Dear Editor: In a recent article in Neuro-Oncology, “Association between radiation dose to neuronal progenitor cell niches and temporal lobes and performance on neuropsychological testing in children: a prospective study,”1 the connection between radiation dose and cognitive outcomes was evaluated. This is an interesting and important area of research that is inherently difficult to study because of the wide variation in treatment protocols. There are several methodological observations that may be important to consider when interpreting the findings of this study. First, the reported relationship between radiation dose and neurocognitive outcomes is difficult to interpret. For example, 74% of the patients were treated with 50–60 Gy radiation, with an overall mean dosage of 42.9 Gy. Only 3 patients (16% of the sample) were treated with 21–49 Gy radiation, and only 2 patients (11% of the sample) were treated with 0–20 Gy radiation. Because of these uneven and small individual sample sizes, it is very difficult to make comparisons among groups in neurocognitive functioning. In addition, there does not appear to be adequate variance within the patient sample (74% of patients were in the highest dose group) to draw conclusions regarding dose effects, particularly for those receiving <50–60 Gy radiation. Second, findings are very difficult to generalize because of the heterogeneity of the sample. Patients ranged in age from 1 to 18 years at the time of radiation, with a standard deviation of 1.1–18.6 years of age. To address this wide age range, the authors state in the Discussion that “Because we used standardized, age-adjusted scores for our analysis of the neurocognitive tests, we would not expect this difference to have an impact on our results.” The authors appear to have used raw scores instead of normative data and standardized scores and controlled for age in the regression analysis. Controlling for age at the time of radiation dose is not equivalent to using normative standardized scores. It is not stated in the manuscript why this method was chosen. Moreover, it appears that controlling for age at the time of radiation is counterintuitive because age at time of radiation moderates cognitive functioning and brain maturational trajectories.2 In other words, age at time of radiation and neurocognitive performance is not random but likely strongly correlated and contributory. The same argument might be made for controlling for presence or absence of chemotherapy; controlling for a variable that is likely to affect an outcome measure in one group but not the other (ie, patient group but not control group) may be artificially masking important and significant variance. On a similar note, neurocognitive testing was attempted at baseline (1–18 years of age) and at around 6, 15, and 27 months after completion of radiation treatment. The time between diagnosis and baseline testing ranged from 14 to 2284 days. It is theoretically and conceptually problematic that some patients received baseline testing years after diagnosis, whereas others were tested 2 weeks after diagnosis. There are many variables that may account for differences in neurocognitive functioning (eg, health status, time away from school, age differences, and chemotherapy effects) beyond radiation dose. This weakens the interpretation of the independent and dependent variables and overall results of the study. The study would be improved by clearly defined and more homogenous groups, to provide more ecologically valid information. The use of statistical covariates likely to affect outcome measures should also be clarified if indeed the results are to inform clinical practice and ongoing research. Conflict of interest. None declared.
Read full abstract