinical stage|| || || || |
| T1||1106 (87.7)||846 (86.4)||260 (92.5)||0.01|
| T2||149 (11.8)||128 (13)||21 (7.5)|| |
| T3||6 (0.5)||6 (0.6)||0|| |
|Median pathological GS||7||7||7||NS|
|Pathological T stage|| || || || |
| T2||921 (73)||712 (72.7)||209 (74.3)||NS|
| T3||313 (24.8)||251 (25.6)||62 (22.1)|| |
| T4||27 (2.2)||17 (1.7)||10 (3.6)|| |
|Pathological N stage|| || || || |
| Nx||279||201 (20.5)||78 (27.7)||0.03|
| N0||965||765 (78.1)||200 (71.2)|| |
| N1||17||14 (1.4)||3 (1.1)|| |
|Positive margin rate (%)||315 (25)||235 (24)||81 (29)||NS|
On univariate analysis race (black vs white, P < 0.01; Hispanic vs white, P= 0.021), preoperative PSA (P < 0.01), clinical stage T3 or higher (vs T1c P= 0.01), biopsy GS ≥ 7 (P < 0.01), pathological GS ≥ 7 (P < 0.01), pathological stage T3 or higher (P < 0.01), margin status (P < 0.01) and statin use (P= 0.01) all significantly increased the risk of BCR. On multivariate analysis, race (black vs white), pathological stage T3 or higher, pathological GS, margin status, and preoperative PSA remained independent predictors of BCR (Table 2). In addition, when controlling for these significant factors on multivariate analysis, as well as the decade of surgery (1990–2000 vs 2001–present), statin use remained a statistically significant independent predictor of BCR (hazard ratio 1.54, P < 0.05) (Table 2).
Table 2. Multivariate Cox regression analysis for BCR
|Variable||Hazard ratio|| P |
|Age at diagnosis||0.99 (0.983–1.0)||0.725|
|Race|| || |
| White||ref|| |
| Black||1.7 (1.1–2.6)||0.03|
| Hispanic||1.4 (0.7–2.5)||0.31|
| Other||1.1 (0.6–1.9)||0.8|
| T2||ref|| |
| T3 or higher||3.26 (2.2–4.8)||<0.001|
|Pathological GS|| || |
| <7||ref|| |
| ≥7||1.84 (1.1–3.3)||0.03|
|Margin status||1.81 (1.3–2.5)||0.001|
|Decade of surgery|
| 1990–2000||ref|| |
| 2001–present||1.2 (0.8–1.9)||0.41|
|Statin use vs non-use||1.5 (1.0–2.2)||0.049|
On Kaplan–Meier survival analysis the 5- and 10-year overall BCR-free survival rates were 82% and 63% respectively. Statin users had significantly lower 5-year BCR-free survival compared with non-users (75% vs 84%) (P < 0.05) (Fig. 2). When broken down into low, medium and high risk groups, 5-year BCR-free survival rates were 90%, 82% and 68% respectively (P < 0.001). Further substratification and Kaplan–Meier survival analysis of statin use within each risk group showed that low risk patients who were statin users had a significantly lower 5-year BCR-free survival rate than non-users: 75% vs 90% (P= 0.001) (Fig. 3). There was no significant difference in BCR-free survival for statin users vs non-users in the medium and high risk groups.
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Statin drugs may have a protective effect against prostate cancer according to several recent epidemiological studies [2,10]. Shannon et al. were among the first to describe the association between statins and prostate cancer and their group showed a 50–65% risk reduction in cases vs matched controls. Similar findings of an inverse association between statin use and prostate cancer risk were reported from a large case–control study utilizing a pharmacy database . Recent meta-analyses and population-based studies contradict these findings and have shown no association between statin use and overall prostate cancer risk [11,12]. Given these emerging data on the potential association between statin use and prostate cancer risk and the limited data on their effect on oncological outcomes following local therapy, particularly RP, we sought to evaluate the role of statin use on BCR in our comprehensive database of surgical patients.
Our findings that statin users have a lower preoperative PSA than non-users and that statin use is an independent predictor of BCR following RP are intriguing and are contrary to the expected protective effect of statins on developing prostate cancer. Despite the fact that there was a higher proportion of patients with GS ≥ 7 for the statin group as well as a slightly higher (but not statistically significant) positive surgical margin rate, when controlling for these and other known predictors of BCR in the multivariate analysis, statin use was still associated with increased risk of BCR (Table 2).
The reduction in PSA by statin medications may explain our present findings of an increased risk of BCR for statin users as well as significantly decreased BCR-free survival, particularly in men who are deemed low risk. When stratified by risk groups, statin users in the intermediate and high risk categories had no significant difference in outcome compared with non-users. However, low risk patients who were taking statins did significantly worse with respect to 5-year BCR-free survival than those who were not using statins (75% vs 88%, P < 0.001). Given the fact that preoperative PSA is a known predictor of BCR and disease-specific survival after treatment, an artificial decrease in PSA for a statin user may be misleading in terms of the cancer’s aggressiveness prior to treatment. In particular, men who are thought to be low risk with a PSA < 10 while taking a statin could potentially harbour more aggressive disease than their low PSA would suggest and therefore are more likely to experience BCR after RP. This suggests that the increased risk of BCR in low risk statin users may be related not to a direct biological effect of the medication causing increased tumour progression and aggressiveness but instead to a decrease in PSA, delay in diagnosis and inaccurate staging of the cancer prior to RP. In essence, a ‘low risk’ statin user in our study cohort therefore shows a BCR rate that is similar to that of an ‘intermediate risk’ patient who does not take statins. Our additional finding of a significantly higher proportion of patients with pathological GS ≥ 7 among statin users vs non-users further supports the rationale that statin use may mask biologically aggressive disease.
Hamilton et al. described an interesting inverse association between PSA and statin use. They found that after starting a statin medication men experienced a 4.1% decline in PSA . Furthermore, looking at biopsy threshold values, they showed that men with pre-statin PSA values of ≥4.0, ≥3.0 and ≥2.5 ng/mL experienced PSA declines of 12.5%, 11% and 9.5% respectively with 39%, 26% and 25%, respectively, of these men falling below the threshold after taking statins . The authors conclude that PSA reduction may therefore complicate prostate cancer detection presumably due to the proportion of men who fell below the standard PSA cut-points for prompting prostate biopsy. Mener et al. have shown similar findings with statin users experiencing a statistically significant 8.04% (–0.29 ng/mL) decline in PSA following statin administration. A recent study  in men undergoing RP also found a reduction in PSA for statin users compared with non-users at the time of surgery. Similar to our findings these authors  observed that statin users presented with more aggressive GS but, in contrast to the present study, there was no difference in biochemical outcome for statin users vs non-users.
Platz et al. analysed 34 989 men from the Health Professionals Follow-up Study specifically looking at statin use and total as well as advanced prostate cancer risk. Despite finding no association between statins and overall risk of prostate cancer, there was a significant reduction in the risk of advanced prostate cancer for statin users . Several other large studies, including the Cancer Prevention II Study, the California Men’s Health Study and a Finnish population study, also showed an association between statin use and a reduced risk of advanced prostate cancer [16–18]. The biological mechanism of action for these findings is unclear; however, theoretically, statins may protect against prostate cancer by interfering with the cholesterol-rich domains in the cell membrane (‘lipid rafts’) that are responsible for intracellular signalling . In addition, cholesterol is a precursor for androgen synthesis, and statin use, which lowers cholesterol, may decrease androgen levels thereby inhibiting tumour growth in prostate cancer cells .
We did not find any evidence to support the hypothesis that statin use may be protective, or leads to a better oncological outcome for prostate cancer patients. These findings are in contrast to the recent study by Hamilton et al. which showed a reduction in risk of BCR for statin users compared with non-users. Interestingly, they found that the protective effect of statin use was more pronounced in men with low body mass index (BMI) and in those who were taking ≥1 dose equivalent of the drug . However, men with BMI >35 kg/m2 in this cohort actually had an increased risk of BCR if they were using a statin medication . The disparities between these results and those of our present study may be due to a difference in patient populations; however, because of the lack of patient BMI data and statin dose in our database we are unable to directly compare these results.
The retrospective nature of the study and short follow-up are limitations, but there are others that should be noted in the context of statin use. The data regarding statin use were taken from the admission and/or discharge record at the time of surgery; therefore the duration of preoperative and postoperative use of a statin could not be determined. In addition serum lipid data were not available to confirm that a durable lipid-lowering effect from statin use was present in these patients preoperatively or postoperatively. Other medications including 5 alpha-reductase inhibitors and NSAIDs have been shown to lower PSA levels and we did not include the use of these medications in our analysis when considering the PSA lowering effect of statins [22,23]. Statin users may have a higher BMI than non-users and some studies suggest that BMI is an independent predictor of BCR [24,25]. While the present study does not control for BMI in the multivariate analysis, we have previously published data demonstrating that BMI is not an independent predictor of BCR and this has been shown in studies from other institutions as well [26,27]. Finally, although the cohort in our database (n= 1261) is large, the overall number of statin users (281) and the number of users in each subgroup were small, and so the analysis may have lacked sufficient power to detect a difference in biochemical outcome in the medium and high risk groups.
In conclusion, the present study shows that statin users may have significantly lower serum PSA levels than non-users at the time of diagnosis of prostate cancer and that statin use is an independent predictor of BCR following RP even when controlling for decade of surgery, stage, GS and PSA. Men who use statins, particularly those classified as low risk, have a significantly decreased BCR-free survival compared with non-users. Further studies should also consider the potential interaction between race and statin use on biochemical outcome particularly for black men as it has been shown that they may present with advanced disease at an earlier age . In addition, future research is needed to validate the present findings and to determine the effect of statin drugs on serum PSA levels and the implications for diagnosis as well as outcome following local therapy.