Impact of Sex on Cardiovascular Outcome in Patients at High Cardiovascular RiskClinical Perspective
Analysis of the Telmisartan Randomized Assessment Study in ACE-Intolerant Subjects With Cardiovascular Disease (TRANSCEND) and the Ongoing Telmisartan Alone and in Combination With Ramipril Global End Point Trial (ONTARGET)
Background—Epidemiological data suggest that sex independently contributes to cardiovascular risk. Clinical trials are often hampered by the enrollment of few female patients.
Methods and Results—The Ongoing Telmisartan Alone and in Combination With Ramipril Global End Point Trial (ONTARGET) and the parallel Telmisartan Randomized Assessment Study in ACE Intolerant Subjects With Cardiovascular Disease (TRANSCEND) included a large proportion of female patients (9378 female versus 22 168 male patients). Differences in male and female patients enrolled in ONTARGET/TRANSCEND were analyzed for the primary 4-fold end point (composite of cardiovascular death, myocardial infarction, stroke, or admission to hospital for heart failure), a secondary 3-fold end point (cardiovascular death, myocardial infarction, stroke), and individual components of the primary composite. Baseline characteristics included age, ethnicity, body mass index, physical activity, tobacco use, alcohol consumption, formal education, clinical diagnosis for study entry, patient history, and concomitant medication. Patients were followed up until death or the end of the study (median, 56 months). Compared with male patients, female patients had a 19% significantly lower risk for the 4-fold end point and 21% for the 3-fold end point (after adjustment for study, treatment, and the above baseline values). Similarly, the adjusted risk for cardiovascular death (17%) and myocardial infarction (22%), but not for stroke and hospitalization for heart failure, was also significantly lower in women. Diabetic female patients were characterized by a higher risk for acute myocardial infarction compared with diabetic male patients, whereas alcohol consumption resulted in significantly lower risk in women.
Conclusions—In our analysis made up of 70.3% male and 29.7% female patients, an ≈20% lower risk for the combined cardiovascular end points in female patients was observed despite treatment with cardioprotective agents. This difference was driven primarily by a significantly lower incidence of myocardial infarction. Thus, we demonstrate in a large interventional trial that sex greatly affects the occurrence of cardiovascular events in patients with vascular disease or high-risk diabetes mellitus.
Sex has been observed to contribute independently to cardiovascular (CV) risk, but the underlying mechanisms are poorly understood. Epidemiological data from the Framingham Heart Study show later onset and lower prevalence of CV disease in women.1 In addition, results from 37 World Health Organization's Monitoring of Trends and Determinants in Cardiovascular Diseases (MONICA) project populations and CV mortality data within Europe demonstrate lower rates of coronary heart disease mortality in women than in men.2,3 Data on sex-related outcomes based on controlled clinical studies with contemporary drug treatments are sparse, mainly because of unequal proportions of male and female patients being enrolled in large-scale CV clinical outcome trials.
Editorial see p 913
Clinical Perspective on p 941
The Ongoing Telmisartan Alone and in Combination With Ramipril Global End Point Trial (ONTARGET; n=25 620)4 and Telmisartan Randomized Assessment Study in ACE Intolerant Subjects With Cardiovascular Disease (TRANSCEND; n=5926 patients)5 recruited >30 000 patients ≥55 years of age with high-risk CV disease or diabetics with signs of end-organ damage. These patients, the largest study group of patients at high CV risk so far, were on guideline-recommended protective treatments. We hypothesized that sex independently affects CV risk in this cohort. Clinical trials, particularly in CV fields, generally face the difficulty of very high proportions of male patients. ONTARGET recruited 26.7% and TRANSCEND recruited 43.0% women. These high percentages of female patients allowed us to attempt to analyze the primary and secondary composite outcomes in male versus female patients (regardless of pharmacological intervention) to examine drug-independent sex-specific outcomes.
The design of the ONTARGET and TRANSCEND trials, including patient eligibility criteria, has been described elsewhere.6 Briefly, the studies included men and women ≥55 years of age who were tolerant of an angiotensin-converting enzyme inhibitor (ONTARGET) or were intolerant of an angiotensin-converting enzyme inhibitor (TRANSCEND) and at high risk of a major CV event on the basis of one or more of the following: history of coronary artery disease, peripheral vascular disease, stroke or transient ischemic attack, or diabetes mellitus with end-organ complications.
Intolerance was defined as previous angiotensin-converting enzyme inhibitor discontinuation by a physician for intolerance with a specific documented cause. The studies were performed in 733 and 630 centers for ONTARGET and TRANSCEND, respectively, in 40 countries. After randomization, patients were followed up at 6 weeks, 6 months, and then every 6 months. The study protocol was approved by the Research Ethics Board of each participating center. All subjects provided written informed consent. The study was performed according to Good Clinical Practice/US Food and Drug Administration rules with rigorous monitoring and adjudication of all end points.
In ONTARGET and TRANSCEND, 25 620 patients and 5926 patients, respectively, underwent randomization. Baseline characteristics were obtained by clinical assessment. The primary outcome was the composite of CV death, myocardial infarction (MI), stroke, or hospitalization for heart failure (HF). Secondary outcomes were the composite outcome of CV death, MI, or stroke (which was the primary outcome of the Heart Outcomes Prevention Evaluation [HOPE] trial).7 Other secondary outcomes included separate counts of nonfatal MI, stroke, and hospitalization for HF. Of the 31 546 patients initially enrolled, 5730 (18.2%) stopped study medication by the fourth year.
Baseline categories were compared for sex differences by means of the χ2 test. For all end points, time to first event was compared between sexes by Cox regression. Analyses were adjusted for study (ONTARGET versus TRANSCEND) and treatment. In addition, in a separate analysis, further adjustment was made for age, body mass index, physical activity, smoking, alcohol consumption, education, ethnicity, previous MI, previous stroke, history of hypertension, presence of diabetes mellitus, clinical diagnosis at study entry, and concomitant medication at baseline. Patients were categorized into 5 age groups: <60, 60 to <65, 65 to <70, 70 to <75, and ≥75 years of age. These cutoffs represent well the age range of patients in the 2 studies; they were all at least 55 years of age. In contrast to inclusion of age as a continuous variable, the categorical approach does not require the relationship between risk and age to be linear and leaves more flexibility in modeling the data. However, for the estimation of the average time delay of events in female compared with male patients, age was included as a continuous variable. In this model, the time delay can be calculated as the ratio of the coefficients for sex and age on the log scale. This value is equivalent to the time span in years a female patient has to be older to have the same hazard as a male patient with otherwise identical risk factors; approximate confidence intervals were calculated with the delta method.
The associations between sex and age and between sex and other risk factors were explored by use of the Cox model including the respective risk factor and interaction with sex. We also explored potential sex differences in patients with an MI and/or stroke during the trial. In these patients, we compared the event rates of CV death, further MI or stroke, or subsequent hospitalization for HF using the same methodology as described above.
Baseline Characteristics of Male Versus Female Patients
Baseline characteristics are shown in Table 1. Female patients were older and were characterized by less physical activity. Men had a higher prevalence of current or former tobacco use and a higher prevalence of alcohol consumption. Although coronary artery disease was more frequent in men than in women, women exhibited a higher prevalence of previous stroke, high-risk diabetes mellitus, and hypertension.
Outcome in Male Versus Female Patients
The annualized rates of the primary outcome, the 3-fold end point (as for HOPE), and single end points are shown in Table 2.
After adjustment for age, ethnicity, body mass index, physical activity, formal education, smoking, alcohol intake, prior MI, prior stroke, history of hypertension, diabetes mellitus, clinical diagnosis at study entry, and concomitant medication at baseline, women had a 19% lower risk for the 4-fold end point (Table 2 and Figure 1A), a 21% lower risk for the 3-fold end point (Table 2 and Figure 1B), and a 22% lower risk for MI in (Table 2 and Figure 1C). In addition, for CV death, there was also a significantly lower risk in women than in men by 17%. In contrast, the risks for both stroke and hospitalization for HF were not different (Table 2). As also depicted in Figure 1A through 1C, further subanalyses of the 4-fold-, 3-fold-, and MI outcomes with regard to sex and age are shown in Table I in the online-only Data Supplement. Although the overall incidence for all end points was significantly lower in women, the sex difference was consistent across age groups with no indication of any sex-by-age interaction (Table I in the online-only Data Supplement). Overall, from the confounders included in the model, age, diabetes mellitus, history of MI or stroke, physical activity, and smoking (depending on the actual end point) exhibited the biggest predictive value. Because the risk in women for the occurrence of both the 4-fold and 3-fold end points was driven mainly by the lower risk for MI, we performed additional analyses concerning this outcome. Thus, patients were further subdivided according to the presence of previous MI. As shown in Table 3 and Figure 1D, previous MI clearly enhanced the risk for experiencing an additional MI in men and women to a similar extent (dashed versus solid lines in Figure 1D). In addition, in patients with previous MI, women had a lower risk than men throughout all age groups (Figure 1D).
Time Delay of Events in Female Compared With Male Patients
The difference in the risk of CV events between female and male patients can be quantified in terms of the average time delay of a specific event for a female patient compared with a male patient with the same risk profile, ie, otherwise identical confounders. The calculated delay for the combined 4-fold end point was 5.7 years and for the 3-fold end point was 6.3 years. Among the analyzed end points, the time delay for experiencing an MI was greatest at 10.7 years (Table II in the online-only Data Supplement).
Sex Differences in Risk Factor Associations With Acute MI
Subanalyses were performed for risk factor associations with the incidence of acute MI (Figure 2). Hypertension, previous MI, stroke, current or former smoking, and physical activity did not affect the incidence of acute MI between female and male patients differently (no indication for interaction between sex and risk factor). In contrast, diabetes mellitus and alcohol consumption affected the outcome of MI between male and female patients differently. In diabetic patients, women were at a higher risk for an acute MI. Conversely, although alcohol consumption was associated with reduced risk for MI in both sexes, this reduction was significantly greater in women (Figure 2).
Tracking Events After an MI or a Stroke
Further analyses were performed to clarify whether an acute MI and/or a stroke predicts the risk for subsequent events. As shown in Table 4, patients experiencing an MI during the course of the studies had higher rates for a second MI than for a stroke; nevertheless, a substantial portion of male and female patients with a previous MI had a subsequent stroke (annualized rates, 1.60% and 2.38%, respectively). Conversely, patients with a stroke had higher rates for a subsequent stroke but also a substantial number of MI (annualized rates, 1.90% for men and 1.92% for women). However, after full adjustment for all potential confounders, there were no sex-specific differences in terms of the annualized rates between patients with primary MI.
Although CV death occurred at comparable rates in patients with MI or stroke, the incidence for HF hospitalization was clearly higher in both female and male patients with MI than in those with stroke (Table 4). Overall, the high annualized rates of 17% to 25% for the composite end point in either patient group (previous MI or stroke) demonstrate the general interdependence of CV end points.
Differences in clinical diagnosis and patient history at study entry were reflected by differences in medication (Table 5). Female patients were treated more frequently with antihypertensives and antidiabetics, whereas statins/fibrates and anticoagulants were prescribed more frequently in male patients. In addition, the number of comedications, including antihypertensives, anticoagulants, antidiabetics, and statins/fibrates, significantly differed between male and female patients at both study entry and the final visit with or without the occurrence of a primary event. The occurrence of the primary event led to higher proportions of both female and male patients receiving 4 concomitant medications (Table 5). Except for anticoagulants in male patients, compared with study entry, medications, particularly antidiabetics and statins/fibrates, were given more frequently at the final visit in those experiencing the primary event.
In this post hoc analysis, we investigated differences in CV outcome related to sex in a nonepidemiological controlled study cohort of >30 000 high-risk patients on optimal standard treatment.
We observed a better outcome in female than in male patients that was robust after adjustment for various confounders (age, ethnicity, body mass index, physical activity, formal education, smoking, alcohol drinking, prior MI, prior stroke, history of hypertension, history of diabetes mellitus, and concomitant medication at baseline). There were significantly lower risks in female patients for the 4-fold end point, the 3-fold end point, and MI of 19%, 21%, and 22%, respectively, which are also directly evident by significantly lower annualized incidence rates in female patients. Our data further indicate that female patients develop CV end points ≈5 to 10 years later than male patients with smaller differences at later ages. Male patients recruited at later ages might be likely to be at lower risk compared with earlier ages because of attrition of higher-risk male patients.
At study entry, the number of comedications differed significantly between male and female patients. This finding does not necessarily represent undertreatment in women because the type of medications reflected the patient history and disease at the time of enrollment. In particular, the higher prevalence of hypertension and diabetes mellitus in female patients was reflected by a higher percentage of drug treatment with antihypertensives and antidiabetics. In contrast, male patients had significantly higher prevalence of MI in their history, which likely resulted in significantly more male patients being treated with statins/fibrates and anticoagulants at study entry. The clinical outcome of patients during the trials largely affected the given medications. The occurrence of the primary event was followed by higher frequencies of antihypertensives, antidiabetics, and statins/fibrates in both male and female patients.
Regardless of sex, age contributes significantly to CV risk. Thus, in both male and female patients, higher age was associated with increased risk for the 4-fold end point, the 3-fold end point, and the individual secondary outcomes of CV death, MI, stroke, and hospitalization for heart failure. CV end points occurred ≈5 years later in women, whereas MI had a ≈10-year delay. Thus, the beneficial outcome in women after adjustment for various factors was largest for MI. The reason for the female-related benefit in particular with regard to MI, but less for stroke, remains unclear but is in line with previous reports in epidemiological investigations on Swedish and Finnish populations.8 In contrast, after adjustment, hospitalization for HF was not different between male and female patients. The recent EuroHeart Failure Survey II also did not demonstrate sex-associated differences.9 A pooled analysis of 5 randomized clinical trials on sex-related differences in origin and outcome (all-cause mortality and death or all-cause hospitalization) in patients with HF likewise did not produce differences based on sex.9 Although adjustment only for study (ONTARGET versus TRANSCEND) and treatment showed a higher incidence of HF hospitalization only in female patients, this sex-related difference disappeared after additional adjustment for age, body mass index, physical activity, smoking, alcohol consumption, education, ethnicity, previous MI, previous stroke, hypertension, diabetes mellitus, clinical diagnosis at study entry, and concomitant medication at baseline. Subanalyses of defined risk factor associations revealed that hypertension, former MI, stroke, smoking, and physical activity are not associated with different risks between the sexes.
However, diabetes mellitus and alcohol consumption showed a sex-associated impact on incidence of acute MI. In agreement with the INTERHEART study,10 a standardized case-control study of acute MI, diabetes mellitus was a stronger risk factor for MI in women compared with men. Morrish et al11 reported that incidence rates of macrovascular diseases were similar between diabetic men and women except for MI in those patients with non–insulin-dependent diabetes mellitus, among whom male patients were characterized by a significantly higher incidence rate. Although it is well acknowledged that diabetes mellitus in general markedly contributes to the risk of CV disease, this risk has previously been found to be greater in women than in men.12 This is in congruence with our recent data. However, Lundberg et al12 investigated diabetes mellitus as a risk factor for acute MI from a population perspective in a region with high risk of CV disease using a population-based acute MI registry with a total of 2432 men and women 35 to 64 years of age. Thus, our study group significantly differs from this cohort by the higher number of investigated patients and the use of a combined data set of 2 large-scale controlled studies with patients on optimal standard therapy for individual cardiometabolic diseases, including diabetes mellitus. Moreover, as in the standardized case-control INTERHEART, we observed a significantly stronger impact of alcohol on reducing the risk for acute MI in women than in men. The underlying mechanism for this finding has not been established so far. Although it is generally agreed that light to moderate alcohol consumption is inversely correlated with CV risk,13,14 our findings should not be used to promote alcohol consumption as a CV preventive strategy because alcohol has also several other adverse health consequences such as cancer and injuries.15,16
Our data are nonrandomized observations from 2 clinical trials, albeit they comprised >30 000 patients. Although epidemiological analyses frequently cover periods of decades, our observation was limited to 5 years. An additional limitation is that only high-risk patients were enrolled in these 2 clinical trials.
With a total of 31 546 patients (men, 70.3%; women, 29.7%), with a median follow-up of 56 months, and after adjustment for potential confounders, we report an ≈20% lower risk for combined CV end points in female patients, which was driven mainly by significantly lower incidence of MI. Thus, we demonstrate that sex largely affects the occurrence of CV events in patients with vascular disease or high-risk diabetes mellitus despite cardioprotective therapy. Previously, the time difference between male and female patients in terms of the occurrence of a CV end point has been estimated to be between 5 and 10 years. This was true for the combined end points but not for MI. Interestingly, despite their lower risk in the total population, diabetic women were at greater risk than diabetic man of developing MI. In contrast, hospitalization for chronic heart failure was, after adjustment, not different between men and women in all age groups.
Taken together, our data clearly point out significant sex-related differences in CV outcome in patients with vascular disease or high-risk diabetes mellitus. Despite appropriately individualized cardioprotective therapy, female patients are at lower net CV risk than male patients and develop CV end points ≈5 to 10 years later.
Sources of Funding
The trials were supported by Boehringer Ingelheim. Boehringer Ingelheim had no role in the design, conduct, or analysis of the study or in the decision to submit the manuscript for publication.
Drs Böhm, Schmieder, and Unger have received consulting, lecture fees, and research grants from Boehringer Ingelheim and other companies manufacturing angiotensin receptor blockers, angiotensin-converting enzyme inhibitors, and other blood pressure–lowering drugs. Dr Koon has received honoraria from Boehringer Ingelheim. Dr Schumacher is an employee of Boehringer Ingelheim working as a statistician. Dr Sleight has received grant support for the ISIS1 to ISIS4, HPS, SEARCH, HOPE, and ONTARGET/TRANSCEND trials from BHF, UK MRC, Canadian MRC, Ontario Heart and Stroke Foundation, AstraZeneca, Aventis, Boehringer Ingelheim, BMS, GSK, Monarch, MSD, National Vitamin E Association, and Roche; speaker fees or Data and Safety Monitoring Board fees from Abbott, AstraZeneca, Aventis, Bayer, Boehringer Ingelheim, Boehringer Mannheim, BMS, Genentech, GSK, Knoll, Medscape, Menarini, Merck, Monarch, MSD, Novartis, Pfizer, Pharmacia, Sanofi, and Servier; and speaker bureau fees or stock/stock options from NIL. Dr Yusuf has received grants, honoraria, and consulting fees from Boehringer Ingelheim, Sanofi, BMS, Servier, and GSK for studies related to CVD prevention. Dr Kappert reports no conflicts.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.111.086660/-/DC1.
- Received December 29, 2011.
- Accepted May 30, 2012.
- © 2012 American Heart Association, Inc.
- D'Agostino RB Sr.,
- Vasan RS,
- Pencina MJ,
- Wolf PA,
- Cobain M,
- Massaro JM,
- Kannel WB
- Tunstall-Pedoe H,
- Kuulasmaa K,
- Mahonen M,
- Tolonen H,
- Ruokokoski E,
- Amouyel P
- Muller-Nordhorn J,
- Binting S,
- Roll S,
- Willich SN
- Yusuf S,
- Teo K,
- Anderson C,
- Pogue J,
- Dyal L,
- Copland I,
- Schumacher H,
- Dagenais G,
- Sleight P
- Teo K,
- Yusuf S,
- Sleight P,
- Anderson C,
- Mookadam F,
- Ramos B,
- Hilbrich L,
- Pogue J,
- Schumacher H
- Nieminen MS,
- Harjola VP,
- Hochadel M,
- Drexler H,
- Komajda M,
- Brutsaert D,
- Dickstein K,
- Ponikowski P,
- Tavazzi L,
- Follath F,
- Lopez-Sendon JL
- Anand SS,
- Islam S,
- Rosengren A,
- Franzosi MG,
- Steyn K,
- Yusufali AH,
- Keltai M,
- Diaz R,
- Rangarajan S,
- Yusuf S
- Ronksley PE,
- Brien SE,
- Turner BJ,
- Mukamal KJ,
- Ghali WA
- Brien SE,
- Ronksley PE,
- Turner BJ,
- Mukamal KJ,
- Ghali WA
Although uncontrolled epidemiological analyses suggest that sex independently contributes to cardiovascular risk, there are limited controlled large-scale clinical studies on sex-specific cardiovascular outcome. Moreover, clinical trials are often hampered by the enrollment of few female patients. In our analyses based on the Ongoing Telmisartan Alone and in Combination With Ramipril Global End Point Trial (ONTARGET) and the parallel Telmisartan Randomized Assessment Study in ACE Intolerant Subjects With Cardiovascular Disease (TRANSCEND) with patients ≥55 years of age, we included a large proportion of female patients (29.7%). In these large interventional trials, we observed that sex greatly affects the occurrence of cardiovascular events in patients with vascular disease or high-risk diabetes mellitus. This ≈20% lower risk for combined cardiovascular events in female patients, who were under optimal, guideline-recommended pharmacological treatment, is driven mainly by a significantly lower incidence of acute myocardial infarction. However, it should be pointed out that diabetic female patients were characterized by a higher risk for acute myocardial infarction compared with diabetic male patients, whereas alcohol consumption resulted in significantly lower risk in women. Sex- and age-dependent analyses demonstrate that female patients develop cardiovascular end points ≈5 to 10 years later despite appropriately individualized cardioprotective therapy in both female and male patients. In these controlled cardiovascular end-point trials, we clearly point out significant sex-related differences, with women being at lower net cardiovascular risk than men except when diabetic. These differences should be acknowledged in daily clinical routine when female and male patients at high cardiovascular risk are treated.