Following the approval of nivolumab for r/r cHL in Japan, this PMS was conducted to obtain data on its safety and effectiveness in the real-world setting. Unlike prior clinical trials, all patients treated with nivolumab for cHL were registered without any specific exclusion criteria. Therefore, the patient population may more closely represent the patients who suffer from cHL and are treated with nivolumab in Japan than in clinical trials. Notably, the overall incidence of TRAEs (63.5%) was not higher than that observed in ONO-4538-15 and CheckMate 205 combined (91.8%) (Table 2). The incidence of grade ≥ 3 TRAEs (29.9%, ESM Table 2) was also similar to that in both trials (ONO-4538-15: 23.5%; CheckMate 205: 25.0%). However, observational studies (including PMS) may underestimate the incidence of TRAEs compared with clinical trials.

The clinical trials excluded patients with prior allo-HSCT. In the present analyses, prior allo-HSCT was found to be a risk factor for hepatic failure/hepatic dysfunction/hepatitis/cholangitis sclerosing (HR 2.80–4.30, ESM Table 4). Liver-related TRAEs are a common symptom of GVHD, which is a potential complication of allo-HSCT. Liver-related TRAEs were observed in eight patients with prior allo-HSCT, and GVHD was reported as a TRAE in two of those patients. This suggests that liver-related TRAEs in patients with prior allo-HSCT may be associated with GVHD. In addition, another study reported a higher incidence rate of abnormal hepatic function in patients with prior allo-HSCT than in patients without prior allo-HSCT (20 vs 7%) [15]. Therefore, the higher incidence of liver-related TRAEs than in the previous clinical trials is partially responsible for the higher incidence of TRAEs in patients with prior allo-HSCT. Liver-related TRAEs were reported in nivolumab-treated patients with other carcinomas and are not new concerns [16,17,18,19]. We also noted that more than 80% of patients recovered/were recovering with appropriate treatment (Table 4). Regarding hepatic failure/hepatic dysfunction/hepatitis/cholangitis sclerosing, the proportion of patients that recovered/were recovering was similar between patients with and without prior allo-HSCT (75.0 vs 86.7%, respectively, ESM Table 7).

Immune system disorders (4.2%) were more frequent in this PMS than in CheckMate 205 (2.5%) and ONO-4538-15 (0%) (ESM Fig. 1). Immune system disorders included TRAEs related to transplantation complications (acute/chronic GVHD in eight patients and engraftment syndrome in one patient). The exclusion of patients with prior allo-HSCT from the clinical trials may contribute to the higher incidence of immune system disorders in this PMS. Prior reports described an increased frequency of GVHD with administration of an anti-PD-1 monoclonal antibody (mAb) administration before and after allo-HSCT [9, 20,21,22,23,24,25,26]. Since detailed information about GVHD is not available, it is difficult to directly compare the prevalence of GVHD in this study with that reported in other studies. It should be noted that GVHD occurred in six of 23 patients with prior allo-HSCT (26.1%), and half of those patients did not recover or died. Considering earlier reports and our data, physicians should pay careful attention to GVHD when administering nivolumab before and after allo-HSCT.

Several TRAEs (renal and urinary disorders, cardiac disorders, and herpes zoster) were more frequent in this PMS than in the prior clinical trial. However, the sample sizes of CheckMate 205 (n = 80) and ONO-4538-15 (n = 17) were too small to detect these TRAEs. Nevertheless, these are not new safety signals for nivolumab having been reported previously [16, 18].

ILD, a TRAE of special interest, was more frequent in this PMS (7.3%) than in CheckMate 205 (3.8%). Pneumonitis and other lung disorders are well recognized among patients treated with immune checkpoint inhibitors [27]. Furthermore, Japanese individuals tend to develop ILD more frequently (4 vs 0.2% for the rest of the world) [28, 29], possibly due to genetic factors [30], which may explain the higher frequency in this PMS. Prior ILD is a potential risk factor for ILD in patients treated with immune checkpoint inhibitors [30]. We found that ILD was numerically, but not significantly, more frequent in patients with prior ILD than in patients without prior ILD (14.7 vs 6.3%, HR 2.48, 95% CI 0.90–6.83; ESM Table 6). Patients with any abnormal chest radiographic findings had a numerically, but not significantly, higher incidence of ILD than patients with no abnormal chest findings (10.9% vs 4.3%, HR 2.54, 95% CI 0.92–7.00). Prior ILD and presence of any chest radiographic abnormalities were associated with the risk of ILD in earlier PMS of patients with lung cancer and melanoma [18, 19]. Therefore, the higher incidence of ILD in this PMS than in the clinical trials may be attributable to genetic factors and/or the inclusion of more patients with risk factors for ILD (e.g., prior ILD). Further data are needed to better understand this association.

Several other TRAEs of special interest were reported here but not in CheckMate 205 or ONO-4538-15 (Table 3). However, these TRAEs of special interest were reported in prior PMS of nivolumab [16,17,18,19] and do not represent new safety signals.

Finally, regarding the effectiveness of nivolumab, the ORR was 61.7% (95% CI 55.8–67.4%), similar to that in CheckMate 205 (66.3%, 95% CI 54.8–76.4%) [8], but numerically lower than that in ONO-4538-15 (81.3%, 95% CI 54.4–96.0%) [10]. A possible limitation of this outcome is that the responses were not centrally assessed. If we consider the overlapping 95% CIs and the limited number of patients in the Japanese phase II trial, the ORR in this PMS is within the range reported in the clinical trials.

The results of this PMS should be interpreted cautiously because of its potential limitations, including the lack of a control group, the relatively short (1 year) follow-up, and lack of central review for safety and effectiveness assessment. Due to the lack of detailed information on GVHD (such as the new onset or worsening of preexisting GVHD, the severity/grade of prior GVHD, the conditioning regimen, the donor’s or recipient’s status, or the presence of prophylaxis), it is difficult to directly compare its prevalence in this PMS with that reported in other studies.

In conclusion, the safety profile of nivolumab in this PMS of Japanese patients with r/r cHL was similar to that reported in earlier clinical trials of r/r cHL and PMS for other malignancies. We identified no new safety concerns, based on TRAEs of special interest, that were not previously reported in clinical trials and/or PMS for other malignancies in Japan. The effectiveness of nivolumab treatment in patients with r/r cHL in this real-world setting was comparable with that reported in the earlier clinical trials. Overall, these results demonstrate that nivolumab is a safe and effective treatment for patients with r/r cHL in the real-world setting in Japan.