During the period from January 1, 2015 to December 31, 2018, patients diagnosed as intestinal SR-aGVHD and received basiliximab treatment after allo-HSCT in the Institute of Hematology, Peking University (PUIH) were enrolled in this study. The major clinical outcomes of these patients had been reported previously as parts of a multicenter real-world study (18), and these patients were further followed in this study. The study was carried out in accordance with the Declaration of Helsinki. This study protocol was approved by the Ethics Committee of Peking University People’s Hospital.

Methylprednisolone (MP) was the first-line treatment for newly diagnosed aGVHD. 2 mg·kg⁻¹·day⁻¹ of MP for patients aged < 10 years or weighing < 30 kg, and the others received MP at a dose of 1 mg·kg⁻¹·day⁻¹. If the patient was diagnosed with SR-aGVHD (36), treatment with basiliximab was initiated. Recommended dose of basiliximab was as followed: 20 mg per dose for adults and children weighing ≥ 35 kg, or 10 mg per dose for children weighing < 35 kg. Basiliximab was administered on +1, +3, +8 days, and once a week thereafter until the aGVHD was less than grade II (18).

Complete response (CR) referred to the complete disappearance of all signs and symptoms of aGVHD in all assessable organs without additional systemic treatment. Partial response (PR) was defined as one stage improvement in one or more organs involved in the signs or symptoms of aGVHD, without progression in other organs and without additional systemic treatment. The overall response rate (ORR) was defined as CR plus PR.

Definition of the low-dose steroid criteria: steroid decreasing to ≤ 0.2 mg/kg in patients < 40 kg of body weight, or ≤ 10 mg/day in patients of > 40 kg of body weight (37).

NRM referred to all deaths without underlying disease recurrence. OS was defined as the time from treatment with basiliximab to death from any cause. Disease free survival (DFS) was defined as the survival duration of continuous CR after basiliximab treatment.

aGVHD grade was based on the Mount Sinai Acute GVHD International Consortium (MAGIC) and Minnesota Acute GVHD Risk Score (36, 38). Hematopoietic Cell Transplantation-Specific Comorbidity Index (HCT-CI) scores were reported according to the results of Sorror et al. (39).

For the comparisons of patient characteristics, categorical variables were calculated using the chi-squared test or Fisher’s exact test. Continuous variables were analyzed using t-tests or Mann–Whitney U tests. The probability of survival was analyzed using the Kaplan–Meier method. The cumulative incidence of ORR, NRM, and GVHD were estimated using competing risk regression analysis.

Hazard ratios (HRs) for steroid decreasing velocity and other clinical outcomes were estimated with ORR, NRM, OS, and DFS in a multivariable analysis using Cox proportional hazards regression with a backward stepwise model selection approach. Independent variables with P values >.1 were sequentially excluded from the model, and P values <.05 (two-tailed) was considered statistically significant. Statistical analyses were performed using SPSS 26 (SPSS Inc./IBM, Armonk, NY, USA), the R software package (version 4.3.0; http://www.r-project.org) and Prism 9 (GraphPad Software).

A total of 314 patients with intestinal SR-aGVHD were enrolled ( Table 1 ). The median time from diagnosis of intestinal aGVHD to basiliximab treatment was 5 (range, 3 to 23) days. The median doses of basiliximab were 4 (range, 2 to 14) doses. The median time from the beginning of basiliximab treatment to achieving ORR and CR was 6 (range, 2 to 65) days and 10 (range, 3 to 95) days, respectively. The cumulative incidence of ORR and CR at day 28 after basiliximab treatment was 70.1% and 58.0%, respectively. The cumulative incidence of ORR and CR at any time after basiliximab treatment was 76.4% and 65.3%, respectively. The probability of OS, DFS, and NRM at 3 years after basiliximab treatment was 66.9% (95% CI 61.6%-71.5%), 64.4% (95% CI 59.1%-69.7%), and 25.9% (95% CI 21.1%-30.9%), respectively.

The median time from the beginning of basiliximab treatment to steroid dose decreased to the 50% of initial dose was 6 (range, 1 to 21) days, and the median time from the beginning of basiliximab treatment to steroid dose decreased to the low-dose steroid was 13 days (range, 2 to 45) days. The median time for steroid dose decreased to the 50% of initial dose and decreased to the low-dose steroid was 5 (range, 1-16) days versus 7.5 (range, 2-21) days (P < 0.001) and 12 (range, 2-33) days versus 15 (range, 5-45) days (P = 0.003), respectively, for patient with and without achieving ORR at day 28. The median time for steroid dose decreased to the 50% of initial dose and decreased to the low-dose steroid was 5 (range, 1-21) days versus 7.5 (range, 3-16) days (P < 0.001) and 12 (range, 2-44) days versus 15 (range, 5-45) days (P = 0.008), respectively, for patient with and without achieving ORR at any time. This suggested that patients who achieved ORR or CR showed more rapid steroid decreasing velocity.

Furthermore, for patient with and without achieving ORR at day 28, the median cumulative dose of steroids at 28 days was 8.4 (range, 1.7-26.6) mg/kg versus 10.1 (range, 1.9-44.8) mg/kg (P = 0.003). For patient with and without achieving ORR at any time, the median cumulative dose of steroids at any time was 8.5 (range, 1.7-37.9) mg/kg versus 10.1 (range, 1.9-44.8) mg/kg (P = 0.006). Meanwhile, the steroid decreasing velocity curves within 28 days in patients achieving ORR at day 28 and at any time were both more rapid than non-ORR patients ( Supplementary Figure 1A , P < 0.001; Supplementary Figure 1B , P < 0.001). These results suggests that patients achieving ORR after basiliximab treatment required fewer steroids dose than non-ORR patients.

ROC curve analysis was performed to estimate the optimal cut-off point of steroid decreasing velocity. The optimal cut-off time of steroid dose decreased to the 50% of initial dose of steroid was 5 days according to the ORR at day 28 ( Figure 1A ), and the optimal cut-off time for decreasing steroid dose to the low-dose steroid was 12 days according to the ORR at any time ( Figure 1B ).

The ORR at day 28 and at any time of steroid dose halving ≤ 5 days group was 83.7% and 88.9%, respectively, which was both significantly higher than those of halving > 5 days group (at day 28: 57.1%, P = 0.001, at any time: 64.6%, P = 0.001, Figures 2A, B ). The ORR at day 28 and at any time of steroid dose decreased to the low-dose steroid ≤ 12 days was 79.5% and 84.1%, respectively, which was also both significantly higher than those of steroid dose decreased to the low-dose steroid >12 days (at day 28: 61.3%, P = 0.001, at any time: 69.3%, P = 0.002, Figures 2C, D ). In addition, the steroid dose halving ≤ 5 days group and steroid dose decreased to the low-dose steroid ≤ 12 days group were also showed better CR rates ( Figures 2A–D ).

Thus, the intestinal SR-aGVHD patients were divided into three groups: rapid (steroid dose decreased to the 50% of initial dose ≤ 5 days and decreased to the low-dose steroid ≤ 12 days), medium (steroid dose decreased to the 50% of initial dose ≤ 5 days or decreased to the low-dose steroid ≤ 12 days) and slow (steroid dose decreased to the 50% of initial dose > 5 days and decreased to the low-dose steroid >12 days) group. The steroid decreasing velocity curve within 28 days in rapid, medium, and slow group were shown in Figure 2G (P < 0.001).

The cumulative incidence of ORR at day 28 was 85.6%, 72.9% and 53.5%, respectively, in rapid, medium, and slow group (P = 0.004, Figure 2E ). The cumulative incidence of ORR at any time was 90.4%, 78.1% and 62.3%, respectively, in rapid, medium, and slow group (P = 0.003, Figure 2F ).

The cumulative incidence of CR at day 28 was 68.3%, 60.4% and 46.5%, respectively, in rapid, medium, and slow group (P = 0.008, Figure 2E ). The cumulative incidence of CR at any time was 75.0%, 68.8% and 53.5%, respectively, in rapid, medium, and slow group (P = 0.003, Figure 2F ).

In patients with grade II intestinal SR-aGVHD at the beginning of basiliximab treatment, the cumulative incidence of ORR at day 28 after basiliximab treatment was 86.1%, 82.3% and 62.2%, respectively, in rapid, medium, and slow group (P = 0.002, Supplementary Figure 2A ). The cumulative incidence of CR at day 28 after basiliximab treatment was 76.4%, 74.2% and 56.8%, respectively, in rapid, medium, and slow group (P = 0.021, Supplementary Figure 2A ). The cumulative incidence of ORR at any time after basiliximab treatment was 90.3%, 85.5% and 66.2%, respectively, in rapid, medium, and slow group (P = 0.001, Supplementary Figure 2B ). The cumulative incidence of CR at any time after basiliximab treatment was 80.6%, 79.0% and 60.8%, respectively, in rapid, medium, and slow group (P = 0.012, Supplementary Figure 2B ).

In patients with grade III–IV intestinal aGVHD at the beginning of basiliximab treatment, the cumulative incidence of ORR at day 28 after basiliximab treatment was 84.4%, 55.9% and 37.5%, respectively, in rapid, medium, and slow group (P = 0.001, Supplementary Figure 2C ). The cumulative incidence of CR at day 28 after basiliximab treatment was 50.0%, 35.3% and 27.5%, respectively, in rapid, medium, and slow group (P = 0.141, Supplementary Figure 2C ). The cumulative incidence of ORR at any time after basiliximab treatment was 90.6%, 64.7% and 55.0%, respectively, in rapid, medium, and slow group (P = 0.004, Supplementary Figure 2D ). The cumulative incidence of CR at any time after basiliximab treatment was 62.5%, 50.0% and 40.0%, respectively, in rapid, medium, and slow group (P = 0.165, Supplementary Figure 2D ).

The results of new-onset infections after basiliximab treatment were showed in Supplementary Table 1 . The rates of new-onset viral infections after basiliximab treatment were 48.1%, 42.7% and 53.5%, respectively, in rapid, medium, and slow group (P = 0.295). The rates of new-onset bacterial infections after basiliximab treatment were 10.6%, 16.7% and 14.9%, respectively, in rapid, medium, and slow group (P = 0.437). The rates of new-onset fungal infections after basiliximab treatment were 5.8%, 4.2% and 0.9%, respectively, in rapid, medium, and slow group (P = 0.141). The rates of any infection (≥ 1 type) were 57.7%, 50.0% and 57.9%, respectively, in rapid, medium, and slow group (P = 0.440), and the rates of multiple infections (≥ 2 types) were 13.5%, 18.8% and 19.3%, respectively, in rapid, medium, and slow group (P = 0.464).

A total of 81 patients died from NRM. The most common cause of NRM was infection (45.7%), followed by GVHD (23.5%). The cumulative incidence of NRM at 1 year after basiliximab treatment was 18.7% (95% CI 11.3%–26.1%), 22.8% (95% CI 14.2%–31.4%) and 32.8% (95% CI 24.1%–41.5%), respectively, in rapid, medium, and slow group (P = 0.035, Supplementary Figure 3A ). The cumulative incidence of GVHD-related mortality at 1 year after basiliximab treatment was 0.9% (95% CI 0.2%–1.7%), 6.4% (95% CI 1.4%–11.5%) and 9.7% (95% CI 4.2%–15.2%), respectively, in rapid, medium, and slow group (P = 0.011, Supplementary Figure 3B ). The cumulative incidence of infection-related mortality at 1 years of basiliximab treatment was 9.6% (95% CI 3.9%–15.3%), 11.7% (95% CI 5.2%–18.3%) and 12.3% (95% CI 6.2%–18.4%), respectively, in rapid, medium, and slow group (P = 0.692, Supplementary Figure 3C ).

A total of 212 intestinal SR-aGVHD patients survived after basiliximab treatment, with a median time of OS was 804 (range, 4-1920) days ( Figure 3A ). The probability of OS at 1 year after basiliximab treatment was 76.9% (95% CI 68.9%–84.9%), 72.7% (95% CI 63.7%–81.7%), and 62.3% (95% CI 53.5%–71.1%), respectively, in rapid, medium, and slow group (P = 0.024, Figure 3C ).

The median time of DFS was 777 (range, 4-1920) days ( Figure 3B ). The probability of DFS at 1 year after basiliximab treatment was 75.0% (95% CI 66.8%–83.2%), 69.8% (95% CI 60.6%–79.0%) and 59.6% (95% CI 50.6%–68.6%), respectively, in rapid, medium, and slow group (P = 0.029, Figure 3D ).

In multivariable analysis, grade III–IV aGVHD at the beginning of basiliximab treatment was the risk factor for having a decreased likelihood of response at day 28 and at any time, while rapid steroid decreasing velocity was associated with a higher response at day 28 and at any time ( Supplementary Table 2 ). In addition, pretransplant chemotherapy, high-risk refined Minnesota aGVHD risk score and HCT-CI score at the beginning of basiliximab treatment were all the risk factors of OS and DFS ( Supplementary Table 3 ). Meanwhile, pretransplant chemotherapy and initial steroid dose ≥ 1mg/kg at the beginning of basiliximab treatment were the risk factors of NRM ( Supplementary Table 3 ). Simultaneously, the rapid group (i.e., the SR-aGVHD patients reducing steroid dose to the 50% of initial dose ≤ 5 days and to the low-dose steroid ≤ 12 days after basiliximab treatment) was a favorable prognostic factor for OS, DFS, and decreased likelihood of NRM ( Supplementary Table 3 ).