Younger age and female gender are determinants of underestimated cardiovascular risk in rheumatoid arthritis patients: a prospective cohort study

Rheumatoid arthritis (RA) patients have a higher risk of developing cardiovascular diseases (CVD) than the general population [1]. It has been previously suggested that this increased risk of CVD is in part due to systemic inflammation seen in patients with RA, although excess in classical risk factors like smoking and obesity also play a role [2].

Current algorithms developed to predict CVD, such as the Systematic Coronary Risk Evaluation score (SCORE), underestimate the risk in RA patients, especially in those patients originally classified as having low or intermediate risk [3]. It is well known that older age leads to higher CV risk in the general population. Likewise, males are at greater risk compared to females. However, the results of a recent meta-analysis suggested that compared to the general population, younger RA patients bear the greatest relative risk of developing CV events, whereas older RA patients seem to have similar relative risks when compared to age-matched counterparts [4]. Furthermore, women with RA seem to be at a greater CV risk than those without RA [5]. Early menopause seems to be a predictor for RA, and in addition, early menopause in women with RA may lead to a higher CVD risk [6‐8]. Therefore, one can suggest that gender and age are likely to have a distinct impact on CVD risk in RA patients than in the general population.

Because the majority of risk calculators such as the SCORE risk charts use standard weighing of gender and age, their unadjusted use to patients with RA would yield inappropriate estimates of the CV risk. Because this hypothesis has been generated from results coming from a meta-analysis, we aimed in this study to confirm in a prospective follow-up inception cohort of RA patients whether age and gender are contributing to the underestimation, and thus misclassification, of CV risk in RA when current risk algorithms developed for the general population are used. Adjusting the impact of these parameters to better suit the RA population has not been part of the present investigation.

According to the results of this study, younger (< 55) RA patients and females with RA are likely to form two subgroups with the largest underestimation of CVD risk, if CV risk algorithms developed for the general population (SCORE) were used. The present results confirm those previously obtained in meta-analysis [4]. RA is associated with an increased risk of developing acute CV events. Following EULAR recommendations [9], patients receiving the diagnosis RA should be further monitored for the presence of risk factors for CVD. Risk management should be initiated by firstly determining the 10-year risk to develop CV events. Various risk calculators have been used so far, yet none of them appears to perform in RA patients as well as in the general population [3]. Attempts to modify these algorithms in order to better fit the RA population, including the addition of inflammatory parameters, disease activity markers, or even genetic markers, have yielded disappointing results [11‐13], although one recent algorithm might lead to some improvement in accuracy [14]. The results of the present study yield the hypothesis of modifying the weight of gender and/or age in current CV risk calculators to better fit RA patients. This hypothesis warrants further attention in the future, yet it does not constitute the aim of the present study.

In the current study, RA females accounted for the vast majority of CV events in the low-risk group (80%), which eventually represented over one third of all CV events in RA women. In comparison, the distribution of CV events in RA men was more in accordance with their predicted risk category, being the lowest (10%) in the low-risk group and highest (68%) in RA men initially assigned to the high-risk group. It seems therefore that at diagnosis, CVD risk is especially underestimated in females, and there are indeed several putative explanations for this. Firstly, previous studies have suggested that females and males with RA are not equally affected by inflammation with respect to CV risk factors. In line with this, we have previously shown that compared to healthy volunteers, the HDL-2 subfraction is declined in females with RA but not in males [5]. Also, it is hypothesized that due to systemic inflammation, females with RA reach menopause earlier than normally expected, which is augmenting their CV risk as compared to non-RA women of similar age [6, 15]. Finally, low-grade inflammation and a disturbed metabolism, as it is the case in diabetes mellitus, may augment the CV risk much more in women (relative risk 3 to 8 times higher) than in men (relative risk 2 to 3 times higher) as compared to the general population [16].

In our study, CV events occurred very often even in young RA patients and accounted eventually for almost 30% of all the registered CV events. Among them, two thirds have been initially assigned to the group of low CV risk at baseline. Age is a very strong predictor of future CV risk. According to the SCORE chart in The Netherlands, women under 55 and men under 50 would nearly always have a low CV risk, independent of the other CV risk factors such as dyslipidemia, hypertension, and smoking status. Consequently, according to the epidemiological data, individuals belonging to this risk category should not initiate drug therapies, as these would have a very limited impact on their CV risk, which is low already. Nevertheless, previous studies performed in RA populations have indicated the presence of atherosclerotic plaques even in younger patients and/or patients assigned to the low CV risk category [17, 18]. Accordingly, one third of RA women who had a low SCORE value and are aged > 49.5 years or/and have a total cholesterol concentration of > 5.4 mmol/l experience high-risk atherosclerosis and would therefore require intensive CVD risk management [19]. As carotid artery intima-media thickness (cIMT) and coronary artery calcification (CAC) score are both surrogate markers of underlying atherosclerosis and are both associated with increased CV risk, this suggests that these patients bear a higher CV risk than the one initially assigned using just the SCORE risk calculator. Of note, cIMT is likely to be more sensitive than CAC in RA patients in order to detect subclinical atherosclerosis associated with high CV risk [20]. Our results strengthen this hypothesis from an epidemiological perspective, as the number of the observed CV events was higher than predicted in the low-CV risk group of patients. These observations might also be explained from a pathophysiological perspective. Atherosclerosis is accelerated in RA patients [21], most probably due to dyslipidemia, which is widely present [5, 22], and due to inflammation during periods of active disease, which may contribute to plaque development and instability/rupture [23]. In other words, it seems that RA patients would need less time (thus would be younger) to reach a critical level of vulnerability of atherosclerotic lesions that would trigger an acute event, as compared to the general population, as suggested by the data of the present study and our previous meta-analysis [4]. We are aware of the low absolute risk in the lower CV risk category patients. Nevertheless, we consider that these results cannot be neglected and should lead to a better performance of risk predictions in the “low-risk” group. Because this is the group where most health gain may be achieved, as these patients have the highest chance of being most often undertreated for their traditional CV risk factors and therefore more prone to develop CVD.

Our study has a few limitations. Firstly, the number of patients included is limited as compared with studies of CVD in the general population. Yet, the cohort used is well and prospectively documented, and also one of the largest single-center cohorts of its kind. Secondly, most of the patients investigated were included before CVD risk management had been implemented, limiting the use of absolute risk predictions. Thirdly, our study investigated if current algorithms have a proper calibration in the subgroups tested (i.e., gender and age-specific subgroups). No statistical comparisons of calibrations between the groups have been made. Finally, data on the use of non-steroidal anti-inflammatory drugs (NSAIDs) and corticosteroids are missing. This might have influenced the results of the study, though some recent reports suggest that the use of NSAIDs does not increase CV incidence in the RA population as compared to osteoarthritis (OA) [24].

In conclusion, the results of our study suggest that the incidence of CVD among women and young RA patients initially assigned to the low-risk category is higher than predicted using current algorithms. Consequently, the CV risk in these subgroups seems underestimated. Whether modifying the weight for the female gender and/or younger age in the risk algorithms would result in better CV risk predictions in RA remains a subject to be investigated in future studies. Alternatively, other strategies (e.g., biomarkers, cIMT, or CAC measurements) aiming at the same goal could be envisaged in order to improve CV risk management in patients with RA and/or other chronic inflammatory conditions.