LAUSR

To the Editor: Life-threatening bleeding is a rare event in immune thrombocytopenia (ITP) and intracranial hemorrhage (ICH) occurs in less than 1% but is associated with a mortality rate of 44% in adults. Even if no controlled study has been conducted in patients with life-threatening bleeding, high-dose corticosteroids, IVIg and platelet transfusions are widely used. In an emergency with uncontrolled bleeding, starting high dose of TPO-RAs can be considered to rapidly increase the platelet count to a level where the risk of severe bleeding is minimized. Such off-label strategy has not been systematically evaluated. To evaluate the safety of this “rescue therapy” we carried out a multicenter retrospective cohort study of ITP patients treated with maximal dose of romiplostim for bleeding emergency from 2017 to 2019 in centers belonging to French national network for adult ITP. We included patients who fulfilled the following criteria:(a) Platelet count <30 × 10/L; (b) Severe bleeding manifestations defined as a score > 8 according to the bleeding score previously reported by our group which takes into account cutaneous, mucosal and visceral bleeding; (c) No response to corticosteroids and IVIg; (d) Rescue therapy with maximal dose of romiplostim (ie, 10 μg/kg body weight). Response (R) and Complete response (CR) were defined according to standardized international criteria. We included 30 patients (15 men) with ITP and severe bleeding manifestations (Table S1, Figure S1). The median age was 65 years [range 18-77]. Twenty-seven patients had newly diagnosed ITP and three chronic ITP. All patients had severe bleeding symptoms, characterized by skin and mucosal bleedings. The median bleeding score was 20 [range 8-33], including intracranial hemorrhage (n = 9), visceral hemorrhage — metrorrhagia, macroscopic hematuria, gastrointestinal or peritoneal bleeding (n = 16) (some patients had several manifestations). Twelve patients had received red blood cell transfusions because of severe anemia related to bleeding. The median platelet count was 2 × 10/L [range 0-10] at the time of severe bleeding. All 30 patients had received corticosteroids and IVIg treatment [median dose 2 g/kg.bw, range 1-4] as first-line therapy, and had failed to achieve any response. Twenty patients (67%) received platelet transfusions. Twenty-six patients had received intravenous vinca alkaloid (VA), in median 8 days (range 3-36) after initial bleeding symptoms: either concomitantly (ie,:1-3 days to romiplostim administration) in 20 cases; or 4-14 days before in six cases (8 mg vinblastine, n = 24 and 1 mg/m2 vincristine n = 2). Four patients did not receive VA including two patients who had a contra-indication to VA. Eight patients had received one or two other concomitant therapies. Romiplostim was started at a median of 9 days [range 3-40] after the first bleeding episode. At that time, bleeding symptoms were severe (bleeding score > 8), and for all but two patients, the platelet count was ≤20 × 10/L. These two patients received repeated platelet transfusions because of cerebral bleeding, with platelet counts 63 and 47 × 10/L, respectively. All patients received a starting dose of romiplostim at 10 μg/kg.bw/week from the first injection (n = 26), or from the second administration after a first attenuated dose of 58 μg/kg.bw (n = 4). During the first month of follow-up, the median peak platelet count among responders was 531 × 10/L [range 70-3840]; for 16 patients, the platelet count was >500 × 10/L and for nine it was >1000 × 10/L. Patients with a platelet count >1000 × 10/L received empirically low-dose aspirin. Three serious adverse events were noted. One 52-year-old woman who was bedridden because of a muscular hematoma, experienced deep-vein thrombosis with an asymptomatic pulmonary embolism 13 days after IVIg treatment and 5 days after the first romiplostim injection; the platelet count was 629 × 10/L at the time of thrombosis. One 70-year-old woman was diagnosed with ischemic stroke in the left parietal and right temporal lobes 13 days after IVIg administration and 10 days after the first romiplostim injection; at that time. Her platelet count was then in normal range (239 × 10/L). No predisposing condition for arterial thrombo-embolic events was found. Lastly, one 67-year-old patient who presented with cerebellar hemorrhage when diagnosed with ITP, died of cerebellar oedema 1 month after initial surgical treatment with external ventricular drains which had to be replaced several times because of infection or occlusion. At that time, cerebellar hemorrhage was stable and complete response was achieved 3 weeks before. We then evaluated response in the 20 patients who received romiplostim and concomitant VA treatment (VA + ROMI group) and compared to an historical cohort of 22 subjects with severe bleeding with absence of response to steroids and IVIg and who received VA without romiplostim (VA group) from 2008 to 2014. The two groups were not closely similar since patients in the VA + ROMI group had more severe bleeding manifestations, with a higher bleeding score (18 [range 8-33] vs 10 [range 8-28]), more ICH (30% vs 5%) and visceral hemorrhage (58% vs 36%) (Table 1). The day of VA administration was considered the starting point for evaluation. During the 14 days period of analysis, steroids were maintained at stable dose in all patients. Additional treatments administered are detailed in Table S1. At day 7 after VA +/− ROMI initiation, there was no significant difference in terms of response between the two groups (VA + ROMI: 70% vs VA: 48%, P = .15), but complete response was Received: 9 September 2020 Revised: 23 October 2020 Accepted: 25 October 2020