The disposition of patients is displayed in Fig. 1; 2076 patients were accepted for registration at 697 sites, of which 162 did not consent to the publication of the survey results. Of the 1879 patients eligible for enrolment from 660 sites, 1632 patients completed the survey. The safety analysis set included 1622 patients, and the effectiveness analysis set included 1467 patients.

Table 1 describes the characteristics of patients in the safety analysis set at the start of treatment. At the start of romiplostim administration, 56.97% of patients were aged ≥ 65 years. Romiplostim was administered to 94.08% of patients as a treatment for chronic ITP; platelet counts were < 2.0 × 104/µL in 60.54% of patients and ≥ 2.0 to < 3.0 × 104/µL in 11.22%. At the time of romiplostim initiation, corticosteroids were administered in 84.40% of chronic ITP patients, and splenectomy was performed in 17.76%.

The exposure to romiplostim in the safety analysis set is shown in Table 2. The mean single dose of romiplostim was stable after 12 weeks from the start of romiplostim administration and remained below 4 µg/kg in nearly 50% of patients; the dose remained below 6 μg/kg in approximately 70% of patients (data not shown). The median romiplostim treatment duration was 190 (range, 1–1203) days. By the end of the survey period, 226 patients had received at least one maximum dose of 10 μg/kg, and 25 patients had received at least one maximum dose of > 10 μg/kg of romiplostim (i.e., exceeding the approved dose); three patients had a suspected ADR of renal disorder (non-serious), neutrophil count increased (non-serious), and platelet count decreased (serious).

The reasons for discontinuation of romiplostim and the major AEs/suspected ADRs that resulted in the discontinuation, dosage at discontinuation, and platelet counts at discontinuation are summarised in Table 3. A total of 64.00% (1038/1622) of patients discontinued romiplostim within 2 years of treatment, and around half of those patients (579/1038 patients) discontinued within 12 weeks. The mean ± standard deviation (SD) dose of romiplostim at the time of discontinuation was 3.87 ± 2.94 μg/kg, and the mean ± SD platelet count was 9.44 ± 13.12 × 104/µL. In this survey, treatment discontinuation with romiplostim occurred in 14.92% of patients due to AEs, and in 6.47% of patients due to suspected ADRs. The most common AEs leading to discontinuation were pneumonia (2.59%), white blood cell (WBC) count increased (1.36%), sepsis (1.17%), anaemia (0.99%), and cerebral haemorrhage (0.99%). The most common suspected ADRs that led to the discontinuation of romiplostim included WBC count increased (0.43%) and cerebral haemorrhage (0.37%). In this survey, 12.45% of patients discontinued treatment with romiplostim because of “inadequate effectiveness”, with a mean ± SD dose of romiplostim at discontinuation being 6.09 ± 3.22 μg/kg and mean ± SD platelet count of 2.63 ± 4.94 × 104/µL; 14.00% of patients discontinued because of “symptom improvement”, with a mean ± SD romiplostim dose and platelet count at the time of discontinuation of 2.33 ± 2.18 μg/kg and 16.38 ± 14.84 × 104/µL, respectively.

The priority survey items of worsening of thrombocytopenia and haemorrhage-related events after discontinuation of romiplostim were evaluated. Of the 1038 patients who discontinued treatment with romiplostim within 2 years, 4.43% (46/1038 patients) had thrombocytopenia recurrence after discontinuation, 80.44% (835/1038 patients) had no recurrence, and 15.13% (157/1038 patients) were unknown. Haemorrhage-related events occurred after discontinuation of romiplostim in 6.07% (63/1038 patients), did not occur in 80.83% (839/1038 patients), and 13.10% (136/1038 patients) of the cases were unknown.

Table 4 highlights the occurrence of AEs and suspected ADRs in the safety analysis set. This survey reported suspected ADRs resulting in death in 2.22% (36/1622) of patients (40 events). Of these, the MedDRA system organ class (SOC) of ‘Infections and infestations’ included pneumonia (n = 2) and sepsis, influenza, and Pneumocystis jirovecii pneumonia (n = 1, each).

Haemorrhagic suspected ADRs were also evaluated as priority survey items. Haemorrhagic suspected ADRs of romiplostim included non-serious and serious haemorrhage. Non-serious/serious haemorrhage occurred in 3.27% (53/1622) of patients. Mean ± SD platelet count at the onset of non-serious/serious haemorrhage was 2.77 ± 4.58 × 104/µL, and 85.07% of these patients had a platelet count of < 5.0 × 104/µL. In these patients, the median onset date was 70.5 (range, 1–680) days. Serious haemorrhage occurred in 1.97% (32/1622) of patients, and 82.93% of patients with serious haemorrhage had a platelet count of < 3.0 × 104/µL. In the safety analysis set (n = 1622), the mean ± SD dose of romiplostim at which haemorrhagic suspected ADRs were observed was 4.31 ± 2.88 μg/kg, which was similar to the most frequent romiplostim dose (4.03 ± 2.91 μg/kg) (Table 2).

Thromboembolic suspected ADRs of romiplostim were another priority survey item evaluated. Fifty-eight events occurred in 3.08% (50/1622) of patients. Of these, platelet counts were available for 50 events; 30.00% (15/50) of which occurred at platelet counts of ≥ 40.0 × 104/µL, 16.00% (8/50) at 20.0–40.0 × 104/µL, 46.00% (23/50) at 5.0–20.0 × 104/µL, and 8.00% (4/50) at < 5.0 × 104/µL. The highest incidence of thromboembolism was observed in 46.00% (23/50) of events with platelet counts between 5.0 and 20.0 × 104/µL, followed by 30.00% (15/50) of events with platelet counts over 40.0 × 104/µL (Table 5). For these thromboembolic events, the median onset date was 118 (range, 9–707) days, 36.21% (21/58) of events occurred within 84 days (12 weeks) from the start of romiplostim administration, and 82.76% (48/58) of events occurred within 364 days (52 weeks). In the safety analysis set (n = 1622), the mean ± SD dose of romiplostim at which thromboembolic suspected ADRs were observed was 4.12 ± 2.75 μg/kg, which was similar to the most frequent mean dose (4.03 ± 2.91 μg/kg) (Table 2).

Increased bone marrow reticulin or myelofibrosis occurring in the safety analysis set was also evaluated. Increased bone marrow reticulin or myelofibrosis AEs and suspected ADRs were reported in 0.49% (8/1622) and 0.37% (6/1622) of patients, respectively; suspected ADRs included reticulin increased (n = 3), myelofibrosis (n = 2), and bone marrow reticulin fibrosis (n = 1) (Table 6). For events of increased bone marrow reticulin or myelofibrosis, the median onset date was 237 (range, 5–547) days, 42.86% (3/7) of events occurred within 84 days (12 weeks) from the start of romiplostim administration, and 85.71% (6/7) of events occurred within 364 days (52 weeks). No cases of irreversible myelofibrosis were identified from the assessment of these six individual case survey forms.

Haematological malignancies and MDS occurring in the safety analysis set were also evaluated. MDS-related AEs and suspected ADRs were reported in 0.92% (15/1622) and 0.49% (8/1622) of patients, respectively. Suspected ADRs included MDS (n = 5) and MDS transformation, blast cell counts increased, and blast cell proliferation (n = 1, each) (Table 6). For MDS-related haematological malignancies, the median onset date was 128 (range 1–662) days, 42.85% (6/14) of events occurred within 84 days (12 weeks) from the start of romiplostim administration, and 85.71% (12/14) of events occurred within 364 days (52 weeks). AEs and suspected ADRs of non-MDS-related haematological malignancies were reported in 1.48% (24/1622) and 0.25% (4/1622) of patients, respectively. Suspected ADRs included myelofibrosis (n = 2), and plasma cell myeloma and acute leukaemia (n = 1, each) (Table 6). For non-MDS-related haematological malignancies, the median onset date was 204 (range 15–573) days, 36.00% (9/25) of events occurred within 84 days (12 weeks) from the start of romiplostim administration, and 76.00% (19/25) of events occurred within 364 days (52 weeks). In this survey, we took into account confounding factors and the potential influence of other factors (age, comorbidities, other treatments, complications). After analysing the reported cases, none showed a reasonable possibility of a causal relationship to romiplostim.

Figure 2 shows mean changes in platelet counts from the start of romiplostim administration to 104 weeks (2 years) excluding platelet counts affecting effectiveness. The mean ± SD platelet count before initiation of romiplostim treatment was 2.84 ± 5.76 × 104/µL, which gradually increased to 9.19 ± 13.01 × 104/µL for the first 4 weeks, and ranged from 10.34 ± 10.72 to 12.38 ± 12.63 × 104/µL from 8 to 104 weeks of romiplostim treatment. The proportions of patients with platelet response (doubling of the baseline platelet counts before romiplostim administration and platelet counts ≥ 5.0 × 104/µL) were 60.03% at 24 weeks, 60.25% at 52 weeks, 63.89% at 76 weeks, and 64.53% at 104 weeks.

Mean changes in platelet counts from the start of romiplostim administration to 104 weeks (2 years), excluding platelet counts affecting the effectiveness of romiplostim. Data are mean ± standard deviation