|Home | About | Journals | Submit | Contact Us | Français|
We appreciate the opportunity to respond to comments by Drs. Gutierrez and Elkind regarding our recent article1. We believe they misunderstood our imaging-based methodology and would like to offer reasons that might explain their discrepant results2.
The authors assert “plaque burden” should be based on intimal area when determining the remodeling threshold. Our definition has been used in other published MRI and intravascular ultrasound studies that demonstrate remodeling justifiably since atherosclerotic changes can extend beyond the internal elastic lamina. Furthermore, co-linearity did not influence our results as suggested by Drs. Gutierrez and Elkind since we defined stenosis based on lumen reduction, unlike the definition used in pathologic specimen studies2,3. Regarding the slope of the regression line, S is simply a variable that can be negative or positive depending on whether the vessel tapers or, uncommonly, enlarges from proximal to distal locations. Others have similarly used a variable to represent slope4. Figure 1 is a schematic to help conceptualize use of this formula, and S would be negative to represent a tapering vessel. Drs. Gutierrez and Elkind are correct that the slope in supplemental Figure II should be negative and we appreciate bringing this to our attention.
In our paper, we stated “MRI measurements were obtained for all plaques detected in the proximal segments of the intracranial arteries” and then listed these segments (e.g., M1, M2). “Lesion site” was defined as the location of the thickest wall “for each segment” and had nothing to do with a proximal location within a segment, as suggested by Drs. Gutierez and Elkind. The reference site could be either proximal or distal to the plaque depending on where the wall was thinnest, as we explained in our paper, and the sign of the slope changed accordingly. Beta coefficients were therefore appropriately positive or negative.
We disagree that selecting only symptomatic patients will bias results from increased wall thickness in neighboring intima. Imaging studies enable wall measurements of an entire vessel segment beyond the plaque site, unlike the specimen studies Drs. Gutierrez and Elkind reference. Furthermore, most plaques included in our analysis were asymptomatic and not advanced high-grade lesions (Results section1) despite our inclusion criteria.
Regarding image resolution, we have previously shown that trained readers can make reasonably accurate visual interpolations of thin structures that overcome MRI resolution limitations5. Furthermore, we believe that this constraint would favor against identifying outward remodeling since the increased outer wall area at the lesion site might be undetectable due to partial volume averaging from inadequate resolution.
As noted in our paper, we were first to report differences in remodeling between anterior and posterior circulations but not to report outward remodeling in brain arteries (references 18–211). Drs. Gutierrez et al reported no outward remodeling in brain arteries2; however, they used a linear model to evaluate potentially non-linear associations (remodeling). Importantly, one cannot identify a threshold for remodeling using a linear term in regression (no inflection point, Figure 2b2). In fact, the regression depicted in Figure 2b seems to show no significant change in lumen area with increasing stenosis, which supports remodeling. Furthermore, although a mixed model was used to adjust within-patient variation, adjustments for between-patient variation (e.g., height, sex, race) are also needed to enable comparability of data points. This would have a different effect than categorizing by size as was done in their study. The prediction of negative lumen areas (Figure 2d2) further highlights inappropriate use of this model. Consequently, we believe the paper by Drs. Gutierrez et al. does not disprove remodeling in brain arteries.
Drs. Qiao and Wasserman have a patent pending (no 13/922,111) for the black blood MRI technique.