You are here

  • Home
  • Archive
  • Volume 9, Issue 4
  • Cardiovascular risk assessment in patients with antiphospholipid syndrome: a cross-sectional performance analysis of nine clinical risk prediction tools

Article Text

Article menu

Download PDFPDF

Autoimmunity
Original research
Cardiovascular risk assessment in patients with antiphospholipid syndrome: a cross-sectional performance analysis of nine clinical risk prediction tools
  1. George C Drosos1,
  2. George Konstantonis1,
  3. Petros P Sfikakis1,2 and
  4. Maria G Tektonidou1,2
  1. 1Cardiovascular Risk Research Laboratory, First Department of Propaedeutic Internal Medicine, National and Kapodistrian University of Athens School of Medicine, Athens, Greece
  2. 2Rheumatology Unit, First Department of Propaedeutic Internal Medicine and Joint Academic Rheumatology Program, National and Kapodistrian University of Athens School of Medicine, Athens, Greece
  1. Correspondence to Professor Maria G Tektonidou; mtektonidou{at}gmail.com

Abstract

Objectives This study aimed to assess the performance of cardiovascular risk (CVR) prediction models reported by European Alliance of Associations for Rheumatology and European Society of Cardiology recommendations to identify high-atherosclerotic CVR (ASCVR) patients with antiphospholipid syndrome (APS).

Methods Six models predicting the risk of a first cardiovascular disease event (first-CVD) (Systematic Coronary Risk Evaluation (SCORE); modified-SCORE; Framingham risk score; Pooled Cohorts Risk Equation; Prospective Cardiovascular Münster calculator; Globorisk), three risk prediction models for patients with a history of prior arterial events (recurrent-CVD) (adjusted Global APS Score (aGAPSS); aGAPSSCVD; Secondary Manifestations of Arterial Disease (SMART)) and carotid/femoral artery vascular ultrasound (VUS) were used to assess ASCVR in 121 APS patients (mean age: 45.8±11.8 years; women: 68.6%). We cross-sectionally examined the calibration, discrimination and classification accuracy of all prediction models to identify high ASCVR due to VUS-detected atherosclerotic plaques, and risk reclassification of patients classified as non high-risk according to first-CVD/recurrent-CVD tools to actual high risk based on VUS.

Results Spiegelhalter’s z-test p values 0.47–0.57, area under the receiver-operating characteristics curve (AUROC) 0.56–0.75 and Matthews correlation coefficient (MCC) 0.01–0.35 indicated moderate calibration, poor-to-acceptable discrimination and negligible-to-moderate classification accuracy, respectively, for all risk models. Among recurrent-CVD tools, SMART and aGAPSSCVD (for non-triple antiphospholipid antibody-positive patients) performed better (z/AUROC/MCC: 0.47/0.64/0.29 and 0.52/0.69/0.29, respectively) than aGAPSS. VUS reclassified 34.2%–47.9% and 40.5%–52.6% of patients classified as non-high-ASCVR by first-CVD and recurrent-CVD prediction models, respectively. In patients aged 40–54 years, >40% VUS-guided reclassification was observed for first-CVD risk tools and >50% for recurrent-CVD prediction models.

Conclusion Clinical CVR prediction tools underestimate actual high ASCVR in APS. VUS may help to improve CVR assessment and optimal risk factor management.

  • Antiphospholipid Syndrome
  • Cardiovascular Disease
  • Atherosclerosis
  • Ultrasonography

Data availability statement

Data are available on reasonable request. The data underlying this article will be shared on reasonable request to the corresponding author.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/rmdopen-2023-003601

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Patients with antiphospholipid syndrome (APS) are at high cardiovascular risk (CVR), including accelerated atherosclerosis. To date, the performance of first-cardiovascular disease (CVD) and recurrent-CVD event risk prediction models reported by European Alliance of Associations for Rheumatology and the European Society of Cardiology (ESC) recommendations to assess CVR has yet to be examined in patients with APS.

    WHAT THIS STUDY ADDS

    • Most of the examined nine risk prediction tools showed inadequate performance in calibration, discrimination and classification accuracy to identify actual high CVR as documented by vascular ultrasound (VUS)-detected atherosclerotic plaques (a risk assessment modality endorsed by the ESC), even in younger patients and those without APS-related high-CVR factors such as triple antiphospholipid antibody positivity.

    • Risk reclassification of patients to high risk due to VUS-detected plaques amounted to up to 50% of non-high-risk cases according to clinical risk prediction tools.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • VUS in addition to the use of risk prediction tools may assist clinicians to optimise CVR assessment in apparently non-high-risk patients with APS.

    Introduction

    Antiphospholipid syndrome (APS) is a systemic autoimmune rheumatic and musculoskeletal disorder (RMD) characterised by recurrent arterial and/or venous thrombosis and obstetric morbidity in the presence of persistently circulating antiphospholipid antibodies (aPL).1 2 Cardiovascular disease (CVD) events represent the most devastating manifestations of this rare disorder affecting primarily young adults, either in its primary form (‘primary APS’, PAPS) or in association with systemic lupus erythematosus (SLE).2 3 Stroke is the most frequent arterial event and, along with myocardial infarction (MI), are the most common causes of death in APS.4

    A high risk of premature and accelerated subclinical atherosclerosis has been also described in APS compared with the general population3 5 6 and at comparable rates to established high-cardiovascular risk (CVR) diseases such as diabetes mellitus (DM).7 8 CVD pathophysiology in APS is characterised by an interplay between aPL-mediated thrombosis, inflammation and atherogenesis.5 9 Traditional atherosclerotic CVR (ASCVR) factors are more prevalent in patients with APS versus other high-CVR RMDs such as rheumatoid arthritis (RA)10 11 and have been independently associated with incident and recurrent vascular events in APS.5 12–14 Hence, the recent European Alliance of Associations for Rheumatology (EULAR) recommendations for CVR management in RMDs prompted for joint assessment of ASCVR and disease-related features to reduce cardiovascular harm in APS.15

    ASCVR can be assessed by clinical prediction tools such as the Systematic Coronary Risk Evaluation (SCORE) and modified SCORE (mSCORE), Framingham risk score (FRS), Pooled Cohort Risk Equations (PCRE), Prospective Cardiovascular Müncher study risk tool (PROCAM) and Globorisk.16 17 The Secondary Manifestations of Arterial disease (SMART) prediction model has been developed to calculate residual ASCVR in patients with previous arterial events.18 A disease-specific risk tool for the prediction of recurrent thrombotic events in APS has been developed, the adjusted Global APS Score (aGAPSS),12 which includes weights for the three major aPL used in the classification criteria for APS,1 along with hypertension and hyperlipidaemia. A derivative equation that incorporates additional ASCVR factors such as obesity, smoking and DM, the aGAPSS for CVD (aGAPSSCVD), has also been tested for the prediction of recurrent arterial events in APS.13

    The presence of subclinical atherosclerotic plaques on vascular ultrasound (VUS) classifies patients at a CVD event-equivalent risk.16 Thus, according to the European Society of Cardiology (ESC),16 19 20 VUS-detected carotid and/or femoral plaques should be considered as an ASCVR modifier for patients classified as non-high risk by clinical prediction tools. Cumulative evidence15 21 indicates that most generic CVR models may miss actual risk in patients with RMDs for whom VUS screening in addition to clinical assessment could help to optimise risk factor management.22 23

    Currently, there is no evidence on the performance of risk prediction tools for ASCVR assessment in APS.15 Therefore, we designed a study encompassing two clinically relevant topics: (1) the performance of six first CVD event (first-CVD) risk prediction models (SCORE, mSCORE, FRS, PCRE, PROCAM, Globorisk) and three recurrent CVD event (recurrent-CVD) risk prediction tools (SMART, aGAPSS, aGAPSSCVD) to identify high-risk patients as defined by the presence of atherosclerotic plaques on VUS and (2) the impact of VUS on ASCVR classification by clinical prediction tools.

    Methods

    Study population

    All adult patients with APS followed up at the Rheumatology Unit of our Department were screened for their eligibility to participate in the study. Exclusion criteria included a history of symptomatic atherosclerotic CVD events, high-ASCVR comorbidities including DM and chronic kidney disease (CKD) and acute illness (see online supplemental data). All APS patients fulfilled the updated Sapporo classification criteria1 and, those with SLE/APS, the SLE International Collaborating Clinics classification criteria.24

    Clinical and laboratory parameters

    Eligible patients were referred for assessment to the Cardiovascular Risk Research laboratory of our Department. The following baseline clinical parameters were cross-sectionally collected during the first visit of patients: sociodemographic data, medical history and current medications; weight, height and body mass index and averaged blood pressure (BP) levels according to ESC guidelines.20 Laboratory variables included blood cholesterol fractions and triglycerides; the type, number and titres of aPL, that is, IgG and IgM anticardiolipin antibodies (aCL), and IgG and IgM anti-beta-2-glycoprotein I antibodies (anti-β2GPI) and lupus anticoagulant (LA) tested according to International Society on Thrombosis and Haemostasis guidelines.25 26 The presence of atherosclerotic plaques on VUS of the carotid and femoral arteries was determined according to a standardised protocol.27 28 Details about the recorded parameters are reported in online supplemental data.

    CVR assessment

    Nine CVR prediction tools were used to classify patients in high-risk versus non-high-risk categories according to established risk cut-offs reported in the literature for each of the examined risk prediction models (see also online supplemental data).12 13 18 28 Specifically, six prediction models designed to assess the risk of a first CVD event (SCORE, mSCORE, FRS, PCRE, PROCAM and Globorisk) and three risk prediction models developed for patients with a previous history of arterial events (SMART, aGAPSS and aGAPSSCVD) were assessed regarding their performance to identify actual ASCVR due to the presence of atherosclerotic plaques. Since atherosclerosis is a disease of the arteries, SMART and the two aGAPSS equations were specifically examined in patients with a previous arterial event (see the Sensitivity analyses section). Following the ESC recommendations on the use of VUS in ASCVR assessment,16 19 patients initially classified by clinical prediction tools as non-high risk were reclassified as high risk after detecting atherosclerotic plaques on VUS.

    Statistical analysis

    Categorical variables were assessed with χ2 test or Fisher’s exact test and are presented as percentages. Student’s t-test and Mann-Whitney U test were used to compare subgroup characteristics for normally and non-normally distributed continuous data, respectively, and are displayed as mean±SD or median and IQR according to data distribution profile.

    Performance measures tested for all risk models included calibration, discrimination and classification accuracy according to Spiegelhalter’s z-test, receiver operating characteristics (ROC)/area under the curve (AUC), and Matthews’ correlation coefficient (MCC), respectively. Sensitivity and specificity were examined at established high-risk cut-offs for each risk prediction model, while the Youden index was used to test the performance of risk tools regarding maximum accuracy corresponding to optimal high-risk cut-offs to identify high ASCVR due to the presence of VUS-detected atherosclerotic plaques. Fleiss kappa was used to examine the agreement between the recurrent-CVD risk tools. Statistical significance was set at a p<0.05. All statistical analyses were performed with Stata V.13.0 (StataCorp).

    Sensitivity analyses

    Five sensitivity analyses were included in the study, as follows:

    1. Considering that recurrent-CVD risk prediction tools are designed to predict different types of recurrent vascular events, the performance of SMART and the aGAPSS models was separately examined in patients with (a) arterial with or without venous events (‘arterial±venous’ group) and (b) isolated arterial events (‘arterial only’ group).

    2. According to the weights employed in the aGAPSS equations, triple aPL positivity for LA, aCL and anti-β2GPI classifies by default patients at high risk.12 13 Thus, performance metrics for aGAPSS and aGAPSSCVD were tested both before and after excluding triple aPL-positive patients (‘non-triple aPL’ group).

    3. To better delineate the performance of risk prediction tools to identify actual high-ASCVR as documented by VUS-detected plaques in the absence of major CVD events, a sensitivity analysis that further excluded patients with non-atherosclerotic MI or stroke on treatment with antihypertensives and/or hypolipidaemics was carried out for all risk models.

    4. The nine risk prediction tools examined in this study employ different lower age limits for clinical application. Furthermore, the classification criteria for APS1 suggested that in patients older than 55 years, a high impact of traditional ASCVR factors on thrombotic events would be expected. To satisfy both conditions, risk reclassification by VUS was additionally explored in a subgroup of patients aged 40–54 years.

    5. Based on current evidence suggesting that patients with PAPS may have a different ASCVR compared with those with SLE-APS,4 5 VUS-guided risk reclassification was separately examined in these two patient subpopulations.

    Results

    Baseline characteristics

    One hundred and twenty-one patients (mean age 45.8±11.8 years; 68.6% women; 62.8% PAPS; 37.2% SLE-APS) were included in the study (online supplemental figure S1). Median systolic BP was 123 (IQR 116–131) mm Hg, mean LDL-cholesterol 111±34 mg/dL and 34.7% of patients were current smokers. Two-thirds of patients had high-titre aPL, while triple aPL positivity was noted in 44.6% of cases. Atherosclerotic plaques were observed in 35.5% of patients (carotid: 12.4%; femoral: 9.9%; both arteries: 13.2%). Baseline characteristics of study participants are shown in table 1.

    View this table:
    Table 1

    Baseline characteristics of study participants (N=121)

    CVR estimates

    Baseline risk estimates according to first-CVD and recurrent-CVD risk tools in patients with and without plaques are shown in table 2. All first-CVD prediction models assigned significantly higher risks (p<0.05) to patients with plaques compared with patients without plaques, except for SCORE and mSCORE. In the ‘arterial only’ group, SMART scored patients with plaques at risks above the established high-risk threshold and significantly higher than patients without plaques (23.2%±12.2% vs 16.5%±7.5%; p=0.033); a similar result was observed in the ‘arterial±venous’ group, though it did not reach statistical significance (p=0.078). No difference was noted for risk estimates calculated with aGAPSS and aGAPSSCVD between patients with and without plaques in ‘arterial±venous’ or ‘arterial only’ and ‘non-triple aPL/arterial±venous’ or ‘non-triple aPL/arterial only’ groups.

    View this table:
    Table 2

    Cardiovascular risk estimates of 121 APS patients by clinical prediction models

    Performance of risk prediction models

    The performance of clinical prediction models is shown in table 3. Results for a sensitivity analysis of performance measures after excluding patients with MI and/or stroke treated with antihypertensive and/or hypolipidaemic agents are included in online supplemental table S2.

    View this table:
    Table 3

    Performance measures of cardiovascular risk prediction tools to identify high ASCVR in patients with APS (N=121)

    Assessment of first-CVD risk prediction models

    Spiegelhalter’s z-test p values ranged from 0.47 to 0.57, indicating moderate calibration. ROC analysis revealed a poor-to-acceptable discriminatory power (AUC 0.59–0.75). MCC values for classification accuracy were 0.11, 0.10, 0.10, 0.22, 0.27 and 0.11 for SCORE, mSCORE, FRS, PCRE, Globorisk and PROCAM, respectively, suggesting no improvement of classification by using mSCORE versus SCORE. Sensitivity ranged from 9.5% to 41.5% and specificity from 82.9% to 96.1%. Optimal high-risk cut-offs to identify high-risk cases due to the presence of VUS-detected plaques according to the highest Youden index were as follows: SCORE: 0.2%; mSCORE: 1.0%; FRS: 3.2%; PCRE: 5.3%; Globorisk: 3%; PROCAM: 1.7%.

    Assessment of recurrent-CVD risk prediction models

    Moderate calibration was observed for all recurrent-CVD prediction models across all groups (p value range 0.47–0.52). For SMART, AUC and MCC were 0.64 and 0.29 in the ‘arterial±venous’ group and 0.66 and 0.33 in the ‘arterial only’ group, respectively, while the optimal high-risk cut-off was identical in both groups.

    aGAPSS and aGAPSSCVD showed marginally poor-to-good discrimination (AUC 0.56–0.69) in all groups and negligible-to-very low classification accuracy (MCC 0.01–0.08) in ‘arterial±venous’ and ‘arterial only’ groups. In ‘non-triple aPL/arterial±venous’ and ‘non-triple aPL/arterial only’ groups, MCC was better for aGAPSSCVD (0.29 and 0.28, respectively) compared with aGAPSS (0.16 and 0.17, respectively). Optimal high-risk cut-offs for aGAPSS and aGAPSSCVD were 14 and 16 points in ‘arterial±venous’ and ‘arterial only’ groups, and 9 and 10 points in ‘non-triple aPL/arterial±venous’ and ‘non-triple aPL/arterial only’ groups, respectively.

    Fleiss kappa for SMART, aGAPSS and aGAPSSCVD was 0.30 in the ‘arterial only’ group and 0.44 in the ‘non-triple aPL/arterial only’ group, indicating fair and moderate agreement, respectively, between recurrent-CVD risk prediction models to predict high CVR.

    Impact of VUS on risk classification

    Figure 1 shows high-risk versus non-high-risk classification of patients by first-CVD and recurrent-CVD risk prediction models. Risk reclassification of patients from non-high risk according to risk prediction models to high risk based on the presence of VUS-detected atherosclerotic plaques is displayed in figures 2 and 3.

    Figure 1
    Figure 1

    Risk classification by first CVD event (A) and recurrent CVD event (B) clinical risk prediction tools of APS patients. *Patients with arterial with or without venous events. †Patients with isolated arterial events. aGAPSS, adjusted Global Antiphospholipid Syndrome Score; aGAPSSCVD, aGAPSS for CVD; CVD, cardiovascular disease; FRS, Framingham Risk Score; mSCORE, modified SCORE; PCRE, Pooled Cohorts Risk Equation; PROCAM, Prospective Cardiovascular Muncher Study calculator; SCORE, Systematic Coronary Risk Evaluation; SMART, Secondary Manifestations of Arterial Disease risk score.

    Figure 2
    Figure 2

    Risk reclassification by vascular ultrasound of APS patients classified as non-high-risk according to first CVD event (A1, A2) and recurrent CVD event (B1, B2) clinical risk prediction tools. *Patients with arterial with or without venous events. †Patients with isolated arterial events. aGAPSS, adjusted Global Antiphospholipid Syndrome Score; aGAPSSCVD, aGAPSS for CVD; CVD, cardiovascular disease; FRS, Framingham Risk Score; mSCORE, modified SCORE; PCRE, Pooled Cohorts Risk Equation; PROCAM, Prospective Cardiovascular Muncher Study calculator; SCORE, Systematic Coronary Risk Evaluation; SMART, Secondary Manifestations of Arterial Disease risk score.

    Figure 3
    Figure 3

    Distribution of risk reclassification by vascular ultrasound among PAPS and SLE-APS patients classified as non-high-risk according to first CVD event (A) and recurrent CVD event (B) clinical risk prediction tools. *Patients with arterial with or without venous events. †Patients with isolated arterial events. aGAPSS, adjusted Global Antiphospholipid Syndrome Score; aGAPSSCVD, aGAPSS for CVD; CVD, cardiovascular disease; FRS, Framingham Risk Score; mSCORE, modified SCORE; PAPS, primary antiphospholipid syndrome; PCRE, Pooled Cohorts Risk Equation; PROCAM, Prospective Cardiovascular Muncher Study calculator; SCORE, Systematic Coronary Risk Evaluation; SLE-APS, systemic lupus erythematosus-associated antiphospholipid syndrome; SMART, Secondary Manifestations of Arterial Disease risk score.

    VUS plus first-CVD risk prediction models

    High-risk classification before applying the results of vascular imaging ranged from 5.8% to 29.3%. VUS-based risk reclassification was 48% for both SCORE and mSCORE, 37.8% for FRS, 44.8% for PCRE, 41.4% for Globorisk and 34.2% for PROCAM; reclassification rates were similar for all risk prediction models in the 40–54 years old subgroup.

    When examined separately, VUS reclassified PAPS versus SLE-APS patients from non-high risk to high risk as follows: SCORE: 52.2% versus 40%; mSCORE: 54.5% versus 34.8%; FRS: 44.3% versus 27%; PCRE: 50% versus 36%; Globorisk: 47.2% versus 31.8%; and PROCAM: 40.3% versus 23.1%, respectively.

    VUS plus recurrent-CVD risk prediction models

    SMART, aGAPSS and aGAPSSCVD assigned 36.8% and 31.3%, 66.7% and 59.6% and 66.7% and 66% of patients to high risk in the ‘arterial±venous’ and ‘arterial only’ groups, respectively. VUS-guided risk reclassification rates of non-high-risk patients in ‘arterial±venous’ versus ‘arterial only’ groups for SMART, aGAPSS and aGAPSSCVD were 41.7% versus 40.7%, 40.5% versus 52.6% and 50% versus 50% for all patients, respectively. In the 40–54 years old subgroup, VUS reclassified ≥50% of apparently non-high-risk cases according to SMART, aGAPSS or aGAPSSCVD in both the ‘arterial±venous’ and ‘arterial only’ groups. Risk reclassification was higher among PAPS (50%–70.6%) compared with SLE-APS (0%–25%) patients for recurrent-CVD risk prediction tools also.

    Application of VUS findings to ‘non-triple aPL/arterial±venous’ and ‘non-triple aPL/arterial only’ groups returned similar reclassification results to those in ‘arterial±venous’ and ‘arterial only’ groups for aGAPSS and aGAPSSCVD in the entire cohort (all APS patients) and in the PAPS and SLE-APS subgroups (online supplemental figures S2 and S3).

    Discussion

    To our knowledge, this is the first study in APS to assess multiple types of CVR assessment tools along with complementary risk classification modalities such as VUS according to EULAR15 and ESC recommendations.16 19 29 Our findings showed inadequate performance of clinical tools alone to identify actual CVR, affecting optimal CVR management to prevent cardiovascular harm in patients with APS.2 15

    Mounting evidence suggests that first-CVD risk prediction models underestimate actual risk in patients with RMDs.15 21 28 30 Our results about SCORE, FRS, PCRE, PROCAM and Globorisk in APS are in line with these observations. Although demonstrating moderate calibration, all models showed low classification accuracy to identify patients at actual high risk. Remarkably, identification of high-risk cases would require to use cut-offs corresponding to very low estimates compared with the thresholds recommended in the general population. Consequently, pragmatic use of these CVR assessment tools to guide CVR management tailored to risk class15 19 seems challenging in APS.

    Failure of generic first-CVD risk models to capture actual ASCVR in RMDs may arise out of biases related to a higher risk attributed to older age and the higher prevalence of traditional CVR factors in RMDs compared with the general population.15 21 30 Previous evidence showed that SLE31 32 and RA32 are independently associated with premature atherosclerotic CVD. In our study, marked reclassification of younger non-high-risk patients to high risk was observed with use of VUS for all generic risk models. These findings indicate that earlier and stricter CVR screening and risk factor management may be required in younger APS patients compared with same age individuals in the general population.

    Modifications of baseline risk assessed by generic models using multipliers has been proposed by the ESC to improve CVR assessment in RMDs,16 29 with mixed results.15 28 30 33 Our results suggest that mSCORE does not perform better than SCORE to predict actual high risk in APS patients. Nevertheless, incorporation of appropriately weighted disease-related risks into existing prediction models, as the inclusion of SLE into QRISK3,28 30 31 instead of applying multiplication factors, could help to optimise the use of measures of association computed to calculate risk of CVD events in APS.

    APS is characterised by a wide range of vascular manifestations.1 3 5 9 From a clinical standpoint, it is unclear whether patients with venous versus those experiencing arterial events should be assessed differently. In our study, this issue was addressed using recurrent-CVD risk prediction tools. SMART and the aGAPSS equations showed comparable calibration, but SMART had considerably greater classification accuracy against aGAPSS and aGAPSSCVD, and high-risk cut-offs of maximum accuracy corresponded to thresholds validated in the general population.18 Results about cut-off points for SMART were similar also after excluding patients with non-atherosclerotic stroke and/or MI, but only in the ‘arterial only’ group. In the same sensitivity analysis, Globorisk performed considerably better among all first-CVD risk prediction models. Even so, optimal high-risk cut-offs were markedly low to suggest the use of Globorisk (or any of the other first-CVD risk tools) to assess ASCVR in APS patients without non-atherosclerotic major CVD events.

    Lack of consideration of disease-related features in CVR assessment of patients with RMDs has been gaining increasing attention.15 21 aGAPSS and aGAPSSCVD are among the few developed CVR prediction tools including both traditional and disease-related risk factors in APS.15 Our findings suggest that both models perform poorly to identify actual high ASCVR in APS. This observation may be related to the weighting of aPL relative to traditional ASCVR factors in the aGAPSS and aGAPSSCVD equations. Indeed, exclusion of triple aPL-positive patients substantially improved the performance of both risk prediction tools. Therefore, use of these models to assess CVR may not apply to all aPL profiles encountered in clinical practice.

    Earlier evidence reported that aGAPSS and aGAPSSCVD estimates do not associate with progression of ASCVR in APS,8 while a recent study examining aGAPSS34 found a thrombotic recurrence cut-off similar to the high-ASCVR threshold identified in our study. Interestingly, both in the latter study and ours, the decision thresholds to assign a patient to high CVR were similar, as well as corresponding to a higher threshold compared with the cut-off reported in the aGAPSS derivation study.12 Nevertheless, in the sensitivity analysis of our study excluding triple aPL positivity as a potential shortcoming of the aGAPSS and aGAPSSCVD equations, aGAPSSCVD performed better compared with aGAPSS in both ‘arterial±venous’ and ‘arterial only’ groups, as well as in patients with and without stroke and/or MI, to identify high-risk patients with plaques. Since both models assign equal weights to aPL, this difference could be attributed to the inclusion of a greater number of ASCVR factors into the aGAPPSCVD. Risk estimates calculated with aGAPSSCVD also correlate with measures of subclinical atherosclerosis such as intima–media thickness,13 but further validation of both aGAPSS equations is needed to better elucidate their role in identifying high ASCVR groups in APS.15

    CVR management in APS may need to address unique aspects of CVD pathophysiology in the syndrome.5 Besides their role in immune-mediated thrombosis, aPL has been shown to also have a proatherogenic effect even in the absence of comorbidities or concomitant ASCVR factors.6 35 Importantly, comprehensive aPL profiling2 5 35 according to the type, titres and number of autoantibodies may be necessary to improve CVR assessment strategies in APS. Likewise, traditional ASCVR factors generally recognised as promoters of atherosclerosis may also act as precipitating factors for first and recurrent thrombotic events in APS.5 9 Therefore, studies examining revisions of existing disease-related risk prediction models or additional risk modifiers to CVR estimates by generic models would help to prevent cardiovascular damage in APS. This line of research would be consistent with EULAR recommendations for CVR management in APS15 and has been introduced for other high-CVR diseases such as DM or CKD.16 19

    VUS has been proposed by the ESC16 to refine CVR estimation in non-high-risk patients. Increasing evidence has also highlighted the importance of VUS for CVR assessment in RMDs.22 28 36 Risk reclassification by VUS in our study ranged from two-fifths of all patients to more than half of the younger subgroup population, indicating that this risk assessment modality may significantly impact CVR management in patients with APS misclassified as non-high risk by clinical risk prediction alone. VUS reclassification was less pronounced in SLE-APS cases, but this finding is most likely related to the smaller representation of this disease subgroup (37%) in our study sample. However, optimal use of VUS to inform CVR prevention in RMDs is currently unclear.15 36 Awaiting further investigation, employment of VUS-guided risk assessment in APS could be reserved for clinical phenotypes expected to be misidentified as non-high risk by risk prediction models such as younger patients or those with PAPS, as shown in this study.

    Strengths of our study include testing a wide range of clinical risk prediction models to identify APS patients at actual high risk as documented by the presence of atherosclerotic plaques in multiple anatomical sites of the carotid and femoral arteries on VUS, as recommended by the ESC.16 19 Assessment of risk prediction tools was examined in a well-characterised, relatively large sample considering the rarity of APS, without high-ASCVR comorbidities. Additionally, performance analysis of risk prediction models incorporated different clinical scenarios, as suggested by EULAR,15 including patients with and without major CVD events in the presence or absence of triple aPL positivity.

    We acknowledge that our study has certain limitations. The cross-sectional design does not allow deducing definite conclusions about the performance of risk models on par with prospectively recorded CVD events over a 10-year interval, particularly regarding calibration. Accounting for the fact that VUS-detected plaques are a surrogate measure of ASCVR,16 our results should be approached with caution, especially considering the complex interplay between aPL and traditional risk factors in both ASCVR and immune-mediated CVR in APS. Nevertheless, in terms of risk-stratified ASCVR factor modification,19 20 the presence of atherosclerotic plaques on VUS has equivalent importance to incident cardiovascular events.16 19 20 Additionally, aGAPSS and aGAPSSCVD have not been specifically designed to assess ASCVR, though our results—considering also the inclusion of traditional ASCVR factors in these models and the high ASCVR in APS6—suggest that recalibration based on studies comprising prospectively recorded atherosclerotic and non-atherosclerotic thrombotic events could help to improve the use of these risk tools in patients with APS. Our study population included only white Europeans and was recruited in an academic hospital centre, limiting conclusive validity and generalisation of our findings permitted by ethnic diversity and primary care settings.

    To conclude, our study showed that generic first-CVD risk prediction models misidentify actual ASCVR in APS without the use of additional risk assessment modalities such as VUS. Among recurrent-CVD risk tools, SMART had acceptable performance only for patients with isolated arterial events, while CVR assessment incorporating disease-related features may require an update of the relative contribution of each of aPL based on the new American College of Cardiology/EULAR classification criteria for APS37 and each of traditional ASCVR factors. Studies further exploring the complexities of CVR assessment in APS are urgently needed.

    Data availability statement

    Data are available on reasonable request. The data underlying this article will be shared on reasonable request to the corresponding author.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and was approved by Laiko General Hospital Scientific Board (SB number 1790). Participants gave informed consent to participate in the study before taking part.

    Acknowledgments

    The authors would like to thank the administrative personnel of the Rheumatology Unit of the First Department of Propaedeutic Internal Medicine of the National and Kapodistrian University of Athens for their general support.

    References

    1. ↵
      1. Miyakis S,
      2. Lockshin MD,
      3. Atsumi T, et al
      . International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost 2006;4:295–306. doi:10.1111/j.1538-7836.2006.01753.x
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Tektonidou MG,
      2. Andreoli L,
      3. Limper M, et al
      . EULAR recommendations for the management of antiphospholipid syndrome in adults. Ann Rheum Dis 2019;78:1296–304. doi:10.1136/annrheumdis-2019-215213
      OpenUrlCrossRefPubMed
    3. ↵
      1. Ambrosino P,
      2. Lupoli R,
      3. Di Minno A, et al
      . Markers of cardiovascular risk in patients with antiphospholipid syndrome: a meta-analysis of literature studies. Ann Med 2014;46:693–702. doi:10.3109/07853890.2014.959559
      OpenUrlCrossRefPubMed
    4. ↵
      1. Cervera R,
      2. Serrano R,
      3. Pons-Estel GJ, et al
      . Morbidity and mortality in the antiphospholipid syndrome during a 10-year period: a multicentre prospective study of 1000 patients. Ann Rheum Dis 2015;74:1011–8. doi:10.1136/annrheumdis-2013-204838
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Tektonidou MG
      . Cardiovascular disease risk in antiphospholipid syndrome: thrombo-inflammation and atherothrombosis. J Autoimmun 2022;128:102813. doi:10.1016/j.jaut.2022.102813
      OpenUrl
    6. ↵
      1. Karakasis P,
      2. Lefkou E,
      3. Pamporis K, et al
      . Risk of subclinical atherosclerosis in patients with antiphospholipid syndrome and subjects with antiphospholipid antibody positivity: a systematic review and meta-analysis. Curr Probl Cardiol 2023;48:101672. doi:10.1016/j.cpcardiol.2023.101672
      OpenUrl
    7. ↵
      1. Kravvariti E,
      2. Konstantonis G,
      3. Tentolouris N, et al
      . Carotid and femoral atherosclerosis in Antiphospholipid syndrome: equivalent risk with diabetes mellitus in a case-control study. Semin Arthritis Rheum 2018;47:883–9. doi:10.1016/j.semarthrit.2017.10.015
      OpenUrl
    8. ↵
      1. Evangelatos G,
      2. Kravvariti E,
      3. Konstantonis G, et al
      . Atherosclerosis progression in antiphospholipid syndrome is comparable to diabetes mellitus: a 3 year prospective study. Rheumatology (Oxford) 2022;61:3408–13. doi:10.1093/rheumatology/keab882
      OpenUrl
    9. ↵
      1. Corban MT,
      2. Duarte-Garcia A,
      3. McBane RD, et al
      . Antiphospholipid syndrome: role of vascular endothelial cells and implications for risk stratification and targeted therapeutics. J Am Coll Cardiol 2017;69:2317–30. doi:10.1016/j.jacc.2017.02.058
      OpenUrlFREE Full Text
    10. ↵
      1. Bolla E,
      2. Tentolouris N,
      3. Sfikakis PP, et al
      . Cardiovascular risk management in antiphospholipid syndrome: trends over time and comparison with rheumatoid arthritis and diabetes mellitus. Lupus Sci Med 2021;8:e000579. doi:10.1136/lupus-2021-000579
    11. ↵
      1. Panopoulos S,
      2. Thomas K,
      3. Georgiopoulos G, et al
      . Comparable or higher prevalence of comorbidities in antiphospholipid syndrome vs rheumatoid arthritis: a multicenter, case-control study. Rheumatology (Oxford) 2021;60:170–8. doi:10.1093/rheumatology/keaa321
      OpenUrlCrossRefPubMed
    12. ↵
      1. Radin M,
      2. Sciascia S,
      3. Erkan D, et al
      . The adjusted global antiphospholipid syndrome score (aGAPSS) and the risk of recurrent thrombosis: results from the APS ACTION cohort. Semin Arthritis Rheum 2019;49:464–8. doi:10.1016/j.semarthrit.2019.04.009
      OpenUrlCrossRefPubMed
    13. ↵
      1. Di Minno MND,
      2. Scalera A,
      3. Tufano A, et al
      . The association of adjusted global antiphospholipid syndrome score (aGAPSS) with cardiovascular disease in subjects with antiphospholipid antibodies. Atherosclerosis 2018;278:60–5. doi:10.1016/j.atherosclerosis.2018.09.010
      OpenUrlPubMed
    14. ↵
      1. Moreno-Torres V,
      2. Royuela A,
      3. Tarín C, et al
      . Impact of cardiovascular risk factors in antiphospholipid syndrome: an observational study from the Spanish national registry. Clin Exp Rheumatol 2022;40:2161–6. doi:10.55563/clinexprheumatol/h2tkx3
      OpenUrl
    15. ↵
      1. Drosos GC,
      2. Vedder D,
      3. Houben E, et al
      . EULAR recommendations for cardiovascular risk management in rheumatic and musculoskeletal diseases, including systemic lupus erythematosus and antiphospholipid syndrome. Ann Rheum Dis 2022;81:768–79. doi:10.1136/annrheumdis-2021-221733
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Piepoli MF,
      2. Hoes AW,
      3. Agewall S, et al
      . European guidelines on cardiovascular disease prevention in clinical practice: the sixth joint task force of the European society of cardiology and other societies on cardiovascular disease prevention in clinical practice (constituted by representatives of 10 societies and by invited Experts)developed with the special contribution of the European association for cardiovascular prevention & rehabilitation (EACPR). Eur Heart J 2016;37:2315–81. doi:10.1093/eurheartj/ehw106
      OpenUrlCrossRefPubMed
    17. ↵
      1. Hajifathalian K,
      2. Ueda P,
      3. Lu Y, et al
      . A novel risk score to predict cardiovascular disease risk in national populations (Globorisk): a pooled analysis of prospective cohorts and health examination surveys. Lancet Diabetes Endocrinol 2015;3:339–55. doi:10.1016/S2213-8587(15)00081-9
      OpenUrl
    18. ↵
      1. Dorresteijn JAN,
      2. Visseren FLJ,
      3. Wassink AMJ, et al
      . Development and validation of a prediction rule for recurrent vascular events based on a cohort study of patients with arterial disease: the SMART risk score. Heart 2013;99:866–72. doi:10.1136/heartjnl-2013-303640
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Mach F,
      2. Baigent C,
      3. Catapano AL, et al
      . 2019 ESC/EAS guidelines for the management of dyslipidaemias: lipid modification to reduce cardiovascular risk. Eur Heart J 2020;41:111–88. doi:10.1093/eurheartj/ehz455
      OpenUrlCrossRefPubMed
    20. ↵
      1. Williams B,
      2. Mancia G,
      3. Spiering W, et al
      . ESC/ESH guidelines for the management of arterial hypertension. Eur Heart J 2018;39:3021–104. doi:10.1093/eurheartj/ehy339
      OpenUrlCrossRefPubMed
    21. ↵
      1. Anyfanti P,
      2. Dara A,
      3. Angeloudi E, et al
      . Monitoring and managing cardiovascular risk in immune mediated inflammatory diseases. J Inflamm Res 2021;14:6893–906. doi:10.2147/JIR.S276986
      OpenUrl
    22. ↵
      1. Colaco K,
      2. Ocampo V,
      3. Ayala AP, et al
      . Predictive utility of cardiovascular risk prediction algorithms in inflammatory rheumatic diseases: a systematic review. J Rheumatol 2020;47:928–38. doi:10.3899/jrheum.190261
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Drosos GC,
      2. Konstantonis G,
      3. Sfikakis PP, et al
      . Lipid management in systemic lupus erythematosus according to risk classifiers suggested by the European society of cardiology and disease-related risk factors reported by the EULAR recommendations. RMD Open 2023;9:e002767. doi:10.1136/rmdopen-2022-002767
    24. ↵
      1. Petri M,
      2. Orbai A-M,
      3. Alarcón GS, et al
      . Derivation and validation of the systemic lupus international collaborating clinics classification criteria for systemic lupus erythematosus. Arthritis Rheum 2012;64:2677–86. doi:10.1002/art.34473
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Vandevelde A,
      2. Chayoua W,
      3. de Laat B, et al
      . Semiquantitative interpretation of anticardiolipin and Antibeta2glycoprotein I antibodies measured with various analytical platforms: communication from the ISTH SSC subcommittee on lupus anticoagulant/Antiphospholipid antibodies. J Thromb Haemost 2022;20:508–24. doi:10.1111/jth.15585
      OpenUrl
    26. ↵
      1. Devreese KMJ,
      2. de Groot PG,
      3. de Laat B, et al
      . Guidance from the scientific and standardization committee for lupus anticoagulant/Antiphospholipid antibodies of the international society on thrombosis and haemostasis: update of the guidelines for lupus anticoagulant detection and interpretation. J Thromb Haemost 2020;18:2828–39. doi:10.1111/jth.15047
      OpenUrlCrossRefPubMed
    27. ↵
      1. Touboul P-J,
      2. Hennerici MG,
      3. Meairs S, et al
      . Mannheim carotid intima-media thickness and plaque consensus (2004–2006–2011). Cerebrovasc Dis 2012;34:290–6. doi:10.1159/000343145
      OpenUrlCrossRefPubMed
    28. ↵
      1. Drosos GC,
      2. Konstantonis G,
      3. Sfikakis PP, et al
      . Underperformance of clinical risk scores in identifying vascular ultrasound-based high cardiovascular risk in systemic lupus erythematosus. Eur J Prev Cardiol 2021;28:346–52. doi:10.1093/eurjpc/zwaa256
      OpenUrlPubMed
    29. ↵
      1. Rossello X,
      2. Dorresteijn JA,
      3. Janssen A, et al
      . Risk prediction tools in cardiovascular disease prevention: a report from the ESC prevention of CVD programme led by the European association of preventive cardiology (EAPC) in collaboration with the acute cardiovascular care association (ACCA) and the association of cardiovascular nursing and allied professions (ACNAP). Eur J Prev Cardiol 2019;26:1534–44. doi:10.1177/2047487319846715
      OpenUrlPubMed
    30. ↵
      1. Sivakumaran J,
      2. Harvey P,
      3. Omar A, et al
      . Assessment of cardiovascular risk tools as predictors of cardiovascular disease events in systemic lupus erythematosus. Lupus Sci Med 2021;8:e000448. doi:10.1136/lupus-2020-000448
    31. ↵
      1. Hippisley-Cox J,
      2. Coupland C,
      3. Brindle P
      . Development and validation of QRISK3 risk prediction algorithms to estimate future risk of cardiovascular disease: prospective cohort study. BMJ 2017;357:j2099. doi:10.1136/bmj.j2099
    32. ↵
      1. Mahtta D,
      2. Gupta A,
      3. Ramsey DJ, et al
      . Autoimmune rheumatic diseases and premature atherosclerotic cardiovascular disease: an analysis from the VITAL registry. Am J Med 2020;133:1424–32. doi:10.1016/j.amjmed.2020.05.026
      OpenUrlPubMed
    33. ↵
      1. Urowitz MB,
      2. Ibañez D,
      3. Su J, et al
      . Modified framingham risk factor score for systemic lupus erythematosus. J Rheumatol 2016;43:875–9. doi:10.3899/jrheum.150983
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Barilaro G,
      2. Esteves A,
      3. Della Rocca C, et al
      . Predictive value of the adjusted global anti-phospholipid syndrome score on clinical recurrence in APS patients: a longitudinal study. Rheumatology (Oxford) 2023;62:1576–85. doi:10.1093/rheumatology/keac485
      OpenUrl
    35. ↵
      1. Di Minno MND,
      2. Emmi G,
      3. Ambrosino P, et al
      . Subclinical atherosclerosis in asymptomatic carriers of persistent antiphospholipid antibodies positivity: a cross-sectional study. Int J Cardiol 2019;274:1–6. doi:10.1016/j.ijcard.2018.06.010
      OpenUrl
    36. ↵
      1. Sade LE,
      2. Akdogan A
      . Imaging for screening cardiovascular involvement in patients with systemic rheumatologic diseases: more questions than answers. Eur Heart J Cardiovasc Imaging 2019;20:967–78. doi:10.1093/ehjci/jez171
      OpenUrlPubMed
    37. ↵
      1. Barbhaiya M,
      2. Zuily S,
      3. Naden R, et al
      . ACR/EULAR antiphospholipid syndrome classification criteria. Ann Rheum Dis 2023;82:1258–70. doi:10.1136/ard-2023-224609
      OpenUrlAbstract/FREE Full Text

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Twitter @MariaTektonidou

    • Contributors GCD: methodology, investigation, resources, data curation, writing–original draft, writing–review and editing, and visualisation. GK: investigation, data curation and formal analysis. PPS: investigation, resources. MGT: conceptualisation, investigation, resources, writing–review and editing, supervision and project administration; guarantor. All authors contributed to data interpretation and gave final approval and agreed to be accountable for all aspects of work ensuring integrity and accuracy.

    • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.