of Colorado Cancer Center
Adam Green, MD Photo from the University
New research suggests that, in the US, early deaths from childhood cancer may be more common than clinical trials suggest.
Researchers found the rate of death within 1 month of diagnosis was higher among patients included in a large national database than among patients enrolled in phase 3 trials.
The data also indicated that early death is more likely in cancer patients under the age of 1 and those belonging to minority racial and ethnic groups.
Adam Green, MD, of the University of Colorado Anschutz Medical Campus in Aurora, and his colleagues conducted this research and reported the results in the Journal of Clinical Oncology.
The researchers analyzed data from the Surveillance, Epidemiology and End Results (SEER) database, which collects about 15% of all cancer outcomes across the US (representing a geographic and socioeconomic cross-section).
The team identified 36,337 patients with pediatric cancer (ages 0 to 19) diagnosed between 1992 and 2011. Of these patients, 555 (1.5%) died within 1 month of diagnosis.
Young age
Overall, the strongest predictor of early death was age below 1 year. The odds ratio (OR) was 4.36 for these patients (P<0.001), 0.75 for patients age 1 to 4, 0.78 for patients age 5 to 9, 1.00 (reference) for those age 10 to 14, and 1.69 for those age 15 to 19.
For hematologic malignancies, the adjusted OR (adjusted by poverty, unemployment, education, and year of diagnosis) was 4.32 for patients younger than 1 (P<0.001), 0.76 for patients age 1 to 4, 0.79 for patients age 5 to 9, 1.00 for those age 10 to 14, and 1.76 for those age 15 to 19 (P=0.002).
“In general, babies are just challenging, clinically, because they can’t tell you what they’re feeling,” Dr Green noted. “Parents and physicians have to pick the ones with cancer from the ones with a cold, without the patient being able to tell you about symptoms that could be diagnostic.”
“Babies tend to get aggressive cancers, it’s hard to tell when they’re getting sick, and some are even born with cancers that have already progressed. These factors combine to make very young age the strongest predictor of early death in our study.”
Race/ethnicity
Black race and Hispanic ethnicity also predicted early death. The OR was 1.48 for black race, 1.00 for white race (reference), and 1.09 for “other” races (P=0.102). The OR was 1.39 for Hispanic patients and 1.00 (reference) for non-Hispanic patients (P=0.007).
For hematologic malignancies, the adjusted OR was 1.68 for black patients (P=0.01) and 1.44 for patients of other, non-white races. The adjusted OR was 1.48 for Hispanic patients (P=0.009).
Dr Green said he hopes future studies will be able to determine the factors responsible for these disparities.
Higher death rates
Dr Green and his colleagues also found the rate of early death due to pediatric cancers is higher than reported in clinical trials, and this was true for all cancer subtypes assessed.
“Most of what we know about outcomes for cancer patients come from clinical trials, which have much more thorough reporting rules than cancer treated outside trials,” Dr Green said. “However, these kids in our study aren’t surviving long enough to join clinical trials.”
The researchers looked at a phase 3 trial (COG AAML0531) of pediatric patients with acute myeloid leukemia (AML) who were studied from August 2006 to June 2010. The early death rate in this trial was 1.6% (16/1022).
In contrast, the SEER database showed an early death rate for pediatric AML patients of 5.9% (15/256) during the same period as the trial and 6.2% (106/1698) for the period from 1992 to 2011. This is almost 4 times as high as the death rate in the trial.
The same effect was observed for acute lymphoblastic leukemia (ALL).
The early death rate for non-infant ALL was 0.7% (13/1790) in a trial (POG 9900) conducted from April 2000 to April 2005, 1.6% (30/1823) in the SEER data covering the same time period, and 1.3% (94/7353) in the SEER data from 1992 to 2011.
For infant ALL, the rates were 2.0% (3/149) in a trial (COG AALL0631) conducted from January 2008 to June 2014, 5.0% (2/40) in the SEER data during the trial period, and 5.4% (12/223) in the SEER data from 1992 to 2011.
“I had a hunch this was a bigger problem than we thought,” Dr Green said. “Now we see that is indeed the case.”
