Sample size calculation for clinical trials using magnetic resonance imaging for the quantitative assessment of
Tobias Saam1, William S Kerwin, Baocheng Chu
1Department of Radiology, University of Washington, Seattle, Washington 98195, USA.
Purpose:
To provide sample size calculation for the quantitative assessment of carotid atherosclerotic plaque using non-invasive magnetic resonance imaging in multi-center clinical trials. METHODS. As part of a broader double-blind randomized trial of an experimental pharmaceutical agent, 20 asymptomatic placebo-control subjects were recruited from 5 clinical sites for a multi-center study. Subjects had 4 scans in 13 weeks on GE 1.5 T scanners, using TOF, T1-/PD-/T2- and contrast-enhanced Tl-weighted images. Measurement variability was assessed by comparing quantitative data from the index carotid artery over the four time points. The wall/outer wall (W/OW) ratio was calculated as wall volume divided by outer wall volume. The percent lipid-rich/necrotic core (%LR/NC) and calcification (%Ca) were measured as a proportion of the vessel wall. For %LR/NC and %Ca, only those subjects that exhibited LR/NC or Ca components were used in the analysis.
Results:
Measurement error was 5.8% for wall volume, 3.2% for W/OW ratio, 11.1% for %LR/NC volume and 18.6% for %Ca volume. Power analysis based on these values shows that a study with 14 participants in each group could detect a 5% change in W/OW ratio, 10% change in wall volume, and 20% change in %LR/NC volume (power = 80%, p < .05). The calculated measurement errors presume any true biological changes were negligible over the 3 months that subjects received placebo.
Conclusion:
In vivo MRI is capable of quantifying plaque volume and plaque composition, such as %lipid-rich/necrotic core and %calcification, in the clinical setting of a multi-center trial with low inter-scan variability. This study provides the basis for sample size calculation of future MRI trials.
