Spectral Analysis of Platelet Oscillations in Cyclic Thrombocytopenia.

Abstract

Abstract 4459 IntroductionCyclic thrombocytopenia (CT) is a rare autoimmune condition characterized by predictable oscillations in platelet count levels. Periodic platelet cycles in CT have not been well characterized. MethodsWe studied three patients with wide platelet count fluctuations to identify patterns consistent with CT. Medical charts were reviewed and platelet levels and treatments were abstracted. Platelet counts were plotted over time for each patient. Spectral analyses were completed using the method of Lomb and Scargle to determine if platelet count fluctuations were random or had a stable period of oscillation. Where stable periods were identified (fixed time from peak to peak) the likelihood of those periods being random was calculated using the false-alarm probability.Table 1:Spectral Analysis (LS Periodogram) ResultsPatientAssessment periodPlatelet counts measurements (N)Primary Period (days)Spectral Power at Primary PeriodFalse-Alarm Probability*I10 years18441.916.7p= 0.00001II11 years14239.313.3p= 0.0003III6 years14023.014.2p= .00012** Platelet count fluctuations in Patient III subsequently became random.*the probability that the identified period is random (P<0.01 is significant). ResultsRegular periods of platelet oscillation were identified for each patient (Table 1). Periods varied between 23 and 42 days and were not random. Patient I had a 2-year remission induced by immune suppressive medications during which time the platelet count was normal and stable. Platelet count oscillation amplitudes were 350 ×109/L both at presentation of CT and at relapse. Gradual building and decaying of oscillations were noted at the onset of relapse, and remission, respectively. Patient IIMatching a sine curve to the platelet data demonstrated that platelet nadirs of <20 ×109/L could be predicted to within 4 days for every cycle up to one year in advance. Patient III had typical CT oscillations for 6 years, but thereafter platelet fluctuations became random and no regular period was observed. Treatment with eltrombopag, a thrombopoietin (TPO) mimetic agent for 5 days resulted in a peak platelet count peak in excess of 1200 ×109/L in this patient. ConclusionIn three patients with CT, we identified regular periodic oscillations in platelet counts that were non-random. Some patients have predictable cycles that may allow timed delivery of treatments such as TPO-mimetic agents. These findings may provide insight into the nature of the autoantibody in CT. Disclosures:Arnold:Hoffman-LaRoche: Research Funding; Amgen: Consultancy, Honoraria.