Quantitative Perfusion, Oxygenation, and CMRO2 Imaging in Post-Acetazolamide Moyamoya Disease Patients
Wendy Ni1,2, Thomas Christen2, and Greg Zaharchuk2

1Department of Electrical Engineering, Stanford University, Stanford, CA, United States, 2Department of Radiology, Stanford University, Stanford, CA, United States

Synopsis

The acetazolamide challenge can be used to assess cerebrovascular reserve and oxygenation in patients with steno-occlusive diseases such Moyamoya, thus enabling evaluation of the quantitative blood oxygen-level dependent (qBOLD) approach of modeling tissue oxygenation. In this study, we mapped post-acetazolamide oxygenation (with transverse relaxation rate R2’), CBF with arterial spin labeling (ASL), and CBV with dynamic susceptibility contrast (DSC). We found that angiographically abnormal tissues are relatively hypoperfused and hypoxic. Finally, we investigated a qBOLD biophysical model for quantitative tissue oxygenation which suggested no difference in the cerebral metabolic rate of oxygen consumption (CMRO2) between normal and affected regions.

Purpose

Acetazo­­lamide (ACZ) injection is a robust clinical perfusion challenge for assessing cerebrovascular reserve1. For steno-occlusive pathologies such as Moyamoya disease, oxygenation mapping may provide additional dialogistic information2. In this study, we perform R2’ relaxometry for oxygenation mapping, multi-delay ASL for measuring cerebral blood flow (CBF), and dynamic susceptibility contrast (DSC) for measuring cerebral blood volume (CBV) after ACZ injection in a cohort of pre-operative Moyamoya disease patients. We also compare our findings with a quantitative BOLD (qBOLD) oxygenation model3 to estimate the cerebral metabolic rate of oxygen consumption (CMRO2) in both affected and non-affected tissues.

Methods

19 Moyamoya disease patients (ages 39±12 y, 14 F) were scanned with informed consent and IRB approval, at 3.0T (MR750W, GE-Healthcare) with ACZ injection (1g IV). Post-ACZ R2’ mapping was performed using GESFIDE4 (TESE/TR 100/2000ms, TE 5-130ms, 40 echoes, resolution 1.9×1.9×1.5mm3, 14 slices). Post-ACZ perfusion was mapped with two sequences: a) multi-delay pseudo-continuous ASL5 (TR/TE 6518/25.1ms, label time 2000ms, 5 equally spaced PLDs 700-3000ms, resolution 3.4×3.4×4mm3, 36 axial slices), offering improved quantitation in areas of slow collateral flow; b) DSC imaging (TR/TE 1800/40ms, resolution 1.9×1.9×5.0mm3 with 20 slices) using 0.1mmol/kg of gadolinium contrast (Multihance, Bracco). Whole-brain T1-weighted 3D IR-FSPGR and TOF MRA of the Circle of Willis were performed.

GESFIDE images were analyzed in native space, with R2’ maps calculated using mono-exponential fitting4. CBF maps were calculated from ASL images6, while DSC images were processed7 using commercial software (RAPID, iSchemaView) before combined ASL-DSC (CAD) correction8 was applied for absolute CBV quantitation. Using a multiparametric qBOLD approach9 with arterial/venous blood volume ratio 30/70 10, we created tissue oxygen saturation (StO2) and CMRO2 maps. Data was analyzed using 3cm annular, mixed-cortical ROIs approximating major arterial territories, with 6 radial segments at two standard 14mm-thick levels. Each ROI was classified “normal” or “abnormal” by blinded inspection of TOF MRA images and axial projection by an experienced neuroradiologist. Comparisons were performed using the 2-tailed t-test with α=0.05.

Results

The results from group-level analysis (N=228) are shown in Table 1. Post-ACZ R2’ and CBF significantly differed between normal and abnormal ROIs (respectively, 16±31% higher and 13±45% lower in abnormal ROIs), while CBV did not (p=0.18).

Evaluating all regions, R2’ significantly decreased with increased perfusion metrics, consistent with better oxygenation in regions with better perfusion (Fig. 2-3): (normal) R2’=-0.027*CBV +3.23, R2=0.06, p<0.001; and (abnormal) R2’=-0.015*CBF +3.94, R2=0.19, p<0.001. The slopes of regression lines were not significantly different between normal and abnormal data, while the intercepts were.

Finally, we calculated post-ACZ StO2 to be mostly over 99% and significantly (p<0.05) lower in abnormal ROIs, though only by a small amount. Post-ACZ CMRO2 did not differ between normal and abnormal regions.

Discussion

Our study found that post-ACZ CBF, measured via multi-delay ASL, discriminated between angiographically normal and abnormal regions. Since diseased tissues experience chronic vasodilation at baseline and have limited ability to further vasodilate, it is reasonable that no differences in post-ACZ CBV were observed.

R2’ imaging showed that the amount of deoxyhemoglobin also differed in normal and abnormal tissues. According to the qBOLD biophysical model presented by Yablonskiy and Haacke3: $$$R_2 \propto Hct\cdot DBV \cdot (1-StO_2)$$$, where Hct is capillary hematocrit, DBV is deoxygenated blood volume and StO2 is tissue oxygen saturation, we conclude that StO2 is lower in abnormal regions, consistent with our expectation that some of them experience chronic under-oxygenation. However, the weak correlation (Fig. 2) suggests that Hct and/or post-ACZ StO2 were not sufficiently constant across data points, and/or CBV was not proportional to DBV. The significant offset indicates sources of R2’ other than oxygenation, some of which have previously been discussed in literature4.

Finally, the StO2 values calculated from the multiparametric qBOLD approach were significantly higher compared to literature11, though StO2 was lower in affected regions, some of which are expected to be under-oxygenated. The source of the high StO2 values appears to be the very high CBV values measured (Table 1). CMRO2 in angiographically normal regions was comparable to measurements obtained via positron emission tomography12. The lack of significant difference between normal and abnormal regions may indicate that most patients in our study are not at late stage of Moyamoya disease where tissue metabolism is reduced.

Conclusion

In this study, we imaged post-ACZ perfusion in Moyamoya disease patients using multi-delay ASL and DSC. We also imaged R2’ for oxygenation quantitation with qBOLD methods. Our data agreed with our expectation that angiographically abnormal tissues are under-oxygenated and less well-perfused during a perfusion challenge, but quantitation of StO2 requires needs further evaluation.

Acknowledgements

NIH R01NS066506, R01NS047607, R21NS087491, NCRR 5P41RR09784. GE Healthcare.

References

1. AS Vagal et al., Am J Neuroradiol, 2009.

2. W Powers, Ann Neurol, 1991.

3. DA Yablonskiy & EM Haacke, Mag Reson Med, 1994.

4. W Ni et al., Mag Reson Med, 2015.

5. R Wang et al., Eur Radiol, 2014.

6. W Dai et al., Mag Reson Med, 2012.

7. M Straka et al., J Mag Reson Imag, 2010.

8. G Zaharchuk et al., Mag Reson Med, 2010.

9. T Christen et al., Mag Reson Med, 2012.

10. D Bulte et al., Proc ISMRM, 2015.

11. V Faoro et al., J Appl Physiol, 2007.

12. H Okazawa et al., JCBFM, 2001.

Figures

Table 1: Post-ACZ measurements showed significant differences in R2’, CBF, Tmax and StO2 between angiographically normal and abnormal ROIs, while CBV and CMRO2 was the same in both regions.

Fig. 1: Example of (a) axial maximum intensity projection (MIP) of TOF MRA indicating an occluded MCA and a stenosed ACA; and (b-c) two levels of analysis with overlaid regions of interest (ROIs) in which the color indicates its angiographic status.

Fig. 2: Linear regression of R2’ vs. CBV found a very weak negative correlation. Separate fitting of normal and abnormal data found significantly different (p<0.05) intercepts but identical slopes: ynormal=-0.024x+2.98 and yabnormal=-0.024x+3.40, potentially indicating a non-perfusion, non-oxygenation source of R2’ affected by disease, or a non-linear relationship in general.

Fig. 3: Linear regression of R2’ vs. CBF also found a slightly stronger negative correlation. Separate fitting found significantly different intercepts and slopes: ynormal=-0.007x+3.23 and yabnormal=-0.017x+4.17.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1432