In observational studies of men with prostate cancer, men on endocrine treatment (ET) have had an increased risk of ischemic heart disease (IHD) and stroke. However, prostate cancer per se may increase risk of IHD and stroke and men on ET may have been at increased risk already prior to initiation of ET. We assessed the incidence of IHD and stroke in men with prostate cancer before and during different endocrine treatments. The hazard ratio (HR) of IHD and stroke in 39,051 men with prostate cancer vs. a matched control population without prostate cancer was assessed by use of Cox proportion hazard models. An increased risk was found among 30,883 men with prostate cancer who did not receive ET, with a HR of 1.08 (95% CI 1.00–1.18) for IHD and 1.10 (95%CI 1.00–1.21) for stroke. In 8,168 men who initiated ET during the observation period, the risk of IHD was significantly higher (p = 0.014), during ET (HR 1.40, 95% CI 1.17–1.67) compared with before initiation of ET (HR of 0.98, 95% CI 0.72–1.33), whereas no such increase was found for stroke. Regardless of treatment, men with prostate cancer had a small increase in risk of IHD and stroke and initiation of ET was associated with a further increase in risk of IHD. Our data underline the importance of a proper indication for ET because many men with low-risk prostate cancer currently receive ET.
Endocrine treatment (ET) is the standard treatment for advanced prostate cancer building on consistent evidence for beneficial effects of ET in locally advanced and metastatic disease.1, 2 However, ET is increasingly used in localized disease,3–5 in prostate-specific antigen (PSA) relapse after curatively intended procedures and as neoadjuvant and adjuvant treatment in men undergoing external beam radiotherapy.1 Men with low-risk disease have a life expectancy close to that of the background population so even relatively small increases in risk of serious adverse events threatens the cost-benefit of ET in this patient group.6 Recently, several observational, register-based studies have reported an increased risk of diabetes, cardiovascular disease (CVD), and stroke in men treated with ET for prostate cancer.7–12
However, these studies were performed among men with prostate cancer only and no data were available for men before and after initiation of ET. Thus, the results from these studies may have been biased because prostate cancer per se may be associated with an increased risk and risk estimates may also be affected by confounding by indication to treatment as men who receive ET may already prior to ET be at increased or decreased risk for these adverse effects.
In a previous work from PCBaSe Sweden,13 based on primary treatment at diagnosis of prostate cancer CVD was investigated in relation to primary treatment. However, that study could not take longitudinal data into account. Men without ET at diagnosis were regarded as untreated for the entire follow-up, which introduces a misclassification of exposure since the proportion of men on ET continuously increase because some men relapse after curative treatment and some men on surveillance experience a progress in their disease.
The aim of this study was to investigate the risk of ischemic heart disease (IHD) and stroke in nearly 40,000 men with prostate cancer with different treatments in comparison with prostate cancer-free control population matched for age and residency. We investigated risk of IHD and stroke before initiation of ET and during ET, and we took prior heart disease and heart medication into consideration.
PCBaSe Sweden is based on the National Prostate Cancer Register (NPCR) of Sweden, which was linked to various other national registers. A more detailed description of NPCR and PCBaSe Sweden is given elsewhere.14–17 In brief, the NPCR contains data on tumour characteristics according to the tumour, node status, metastasis TNM classification,18 histological grade, date of diagnosis, serum levels of prostate-specific antigen (PSA) at the time of diagnosis, and primary treatment delivered or decided within 6 months after diagnosis for 98% of all men with prostate cancer in Sweden since 1998. In this study NPCR was linked to Cause of Death Register, the Prescribed Drug Register, the National Hospital Discharge Register and the Day Surgery Register. These registers contain information on cause and date of death, all dispensed prescribed drugs from pharmacies, discharge diagnoses, and surgical procedures.19 Socioeconomic characteristics were assessed by record linkages to the 1960–1990 five-yearly Census Databases and based on socioeconomic status (SES). SES is based on occupational group and stratifies men into white-collar worker, blue-collar worker, not gainfully employed, and unknown.20 The Swedish Central Ethics Committee approved of the project.
Medication among men with prostate cancer and population controls with cardiovascular drugs at the start of the observational period was registered from the Prescribed Drug Register in seven categories: warfarine, platelet aggregation inhibitors, lipid modifying agents, beta-blocking agents, angiotensin converting enzyme (ACE)-inhibitors, diuretics, and digitalis. All medical ET initiated during the study period was registered in the Prescribed Drug Register and orchiectomy was registered in the National Hospital Discharge Register or the Day Surgery Register. ET was separated into four different categories: GnRH agonists with or without flare prophylaxis [defined as an antiandrogen (AA) used for a maximum of 30 days], GnRH agonists with AA longer than 30 days, AA as monotherapy, orchiectomy, and other treatments (e.g., estrogens and other treatments that could not be fitted into any of the three previous categories).
Diagnoses of IHD (ICD10:I20-25) and stroke, both nonfatal and fatal (ICD10:I60-64, G45) were available from the National Hospital Discharge Register. The quality of these data has been assessed previously and the validity was between 94% and 96% when re-examined by an expert committee.21–23 Five risk groups were used to described the tumours; low risk: T1-2, Gleason score = 6 and PSA < 10 ng/mL. Intermediate risk: T1-2, Gleason score 7 and/or PSA 10 to < 20 ng/mL. High risk: T3-4 and/or Gleason score 8–10 and/or PSA 20 to <50 ng/mL. Regionally metastatic disease: N1 and/or PSA 50 to <100 ng/mL in the absence of distant metastases (M0 or Mx). Distant metastases: M1 and/or PSA ≥ 100 ng/mL.24
We compared two cohorts, one with men with prostate cancer that were not treated with ET at start of follow-up, and another with these men's individually matched population-based controls. We compared the incidence of IHD and stroke between (i) all men with prostate cancer irrespective of treatment status regarding ET with their corresponding population controls, (ii) all men with men with prostate cancer not treated with ET and the corresponding population controls and finally, (iii) all men with prostate cancer for the time period they were not treated, respectively, when they were treated with ET and their corresponding population controls. Thus, men who started ET during the observation period were analysed in the unexposed group before ET, and when ET had been initiated they were analysed as exposed to ET. The population controls were used as reference during the respective calendar periods of exposure for unexposed as well as exposed men.
Men with prostate cancer
We included (i) all men diagnosed with prostate cancer and registered in NPCR between November 1, 2005 and December 31, 2006 and (ii) men diagnosed before November 1, 2005 who had not received ET at the time of diagnosis according to NPCR, who had not received ET between July 1, 2005 and November 1, 2005 according to the Prescribed Drug Register, and who had not undergone orchiectomy before November 1, 2005 according to the National Hospital Discharge Register or the Day Surgery Register (Fig. 1). We assumed that the men without ET between July 1, 2005 and November 1, 2005 were previously not treated with ET, since the maximum duration of dispensed prescription at a pharmacy is 100 days and 2/3 of the amount of the previously dispensed medication must have been consumed before the patient can buy more of the same medication. We performed a waiting time distribution according to Hallas et al.25 This showed that the curve level off after 4 months. By this design and the results from the waiting time distribution test, we were confident that only men untreated with ET at start of follow-up were included. The start date of follow-up was based on the start of the Prescribed Drug Register which was July 1, 2005 (Fig. 1).
For each man with prostate cancer, a population control was selected from the general population sets of men matching the man with prostate cancer for age (±1 year) and county of residency. To be eligible as a population control the man must be alive on the November 1, 2005.
The endpoints considered were occurrence of IHD or stroke and only the first of any event was allowed for each subject. Hazard ratio (HR) was used as the estimate of relative risk and was modeled in Cox proportional Hazard models. All HRs refer to a comparison with the population-based control cohort unless another comparison is specifically defined. The initiation of ET was considered as a time-dependent covariate. In the analysis of IHD, we censored for end of follow-up and death and ignored events of stroke. Similarly, we ignored events of IHD in the analysis of stroke. All models were adjusted for age (logarithmized), number of previous IHD (stroke), time since last event of IHD (stroke), socioeconomic status (high vs. others), and medication for heart disease at baseline (yes/no). The association of ET to outcomes was analysed by using the information on day of ET onset as a time-dependent covariate in the Cox proportional hazards model. The effect of ET was estimated as the hazard ratio for the time dependent covariate.
Statistical significance was evaluated in Cox-regression analysis of patients from date of Pca to end of follow-up and using date of ET start as time-dependent covariate.
In total 39,051 men with incident Pca were included, with a mean follow-up of 1.9 years (SD ± 0.4). At the beginning of follow-up, there were virtually no differences in the distribution among men with prostate cancer and controls in age, number of previous IHD and strokes, and medication although the men with prostate cancer more frequently had a high SES (52.2%) than controls (46.0%; Table 1). Of the 39,051 men with prostate cancer, ET was initiated in 8,168 (20.9%) men during the observation period (Table 2); 3,203 (39.2%) of these men had their Pca diagnosed before our observation period started and 4,965 (60.8%) during our observation period. Of these 4,965 men diagnosed during the observational period and treated with ET, 1,125 (23%) were men with a low- or intermediate-risk disease, the majority of them had received ET within 6 months of diagnosis (data not shown).
Table 1. Mean follow-up time, age at diagnosis, previous episodes of ischemic heart disease and stroke, medication and socioeconomic status for men with prostate cancer and population controls in PCBaSe Sweden
Table 2. Time since date of diagnosis of prostate cancer, tumor characteristics at date of diagnosis and primary treatment for men in PCBaSe Sweden
There was a modest but statistical significantly increase in risk of IHD and stroke for all men with prostate cancer compared with the control population, HR 1.12 (95% CI 1.05–1.21) for IHD and HR 1.11 (95% CI 1.02–1.21) for stroke (Table 3). Men with prostate cancer with no ET also had a somewhat higher risk of IHD 1.08 (95% CI 1.00–1.18) and stroke 1.10 (95% CI 1.00–1.21). Among men who initiated ET during the observation period the risk of IHD was higher after initiation of ET than before ET, HR 0.98 (95% CI 0.72–1.33) before ET and HR 1.40 (95% CI 1.17–1.67) during ET. The increase in risk from 0.98 to 1.40, was statistically significant (p = 0.014; Table 3). Stroke data indicated a decreased risk after initiation of ET with a HR before ET of 1.26 (95% CI 0.88–1.80) and a HR after initiation of ET of 1.11 (95% CI 0.91–1.37), however, this decrease was not statistically significant (p = 0.54; Table 3).
Table 3. Incidence and hazard ratio of ischemic heart disease and stroke for men with prostate cancer and population controls in PCBaSe Sweden
A separate analysis after excluding men with prevalent IHD/stroke gave the same changes in HR as for the full cohort but the confidence intervals were wider (data not shown).
In analyses of the four ET categories separately, a statistically significantly increased risk of IHD was found for men treated with GnRH agonists with a HR of 1.63 (95% CI 1.15–2.32).
We further analysed if medication for IHD or stroke modified risk, and we took previous IHD, stroke and socioeconomic status in to account in the analysis (Table 4). Men with medications against CVD had a higher risk to develop IHD or stroke when ET was initiated in comparison with men without such medications, thus no protective effect of these medications could be observed. HR for IHD for men on warfarin when ET was initiated was 2.96 (95% CI 1.03–8.51) compared with men without warfarin with a HR of 1.30 (95% CI 1.01–1.68; Table 4).
Table 4. Hazard ratios for ischemic heart disease and stroke after initiation of endocrine treatment according to use of heart medications
In this nationwide, population-based study of men with prostate cancer, we found a modestly increased risk of IHD and stroke in comparison with an age-matched control population when all men with prostate cancer were studied. The risk of IHD increased after initiation of ET but no such risk increase after initiation of treatment could be found for stroke. When subtypes of ET were considered, increased risks of IHD were associated with GnRH agonists with or without flare prophylaxis. Men on medication for IHD or stroke had a higher risk for these events after initiation of ET than men who received ET but were not on these drugs.
Previous studies show inconsistent results on the association between ET for prostate cancer and risk of IHD and stroke. Several recent investigations have reported an association between ET and CVD. In a study from the Veterans Healthcare Administration of men of all ages with local or regional prostate cancer (metastatic disease was excluded), Keating et al. found that the use of GnRH agonists was associated with increased risk of coronary heart disease and stroke.12 They found that GnRH agonists also increased the risk of stroke, but we saw no such increase when risk of stroke for all ET's was assessed, but when we specifically investigated GnRH agonists, we saw a small nonsignificant increase in stroke.
In a report by Saigal et al., the use of ET was associated with a 20% higher risk of serious cardiovascular morbidity when all stages of prostate cancer were studied.9 In a recent study also from PCBaSe Sweden, based on primary treatment given or planned within 6 months after date of diagnosis, risks of IHD and stroke were elevated with SIR of 1.32 (95% CI 1.27–1.36) for IHD and 1.26 (95% CI1.21–1.30) for stroke.13 Importantly, 40% of the men receiving ET had metastatic disease in this study. In contrast, Alibhai et al. performed a matched cohort study of men with prostate cancer comparing men treated with ET to men untreated with ET and found no excess risk of myocardial infarction or congestive heart failure and they found a decreased risk of stroke.10 In this study, men with metastatic disease were excluded and further stages of the disease were not given.
The studies by Keating et al, Saigal et al. and Alibhai et al. used men with prostate cancer as reference population and the previous study from PCBaSe Sweden had a high proportion of men with metastatic disease. Prostate cancer may per se increase risk of IHD and stroke, as indicated by our study, thus, the use of prostate cancer cases without ET as controls may lead to risk estimates difficult to interpret. Furthermore, there may be confounding by indication to treatment if analysis is restricted to men who receive primary ET because they may be at increased risk of IHD and stroke already prior to ET. Also, as our results show, it is important to have information on presence of IHD and stroke before initiation of ET.
Use of GnRH agonists was associated with significantly increased risk of IHD but use of other ETs was not. However, the ET most commonly used was GnRH agonists so the power to detect an increase in risk was much higher in this group. Furthermore, we do not know in which prostate cancer risk category these men were in when ET was initiated. If a large tumour burden is associated with increased risk of IHD and stroke, our results may partly reflect a selection bias as men with comparatively more advanced disease may more frequently have received GnRH agonists, whereas men with a lower tumour burden may more often have received AA.
We were able to control for use of drugs used for IHD and stroke. Men on ET and who were also on these drugs had a higher risk of IHD and stroke compared with men who were not on these drugs. Thus, in this setting, these medications appear as indicators of more severe cardiovascular problems and of predisposition to IHD and stroke following ET.
In the present study, ∼7% of the men in the low and 25% of the men in the intermediate prostate cancer risk category diagnosed during the observational period received ET within 6 months after diagnosis. As there are no data on a beneficial effect of ET in these risk categories, our data show an unwarranted use of ET for men in these categories in accordance with what was recently shown in CaPSURE.5
This study was nationwide, population-based with a population-based control cohort, and we had good information on what kind of ET the men received and when it was initiated both primarily and after an initial period of surveillance. We had individual data on medications and history of IHD and stroke. Our study design minimized the risk that the results would merely reflect a confounding by indication for treatment. Because of the high-capture rate (98%) in NPCR, we were able to include virtually all men with prostate cancer in Sweden who initiated ET for any cause.
Our study also had some limitations. The observation period was short with a mean follow-up time of 2 years. However, adverse events related to ET occur soon after initiation of treatment,9 and we had complete data on risk before and after initiation of ET.
The number of events for each category of ET was not sufficiently high to assess risk with high precision for each of these categories. We do not know in which stage these men were in when they received their ET as deferred treatment. Furthermore, men with prostate cancer may be more frequent users of health care than the background population and thus be more likely to be diagnosed with IHD and stroke. However, our data on IHD and stroke were based on inpatient admissions and a primary diagnosis of IHD or stroke requiring in-patient care are likely to be identified also among men with less frequent use of health care.
In conclusion, this large population-based study with population controls shows that regardless of treatment prostate cancer per se is associated with increased risk of IHD and stroke. Importantly, our study design allowed us to conclude that initiation of ET was linked to an increased risk of IHD but not stroke. Additional research is needed to elucidate which ET that is associated with least treatment-related morbidity and maximum benefit for each risk category of prostate cancer. It is also important to study how risk of IHD and stroke is related to the pre-existing morbidity and if it is possible to prevent the ET-related morbidity.
This project was made possible by the continuous work of the National Prostate Cancer Register of Sweden (NPCR) steering group: Pär Stattin chairman, Anders Widmark, Stefan Carlsson, Magnus Törnblom, Jan Adolfsson, Anna Bill-Axelson, Ove Andrén, David Robinson, Bill Pettersson, Jonas Hugosson, Jan-Erik Damber, Ola Bratt, Göran Ahlgren, Lars Egevad and Roy Ehrnström. The funding organizations had no influence in the design and conduct of the study, collection, management, analysis, and interpretation of data, and preparation, review, or approval of the manuscript.