QRISK3: A New Marker for Thrombotic Events in Essential Thrombocythemia and Polycythemia Vera at Diagnosis and during Follow up

Introduction:

Thrombotic events (TE) are the most common cause of morbidity and mortality in patients with myeloproliferative neoplasms (MPN). Few specific predictive models have been developed in this to predict the risk of thrombosis, with patients primarily stratified according to age and history of previous thrombotic events (TE), so called ELN criteria. The QRISK3 model is a validated tool for assessing cardiovascular risk in the general population, but its potential role in MPN patients requires further assessment.

Methods:

We analysed electronic health record data from patients with essential thrombocythaemia (ET) and polycythaemia vera (PV), treated at our tertiary referral centre. Patients were required to have a minimum follow up period of 3 months. QRISK3 score was calculated using a web tool (https://qrisk.org/index.php) at diagnosis and at the time of TE. Receiver operating characteristic (ROC) analysis was performed to assess the role of QRISK3 in predicting TE at diagnosis, while the best threshold was identified through Youden’s test, considering the median QRISK3 at the time of TE. A commonly used QRISK3 score of >7.5%, was used as a cut-off to define high-risk patients.

Results:

Data from 937 patients (490 ET and 447 PV) with a median follow-up of 85 [3-481] months for ET and 95 [3-405] for PV were analysed. Median ages were 47 years [25-85] in ET and 46 [25-84] in PV. At diagnosis 355 ET patients (72.4%) in and 312 PV patients (69.8%) were low risk, as per ELN criteria.

The median baseline QRISK3 score was 3.2% for ET and 3.5% for PV (p=ns). QRISK3 scores at diagnosis were higher in patients categorized as high vs low-risk using standard ELN criteria, in both cohorts (4.2 vs. 2.4 [ET] and 8.8 vs. 2.8 [PV], p<0.001). 52 TE occurred in 46 ET individual patients. There were 21 (40%) arterial events (cerebrovascular accidents [CVA] n=14, myocardial infarctions [MI] n=8, and retinal artery occlusion n=1) and 31 (60%) venous (4 portal vein, pulmonary embolism [PE] n=2, 2 cerebral venous, 21 other venous thromboembolism [VTE]). There were 73 TE in 62 PV patients during follow-up (higher than ET cohort, p=0.032). In PV, 38 (52%) were arterial events (CVA n=27, MI n=16, and 35 (48%) were venous (splenic/portal vein n=5, PE n=5, cerebral venous n=3, other VTE n=17).

Using ROC analysis, QRISK3 of >7.5%, resulted in a sensitivity of 65% (CI 95%, 51–77%) and a specificity of 81% (CI 95%, 77–84%) for predicting the occurrence of TE in the ET cohort. For PV patients, a threshold of >7.5% showed a sensitivity of 42% (CI 95%, 30–54) and specificity of 78% (CI 95%, 73–82), while the optimal threshold for improving sensitivity in PV was >4.5% (sensitivity 88%, specificity 63%). Of 412 ET patients aged <65 years at diagnosis (age threshold as per BCSH guideline [median age 44 years, range 25-64]), 57 (13.8%) had prior/concurrent thrombosis and would be considered high risk. Excluding these patients, 17.5% (62) of patients recorded a QRISK3 score ≥7.5% and could be considered high-risk. In the PV cohort, 388 (86.7%) patients were aged <65 years at diagnosis (median 44 yrs, 25-64). 76 (19.6%) had a TE before or at diagnosis. Excluding these patients, 11.5% (36) of 312 patients <65 years had a QRISK3 score ≥7.5% at diagnosis and could be considered high-risk.

In the ET cohort, regarding the 46 patients with a TE occurring during follow-up, QRISK3 ≥7.5% identified those with increased risk (30 with QRISK3 ≥7.5% vs. 16 <7.5%, p<0.001). There was no significant difference in time to thrombosis, with these occurring at a median time from diagnosis of 31 [2-324] months in QRISK3 <7.5% vs. 41 [8-113] months in ≥7.5% (p=0.824). In PV, QRISK3 ≥7.5% calculated at diagnosis does not predict occurrence of TE (35 vs. 27, p=ns). However, median QRISK3 score at the time of thrombosis was 8 [2.3-26.7], with 47 out of 62 patients having a score ≥7.5%, suggesting the potential for use as a dynamic threshold to monitor identification of high-risk patients.

Conclusion:

We have demonstrated that the QRISK3 score predicts future TE in patients with ET and PV and should be considered for incorporation into assessment at diagnosis and during follow-up. Calculating QRISK3 also allows better definition of patients who require focused treatment of comorbidities and perhaps earlier intervention with cytoreductive therapies. Future studies could compare the use of QRISK3 with existing predictive tools such as IPSET-T in ET.