Abstract
Background
Immune-mediated thrombotic thrombocytopenic purpura (iTTP) is a rare hematologic disorder with improved survival due to advancements in treatment. However, long-term cardiovascular morbidity and mortality remain significant. Established cardiovascular risk calculators, such as the 2008 Framingham Heart Study (FHS) global cardiovascular disease (CVD) and the American College of Cardiology/American Heart Association (ACC/AHA) atherosclerotic CVD (ASCVD) risk estimators, may not adequately account for the elevated and unique cardiovascular risks in iTTP survivors.
Objectives
To evaluate the discrimination and calibration of the ACC/AHA ASCVD and FHS global CVD models in predicting major adverse cardiovascular events (MACEs) among iTTP survivors.
Methods
This retrospective study analyzed 135 iTTP survivors from Johns Hopkins University (1994-2024). Presence of MACEs, including myocardial infarction, stroke, and cardiac revascularization, was the primary outcome and was assessed during clinical remission. Discriminatory ability of the model was assessed using c-statistics, while calibration was evaluated with Hosmer-Lemeshow tests and calibration plots. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were also calculated.
Results
MACEs occurred in 37.8% of the cohort over a median follow-up of 3.8 years. The ASCVD and FHS models demonstrated poor discrimination (c-statistics, 0.54 and 0.52, respectively) and poor calibration, with observed MACE rates exceeding predicted probabilities (Hosmer–Lemeshow P < .05). The ASCVD model showed sensitivity of 56.5%, specificity of 49.4%, PPV of 36.6%, and NPV of 64.9%, while the FHS model showed sensitivity of 69.6%, specificity of 39.3%, PPV of 37.2%, and NPV of 67.9%.
Conclusion
Standard cardiovascular risk models inadequately predict MACE risk in iTTP survivors, underscoring the need for tailored tools that incorporate iTTP-specific factors to improve cardiovascular risk stratification and management.
Keywords: calibration, cardiovascular, MACE, risk prediction, thrombotic thrombocytopenic purpura
Essentials
-
•
Immune-mediated thrombotic thrombocytopenic purpura (iTTP) raises long-term risk of major adverse cardiovascular events (MACEs).
-
•
This study followed 135 iTTP survivors treated at Johns Hopkins >30 years (1994-2024).
-
•
Standard risk models missed about one-third of MACEs in iTTP survivors.
-
•
New risk models that consider iTTP-specific factors are needed to guide prevention strategies.
1. Introduction
Immune-mediated thrombotic thrombocytopenic purpura (iTTP) is a rare and potentially fatal hematologic disorder characterized by recurrent episodes of microangiopathic hemolytic anemia, thrombocytopenia, and ischemic damage to various organs [1]. This disorder arises due to an acquired deficiency of ADAMTS-13, a von Willebrand factor-cleaving protease that leads to the accumulation of large von Willebrand factor multimers, which consequently causes platelet aggregation and microvascular thrombosis [2]. While advancements in treatment, such as plasma exchange and immunosuppression, have significantly improved survival from acute iTTP episodes, long-term morbidity and mortality among survivors remain substantial concerns [3].
Cardiovascular diseases (CVDs) have emerged as a leading cause of death among iTTP survivors [4]. Studies have shown that iTTP survivors exhibit higher incidence of major adverse cardiovascular events (MACEs), including myocardial infarction (MI), stroke, and cardiac revascularization, than age- and sex-matched controls from the general population [5,6]. We previously demonstrated that up to 23.8% of iTTP survivors experience MACEs during clinical remission, with stroke being the most common event, at a median follow-up of 7.6 years after iTTP diagnosis [7]. The risk of these events is not only higher in these patients, but it also occurs at a younger age in iTTP survivors, with the median age for first stroke and MI occurring approximately 10 years earlier in men and 20 years earlier in women than in the general population [4]. This elevated risk is partly attributable to higher prevalence of cardiovascular risk factors such as hypertension, diabetes mellitus, hyperlipidemia, and chronic kidney disease in iTTP population [8]. iTTP-specific risk factors such as lower levels of ADAMTS-13 in remission are also associated with vascular disease, particularly stroke [9].
Given the heightened cardiovascular risk in iTTP survivors, it is crucial to evaluate the predictive value of established and widely applied cardiovascular risk tools such as the 2008 Framingham Heart Study (FHS) global cardiovascular disease (CVD) risk calculator and the American College of Cardiology/American Heart Association (ACC/AHA) atherosclerotic CVD (ASCVD) risk estimator plus [10,11]. Although both of these tools have been well validated in general and high-risk populations, their applicability and accuracy in predicting cardiovascular risk in iTTP patients have not been thoroughly investigated. These models may not be able to account for the unique pathophysiological mechanisms and heightened risk in iTTP survivors. This study assessed the predictive accuracy of the ACC/AHA ASCVD risk estimator plus and the 2008 FHS global CVD risk calculator in estimating the risk of major cardiovascular events in iTTP survivors. By comparing predicted risks with observed cardiovascular outcomes in this patient cohort, we determined whether these tools can be effectively utilized or if there is a need for developing iTTP-specific risk assessment models. We hypothesize that established cardiovascular risk calculators (the 2008 FHS global CVD and the ACC/AHA ASCVD risk calculators) would demonstrate suboptimal performance (in terms of discrimination and calibration) in predicting MACEs in iTTP survivors compared with their reported accuracy in general and high-risk populations.
2. Methods
2.1. Study cohort
The study cohort included adult patients (aged ≥18 years) with iTTP who were followed at Johns Hopkins University between 1994 and 2024 who survived an acute episode of iTTP and had ≥1 year follow-up. iTTP was diagnosed based on ≥1 confirmed episode of thrombotic microangiopathy with microangiopathic hemolytic anemia (hemoglobin < 10 g/dL with schistocytes and markers of hemolysis), thrombocytopenia with a platelet count < 150 × 109/L, along with an ADAMTS-13 activity level <10% (or 10%-20% with an anti-ADAMTS-13 antibody or inhibitor). The majority of these patients have subsequently enrolled in the Johns Hopkins Thrombotic Microangiopathy Registry, which was established in 2014, also as a prospective cohort of patients with thrombotic microangiopathy treated at the Johns Hopkins Hospital. Patients were followed until death or last clinical contact. The study was approved by the institutional review board at Johns Hopkins University.
2.2. Data management and study outcomes
The baseline visit (visit of the risk score calculation) was the participant’s earliest visit that had no missing values for the risk factors necessary to generate a risk score (Figure 1). We extracted data from the electronic medical record, including patient demographics, systolic blood pressure, diastolic blood pressure, total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, details of initial iTTP presentation, diagnosis and treatment, and presence of comorbidities, including hypertension, diabetes mellitus, smoking status, obesity (defined as body mass index > 30 kg/m2), history of hyperlipidemia, atrial fibrillation, chronic kidney disease (defined as a glomerular filtration rate < 60 mL/min/1.73 m2 persisting ≥3 months), any other vascular disease including peripheral artery disease and coronary artery disease, systemic lupus erythematosus, and other autoimmune diseases at any point since diagnosis of iTTP.
Figure 1.
Study flow and patients included in the analysis. CVD, cardiovascular disease.
The primary study outcome was incident MACEs occurring within 10 years of the baseline visit used for risk score calculation. MACE events were only counted during iTTP remission (platelet count ≥ 150 × 109/L and lactate dehydrogenase <1.5 times upper limit of normal for ≥30 days after cessation of therapeutic plasma exchange [12]). MACE events occurring during acute iTTP episodes were excluded, as they are recognized complications of the disease and are not systematically evaluated or documented consistently in the medical record [7]. Events occurring beyond 10 years from the baseline visit were censored. Similar to prior studies in iTTP, MACE was defined as a composite of nonfatal MI, fatal MI, stroke, and cardiac revascularization (percutaneous coronary intervention or coronary artery bypass grafting surgery) [7]. Nonfatal MI was defined as either non-ST–elevation MI or ST-elevation MI that did not lead to death during the reference hospitalization/clinical encounter. Non-ST–elevation MI and ST-elevation MI were confirmed with troponin elevation, serial electrocardiogram evaluation, and documented symptoms such as angina on chart review. Stroke was defined as documented new neurological deficit(s) with corresponding ischemic lesion(s) on brain magnetic resonance imaging [13]. The determination of stroke was restricted to patients with focal neurologic symptoms lasting >24 hours consistent with the World Health Organization definition of stroke [14]. Transient ischemic attacks, defined as documented neurologic symptoms occurring for <24 hours, were not included in the primary endpoint as these may be inconsistently documented. Cardiac revascularization procedures were identified from the medical record. We included only MACE events occurring during a clinical remission as defined in Cuker et al. [12].
2.3. Calculation of predicted CVD risk
Two established CVD risk tools were utilized to predict cardiovascular risk; the 2008 FHS global CVD function and the ACC/AHA ASCVD risk estimator plus [10,11]. Predicted risk scores were generated using the respective online risk calculators. Risk factors included in the FHS calculator are sex, age, diabetes mellitus, smoking, systolic blood pressure, total cholesterol, HDL cholesterol, and antihypertensive treatment [15]. Risk factors included in the ACC/AHA ASCVD risk estimator plus are sex, race, age, diabetes mellitus, smoking, systolic blood pressure, total cholesterol, HDL cholesterol, LDL cholesterol, and antihypertensive treatment [10].
2.4. Statistical analysis
Baseline demographic and clinical characteristics of the iTTP cohorts, including those with and without MACEs, were summarized. Continuous variables were expressed as median (interquartile range) and categorical variables as numbers and percentages. Comparative analyses between groups were conducted using the chi-squared test for categorical variables and the Mann–Whitney U-test for continuous variables. Baseline characteristics included demographic data (age, sex, race, and ethnicity), clinical parameters (total cholesterol, HDL and LDL cholesterol levels, body mass index, systolic blood pressure), and medical history (antihypertensive medication use, smoking status, presence of diabetes mellitus, lupus or other autoimmune diseases, known vascular disease, deep vein thrombosis, and chronic kidney disease). Follow-up time was calculated from the baseline visit until the first MACE event, last clinical contact in case of loss to follow-up, or date of death.
The discrimination of the ACC/AHA ASCVD risk estimator plus and the FHS global CVD calculator was evaluated using the c-statistic, analogous to the area under the receiver operating characteristic (ROC) curve, by comparing the observed MACE events in our cohort versus the predicted 10-year incidence. Patients without MACEs or censored prior to 10-year follow-up were considered as no events. A c-statistic of 0.5 indicates no discrimination (equivalent to random chance), whereas a value of 1.0 represents perfect discrimination. The closest top-left point method to the ROC curve was applied to determine the optimal elevated-risk threshold, minimizing the distance to the ideal point on the ROC curve [16]. At this optimal threshold, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for both risk scores.
For both the FHS and ACC/ASCVD risk functions, calibration was assessed using the Hosmer–Lemeshow test, with a P value < .05 indicating poor fit/calibration. The iTTP cohort was divided into deciles of predicted 10-year risk of CVD according to each risk function and observed versus predicted probabilities of cardiovascular events across deciles of predicted 10-year risk and compared using calibration plots showing mean predicted risk against the mean observed event rate for each decile. Calibration was then recalculated in a similar manner where participants were grouped in the following 3 categories of predicted risk instead of deciles: <5%, 5% to 20%, and >20%. According to the 2013 ACC/AHA Guideline on the assessment of cardiovascular risk, a predicted 10-year ASCVD risk of <5% is considered “low risk,” 5% to <7.5% is “borderline risk,” 7.5% to <20% is “intermediate risk,” and ≥20% is classified as “high risk” for which intensive preventive therapy is generally recommended (eg, statin therapy). For our analysis, we consolidated the 5% to 7.5% and 7.5% to 20% subcategories into a single “moderate risk” group (5%-20%) for ease of interpretation and to ensure sufficient sample size within each group [10].
Since the CVD risk estimators were derived and validated in a population aged 40 to 79 without prior CVD events, we performed additional sensitivity analyses restricted to the subgroup of patients with iTTP between 40 to 79 years of age and without prior history of stroke or MI temporally related or unrelated to acute iTTP episodes.
3. Results
3.1. Patient cohort and incident MACE risk
The cohort comprised 135 patients with iTTP with a median observation period of 3.8 years (IQR, 1.2-8.0). During clinical remission, 51 patients (37.8%) experienced MACEs, of whom 46 (34.1%) had a MACE event within 10 years of their baseline visit. The median time from baseline evaluation to the first MACE event was 2.3 years. Among these 46 patients, stroke was the most common event (n = 34, 73.9%), followed by nonfatal MI (n = 9, 19.6%) and cardiac revascularization (n = 3, 6.5%). Mean age at baseline was 47.7 ± 17.4 years, 71.8% were women, and 65.2% self-identified as Black. Demographic and clinical characteristics of the cohort are summarized in Table 1. A significantly higher proportion of the MACE group had a history of chronic kidney disease (50% vs 29.2%, P = .03). Other characteristics including parameters such as systolic blood pressure, total and LDL cholesterol, and rate of comorbidities such as hypertension, diabetes mellitus, smoking, and lupus or autoimmune diseases, were not significantly different between the 2 groups (Table 1).
Table 1.
Baseline characteristics of TTP patients.
| Characteristic | Total cohort (N = 135) | Cohort with MACE (n = 46) | Cohort without MACE (n = 89) | P value |
|---|---|---|---|---|
| Age, y, median (IQR) | 60.0 (44.0, 70.0) | 62.0 (48.75,70.0) | 58.0 (41.0, 70.0) | .41 |
| Sex, % | .82 | |||
| Male | 28.2% | 30.4% | 27.0% | |
| Female | 71.8% | 69.6% | 73.0% | |
| Race, % | .91 | |||
| Black | 65.2% | 67.4% | 64.0% | |
| White | 29.6% | 28.3% | 30.4% | |
| Other | 5.2% | 4.3% | 5.6% | |
| Ethnicity, % | .78 | |||
| Not Hispanic or Latino | 79.2% | 76.1% | 80.9% | |
| Hispanic or Latino | 2.2% | 2.1% | 2.2% | |
| Other/unknown | 18.6 % | 21.8% | 16.9% | |
| Total cholesterol, mg/dL, median (IQR) | 170.0 (145.5, 205.8) | 169.0 (147.0, 205.0) | 173.0 (145.0, 208.0) | .68 |
| HDL cholesterol, mg/dL, median (IQR) | 46.0 (40.0, 59.0) | 46.0 (39.2, 56.8) | 46.0 (40.0, 59.0) | .57 |
| LDL cholesterol, mg/dL, median (IQR) | 93.0 (80.0, 121.8) | 92.5 (81.0, 116.5) | 94.0 (78.8, 124.0) | .82 |
| Systolic blood pressure, median (IQR) | 129.0 (118.0, 145.0) | 135.0 (120.0, 151.0) | 127.5 (118.0, 140.8) | .13 |
| Diastolic blood pressure, median (IQR) | 75.0 (67.0, 83.5) | 74.0 (66.0, 85.0) | 75.0 (68.0, 82.8) | .68 |
| Antihypertensive medication, % | 63.0% | 71.7% | 58.4% | .18 |
| Smoker, % | 48.1% | 52.2% | 46.1% | .54 |
| Diabetes mellitus, % | 28.9% | 34.8% | 25.8% | .38 |
| Lupus or any autoimmune disease, % | 29.6% | 28.2% | 30.3% | .48 |
| History of DVT, % | 19.2% | 28.3% | 14.6% | .09 |
| History of CKD, % | 36.3% | 50% | 29.2% | .03 |
CKD, chronic kidney disease; DVT, deep vein thrombosis; HDL, high-density lipoprotein; IQR, interquartile range; LDL, low-density lipoprotein; MACE, major adverse cardiovascular event; TTP, thrombotic thrombocytopenic purpura.
3.2. Discrimination of cardiovascular risk prediction scores
The predictive performance of the ACC/AHA ASCVD and FHS global CVD models for cardiovascular risk in the iTTP cohort was assessed using c-statistics. The ACC/AHA ASCVD model demonstrated a c-statistic of 0.54 (95% CI, 0.46-0.65), indicating poor discrimination in predicting cardiovascular events. Similarly, the FHS global CVD model showed a c-statistic of 0.52 (95% CI, 0.41-0.63), suggesting limited predictive ability. Using the closest top-left point method, the optimal thresholds were identified as 6.50% for the ACC/AHA ASCVD model and 5.92% for the 2008 FHS global CVD model. The sensitivity, specificity, PPV, and NPV at these thresholds are summarized in Table 2. The FHS global CVD model demonstrated better sensitivity (69.6% vs 56.5%) and NPV (67.9% vs 64.9%), while the ACC/AHA ASCVD model exhibited slightly higher specificity (49.4% vs 39.3%). Both models demonstrated suboptimal predictive accuracy.
Table 2.
Validation metrics.
| Metric | ACC/AHA ASCVD function (optimal threshold, 6.50%) | 2008 FHS global CVD function (Optimal threshold, 5.92%) |
|---|---|---|
| Sensitivity | 56.5% (95% CI, 46.4%-67.1%) | 69.6% (95% CI, 56.4%-80.8%) |
| Specificity | 49.4% (95% CI, 39.4%-59.0%) | 39.3% (95% CI, 29.6%-50.0%) |
| PPV | 36.6% (95% CI, 27.8%-45.3%) | 37.2% (95% CI, 26.7%,47.7%) |
| NPV | 64.9% (95% CI, 55.2%-73.8%) | 67.9% (95% CI, 58.1%-77.4%) |
ACC/AHA, American College of Cardiology/American Heart Association; ASCVD, atherosclerotic cardiovascular disease; FHS, Framingham Heart Study; CI, confidence interval; CVD, cardiovascular disease risk calculator; NPV, negative predictive value; PPV, positive predictive value.
3.3. Calibration of models and observed versus predicted risk
Model calibration was assessed using the Hosmer–Lemeshow test. Both models exhibited significant differences between predicted and observed risks, indicating poor calibration in the iTTP cohort. The ACC/AHA ASCVD model had a Hosmer–Lemeshow chi-squared statistic of 44.2 (df = 8, P < .0001), while the 2008 FHS global CVD model had a chi-squared statistic of 32.9 (df = 8, P < .0001). Observed risk consistently exceeded predicted risk across all deciles of predicted risk for both models, as shown in Figure 2. Additionally, Figure 3 illustrates observed and predicted 10-year risk for each model by predicted risk category (<5%, 5%-20%, and >20%), demonstrating that observed risk surpassed predicted risk across all groups.
Figure 2.
Observed and predicted 10-year risk by decile of predicted risk for (A) American College of Cardiology/American Heart Association atherosclerotic cardiovascular disease risk estimator plus; (B) Framingham Heart Study global cardiovascular disease risk calculator.
Figure 3.
Observed and predicted 10-year risk by predicted risk group (Low Risk, <5%; Moderate Risk, 5%-20%; High Risk, >20%). (A) American College of Cardiology/American Heart Association atherosclerotic cardiovascular disease risk estimator plus; (B) Framingham Heart Study global cardiovascular disease risk calculator.
3.4. Sensitivity analysis
In a sensitivity analysis of 85 patients aged 40 to 79, MACEs occurred in 30 patients (Table 3). The discrimination of both risk prediction tools remained poor, with the ACC/AHA ASCVD function yielding a c-statistic of 0.51 (95% CI, 0.38-0.63) and the FHS global CVD function showing a c-statistic of 0.52 (95% CI, 0.40-0.65). Calibration analysis continued to demonstrate significant underestimation of observed cardiovascular risk across all predicted risk groups (Figures 4 and 5). The ACC/AHA ASCVD model had a Hosmer–Lemeshow chi-squared statistic of 43.6 (df = 8, P < .0001), while the 2008 FHS global CVD model had a chi-squared statistic of 39.8 (df = 8, P < .0001). At the optimized risk thresholds, sensitivity, specificity, and predictive values were similarly suboptimal (Table 4). The ASCVD model had a sensitivity of 54.0% and specificity of 50.9%, while the FHS model demonstrated a sensitivity of 53.3% and specificity of 56.4%.
Table 3.
Baseline characteristics of TTP patients in sensitivity analysis.
| Characteristic | Total cohort (N = 85) | Cohort with MACE (n = 30) | Cohort without MACE (n = 55) | P value |
|---|---|---|---|---|
| Age, y, median (IQR) | 54.0 (44.0,64.0) | 56.0 (45.0, 64.8) | 54.0 (44.0, 63.0) | .90 |
| Sex, % | .51 | |||
| Male | 30.6% | 36.7% | 27.3% | |
| Female | 69.4% | 63.3% | 72.7% | |
| Race, % | .44 | |||
| Black | 62.4% | 56.7% | 65.4% | |
| White | 34.1% | 36.7% | 32.8% | |
| Other | 3.5% | 6.6% | 1.8% | |
| Ethnicity, % | .29 | |||
| Not Hispanic or Latino | 76.5% | 70.0% | 80.% | |
| Hispanic or Latino | 1.2% | 3.3% | 0% | |
| Other/unknown | 22.3% | 26.7% | 20% | |
| Total cholesterol, mg/dL, median (IQR) | 173.0 (145.0, 210.8) | 169.0 (150.0, 205.0) | 177.0 (142.5, 211.5) | .71 |
| HDL cholesterol, mg/dL, median (IQR) | 46.0 (41.0, 62.0) | 46.0 (41.0, 59.8) | 47.0 (40.5,65.0) | .25 |
| LDL cholesterol, mg/dL, median (IQR) | 94.0 (79.8, 124.0) | 94.0 (81.0, 106.8) | 96.0 (76.5, 131.3) | .51 |
| Systolic blood pressure, median (IQR) | 131.0 (120.0, 146.8) | 144.0 (122.0, 162.0) | 128.0 (120.0, 140.0) | .05 |
| Diastolic blood pressure, median (IQR) | 73.5 (67.0, 82.0) | 71.0 (64.0, 85.0) | 74.0 (68.0, 81.0) | .62 |
| Antihypertensive medication, % | 67.1% | 70% | 65.5% | .85 |
| Smoker, % | 50.6% | 50.0% | 50.9% | .91 |
| Diabetes mellitus, % | 32.9% | 36.7% | 30.9% | .76 |
| Lupus or any autoimmune disease, % | 22.4% | 13.3% | 27.3% | .22 |
| History of DVT, % | 17.6% | 23.3% | 14.5% | .41 |
| History of CKD, % | 38.8% | 40.0% | 38.2% | .77 |
CKD, chronic kidney disease; DVT, deep vein thrombosis; HDL, high-density lipoprotein; IQR, interquartile range; LDL, low-density lipoprotein; MACE, major adverse cardiovascular event; TTP, thrombotic thrombocytopenic purpura.
Figure 4.
Observed and predicted 10-year risk by decile of predicted risk in sensitivity analysis. (A) American College of Cardiology/American Heart Association atherosclerotic cardiovascular disease risk estimator plus; B. Framingham Heart Study global cardiovascular disease function.
Figure 5.
Observed and predicted 10-year risk by predicted risk group in sensitivity analysis (Low Risk, <5%; Moderate Risk, 5%-20%; High Risk, >20%). (A) American College of Cardiology/American Heart Association atherosclerotic cardiovascular disease risk estimator plus; (B) Framingham Heart Study global cardiovascular disease risk calculator.
Table 4.
Validation metrics in sensitivity analysis.
| Metric | ACC/AHA ASCVD function (optimal threshold, 8.2%) | 2008 FHS global CVD function (optimal threshold, 13.6%) |
|---|---|---|
| Sensitivity | 54.0% (95% CI, 35.5%-71.4%) | 53.3% (95% CI, 34.8%-71.0%) |
| Specificity | 50.9% (95% CI, 38.2%-63.8%) | 56.4% (95% CI, 43.4%-69.3%) |
| PPV | 37.2% (95% CI, 22.0%-53.2%) | 40.0% (95% CI, 25.5%-56.7%) |
| NPV | 66.6% (95% CI, 52.3%-80.5%) | 68.9% (95% CI, 54.5%-82.6%) |
ACC/AHA, American College of Cardiology/American Heart Association; ASCVD, atherosclerotic cardiovascular disease; FHS, Framingham Heart Study; CI, confidence interval; CVD, cardiovascular disease risk calculator; NPV, negative predictive value; PPV, positive predictive value.
We also compared the characteristics of patients included in the sensitivity analysis (n = 85) and those who were excluded from the sensitivity analyses (n = 50) based on prior cardiovascular events or age outside the range of 40 to 79 years (Supplementary Table).
4. Discussion
iTTP survivors are at increased risk of experiencing MACEs, and CVD is a leading cause of morbidity and mortality in this population [4]. We demonstrate for the first time that 2 commonly used cardiovascular risk assessment tools, the ACC/AHA ASCVD risk estimator plus and the and FHS global CVD (2008) calculator, do not accurately predict the risk of MACEs among iTTP survivors in clinical remission. Both models showed poor discrimination, with the ACC/AHA ASCVD model having a c-statistic of 0.54 and the FHS global CVD model a c-statistic of 0.52, and poor calibration evidenced by calibration plots that demonstrated observed cardiovascular risk consistently exceeded predicted risk across most deciles and risk categories. A c-statistic <0.7 is considered poor, 0.7 to 0.8 is moderate (acceptable), and >0.8 is considered good discrimination. A well-calibrated model will produce predictions that align well with observed outcomes across deciles or categories of risk. In contrast, poor calibration indicates a consistent mismatch, such as systematic underestimation or overestimation of risk [17,18]. These results highlight the inadequacy of using traditional cardiovascular risk models to stratify iTTP patients who may require intensified preventive strategies to reduce elevated cardiovascular risk, such as statins, aspirin or ADAMTS-13-directed therapies. Notably, in the sensitivity analysis restricted to individuals aged 40 to 79 years, systolic blood pressure was higher in the MACE group and approached statistical significance (P = .05), suggesting a possible contributory role of elevated systolic blood pressure in cardiovascular risk among older iTTP survivors. Our results also suggest that the increased cardiovascular risk in iTTP survivors is not fully explained by traditional risk factors, and further research is needed to identify more tailored approaches to predict and manage cardiovascular risk in this vulnerable population.
Previous studies have highlighted the heightened incidence of MACEs in iTTP survivors compared with the general population. Factors such as increasing age, diabetes mellitus, and Black race have been identified as risk factors for MACEs [4,7]. This study is novel in addressing the applicability of cardiovascular risk calculators specifically in iTTP survivors, a high-risk population that has previously been overlooked in this context. The applicability of CVD risk prediction models across diverse populations, particularly those different from the original development cohorts, has been a subject of discussion [19]. Among these tools, the FHS models have been the most commonly utilized tools for estimating an individual’s absolute risk of developing a specific CVD outcome within a defined timeframe [20,21]. The introduction of the ACC/AHA guidelines on CVD risk assessment and cholesterol management brought a pivotal shift in prevention strategies for the general population by introducing a novel risk prediction model tailored to a broader and more representative population focusing on ASCVD [[22], [23], [24]]. However, the applicability and performance of these established models in the context of thrombotic and autoimmune disorders such as iTTP remain poorly understood. The ACC/AHA guidelines emphasize the importance of validating these tools in varied settings. Our findings highlight the limitations of these traditional risk prediction tools when applied to individuals with iTTP [25].
The elevated cardiovascular risk observed in iTTP survivors is likely due to the interplay of various pathophysiological mechanisms that go beyond conventional cardiovascular risk factors. Cardiovascular events occur earlier and more often in iTTP survivors [7], likely reflecting a phenotype of accelerated vascular aging [5], which is early onset and rapid progression of structural and functional vascular changes such as increased arterial stiffness, endothelial dysfunction, and early atherosclerosis [5]. This may be driven by persistent low-grade endothelial injury, chronic inflammation, and persistent ADAMTS-13 deficiency, which all contribute to heightened susceptibility to ischemic events such as MI and stroke [26]. A similar pattern of early-onset vascular complications has been observed in congenital TTP, which is caused by a severe inherited deficiency of ADAMTS-13, and in sickle cell disease, in which repeated vaso-occlusive episodes and chronic inflammation promote endothelial damage. While the underlying etiologies differ, both congenital TTP and sickle cell disease demonstrate how chronic vascular injury and thrombotic tendencies can drive early cardiovascular morbidity. ADAMTS-13 plays a central role in preserving vascular integrity by preventing the accumulation of ultra-large von Willebrand factor multimers [27]. In context of iTTP, persistent partial or complete ADAMTS-13 deficiency leads to accumulation of larger von Willebrand factor multimers, which further causes enhanced platelet aggregation, complement activation, and endothelial dysfunction and is subsequently a risk factor for stroke and all-cause mortality in iTTP remission [28,29]. This premise is further supported by large population-based studies that have established an association between reduced ADAMTS-13 activity and increased risks of stroke, silent cerebral infarction, and all-cause mortality in the general population (without iTTP) [9].
To improve cardiovascular risk prediction in iTTP survivors, there is a need to develop disease-specific models that account for the unique pathophysiology of this condition. ADAMTS-13 activity during clinical remission, number of prior acute TTP episodes, therapy received for TTP (eg, plasma exchange, corticosteroids, rituximab, caplacizumab, etc.), and concurrent autoimmune conditions are all iTTP-specific factors that may contribute to cardiovascular risk in this population. Indeed, there has been a recent call for inclusion of TTP in primary prevention of stroke guidelines, along with a recommendation for inclusion of ADAMTS-13 activity in the work-up of cryptogenic stroke [30]. By integrating these and perhaps other biomarkers, future cardiovascular risk models could better capture the complex interplay of thrombotic, inflammatory, and endothelial factors in iTTP, providing a more tailored approach to risk stratification.
Limitations of the study include the retrospective design, which led to some patients being excluded from the analysis due to missing data on indicators required for the FHS or AHA/ACC risk estimators. This may introduce selection bias for a higher risk group who were thought to be candidates for testing for cholesterol disorders by their treating clinicians. The use of lipid-lowering therapy was not systematically documented in this cohort, although most patients had lipid levels below typical thresholds to institute therapy. The follow-up time of our study also spans an interval in which the standard of care for iTTP management as well as prevention of CVD has changed. Additionally, routine ADAMTS-13 monitoring therapy was not started until after 2018, when observational studies regarding the utility of preemptive rituximab were published [31]. Therefore, we are unable to incorporate average remission ADAMTS-13 activity into the present analysis, which needs to be done in future prospective studies. The reliance on a single-center cohort at a tertiary center may not capture the heterogeneity of iTTP survivors managed at other institutions with differing demographic profiles, burden of comorbidities, and varying treatment protocols or follow-up practices. The modest sample size, which is reflective of the rarity of iTTP, also limits statistical power. Many patients with iTTP are <40 years, which is the lower age limit for many CVD risk calculators, or have had events such as stroke during iTTP episodes. These patients were included in our primary analysis. The ACC/AHA ASCVD and FHS global CVD models were developed for the general population and may inherently bias estimates toward the null hypothesis when applied to a high-risk population like iTTP survivors. Despite these limitations, the study provides valuable insights into the inadequacies of current cardiovascular risk calculators in this unique patient cohort and underscores the need for tailored risk assessment tools. In conclusion, the 2008 FHS global CVD function and the ACC/AHA ASCVD function markedly underestimate cardiovascular risk in iTTP survivors. Relying on general population CVD risk calculators for iTTP survivors may lead to undertreatment of cardiovascular risk factors, potentially contributing to poor long-term outcomes in this vulnerable population.
Additional tools to help risk-stratify iTTP patients at highest risk for cardiovascular morbidity need to be developed and validated to allow personalized cardiovascular risk management in the expanding population of iTTP survivors. Until these are available, it is likely prudent to aggressively optimize traditional cardiovascular risk factors for all patients with iTTP.
Acknowledgments
Funding
S.C. is supported by National Institutes of Health’s National Heart Lung and Blood Institute grants (K99HL150594 and R00HL172303) and an American Society of Hematology Scholar Award. S.S. is supported by the Robert A. Winn Excellence in Clinical Trials Career Development Award.
Author contributions
B.J. drafted the manuscript and performed statistical analyses. J.B. collected registry data and reviewed the manuscript. J.M, V.N., and A.L. critically reviewed and edited the manuscript. S.C. and S.S. conceptualized the study, performed analyses, and critically reviewed and edited the manuscript.
Relationship disclosure
S.C. reports advisory board participation or consultancy for Alexion, Sanofi Genzyme, Novartis, Kyowa Kirin, Star, Sobi, and Takeda; steering committee participation for Takeda and Sobi; and her institution has received research funding on her behalf from Sanofi, Sobi, and Takeda. S.S. reports advisory board participation or consultancy for Alexion, Sanofi Genzyme; he has received clinical trial research funding from Novartis.
Footnotes
Handling Editor: Dr John Semple
The online version contains supplementary material available at https://doi.org/10.1016/j.rpth.2025.103005
Supplementary material
References
- 1.Sukumar S., Lämmle B., Cataland S.R. Thrombotic thrombocytopenic purpura: pathophysiology, diagnosis, and management. J Clin Med. 2021;10:536. doi: 10.3390/jcm10030536. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Vesely S.K., George J.N., Lämmle B., Studt J.D., Alberio L., El-Harake M.A., et al. ADAMTS13 activity in thrombotic thrombocytopenic purpura–hemolytic uremic syndrome: relation to presenting features and clinical outcomes in a prospective cohort of 142 patients. Blood. 2003;102:60–68. doi: 10.1182/blood-2003-01-0193. [DOI] [PubMed] [Google Scholar]
- 3.George J.N. TTP: long-term outcomes following recovery. Hematology Am Soc Hematol Educ Program. 2018;2018:548–552. doi: 10.1182/asheducation-2018.1.548. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Sukumar S., Brodsky M., Hussain S., Yanek L., Moliterno A., Brodsky R., et al. Cardiovascular disease is a leading cause of mortality among TTP survivors in clinical remission. Blood Adv. 2022;6:1264–1270. doi: 10.1182/bloodadvances.2020004169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sukumar S., Mazepa M.A., Chaturvedi S. Cardiovascular disease and stroke in immune TTP–challenges and opportunities. J Clin Med. 2023;12:5961. doi: 10.3390/jcm12185961. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Page E.E., Kremer Hovinga J.A., Terrell D.R., Vesely S.K., George J.N. Thrombotic thrombocytopenic purpura: diagnostic criteria, clinical features, and long-term outcomes from 1995 through 2015. Blood Adv. 2017;1:590–600. doi: 10.1182/bloodadvances.2017005124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Brodsky M.A., Sukumar S., Selvakumar S., Yanek L., Hussain S., Mazepa M.A., et al. Major adverse cardiovascular events in survivors of immune-mediated thrombotic thrombocytopenic purpura. Am J Hematol. 2021;96:1587–1594. doi: 10.1002/ajh.26341. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Selvakumar S., Liu A., Chaturvedi S. Immune thrombotic thrombocytopenic purpura: spotlight on long-term outcomes and survivorship. Front Med (Lausanne) 2023;10 doi: 10.3389/fmed.2023.1137019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Upreti H., Kasmani J., Dane K., Braunstein E.M., Streiff M.B., Shanbhag S., et al. Reduced ADAMTS13 activity during TTP remission is associated with stroke in TTP survivors. Blood. 2019;134:1037–1045. doi: 10.1182/blood.2019001056. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Goff D.C., Lloyd-Jones D.M., Bennett G., Coady S., D’Agostino R.B., Gibbons R., et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129:S49–S73. doi: 10.1161/01.cir.0000437741.48606.98. [DOI] [PubMed] [Google Scholar]
- 11.Kannel W.B., McGee D., Gordon T. A general cardiovascular risk profile: the Framingham study. Am J Cardiol. 1976;38:46–51. doi: 10.1016/0002-9149(76)90061-8. [DOI] [PubMed] [Google Scholar]
- 12.Cuker A., Cataland S.R., Coppo P., de la Rubia J., Friedman K.D., George J.N., et al. Redefining outcomes in immune TTP: an international working group consensus report. Blood. 2021;137:1855–1861. doi: 10.1182/blood.2020009150. [DOI] [PubMed] [Google Scholar]
- 13.González R.G. Clinical MRI of acute ischemic stroke. J Magn Reson Imaging. 2012;36:259–271. doi: 10.1002/jmri.23595. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Coupland A.P., Thapar A., Qureshi M.I., Jenkins H., Davies A.H. The definition of stroke. J R Soc Med. 2017;110:9–12. doi: 10.1177/0141076816680121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Medscape Framingham Risk Score (2008) 2020. https://reference.medscape.com/calculator/252/framingham-risk-score-2008
- 16.Perkins N.J., Schisterman E.F. The inconsistency of “optimal” cutpoints obtained using two criteria based on the receiver operating characteristic curve. Am J Epidemiol. 2006;163:670–675. doi: 10.1093/aje/kwj063. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nahm F.S. Receiver operating characteristic curve: overview and practical use for clinicians. Korean J Anesthesiol. 2022;75:25–36. doi: 10.4097/kja.21209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Huang Y., Li W., Macheret F., Gabriel R.A., Ohno-Machado L. A tutorial on calibration measurements and calibration models for clinical prediction models. J Am Med Inform Assoc. 2020;27:621–633. doi: 10.1093/jamia/ocz228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Triant V.A., Perez J., Regan S., Massaro J.M., Meigs J.B., Grinspoon S.K., et al. Cardiovascular risk prediction functions underestimate risk in HIV infection. Circulation. 2018;137:2203–2214. doi: 10.1161/CIRCULATIONAHA.117.028975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.D’Agostino R.B., Sr., Grundy S., Sullivan L.M., Wilson P., CHD Risk Prediction Group Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA. 2001;286:180–187. doi: 10.1001/jama.286.2.180. [DOI] [PubMed] [Google Scholar]
- 21.D’Agostino R.B., Pencina M.J., Massaro J.M., Coady S. Cardiovascular disease risk assessment: insights from Framingham. Glob Heart. 2013;8:11–23. doi: 10.1016/j.gheart.2013.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Arnett D.K., Blumenthal R.S., Albert M.A., Buroker A.B., Goldberger Z.D., Hahn E.J., et al. 2019 ACC/AHA guideline on the primary prevention of cardiovascular disease: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation. 2019;140:e596–e646. doi: 10.1161/CIR.0000000000000678. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Nayor M., Vasan R.S. Recent update to the US cholesterol treatment guidelines: a comparison with international guidelines. Circulation. 2016;133:1795–1806. doi: 10.1161/CIRCULATIONAHA.116.021407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Stone N.J., Robinson J.G., Lichtenstein A.H., Bairey Merz C.N., Blum C.B., Eckel R.H., et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129:S1–S45. doi: 10.1161/01.cir.0000437738.63853.7a. [DOI] [PubMed] [Google Scholar]
- 25.Goff D.C., Lloyd-Jones D.M., Bennett G., Coady S., D’Agostino R.B., Sr., Gibbons R., et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2014;63:2935–2959. doi: 10.1016/j.jacc.2013.11.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Borogovac A., George J.N. Stroke and myocardial infarction in hereditary thrombotic thrombocytopenic purpura: similarities to sickle cell anemia. Blood Adv. 2019;3:3973–3976. doi: 10.1182/bloodadvances.2019000959. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Joly B.S., Coppo P., Veyradier A. Thrombotic thrombocytopenic purpura. Blood. 2017;129:2836–2846. doi: 10.1182/blood-2016-10-709857. [DOI] [PubMed] [Google Scholar]
- 28.Page E.E., Kremer Hovinga J.A., Terrell D.R., Vesely S.K., George J.N. Clinical importance of ADAMTS13 activity during remission in patients with acquired thrombotic thrombocytopenic purpura. Blood. 2016;128:2175–2178. doi: 10.1182/blood-2016-06-724161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Tsai H.M. Thrombotic thrombocytopenic purpura: a thrombotic disorder caused by ADAMTS13 deficiency. Hematol Oncol Clin North Am. 2007;21:609–632. doi: 10.1016/j.hoc.2007.06.003. v. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Picod A., Joly B.S., Veyradier A., Coppo P. Thrombotic thrombocytopenic purpura: a missed cause in stroke prevention guidelines. Stroke. 2025;56:1104–1105. doi: 10.1161/STROKEAHA.124.049952. [DOI] [PubMed] [Google Scholar]
- 31.Jestin M., Benhamou Y., Schelpe A.S., Roose E., Provôt F., Galicier L., et al. Preemptive rituximab prevents long-term relapses in immune-mediated thrombotic thrombocytopenic purpura. Blood. 2018;132:2143–2153. doi: 10.1182/blood-2018-04-840090. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.