Paper Details

ABSTRACT

Background: Although pediatric acute lymphoblastic leukemia (ALL) outcomes have improved dramatically, early relapse (<36 mos in complete remission [CR]) remains difficult to treat, with fewer than 50% of children surviving (Rheingold, Leukemia 2024). We previously found that area-level poverty exposure was associated with early relapse among children treated on frontline Dana-Farber ALL Consortium trials (Bona, PBC 2016). Whether a modifiable, household-level poverty exposure—household material hardship (HMH)—is similarly associated with early relapse is unknown. The DFCI ALL Consortium Protocol 16-001 (NCT03020030) was the first trial to prospectively collect parent-reported HMH to evaluate associations with disease outcomes. 

Methods: DFCI 16-001 enrolled children 1-<22 yrs with de novo B- or T-cell ALL at 8 US and Canadian centers (2017-2022). Participants <18 yrs were eligible to opt-in to a correlative HMH study at initial trial consent; parent/guardians completed baseline surveys within 32-days of enrollment to report HMH exposure (food, housing, or utility insecurity; scored by domain 0-3). The analytic cohort included US participants, <18 years, evaluable for early relapse with available HMH data. The primary exposure was HMH, dichotomized as 0-1 (none/mild) vs >2 (severe) domains (Frank, Pediatrics 2010). Trial covariates included sex (male vs female), age (yrs), immunophenotype (B- vs T-cell), presenting WBC (<50K vs >50K), CNS status (1, 2, 3, traumatic), trial-defined ALL biology (favorable [ETV6::RUNX1, hyperdiploid with trisomies 4,10)] vs adverse [IKZF1 deleted, KMT2A rearrangement, TCF3::HLF, hypodiploid, iAMP21] vs other), NCI risk group (standard vs high), trial-defined final risk group treatment arm (low, intermediate, high, very high risk), end-induction minimal residual disease (low vs high), and race/ethnicity (Hispanic, non-Hispanic Asian, non-Hispanic Black, non-Hispanic Other [other, multiracial, Alaskan/AI], non-Hispanic White, missing). The primary outcome was the rate of early relapse (<36 months in CR) compared by one-sided Fisher’s exact test. Associations with early relapse were evaluated with univariable logistic regression. Two multivariable models were built including HMH and covariates significant in univariable analyses (p<0.10) (Model 1: final risk group; Model 2: age, presenting WBC, biology, and end-induction MRD). Secondary outcomes included event-free (EFS) and overall survival (OS) estimated using Kaplan-Meier methods, compared with one-sided log rank tests.

Results: The analytic cohort included 256 patients with a median follow-up of 5.1 yrs, median age 5.7 yrs, 50% male, 21.2% Hispanic, 8.6% non-Hispanic Black. Forty-five percent had NCI high-risk disease, 21.5% adverse biology and 37.9% favorable biology. Final risk group treatment included 43% low, 21% intermediate, 18% high, 18% very high. At the household-level, 63.7% of children had no HMH exposure; 25.4% were exposed to housing insecurity, 21.5% food insecurity, and 15.2% utility insecurity—resulting in 80.9% with none/mild HMH and 19.1% with severe HMH.

Overall, 6.6% of the cohort had early relapse; early relapse was more frequent among those exposed to severe vs none/mild HMH (14.3% vs 4.8% p=0.026), with a significant dose-response across HMH score (p=0.029). In univariable logistic regression, children with severe HMH had higher odds of early relapse (OR 3.28 [95% CI 1.18, 9.12], p=0.023). In multivariable modeling, severe HMH maintained statistically significant higher odds of early relapse in both models (Model 1: OR 3.07 [95% CI 1.05, 8.93], p=0.040; Model 2: OR 3.07 [95% CI 1.01, 9.35], p=0.048). Children exposed to severe HMH experienced non-significant but numerically inferior 3-year EFS (86.0% vs 93.5%; p=0.070) and OS (91.1% vs 98.3%; p=0.120).

Conclusion: Among US children with ALL treated on a frontline multicenter trial at tertiary care centers, severe HMH exposure at the time of trial enrollment was independently associated with 3-fold higher odds of early relapse. This association was consistent after adjusting for disease and treatment characteristics. Work is ongoing to define mechanisms underlying this association, including investigation of HMH-associated treatment resistance and differential care delivery mechanisms. Concurrently, investigators are evaluating the first poverty-targeted intervention for childhood cancer (RISE) as an embedded aim of the successor DFCI ALL Consortium trial (25-001).