Platelet variability index: a measure of platelet count fluctuations in patients with immune thrombocytopenia

Abstract

Fluctuations in platelet count levels over time may help distinguish immune thrombocytopenia (ITP) from other causes of thrombocytopenia. We derived the platelet variability (PVI) score to capture both the fluctuations in platelet count measurements and the severity of the thrombocytopenia over time. Raw PVI values, ranging from negative (less severe thrombocytopenia and/or low fluctuations) to positive (more severe thrombocytopenia and/or high fluctuations) were converted to an ordinal PVI score, from 0 - 6. We evaluated performance characteristics of the PVI score for consecutive adults with thrombocytopenia from the McMaster ITP Registry. We defined patients with definite ITP as those who achieved a platelet count response after treatment with intravenous immune globulin or high dose corticosteroids; and possible ITP as those who never received ITP treatment or did not respond to treatment. Of 841 thrombocytopenic patients, 104 had definite ITP, 398 had possible ITP, and 339 had non-ITP thrombocytopenia. The median PVI score was 5 (interquartile range [IQR] 5, 6) for definite ITP; 3 (1, 5) for possible ITP; and 0 (0, 2) for non-ITP. A high PVI score correlated with the diagnosis of definite ITP even when calculated at the patient's initial assessment, before any treatment had been administered. Platelet count fluctuations alone contributed to the specificity of the overall PVI score. The PVI score may help clinicians diagnose ITP among patients with thrombocytopenia.

Development and internal validation of a clinical prediction model for the diagnosis of immune thrombocytopenia

“…46 Previous studies have shown that a very low platelet count level and variability from one platelet count measurement to the next can help distinguish between ITP from non-immune thrombocytopenia. 25 Maximum MPV reflects a high IPF (percentage of reticulated circulating platelets) and has also been associated with the ITP diagnosis in previous studies. 8,47 We then Abbreviations: CI, confidence interval; ITP, immune thrombocytopenia.…”

“…The performance of the tool in patients who previously received treatment compared to treatment‐naïve patients requires further study. Another limitation was that one of the predictors, platelet count variability, required at least three platelet count measurements to provide information on platelet fluctuation 25 . In our study population, 523/919 (56.9%) patients with thrombocytopenia had at least three platelet count measurements within 5 years before initial medical consultation; thus, the model could not be applied to all patients (and this proportion may be different in other clinics).…”

“…We developed the Predict‐ITP Tool first by identifying clinical characteristics that were plausibly related to the diagnosis of ITP and determining their independent association with the ITP diagnosis by logistic regression 46 . Previous studies have shown that a very low platelet count level and variability from one platelet count measurement to the next can help distinguish between ITP from non‐immune thrombocytopenia 25 . Maximum MPV reflects a high IPF (percentage of reticulated circulating platelets) and has also been associated with the ITP diagnosis in previous studies 8,47 .…”

“…We compiled a list of candidate predictor variables that were plausibly associated with the ITP diagnosis, including age, sex, bleeding grades (minor and major) defined using the ITP Bleeding Scale 24 (Table S1 in supporting information), previous ITP treatments, lowest platelet count, platelet count variability 25 (which requires three or more platelet count measurements per patient), and mean platelet volume (MPV). Initially, we included platelet antibody testing in the model, but because this test was not routinely available in most practice settings, it was subsequently removed.…”

“…Initially, we included platelet antibody testing in the model, but because this test was not routinely available in most practice settings, it was subsequently removed. The platelet variability index 25 was a measure that we derived to capture both the severity of the thrombocytopenia and the magnitude of the platelet count variability. In this study, we used both components (thrombocytopenia severity and platelet count variability) separately to independently evaluate their contribution.…”