Disease activity influences the reclassification of rheumatoid arthritis into very high cardiovascular risk

This was a cross-sectional study that included 1279 patients with RA. They were all 18 years or older and were included in the study if they fulfilled the 2010 ACR/EULAR classification criteria for RA [12]. For the purpose of inclusion in the present study, RA disease duration needed to be ≥ 1 year. Patients taking glucocorticoids were included only if they were taking an equivalent dose ≤ 10 mg/day of prednisone. However, patients were excluded if they had diabetes, a history of cancer or any other chronic disease, evidence of active infection, or a glomerular filtration rate < 60 ml/min/1.73 m². The study protocol was approved by the Institutional Review Committees at Hospital Universitario de Canarias, Hospital Doctor Negrín, and Hospital Marqués de Valdecilla, Spain (approval number 17/2012). All subjects provided informed written consent.

The subjects completed a CV risk factor and medication use questionnaire and underwent a physical examination. Weight, height, body mass index, abdominal circumference, and systolic and diastolic blood pressure (measured with the participant in a supine position) were assessed under standardized conditions. Information regarding smoking status (current smoker versus non-smoker) and hypertension was obtained from the questionnaire. Medical records were reviewed to ascertain specific diagnoses and medications. Dyslipidemia was defined if one of the following was present: total cholesterol > 200 mg/dl, triglycerides > 150 mg/dl, HDL cholesterol < 40 in men or < 50 mg/dl in women, or LDL cholesterol > 130 mg/dl. Cholesterol, triglycerides, and HDL cholesterol were measured using the enzymatic colorimetric assay. LDL cholesterol was calculated using the Friedewald formula. A standard technique was used to measure the erythrocyte sedimentation rate (ESR) and high-sensitivity C-reactive protein (CRP). Disease activity in patients with RA was measured using the Disease Activity Score (DAS28) in 28 joints [13], the Clinical Disease Activity Index (CDAI) [14], and the Simple Disease Activity Index (SDAI) [15]. Patients with RA were defined as being in clinical remission (DAS28 < 2.6) or having low (DAS28 in the range of 2.6 to 3.2), moderate (DAS28 of > 3.2 to 5.1), or high disease activity (DAS28 > 5.1) as previously described [16]. Function, pain, and patient global estimate of status was measured through the Routine Assessment of Patient Index (RAPID) [17], and disease disability through the Health Assessment Questionnaire (HAQ) score [18].

A carotid ultrasound examination was used to assess cIMT in the common carotid artery and to detect focal plaques in the extracranial carotid tree in patients with RA [19]. A commercially available scanner, the Esaote Mylab 70 (Genoa, Italy), equipped with a 7–12-MHz linear transducer and an automated software-guided radiofrequency technique, Quality Intima Media Thickness in real-time (QIMT, Esaote, Maastricht, Holland), was used for this purpose. As previously reported [19], based on the Mannheim consensus, plaque criteria in the accessible extracranial carotid tree (common carotid artery, bulb, and internal carotid artery) were defined as follows: a focal protrusion in the lumen measuring at least cIMT > 1.5 mm, a protrusion at least 50% greater than the surrounding cIMT, or arterial lumen encroaching > 0.5 mm [20].

Demographic and clinical characteristics are shown as frequencies for binary variables.

Continuous variables data are expressed as mean ± standard deviation (SD) or as a median and interquartile range (IQR) for non-normally distributed variables. Patients and controls with carotid plaques based on ultrasound assessment were reclassified into very high-SCORE risk category. Subjects without plaques were maintained in their original SCORE category. cIMT was not used to determine reclassification because according to current guidelines cIMT is not considered an unequivocal CVD on imaging [10]. Univariate differences between reclassified and non-reclassified patients were assessed through Student’s t, Mann-Whitney U, chi-square, or Fisher’s exact tests according to normal distribution or the number of subjects. Logistic regression analysis adjusted for the variables with a p value < 0.20 in the univariate analysis was performed to assess the relationship between RA disease-related data and the presence of reclassification. An all-sets logistic regression model was constructed to describe the most parsimonious combination of risk reclassification predictors according to Akaike Information Criteria, Schwarz Bayesian Criterion, area under the curve (AUC), and Hosmer-Lemeshow goodness-of-fit statistics. For characteristics associated with reclassification and that were included in the predictive model, sensitivity versus false-positive frequency (1-specificity) was analyzed using receiver-operating characteristic curves (ROC). To determine the optimal cutoff value of baseline characteristics in predicting reclassification, we calculated the Youden index using the following formula: sensitivity + specificity − 1, with the maximum obtained value corresponding to the optimal cutoff point. To estimate the increase in prediction accuracy between models, we used logistic regression to calculate the ROC curves and the area under the ROC curves (AUC). The SCORE AUC was thus considered the reference and was compared to the other model when adding RA-related data (age, dyslipidemia, hypertension, and DAS28 score). A comparison of ROC curves to test the statistical significance of the difference between the areas under two dependent ROC curves (derived from the same cases) was conducted using the method of DeLong et al. [21]. Reclassification differences between models were studied through the net reclassification index (NRI) and integrated discrimination improvement (IDI) as previously described [22]. Similarly, calibration of the models was calculated using the Hosmer-Lemeshow goodness-of-fit test by grouping individuals on the basis of deciles [23, 24]. All the analyses used a 5% two-sided significance level and were performed using SPSS software, version 24 (IBM Corp.). A p value < 0.05 was considered statistically significant.

A total of 1279 patients with RA were included in this study. Demographic and disease-related characteristics of the participants are shown in Table 1. Mean ± SD age was 58 ± 13 years and 72% of the patients were female. Current smoking, obesity, dyslipidemia, and hypertension were present in respectively 25, 28, 48, and 37% of the patients. Disease duration was 5 (IQR 2–12) years. RA patients had moderately active disease as shown by the mean DAS28 (3.09 ± 1.49). Half of the patients (50%) were taking prednisone (the median dose of those patients on prednisone was 5 [IQR 4–6] mg/day at the time of the study). Fifty-nine percent of patients were found to be positive for rheumatoid factor and 53% for anti-citrullinated protein antibody (ACPA). Besides, 77% were on disease-modifying antirheumatic drugs and 13% were receiving anti-TNF-alpha therapy. Regarding carotid ultrasound assessment, 53% of the patients with RA had carotid plaques and the average cIMT in patients was 690 ± 140 μm. Additional information regarding RA patient characteristics is shown in Table 1.

Following SCORE risk chart stratification, 533 (42%) patients were included in the low risk category. Contrary, 487 (39%), 139 (11%), and 103 (8%) of the patients were included in the moderate-, high-, and very high-risk categories, respectively. Following carotid ultrasound assessment, 54% of the patients were found to have carotid plaque and consequently were reclassified into the very high SCORE category risk. Specifically, 26% of the patients in the low category, 65% of those in the moderate risk category, and 86% of the ones in the high category moved into the very high SCORE risk category (Table 2).

Several differences were observed in the recorded characteristics of patients with RA who were reclassified following the carotid ultrasound assessment and those who were not reclassified (Table 3). Reclassified patients were older (62 ± 9 vs. 54 ± 14 years, p = 0.000), and more commonly had hypertension (47% vs. 28%, p = 0.000) and dyslipidemia (57% vs. 40%, p = 0.000).

As expected, patients with RA who were reclassified following a carotid ultrasound had greater cIMT than those who were not reclassified (740 ± 140 vs. 650 ± 140 μm, p = 0.000). However, CRP values did not reveal any differences between reclassified and non-reclassified patients.

Regarding RA-related features, some differences were observed between these two groups of patients (Table 3). Although disease duration was not found to be more frequent in the reclassified patients, both rheumatoid factor and ACPA were more commonly found in the reclassified patients compared to those non-reclassified. The odds ratio (OR) of rheumatoid factor and ACPA for reclassification were, respectively, 1.58 (95% CI 1.19–2.11) and 1.46 (95% CI 1.09–1.94). These associations were observed after multivariable analysis adjusting for age, dyslipidemia, hypertension, and statins and aspirin intake. Additionally, the DAS28-ESR score was statistically significantly associated with reclassification after fully multivariable analysis (OR 1.09 [95% CI 1.02–1.19], p = 0.028). It was observed a similar trend for DAS28-CRP and CDAI scores although, in these cases, statistical significance was not reached. Other associations that were significant in the univariable analysis, like those related to the presence of erosions and current use of NSAIDs, did not remain significant after adjustment for covariables (Table 3).

A predictive model was constructed only for those patients with RA who had been included in the low and moderate risk SCORE categories (< 5%) prior to a carotid ultrasound assessment (Table 4). These variables conjointly represented the most parsimonious model capable of predicting the reclassification of patients with RA into the very high CV risk category (Table 4): age, the presence of hypertension and dyslipidemia, and DAS28-ESR score. Moreover, an age exceeding 54 years and a DAS28-ESR score ≥ 2.6 were the cutoffs among the continuous variables that reached the highest Youden indices.

Table 5 represents the discrimination, reclassification, and calibration assessment of the model using clinical data (age, dyslipidemia, hypertension, and DAS28-ESR score) versus the reference SCORE model. The SCORE, which included traditional CV risk factors, showed a statistically significant discrimination of reclassification (AUC 0.766, 95% CI 0.737–0.794). The AUC of the model, which contained SCORE plus age, dyslipidemia, hypertension, and DAS28-ESR, was found to have higher discrimination (0.775, 95% CI 0.747–0.803) although statistical significance was not reached (p = 0.23). The addition of clinically related data represented a significant change in NRI versus the SCORE reference model (NRI 0.07, 95% CI 0.01–0.12, p = 0.018). Similarly, IDI was significantly higher in this model compared to that of the SCORE reference (0.07, 95% CI 0.05–0.08, p = 0.000). Model calibration (through a Hosmer-Lemeshow chi-square test) was found to be optimal in the final model (p = 0.80).