Derivation and external validation of the PLASMIC score for rapid assessment of adults with thrombotic microangiopathies: a cohort study

Background Among the syndromes characterised by thrombotic microangiopathy, thrombotic thrombocytopenic purpura is distinguished by a severe deficiency in the ADAMTS13 enzyme. Patients with this disorder need urgent treatment with plasma exchange. Because ADAMTS13 activity testing typically requires prolonged turnaround times and might be unavailable in resource-poor settings, a method to rapidly assess the likelihood of severe ADAMTS13 deficiency is needed. Methods All consecutive adult patients presenting to three large academic medical centres in Boston, MA, USA, with thrombotic microangiopathy and a possible diagnosis of thrombotic thrombocytopenic purpura between Jan 8, 2004, and Dec 6, 2015, were included in an ongoing multi-institutional registry (the Harvard TMA Research Collaborative). Univariate analysis was used to identify covariates for a logistic regression model predictive of severe ADAMTS13 deficiency (≤10% activity). A clinical point score was generated, and its diagnostic performance was assessed using internal and external validation cohorts and compared to clinical assessment alone. Findings 214 patients with thrombotic microangiopathy were included in the derivation cohort. A seven-component clinical prediction tool, termed the PLASMIC score, was developed and found to reliably assess the pretest probability of severe ADAMTS13 deficiency (C statistic 0·96, 95% CI 0·92–0·98). Our diagnostic model was reproducibly accurate in both the internal (0·95, 0·91–0·98) and external (0·91, 0·85–0·95) validation cohorts. The scoring system also more consistently diagnosed thrombotic microangiopathy due to severe ADAMTS13 deficiency than did standard clinical assessment, as measured by C statistic (0·96, 95% CI 0·92–0·98 for PLASMIC vs 0·83, 0·77–0·88 for clinical assessment; p<0·0001) and mean Brier score (0·065 for PLASMIC vs 0·111 for clinical assessment; mean paired difference 0·05, 95% CI 0·01–0·08; p<0·0001). When utilised in addition to clinical assessment, the PLASMIC score contributed significant discriminatory power (integrated discrimination improvement 0·24, 95% CI 0·11–0·37). Interpretation We have developed and validated a clinical prediction tool—the PLASMIC score—to stratify patients with thrombotic microangiopathy according to their risk of having severe ADAMTS13 deficiency. We have shown that this scoring system is superior to standard clinical assessment in addressing the diagnostic challenge presented by thrombotic microangiopathy. Its use, together with clinical judgment, may facilitate treatment decisions in patients for whom timely results of ADAMTS13 activity testing are unavailable. Funding The Luick Family Fund of Massachusetts General Hospital. Among the syndromes characterised by thrombotic microangiopathy, thrombotic thrombocytopenic purpura is distinguished by a severe deficiency in the ADAMTS13 enzyme. Patients with this disorder need urgent treatment with plasma exchange. Because ADAMTS13 activity testing typically requires prolonged turnaround times and might be unavailable in resource-poor settings, a method to rapidly assess the likelihood of severe ADAMTS13 deficiency is needed. All consecutive adult patients presenting to three large academic medical centres in Boston, MA, USA, with thrombotic microangiopathy and a possible diagnosis of thrombotic thrombocytopenic purpura between Jan 8, 2004, and Dec 6, 2015, were included in an ongoing multi-institutional registry (the Harvard TMA Research Collaborative). Univariate analysis was used to identify covariates for a logistic regression model predictive of severe ADAMTS13 deficiency (≤10% activity). A clinical point score was generated, and its diagnostic performance was assessed using internal and external validation cohorts and compared to clinical assessment alone. 214 patients with thrombotic microangiopathy were included in the derivation cohort. A seven-component clinical prediction tool, termed the PLASMIC score, was developed and found to reliably assess the pretest probability of severe ADAMTS13 deficiency (C statistic 0·96, 95% CI 0·92–0·98). Our diagnostic model was reproducibly accurate in both the internal (0·95, 0·91–0·98) and external (0·91, 0·85–0·95) validation cohorts. The scoring system also more consistently diagnosed thrombotic microangiopathy due to severe ADAMTS13 deficiency than did standard clinical assessment, as measured by C statistic (0·96, 95% CI 0·92–0·98 for PLASMIC vs 0·83, 0·77–0·88 for clinical assessment; p<0·0001) and mean Brier score (0·065 for PLASMIC vs 0·111 for clinical assessment; mean paired difference 0·05, 95% CI 0·01–0·08; p<0·0001). When utilised in addition to clinical assessment, the PLASMIC score contributed significant discriminatory power (integrated discrimination improvement 0·24, 95% CI 0·11–0·37). We have developed and validated a clinical prediction tool—the PLASMIC score—to stratify patients with thrombotic microangiopathy according to their risk of having severe ADAMTS13 deficiency. We have shown that this scoring system is superior to standard clinical assessment in addressing the diagnostic challenge presented by thrombotic microangiopathy. Its use, together with clinical judgment, may facilitate treatment decisions in patients for whom timely results of ADAMTS13 activity testing are unavailable.