Oral prednisolone is the conventional initial treatment for immune thrombocytopenia (ITP) [1–3]. It is able to increase platelet counts to 50 × 109 L−1 in two-thirds of patients within 1 week . However, in a non-randomized trial, pulse methylprednisolone not only resulted in a higher response, but also faster recovery . Furthermore, in a single-arm study, pulse dexamethasone showed 85% response rate with platelets above 100 × 109 L−1 after 1 week . Subsequent reports employed repeated pulse treatments and confirmed these excellent responses [6,7]. Comparing these two high-dose steroids, dexamethasone is cheaper, available in oral forms and does not require electrocardiogram monitoring. However, the efficacy of steroids must be weighed against the multiple toxicities that can arise through long-term use because adult ITP usually shows a chronic course. A single dexamethasone pulse that yields a durable response in 50% of patients  would possibly allow more rapid steroid reduction. Despite promising data, randomized controlled trials have been lacking.
This is a prospective randomized open-labelled trial. The inclusion criteria were adult primary ITP in Chulalongkorn Hospital with platelet counts below 20 × 109 L−1, or below 30 x 109 L−1 with clinical bleeding (hemorrhagic score ≥ 2 ). The exclusion criteria were secondary ITP, life-threatening bleeding (hemorrhagic score 5), steroid contraindications or steroid refractoriness. The subjects were assigned using shuffled concealed envelopes. The first arm received 10 mg dexamethasone every 6 h (orally or intravenously) for 4 days, followed by oral prednisolone 30 mg day−1. The second arm was given oral prednisolone 60 mg day−1. If severe thrombocytopenia persisted beyond 5 days, prednisolone patients could be transferred to dexamethasone treatment. Intravenous immunoglobulin was used in the event of life-threatening bleeding. Both regimens were continued for 14 days before tapering dosages. Prednisolone was decreased by 10 mg day−1 every 2 weeks until the dosage reached 30 mg day−1, and thereafter by 5 mg day−1 per month if platelet counts were above 50 × 109/L. Patients’ data were recorded after day 5, day 15 and 1 month of therapy and then monthly for 6 months.
The primary outcome was satisfactory response, defined as platelet count over 50 × 109 L−1 without clinical bleeding after day 5 of treatment. The secondary outcome was the average monthly doses of steroid during 6 months of follow-ups. For this, dexamethasone dose was converted to a prednisolone equivalence by dividing by 0.15. There was no intention to look at the long-term outcome of the patients.
During 2006–2007, 45 patients were screened. Nine patients were excluded due to refusal (3), prior treatments (2), HIV (2), systemic lupus erythematosus (1) or refractoriness (1). Eighteen patients were randomized to each arm. Baseline characteristics were similar between these two arms (Table 1). The mean age was 42 years. All new cases were symptomatic for less than 1 month and had not previously been treated. All relapsed patients had received only steroids with current doses of 5 mg day−1 prednisolone or lower, and none had been splenectomized. In the dexamethasone and prednisolone groups, five and six patients, respectively, displayed mild transaminase elevation.
|Dexamethasone (N = 18)*||Prednisolone (N = 18)*||P-value†|
|Sex, female||13 (72.2%)||15 (83.3%)|
|Age (year)||44.9 ± 19.5||39.5 ± 15.1||0.346|
|Bleeding score Grade 1, 2, 3, 4, 5||0, 14, 4, 0, 0 (2.3 ± 0.8)||2, 10, 4, 2, 0 (2.2 ± 0.4)||0.631|
|Baseline Platelet count (×109 L−1)||8.5 ± 8.6||10.3 ± 8.5||0.541|
|Relapsed ITP (Cases)|
(Durations of disease)
(4, 4, 3 years)
(3, 1 year)
|On day 5|
|Platelet ≥ 50 × 109 L−1‡||16 (88.8%)||6 (33.3%)||0.001|
|Platelet ≥ 30 × 109 L−1||17 (94.4%)||11 (61.1%)||0.041|
|Platelet count (×109 L−1)||117.4 ± 95.0||51.9 ± 54.1||0.016|
|Bleeding score ||0.4 ± 0.51||0.4 ± 0.51|
|Bleeding score > 1||1||0|
|On day 15|
|Platelet count (×109 L−1)||183.2 ± 102.5||114.5 ± 96.9||0.046|
|Bleeding score > 0||0||0|
|At Month 1 Platelet count (×109 L−1)||202.4 ± 86.0||202.6 ± 165.3||0.998|
|Steroid doses (mg)§|
|Month 1||1697 ± 0||1720 ± 135.6||0.473|
|Month 2||214 ± 0||940 ± 375||<0.001|
|Month 3||105 ± 0||621 ± 105||<0.001|
|Average (6 months)¶||382 ± 36||639 ± 87||<0.001|
Satisfactory responses were significantly higher in the dexamethasone arm (Table 1). Of the five relapsed cases, four responded (two from each arm). The mean platelet count on day 5 was significantly higher in dexamethasone-treated patients when compared with those receiving prednisolone. Despite this, the bleeding scores decreased similarly . Four cases showed fresh hemorrhage (scores > 0) and platelet counts below 10 × 109 L−1. The three patients in the prednisolone arm were changed to pulse dexamethasone on day 6 followed by 30 mg day−1 of prednisolone. All three of these subsequently responded to the second-line treatment. The non-responding patient in the dexamethasone arm suffered from severe mucosal bleeding and intravenous immunoglobulin was given to raise the platelet count while waiting for the effects of steroid treatment. Despite these alterations, the mean platelet counts were still significantly higher in the dexamethasone arm on day 15 according to the intention-to-treat analysis.
From day 15, no patients from either arm exhibited any clinical bleeding (score 0). Mean platelet counts after the second week, and up to 6 months, rose similarly in both arms. One patient in the dexamethasone group had a relapse during steroid dose tapering. A colchicine and dapsone combination was added and gave a good response. Two other cases responded slowly and as a consequence colchicine was given. One case in the prednisolone arm subsequently became pregnant and had a recurrence. Two monthly pulses of dexamethasone were administered, resulting in partial and non-sustained responses. All three patients that were switched from prednisolone to dexamethasone had steroid reduction according to the planned schedule.
The total administered steroids in both arms were the same in the first month. However, the doses were appreciably different at later time points (Table 1). No patient discontinued treatment due to steroid toxicity. One patient in the dexamethasone group developed hyperglycemia. Seven of the patients (19%) (three on dexamethasone) complained of dyspepsia. Six patients (17%) (four on dexamethasone) had acne.
Our data demonstrate the superior effect of high-dose dexamethasone on platelet counts over standard dose prednisolone in the acute management of patients with severe thrombocytopenia. The primary outcome (i.e. platelet counts above 50 × 109 L−1 on day 5) was set a priori because the margin of safety over the threshold of 30 × 109 L−1 permits patients to be discharged early. In addition, alternative analyses also found that the platelet counts were significantly higher after dexamethasone on days 5 and 15. Therefore, pulse dexamethasone is the drug-of-choice for severe ITP. The second-line therapy and the small sample size preclude the analysis for the difference in clinical bleeding.
Furthermore, this study illustrated that initial high-dose dexamethasone made possible the more rapid reduction of prednisolone, without influencing the long-term effect on platelet counts. Nevertheless, this marked difference does not result in fewer steroid side-effects, probably due to the inadequate power of this study. Despite this, the adverse reactions that arose in both arms were manageable. We found occasional patients, outside this study, who had to stop high-dose dexamethasone due to severe dizziness, agitation and insomnia. More data are now needed to assess the incidence of uncommon adverse events from pulse dexamethasone. In conclusion, the study suggests that pulse dexamethasone followed by half-dose prednisolone is the preferred primary treatment of ITP.