Severe toxicities in survivors of pediatric and adolescent acute lymphoblastic lymphoma: A population-based study in Ontario, Canada

Abstract Cure rates for children, adolescents, and young adults with acute lymphoblastic leukemia (ALL) have improved dramatically and have recently risen further with the introduction of immunotherapy. A renewed focus on decreasing treatment late effects is thus warranted. The Ponte di Legno (PdL) Severe Toxicity (ST) Working Group identified 21 late ST considered “unacceptable”. However, no comprehensive data on the incidence and predictors of these ST are available. Given the rarity and long lag time of certain ST, clinical trial cohorts are unsuited to their study. We leveraged healthcare data in Ontario, Canada to quantify the incidence of the PdL ST among ALL survivors at a population-level. Patients diagnosed with ALL aged 0-18 years between 1986-2018 and treated at one of Ontario's five pediatric tertiary institutions were identified through a provincial childhood cancer registry. Patients must have survived at least 5 years from a pediatric cancer event, such as an original ALL diagnosis or relapse. Survivors were matched 1:5 to general Ontario population controls without cancer based on birth year and month, sex, and area of residence. Survivors and controls were linked deterministically to multiple population-based healthcare databases, including those identifying hospitalizations, emergency department visits, and physician encounters. PdL ST were identified using healthcare data, either directly through validated algorithms (e.g. congestive heart failure) or through proxy measures (e.g. hearing assistive devices as a proxy for hearing loss). We compared the incidence of 20 individual ST between survivors and controls, as well as the incidence of any ST, using Gray's test and unadjusted hazard ratios (HR). We also compared the mean cumulative number of ST between survivors and controls. Finally, we evaluated the impact of demographic, disease, and treatment-related factors [e.g. relapse, bone marrow transplant (BMT)]. We included 2,457 ALL survivors and 12,285 matched controls. Of survivors, 167 (6.8%) had a history of relapse and 129 (5.3%) a history of BMT. Median age at diagnosis was 4 years (range 0–18); median age at max follow up was 24 years (range 6–53). Thirteen of the 20 ST assessed had significantly increased incidence in survivors vs controls. For example, at 20 years from index, survivors had a higher incidence of heart failure (1.3±0.3% vs 0.07±0.04%), insulin-dependent diabetes (3.3±0.6% vs 1.4±0.2%), osteonecrosis (0.8±0.2% vs 0.04±0.03%)(p<0.0001 for all). Survivors were also at higher risk of vocal cord paralysis, second cancers, cognitive dysfunction, seizures, and need for mobility devices. ST such as hepatic or renal failure were not elevated. The most common PdL ST was hospitalization for mental health concerns, though this was not significantly elevated among survivors (5.0±0.6% vs 5.8±0.3%;p=0.10). The 20-year cumulative incidence of any ST, excluding psychiatric hospitalizations, was 17.5±1.0% among survivors vs. 7.1±0.3% among controls [HR 2.5, 95% confidence interval (95CI) 2.2–2.9; p<0.0001)]. Among survivors, the incidence of any ST was dramatically higher amongst those with a history of relapse vs no relapse (47.5±5.6% vs 15.3±1.0%; p<0.0001) and those with a history of BMT vs no BMT (67.4±7.3% vs 15.1±1.0%; p<0.0001). Of note, survivors without a history of relapse and/or BMT were still at significantly higher risk than population controls. For example, patients with no history of BMT were at 2.3 times higher hazard of any ST (95CI 2.0-2.6; p<0.0001). The mean number of ST per patient at 20 years was 0.3 (95CI 0.3-0.3) among survivors vs 0.1 (95CI 0.1-0.1) among controls but was 0.9 (95CI 0.7-1.2) among those with a history of relapse and 1.4 (95CI 1.1-1.9) among those who underwent BMT. ALL survivors experience a significantly higher burden of PdL-defined ST compared to the general population. While decreased rates of relapse and BMT from the incorporation of immunotherapy into treatment will result in a lower future burden, those without relapse or BMT are still at increased risk. Our results support current efforts to remove elements of traditional chemotherapy in an era of frontline immunotherapy to not only maintain event-free survival but to improve severe toxicity-free survival. Quantifying the burden of late toxicities at a population level is a critical first step toward identifying high-risk groups, informing long-term care, and ultimately reducing the impact of these ST.