Published ahead of print on April 7, 2011
doi: 10.3174/ajnr.A2493

American Journal of Neuroradiology 32:E93, May 2011
© 2011 American Society of Neuroradiology

S.P. Kloska^a
aDepartment of Neuroradiology
University of Erlangen-Nuremberg
Erlangen, Germany
and
Department of Clinical Radiology
University of Muenster
Muenster, Germany

I have read with great interest the recently published article of Sharma et al.¹The study compared CT angiographic source imaging (CTA-SI) performed by a modern multisection CT scanner with cerebral blood flow (CBF) and cerebral blood volume (CBV) maps derived from dynamic CT perfusion in patients with acute ischemic stroke. The results demonstrate the strong correlationbetween the lesion volume of CTA-SI and CBF (r = 0.89, P < .001) in contrast to the weak correlation between the lesion volume of CTA-SI and CBV (r = 0.5, P< .001). The authors concluded that CTA-SI is CBF- rather than CBV-weighted.

Although it was probably not the intention of the authors, this conclusion suggests that CTA source data with modern multisection CT scanners somehow include CBF information. CBF can be calculated only from dynamic CT data acquired during the first pass of a contrast bolus. Hence, I consider the conclusion of Sharma et al¹ critical, and it is the opposite of those of previous reports of the blood volume basis of CTA-SI.² In my opinion, the reported results of apparent CBF-weighted CTA-SI in the article of Sharma et al need further discussion and detailed consideration of the underlying principle of CTA-SI.

CBV relates to the area under the curve and is expressed by the following equation:

where c(t) is the tissue concentration and v(t) is the vascular concentration of the marker at certain time points (t). For CT, this is only fulfilled with a dynamic CT data acquisition. As cited by the authors, CTA-SI, like the related techniques of perfusion-weighted CT and perfused blood volume (PBV) imaging, is based on the principle that was originally reported by Hamberg et al.² With the assumption of vascular and tissue contrast steady-state, the above calculation of CBV can be reduced to the following equation:

where max c(t) is the maximum tissue concentration and max v(t) is the maximum vascular concentration of the marker. This consideration eliminates the need for a dynamic CT data acquisition and is the basis of CTA-SI. Blood volume information can be extracted from CTA data when the data acquisition is performed at the plateau phase of contrast injection with vascular and tissue-contrast steady-state. With former generations of CT scanners, dedicated injection protocols in CTA-SI were not that crucial because the slow scanning time “automatically” resulted in appropriate bolus configuration to fulfill the algorithm of CTA-SI. In contrast, the fast scanning times with modern CT scanners require very accurate bolus timing for CTA-SI. Sharma et al¹ used an injection protocol with a delay between 5 and 10 seconds, up to 90 mL of 300-mg iodine/mL concentrated contrast agent, a flow rate of 5 mL/s, and no saline flush. With the fast scanning time of the 64-section CT scanner used, this injection protocol results in a distinct arterial contrast. However, the mandatory assumptionfor CTA-SI with vascular and tissue contrast steady-state according to the considerations of Hamberg et al² is thereby violated. Hence, the size of the CTA-SI lesion is overestimated.

It is not surprising that Sharma et al¹ found a better correlation between CTA-SI with CBF, a parameter estimating infarct core and penumbra, and not with CBV, a parameter reflecting the infarct core. However, this result does not imply causality but is the consequence of the injection protocol used. It has been recently demonstrated that using an injection protocol with individualized delay for peak enhancement in the superior sagittal sinus, 80 mL of 370-mg iodine/mL concentrated contrast agent, a flow rate of 4 mL/s, and 50 mL of saline flush resulted in a strong correlation between the lesion volume of PBV with CBV (r = 0.922, P < .01) when performed by a modern multisection CT scanner.³ In consequence, the results of Sharma et al do not contradict the recommendations of the American Heart Association for CTA-SI⁴ but point out the importance of appropriate injection protocol in conjunction with modern multisection CT scanners to fulfill the requirements for the appropriate use of CTA-SI.

References

1. Sharma M, Fox AJ, Symons S, et al.CT angiographic source images: flow- or volume-weighted? AJNR Am J Neuroradiol 2011;32: 359–64[Abstract/Free Full Text]
2. Hamberg LM, Hunter GJ, Kierstead D, et al.Measurement of cerebral blood volume with subtraction three-dimensional functional CT. AJNR Am J Neuroradiol 1996;17: 1861–69[Abstract]
3. Wittkamp G, Buerke B, Dziewas R, et al.Whole brain perfused blood volume CT: visualization of infarcted tissue compared to quantitative perfusion CT. Acad Radiol 2010;17: 427–32[CrossRef][Medline]
4. Latchaw RE, Alberts MJ, Lev MH, et al.Recommendations for imaging of acute ischemic stroke: a scientific statement from the American Heart Association. Stroke 2009;40: 3646–78[Free Full Text]

Reply

Published ahead of print on April 7, 2011
doi: 10.3174/ajnr.A2510

American Journal of Neuroradiology 32:E94, May 2011
© 2011 American Society of Neuroradiology

R.I. Aviv^a, M. Sharma^a and T.-Y. Lee^a
aDepartment of Medical Imaging
University of Toronto and Sunnybrook Health Sciences Centre
Toronto, Ontario, Canada

We thank Dr Kloska for his insightful remarks on our study, which showed that current CT angiographic source imaging (CTA-SI) may be flow- not volume-weighted.¹ We wish to clarify a number of points, specifically as they relate to the technical differences between the study he cites² and ours. First, we would like to clarify his observations; our contrast injection is routinely followed by 40 mL of normal saline injected at the same rate as the initial contrast bolus. The divergence of our results from those reported in his recent study is most likely attributable to the difference in the order in which CTA and CT perfusion (CTP) were performed in each study.² In our protocol, CTA was performed while the patient was contrast-naïve. This, coupled with the fact that contrast does not have enough time to reach all the blood vessels in the ischemic/infarct region, suggests that the CTA-SI reflects the arrival time difference between the normal and ischemic regions.

In general, it is always true that a region with a long arrival time would have low blood flow. This explains why the CTA defect in our article is correlated to the blood flow defect. On the other hand in the study cited by Dr Kloska,² CTA was performed after injection of contrast for a CTP acquisition. By the time CTA is complete, the contrast injected for the CTP study would have time to reach all the blood vessels within the ischemic region, accounting for a smaller CTA-SI defect than when the brain is contrast-naïve.

Dr Kloska suggests that a time delay based on maximum enhancement in the sagittal sinus may account for differences in the 2 studies. We do not find this explanation convincing because the time to maximum sagittal sinus enhancement in a patient with stroke may not be that different from that in a healthy subject because the sagittal sinus drains blood from the whole brain including the normal hemisphere. Indeed, it is the cortical veins, not the superior sagittal sinus, where contrast delay is observed in balloon test occlusions.³ We do not think that the slightly longer delay used in their study would allow enough time for contrast to reach all blood vessels in the ischemic/infarct area; this belief strengthens our initial assertion that differences are related to CTA/CTP order and not to the appropriateness of the contrast injection protocol.

CTA-SI images can be either blood flow– or blood volume–weighted. For a region in the brain where maximum contrast arrival time (Tmax) is shorter than the delay time of the acquisition of the CTA study, the region in the CTA-SI will be blood volume–weighted. Conversely, if Tmax is longer than the delay time of the CTA acquisition, the region in the CTA-SI will be blood flow–weighted. “Tmax” is defined as the time-to-peak of the deconvolved impulse residue function (IRF) as in the Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution study.⁴ A significant portion of appearance time (T0) is the contrast arrival/T0 at the brain region relative to that of the input arterial function used to calculate the IRF in the deconvolution. We randomly selected 2 acute stroke studies from our data base and found that T0 and Tmax can be as long as 14 seconds and 22 seconds, respectively.

In summary, the possible dual nature of CTA-SI defects, either ischemia or infarct, means that it has to be interpreted together with Tmax or T0 maps and correlated with the scanning time after injection of contrast. This consideration points to the potential pitfall of the use of CTA-SI by itself in acute stroke diagnosis.

References

1. Sharma M, Fox AJ, Symons S, et al.CT angiographic source images: flow- or volume-weighted? AJNR Am J Neuroradiol 2011;32: 359–64[Abstract/Free Full Text]
2. Wittkamp G, Buerke B, Dziewas R, et al.Whole brain perfused blood volume CT: visualization of infarcted tissue compared to quantitative perfusion CT. Acad Radiol 2010;17: 427–32[CrossRef][Medline]
3. Abud DG, Spelle L, Piotin M, et al.Venous phase timing during balloon test occlusion as a criterion for permanent internal carotid artery sacrifice. AJNR Am J Neuroradiol 2005;26: 2602–09[Abstract/Free Full Text]
4. Albers GW, Thijs VN, Wechsler L, et al.Magnetic resonance imaging profiles predict clinical response to early reperfusion: the Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution (DEFUSE) study.Ann Neurol 2006;60: 508–17[CrossRef][Medline]