Preston MA, Wilson KM, Markt SC, Ge R, Morash C, Stampfer MJ, Loda M, Giovannucci E, Mucci LA, Olumi AF. 5α-Reductase Inhibitors and Risk of High-Grade or Lethal Prostate Cancer. JAMA Intern Med. 2014;174(8):1301-1307. doi:10.1001/jamainternmed.2014.1600
5α-Reductase inhibitors (5ARIs) are widely used for benign prostatic hyperplasia despite controversy regarding potential risk of high-grade prostate cancer with use. Furthermore, the effect of 5ARIs on progression and prostate cancer death remains unclear.
To determine the association between 5ARI use and development of high-grade or lethal prostate cancer.
Design, Setting, and Participants
Prospective observational study of 38 058 men followed up for prostate cancer diagnosis and outcomes between 1996 and 2010 in the Health Professionals Follow-up Study.
Use of 5ARIs between 1996 and 2010.
Main Outcomes and Measures
Cox proportional hazards models were used to estimate risk of prostate cancer diagnosis or development of lethal disease with 5ARI use, adjusting for possible confounders including prostate specific antigen testing.
During 448 803 person-years of follow-up, we ascertained 3681 incident prostate cancer cases. Of these, 289 were lethal (metastatic or fatal), 456 were high grade (Gleason sum [GS] 8-10), 1238 were GS 7, and 1600 were low grade (GS 2-6). A total of 2878 (7.6%) men reported use of 5ARIs between 1996 and 2010. After adjusting for confounders, men who reported ever using 5ARIs over the study period had a reduced risk of overall prostate cancer (hazard ratio [HR], 0.77; 95% CI, 0.65-0.91). 5ARI users had a reduced risk of GS 7 (HR, 0.67; 95% CI, 0.49-0.91) and low-grade (GS 2-6) prostate cancer (HR, 0.74; 95% CI, 0.57-0.95). 5ARI use was not associated with risk of high-grade (GS 8-10) prostate cancer (HR, 0.97; 95% CI, 0.64-1.46) or lethal disease (HR, 0.99; 95% CI, 0.58-1.69). Increased duration of use was associated with significantly lower risk of overall prostate cancer (HR for 1 year of additional use, 0.95; 95% CI, 0.92-0.99), localized (HR, 0.95; 95% CI, 0.90-1.00), and low-grade disease (HR, 0.92; 95% CI, 0.85-0.99). There was no association for lethal, high-grade, or grade 7 disease.
Conclusions and Relevance
While 5ARI use was not associated with developing high-grade or lethal prostate cancer, it was associated with a reduction in low-grade, GS 7, and overall prostate cancer. Because the number of patients with high-grade or lethal prostate cancer in our cohort was limited, we cannot rule out potential risk of harm with 5ARI use.
Prostate cancer is the second leading cause of cancer mortality in US men and the most commonly diagnosed noncutaneous malignancy.1 North American men have a 1 in 6 lifetime risk of prostate cancer diagnosis with a median age at diagnosis of 67 years.1 Owing to the high incidence of prostate cancer and the significant personal and public costs associated with diagnosis and treatment, preventive measures would have a powerful impact on reducing associated morbidity and mortality. In addition, prostate cancer is well suited for chemoprevention efforts because of its long latency period and multistep pathogenesis.2
5α-Reductase inhibitors (5ARIs) have been suggested as chemopreventive agents for prostate cancer. The enzyme 5α reductase facilitates conversion of testosterone to the biologically active form, dihydrotestosterone, and is important in prostate development and maintenance.3,4 There are 2 primary isotypes: 5α-reductase type 1 and 5α-reductase type 2.3,5 5α-Reductase inhibitors such as finasteride (type 2 inhibitor) and dutasteride (type 1 and 2 inhibitor) inhibit conversion of testosterone, resulting in lower intraprostatic levels of dihydrotestosterone and subsequent induction of apoptosis.3,5,6 This inhibition results in prostate size reduction and is the basis for 5ARIs being widely used by men with lower urinary tract symptoms due to benign prostatic hyperplasia.7,8
Two large randomized clinical trials examined the efficacy of 5ARIs for prevention of prostate cancer and showed a 23% to 25% reduction in prostate cancer incidence compared with placebo.9,10 The reduction in prostate cancer was predominantly due to decreased incidence of low-grade (Gleason sum [GS] ≤6) cancer. There was an unexpected increase in high-grade cancer (GS 8-10) in the treatment group.11 Because of this, use of 5ARIs for chemoprevention was not endorsed. Criticisms of these trials include poor generalizability due to the high cumulative incidence of prostate cancer associated with mandated end-of-study biopsy, relatively short duration, and inability to evaluate the effect of 5ARIs on incidence of metastatic or fatal prostate cancer. While overall survival between groups was similar in the Prostate Cancer Prevention Trial (PCPT), implications for the increased incidence of high-grade cancer on cancer-specific survival remain unclear.12- 17
The US Food and Drug Administration (FDA) recently evaluated the potential risks and benefits of 5ARIs in prostate cancer chemoprevention.11,18 They concluded that finasteride and dutasteride do not have a favorable risk-benefit profile for the proposed use of chemoprevention in healthy men and that the incidence of metastatic disease and prostate cancer–specific morbidity and mortality has not been evaluated.11
To investigate the relationship between 5ARI use and high-grade or lethal prostate cancer, we performed a prospective cohort study among 38 000 US men. We hypothesized that 5ARI use may reduce risk of developing low-grade prostate cancer, while increasing the incidence of high-grade and lethal prostate cancer. We demonstrate that while 5ARIs were associated with decreased risk of low- and intermediate-grade prostate cancer, they were not associated with increasing or decreasing risk of high-grade or lethal prostate cancer.
The Health Professionals Follow-up Study (HPFS) was approved by the Human Subjects Committee at the Harvard School of Public Health. Informed written consent was obtained from all participants. The HPFS is a prospective cohort of 51 529 US male health professionals who were aged 40 to 75 years at baseline in 1986. The men responded to a mailed baseline questionnaire in 1986, which collected information on age, height and weight, ancestry, medications, disease history, physical activity, lifestyle factors, and diet. Follow-up questionnaires were sent biennially from 1988 to 2010; information on prostate-specific antigen (PSA) testing was collected beginning in 1994. From 2000, all men with prostate cancer received yearly prostate cancer–specific questionnaires to ascertain disease progression and metastases. This cohort has been used to answer questions related to epidemiology and outcomes of prostate cancer.19- 23
We restricted the study population to men in the HPFS who completed the study questionnaire in 1996 when questions on 5ARI use were first included (n = 43 219). A total of 5161 men who were diagnosed as having cancer (except nonmelanoma skin cancer) prior to 1996 were excluded. The remaining 38 058 men were followed up prospectively for 5ARI use, cancer incidence, metastases, and mortality until 2010.
The primary exposure variable was 5ARI use during the study period. The question regarding use of finasteride (Proscar; Merck) was added to the questionnaire in 1996 and repeated biennially since then. In 2004, the question was modified to include “Proscar, Propecia, and Avodart.” Finasteride is available in 5 mg (Proscar) and 1 mg (Propecia [Merck] for male-pattern baldness) formulations, and dutasteride (Avodart; GlaxoSmithKline) is available in a 0.5 mg formulation. Information on specific type or dose was not provided, but finasteride (5 mg) use likely comprised most of the exposure in this cohort owing to earlier FDA approval (Proscar, 1992; Propecia, 1997; and Avodart, 2002) and Propecia typically targeting a younger population.24 Duration of exposure was calculated by summing use across the 2-year periods encompassed by the biennial questionnaires.
Development of high-grade or lethal prostate cancer during the study period were the primary outcomes. Total prostate cancer incidence was studied as a secondary outcome. Information on pathology, treatment, PSA at diagnosis, and metastases were abstracted from medical records and pathology reports. Prostate cancer was categorized as low grade (GS 2-6), intermediate grade (GS 7), or high-grade (GS 8-10) based on radical prostatectomy or biopsy pathology reports. Of 3294 (89%) cases with available GS information, 58% were from biopsy and 42% from radical prostatectomy specimens. Among men with both biopsy and radical prostatectomy GS data (n = 1345), 27% (n = 360) were upgraded, 4% (n = 49) were downgraded, and 70% (n = 936) were unchanged on final pathologic evaluation. Advanced cancers were those that had spread beyond the prostate, including to the seminal vesicles, lymph nodes, or bone (TNM stage T3b, T4, N1, M1), either at diagnosis or during follow-up, and those that resulted in prostate cancer death. Lethal cancers were a subset of advanced cancers and included only those that caused death or metastases. Nonadvanced cancers were stage T1 or T2 and N0, M0 at diagnosis and did not progress to nodal or distant metastases during follow-up. Cases with missing cancer stage or grade information were included in the analysis for total prostate cancer but excluded from analyses for stage or grade. Deaths were obtained from next of kin, the postal system, and the National Death Index, with a previously reported sensitivity of greater than 98%,25 and cause of death was assigned by an end points committee of 4 physicians by reviewing medical history, medical records, registry information, and death certificates.
Each subject contributed person-time from questionnaire return in 1996 until prostate cancer diagnosis, death, or study conclusion in January 2010. Incidence rates and incidence rate ratios for prostate cancer were calculated for 5ARI ever users vs 5ARI never users. In all analyses, 5ARI use was treated as a time-dependent covariate and was updated in each questionnaire cycle.
Cox proportional hazards models using age as the time scale were used to estimate the hazard ratio (HR) of prostate cancer for 5ARI users compared with never users. All models were stratified by age and time period to control as finely as possible for confounding by age, calendar time, and any possible 2-way interactions between these 2 timescales. To adjust for potential confounding, multivariable models also included smoking history (never, former quit >10 years ago, former quit <10 years ago, current), race (white, African American, Asian American, other), family history of prostate cancer in brother or father (yes or no), vigorous physical activity (quintiles), body mass index (6 categories), height (quartiles), diabetes mellitus (yes or no), PSA testing (yes or no in prior questionnaire cycle), intensity of PSA testing (variable indicating men who had testing in >50% of time periods), physical examinations (yes or no in prior 2 years), prostate biopsy or rectal ultrasonography (yes or no in prior 2 years), vasectomy, and use of other medications (statins, digoxin, α-blockers, saw palmetto, aspirin, nonsteroidal anti-inflammatory drugs, multivitamins; all yes or no). All covariates except race and height were updated in each questionnaire cycle. We used missing indicator variables for missing covariate information. Because 5ARI use may influence carcinogenesis over time, we analyzed use of 5ARIs as ever vs never as well as by duration of use (<4 years and ≥4 years). Duration of use was also analyzed as a continuous variable.
To investigate possible diagnostic bias associated with PSA testing, we repeated the primary analysis restricted to those men who reported PSA testing between 1994 and 1996. To account for the effect of lower urinary tract symptoms from benign prostatic hyperplasia leading to medical attention, PSA testing, digital rectal examinations, and prostate biopsy, we repeated the analysis using α-blockers, an alternate medication for treatment of benign prostatic hyperplasia, as the exposure variable.
95% Confidence intervals were calculated, and P values were 2-sided, with statistical significance set at P < .05. We tested the proportional hazards assumption by comparing models with and without interaction terms between 5ARI use and age using log-likelihood tests. The assumption was met for all outcomes. Statistical analyses were performed using SAS version 9.2 (SAS Institute Inc).
During 14 years (448 803 person-years) of follow-up, 3681 men in the cohort were diagnosed as having prostate cancer. Of these, 289 were lethal (metastatic or fatal), 456 were high grade (GS 8-10), 1238 were GS 7, and 1600 were low grade (GS 2-6). Stage was missing in 8% of ever and never users. Grade was missing in 12% and 11% of ever and never users, respectively. Ever use of 5ARIs was reported by 2878 men (7.6%) between 1996 and 2010. 5α-Reductase inhibitor users were more likely to be older, have had a PSA test, digital rectal examination, and prostate biopsy or rectal ultrasonography. Men who had ever used 5ARIs were also more likely to be receiving an α-blocker or statin or take multivitamins than those who never used 5ARIs (Table 1). The overall mortality per 100 000 person-years was 1746 among 5ARI never users and 1567 among ever users.
We found no significant association between 5ARI use and high-grade (HR, 0.97; 95% CI, 0.64-1.46) or lethal (HR, 0.99; 95% CI, 0.58-1.69) disease after multivariate adjustment, including physical examination, PSA testing, and prostate biopsy or rectal ultrasonography (Table 2). When the definition of high grade was expanded to include GS 4 + 3 = 7 (>50% GS 4) prostate cancer, there was no associated increased risk with 5ARI use (HR, 1.07; 95% CI, 0.79-1.45). Including locally advanced disease (TNM stage T3b-T4), commonly a precursor of lethal disease, again revealed no associated increased risk (HR, 0.80; 95% CI, 0.51-1.23).
The age-adjusted HR of overall prostate cancer for ever use of 5ARIs compared with never use was 1.05 (95% CI, 0.89-1.23). After multivariate adjustment, we observed a 23% lower risk of overall prostate cancer for men who had ever used 5ARIs (HR, 0.77; 95% CI, 0.65-0.91) (Table 2). There was also a significantly decreased risk of intermediate (GS 7) (HR, 0.67; 95% CI, 0.49-0.91), and low-grade (GS 2-6) prostate cancer (HR, 0.74; 95% CI, 0.57-0.95).
We observed a statistically significant reduction in overall, localized, GS 7, and GS 2 to 6 prostate cancer when comparing 5ARI users of less than 4 years with those who reported never using 5ARIs. Hazard ratio estimates were similar, but not statistically significant, for users of 4 or more years compared with never users. The HR of all types of prostate cancer did not appear to vary by duration of 5ARI use of 4 or more years; however, the number of long-term users was low (Table 3).
When analyzed as a continuous variable, increased duration of use was associated with significantly lower risk of overall prostate cancer (HR for 1 year of additional use, 0.95; 95% CI, 0.92-0.99), localized (HR, 0.95; 95% CI, 0.90-1.00), and low-grade disease (HR, 0.92; 95% CI, 0.85-0.99). There was no association for lethal, high-grade, or grade 7 disease.
To evaluate the impact of PSA screening we also conducted the analysis among only men who underwent PSA testing in 1996. This revealed very similar results to the primary analysis. In this subgroup, we found no association with high-grade or lethal disease; however, use of 5ARIs remained significantly associated with a reduced risk of overall disease, localized, and low- and intermediate-grade disease (Table 2).
To evaluate the possibility that differences in diagnostic intensity (ie, physician visits, PSA testing, prostate biopsy) between 5ARI users and nonusers might affect our results (beyond adjusting for these variables), we repeated the analysis using α-blocker use as the primary exposure variable, as the patient population and symptomatology were assumed to be similar (Table 2). While 5ARI use was significantly associated with a reduced risk of overall prostate cancer (HR, 0.77; 95% CI, 0.65-0.91), α-blocker use was not significantly associated with risk (HR, including 5ARI use, 0.93; 95% CI, 0.83-1.03). High-grade or lethal prostate cancer was not significantly associated with use of either 5ARIs or α-blockers.
Investigating the role of 5ARIs in the development of lethal prostate cancer is a critical end point to understanding the potential harms and implications of 5ARI use. In this prospective cohort study, 5ARI use was not associated with increased or decreased risk of high-grade or lethal prostate cancer. However, the number of patients with high-grade or lethal prostate cancer in our cohort was limited, leading to wide confidence intervals in our analyses. Therefore, we cannot definitively rule out potential risk of harm with 5ARI use. We found that 5ARI use was associated with significantly decreased risk of overall, low-grade, and GS 7 prostate cancer. Notably, our results showed a 23% reduction in overall prostate cancer, which is similar to the previously published findings of 2 randomized trials studying the effect of 5ARIs on prostate cancer incidence.9,10 The reduced risk of GS 7 disease with 5ARI treatment is a novel finding not observed in these trials and may be due to general upgrading of biopsy pathologic criteria over the study period.
It was initially reported in the Reduction by Dutasteride of Prostate Cancer Events (REDUCE) trial that there was no increase in high-grade cancer. However, subsequent pathologic reassessment by the FDA detected an absolute increase of 0.5% in incidence of high-grade (GS 8-10) cancer with dutasteride use.11 Similarly, the FDA review reported an absolute increase of 0.7% in incidence of high-grade (GS 8-10) cancer with finasteride use in the Prostate Cancer Prevention Trial (PCPT).10,11 As a result, use of 5ARIs for chemoprevention of prostate cancer has not been widely embraced.
In our study, while we found similar reduction in overall prostate cancer as the PCPT and REDUCE trials, we found no increased risk of high-grade disease. There remained a decreased risk of overall disease and no increased risk of high-grade prostate cancer in a PSA-screened subset. Potential explanations for the discrepant findings could be mandated prostate biopsies increasing the likelihood of cancer detection or insufficient statistical power or follow-up in our cohort. It is theorized that prostate cancer reduction with 5ARI use is due to detection bias as users typically have lower PSA levels and are therefore less likely to undergo biopsy. However, 5ARI users in our study were more likely to have had PSA testing (99% vs 93%) or prostate biopsy (24% vs 9%) than nonusers. Therefore detection bias does not explain the lower risk of localized and low-grade disease observed.
It has been proposed that the increased risk of high-grade disease found in the randomized trials might be due to prostate size reduction leading to better biopsy sampling and higher likelihood of cancer detection (ie, reducing the volume of benign prostate tissue with 5ARI therapy).26,27 Others have suggested that sensitivity of PSA as a biomarker is improved with 5ARI therapy, which may have led to higher rates of high-grade prostate cancers in the PCPT and REDUCE trials.28,29 Among patients in the PCPT who underwent prostatectomy, the finasteride-associated increase in high-grade disease (GS ≥7) at biopsy (42.7% finasteride vs 25.4% placebo; P < .001) was diminished at prostatectomy (46.4% finasteride vs 38.6% placebo; P = .10).27 However, no post hoc analyses conducted to date have been able to definitively prove that 5ARIs do not increase high-grade cancer.
5α-Reductase inhibitors are commonly used to reduce symptoms and consequences of benign prostatic hyperplasia.8,9 Reassuringly, long-term follow-up from the PCPT found no significant difference in the rates of overall survival between finasteride and placebo; however, prostate cancer-specific survival could not be assessed. It is possible that a difference in cancer-specific survival could exist and be overshadowed by deaths from other causes, given the prolonged natural history of prostate cancer, a finding documented in other trials.30 Our results support findings from other studies that show 5ARI users do not appear to be at increased risk of high-grade or lethal prostate cancer.14,16 5α-Reductase inhibitors seem to predominantly protect against GS 6 prostate cancer with minimal influence on GS 8 to 10 disease. This may represent inherently different biologic mechanisms between low- and high-grade prostate cancer and potentially, a role for 5ARI treatment in active surveillance.31- 33
Strengths of this study include the prospective nature of the cohort, large sample size, prolonged follow-up, high degree of subject response, PSA data, and detailed long-term prostate cancer–specific outcomes. In particular, the ability to study lethal prostate cancer is a powerful and unique aspect. There are limitations of this observational study. There was limited central pathologic review of GS (11% of cases). This may result in misclassification of cases by grade, particularly GS 7. It is unlikely, however, that there would be significant discrepancy in assignment of GS 8 to 10 because pathologic criteria for this diagnosis is subject to less interobserver variation. Medication history was reported biennially, which may introduce random error into calculation of exposure duration. Survey data were self-reported but remain high quality owing to the study participants being knowledgeable and committed. The HPFS consists predominantly of white men (91% white) and thus may limit generalizability of study findings. Finally, our study does not take into account that variable degrees of prostatic 5ARI expression may have an effect on efficacy of 5ARIs.5
We found that while 5ARIs were associated with significantly reduced risk of low- and intermediate-grade prostate cancer, they were not associated with increased or decreased risk of high-grade or lethal prostate cancer. However, the number of patients with high-grade or lethal prostate cancer was limited, leading to wide confidence intervals in our analyses. Therefore, we cannot definitively exclude potential risk of harm with 5ARI use.
Accepted for Publication: November 9, 2013.
Corresponding Author: Aria F. Olumi, MD, Department of Urology, Massachusetts General Hospital, 55 Fruit St, Yawkey 7E, Boston, MA 02114 (Olumi.Aria@mgh.harvard.edu).
Published Online: June 2, 2014. doi:10.1001/jamainternmed.2014.1600.
Author Contributions: Drs Preston and Wilson had full access to all of the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis. Drs Preston and Wilson contributed equally to manuscript.
Study concept and design: Preston, Morash, Giovannucci, Mucci, Olumi.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Preston, Wilson, Loda, Mucci, Olumi.
Critical revision of the manuscript for important intellectual content: Wilson, Markt, Ge, Morash, Stampfer, Giovannucci, Mucci, Olumi.
Statistical analysis: Preston, Wilson, Markt, Olumi.
Obtained funding: Preston, Morash, Loda, Giovannucci, Mucci, Olumi.
Administrative, technical, or material support: Stampfer, Loda, Olumi.
Study supervision: Morash, Stampfer, Loda, Mucci, Olumi.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by funding from the National Cancer Institute at the National Institutes of Health (NIH) (grants P01 CA055075 and CA133891); NIH training grants T32 CA09001 (Dr Wilson) and R25 CA098566 (Dr Markt); the Prostate Cancer Foundation (Dr Mucci); NIH grant R01-DK 091353 (Dr Olumi); and the Urology Care Foundation Research Scholar Program and the Robert J. Krane, MD, Urology Research Scholar Fund (Dr Preston).
Role of the Sponsors: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Previous Presentations: This study was presented at the American Urological Association Meeting; May 5, 2013; San Diego, California; and at the Canadian Urological Association Meeting; June 23, 2013; Niagara Falls, Ontario, Canada.
Additional Contributions: We are grateful to the ongoing participation of men in the Health Professionals Follow-up Study (UM1 CA167552) and would like to acknowledge our colleagues (Lauren McLaughlin, BA, and Siobhan Saint-Surin, MA) working on these studies for their valuable help. We are grateful to Walter Willett for his expert comments and advice. In addition we would like to thank the following state cancer registries for their help: Alabama, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Idaho, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Nebraska, New Hampshire, New Jersey, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, Tennessee, Texas, Virginia, Washington, Wyoming.
Additional Information: This study was approved by the Connecticut Department of Public Health (DPH) Human Investigations Committee. Certain data used in this publication were obtained from the DPH. The Connecticut DPH information was just used for disease follow-up among people in the study from Connecticut.