Quantification of the risk associated with positive antiphospholipid (aPL) antibodies has been problematic since previous assessments have relied on observing relatively small study populations or meta-analyses of collected larger groups. There would be a significant benefit to a tool that would permit analysis of clinical outcomes of large patient cohorts. We applied a novel analytical information system, named “Clinical Looking Glass” (CLG), which aggregates clinical, diagnostic, therapeutic and outcomes data from a single large academic health care center to address this question.
CLG was employed to collect all patients at a large tertiary care institution for whom antiphospholipid antibodies including lupus anticoagulant (LAC), anticardiolipin (aCL) antibodies and anti- β2-glycoproteinI (aβ2GPI) were performed. The immunoassays were grouped by isotype and values of >30U/L (>99th percentile) were considered positive. An untested control group was derived from outcomes data on all institutional Pioneer Accountable Care Organization patients, a cohort, whose coordinated care is managed by a single institution, permitted accurate and robust follow-up data. We used the CLG to track patients for a period of 583 days (maximum data available for control group) following their individual test results to identify a predefined thrombotic outcome. The outcome was defined by any encounter (inpatient, outpatient or ED), subsequent to the initial lab value date, which demonstrated a new thrombotic event.
Using CLG we were able to evaluate 20,593 unique patients who had some form of LA testing performed. The aCL assays were performed on the greatest number of patients (18,201) followed by the LAC (11,267) with the fewest number of patients tested for aβ2GPI (7,914). A total of 5,660 patients had testing for all three. Of all 11,267 patients having LAC testing performed, 754 patients had at least one positive result. Of these 25.9% went on to develop a thrombotic event during the follow-up period compared to 15.0% of LAC negative patients (p value <0.001) and only 1.64% of the ACO control group suffering an event (p value <0.001). The relative risk associated with LAC positivity over the control group was 14.75 (95% CI 13.6-19.1). All other APL antibodies also demonstrated statistically increased risk of thrombosis over the examined cohort control (RR ranging from 6.6-11.2), but each of these to a significantly lesser degree than when compared directly to the LAC (results summarized in adjoining table and graph).
This first large study of aPL assays with prospective data from a single clinical information system confirms previous observations from smaller studies that LAC is the most significant laboratory predictor of future thrombotic events. However, in contrast to previous studies, all aPL antibodies, including IgA, demonstrated a statistically increased risk over a control population with LAC positivity having statistically greater risk than all others. Interestingly, aCL IgM was the weakest predictor of a future thrombotic event. Additionally we demonstrate the utility of clinical analytical software tools to offer very powerful ways to test prior assumptions and obtain novel results on large cohorts of patients in “real world” settings.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.