In this issue, Mattano et al. report on the development of osteonecrosis among patients with acute lymphoblastic leukemia (ALL) treated in the Children’s Oncology Group trial AALL0232 [1].

Apart from analyzing the incidence and risk factors for osteonecrosis, the authors also analyze the impact of osteonecrosis on event-free survival (EFS) and overall survival (OS). Surprisingly, both EFS and OS were superior among patients who developed osteonecrosis compared to those who did not. Osteonecrosis was seen predominately among patients ≥10 years of age, constituting 93% of the study cohort, a demographic associated with poor prognosis in childhood ALL. Furthermore, the study underscored that both Capizzi methotrexate and dexamethasone were associated with an increased risk of osteonecrosis. Despite the lack of previous association of these two agents with superior survival, patients who developed osteonecrosis exhibited a survival advantage [2]. Thus, the question arises: How does osteonecrosis lead to improved survival outcomes? The answer to this may partly, if not fully, lies in the analysis methodology and could be attributed to immortal time bias (ITB) or guaranteed time bias [3, 4]. ITB is indeed a primary concern of such analyses because patients who experience toxic death, progression, or relapse earlier are not expected to develop delayed toxicities such as osteonecrosis. Conversely, patients who remain in the study for a long duration have more time to experience toxicity. In the present study, the median time to osteonecrosis was 455 days. This suggests that within this timeframe, patients certainly cannot experience certain events such as death. Consequently, this period can be regarded as an immortal time for patients in this group, directly influencing the analysis of EFS and OS. In addition, the study highlights a lower cumulative incidence of relapse (CIR) and a longer median time to relapse among patients with osteonecrosis compared to those without osteonecrosis. Again, CIR in the osteonecrosis group may be confounded by the absence of induction/remission deaths, which are competing events. The difference in time to relapse can be explained by the fact that patients who experience relapse early during therapy may not survive long enough to develop osteonecrosis due to high-risk nature of relapse.