Methods
This retrospective chart review evaluated pediatric patients with mature
B/T-cell NHL who received intravenous HD-MTX (3
g/m2/dose intravenously over 3 hours) from October 1,
2010 through June 30, 2020 at Texas Children’s Hospital. Patients were
excluded from data analysis if they received an empiric methotrexate
dose reduction. All data were abstracted from the electronic health
record, including diagnosis, race/ethnicity, age, height and weight at
diagnosis; dose and timing of HD-MTX administration; relevant laboratory
details (i.e., serum creatinine, methotrexate levels, urine pH and
specific gravity); select supportive care interventions (i.e.,
leucovorin, hydration, alkalization, glucarpidase, concomitant
medications); tumor lysis syndrome; and vital status. Body surface area
and body mass index were calculated from extracted variables. The
primary outcome was delayed methotrexate clearance, defined by a
methotrexate level above the protocol-defined clearance threshold at
hour 48. Secondary outcomes included the time to clearance, frequency of
serum creatinine elevations (i.e., an increase ≥25% after start of
methotrexate), and incidence of Common Terminology Criteria for Adverse
Events (version 5.0) grade ≥3 mucositis. Planned exploratory analysis
included evaluation of risk factors associated with delayed clearance of
methotrexate, serum creatinine elevations, and mucositis.
The frequency of each methotrexate-associated toxicity was summarized by
the number of methotrexate doses resulting in each toxicity and number
of patients who experienced the toxicity following any HD-MTX. Odds
ratios (ORs) and corresponding 95% confidence intervals (CIs) for the
associations between clinical characteristics and the development of
each outcome (delayed clearance, creatinine elevations, and mucositis)
were calculated using multivariable mixed-effects logistic regression to
account for repeated measures following each methotrexate infusion
within each patient. A mixed-effects linear regression model was
constructed to evaluate associations with time to methotrexate
clearance. A step-wise selection process was used to identify a
parsimonious model including age, race/ethnicity, and other clinical
features associated (p-value<0.2) with one or more of the
outcomes in unadjusted models. All analyses were conducted using Stata
SE version 15.1 (College Station, TX) at a two-sided significance level
of 0.05. This study was approved by Institutional Review Board at Baylor
College of Medicine.