Uncertainty in Low-Level Bone Marrow Dose Estimates from Pediatric CT
Abstract
Purpose
A recent controversial study in NEJM reported associations between hematologic malignancy and low estimated bone marrow dose (≤1 mGy) from pediatric CT. Accurate interpretation of such findings requires a clear understanding of uncertainty in organ dose estimation at very low dose levels. This study quantifies variability in pediatric CT bone marrow dose (BMD) estimates across scanners, protocols, and patient characteristics.
Methods
Bone marrow dose was estimated using two widely used CT dosimetry tools, MIRDct and CT-Expo. Simulations were performed across combinations of scanner model (4 in MIRDct; 219 in CT-Expo), patient age (0, 1, 5, 10, and 15 years in MIRDct; baby and child in CT-Expo), sex, scan protocol (chest, abdomen/pelvis, and brain), and tube potential (80, 100, and 120 kVp in MIRDct; 120 kVp in CT-Expo). Variability in scan dose corresponding to a nominal BMD of 1 mGy was characterized using the ratio of dose-length-product to BMD (DLP/BMD) and its interquartile range (IQR). Multivariable linear regression was used to assess the relative influence of scanner model, protocol, age, sex, and kVp on BMD normalized by DLP.
Results
Across 361 MIRDct simulations, the IQR of DLP/BMD was 152.8 mGy·cm/mGy and 89.7 mGy·cm/mGy for 16-cm and 32-cm DLP, respectively, with corresponding medians of 117.7 and 76.5 mGy·cm/mGy. For CT-Expo, the IQR of DLP16/BMD across 2,628 combinations was 80.06 mGy·cm/mGy, with a median of 91.73 mGy·cm/mGy. Patient age (p<0.001), scanner model (p<0.001, except iCT), and scan protocol (p<0.001) significantly affected BMD normalized by DLP, while kVp (p=0.41) and sex (p=0.41) were not significant contributors.
Conclusion
At low estimated bone marrow doses (≤1 mGy) in pediatric CT, substantial variability in scan dose exists across scanners, protocols, and patient ages. These findings highlight the importance of accounting for dose uncertainty when interpreting epidemiologic risk estimates derived from nominal low-dose pediatric CT exposures.