Assessing Perfusion in Steno-Occlusive Cerebrovascular Disease Using Transient Hypoxia-Induced Deoxyhemoglobin as a Dynamic Susceptibility Contrast Agent

Main Findings

Resting cerebral perfusion metrics in patients with cerebrovascular steno-occlusive disease calculated with THx-dOHb contrast were similar to those generated with GBCA. In particular, the spatial distributions of the respective perfusion metrics MTT, rCBF, and rCBV were similar in both GM and WM. Both contrast agents were also similar, enabling the discrimination between the affected and the unaffected hemispheres. Furthermore, the distributions of hemodynamic metrics between hemispheres were consistent with the locations of large-vessel pathology. We note that to eliminate any methodological, temporal, and disease variability, each participant was used as his or her own control. These findings, therefore, show that resting perfusion metrics obtained noninvasively with THx-dOHb can be of clinical value, particularly after considering the benefits and risks compared with GBCA.

Summary of Findings

Perfusion metrics MTT, rCBF, and rCBV were compared between GBCA and THx-dOHb in GM and WM. The perfusion metrics generated using the 2 contrast agents were not significantly different from one another, except for WM MTT. The WM MTT for THx-dOHb was significantly lower in comparison with GBCA (P < .05). Nevertheless, the individual participant perfusion maps are spatially very similar. Indeed, the results from the Bland-Altman plot analysis showed that the average discrepancy (constant bias) between the two agents is small for all perfusion metrics, with a narrow confidence interval, except for a constant bias for WM MTT in which the GBCA MTT is consistently greater than THx-dOHb. Furthermore, the differences between perfusion metrics in GM or WM have no proportional bias, with the slope of the regression line not significantly different from zero. The ratio of affected and unaffected hemispheres were the same for all subjects, with each contrast except for WM rCBV.

We suggest that these findings support the continued exploration of THx-dOHb for DSC imaging as (i) there is high agreement between metrics obtained using DSC imaging with GBCA and (ii) THx-dOHb is able to discriminate between affected and unaffected hemispheres in patients with steno-occlusive disease.

Advances over Current Knowledge

This study in patients with steno-occlusive disease follows previous investigations that demonstrated the utility of THx-dOHb as a contrast agent. Poublanc et al⁷ were the first to use THx-dOHb to calculate and map voxelwise resting perfusion metrics. The investigators applied sequential gas delivery¹³ and prospective end-tidal gas targeting¹² to generate abrupt transient isocapnic reductions in oxygen, to create pharmacokinetic profiles with dOHb that are similar to those obtained from an intravascular injection of GBCA, thereby enabling a similar approach to the calculation of perfusion metrics. Subsequently Sayin et al¹¹ found that the perfusion metrics generated by the 2 contrast agents were similar when assessed in the same healthy individuals. In this study, we extended these observations to show similarity in pattern as well as in the interpretation of perfusion metrics in patients with cerebrovascular pathology in the form of steno-occlusive disease.

THx-dOHb has several advantages over GBCAs as a contrast agent for DSC MR imaging. GBCA is an intravascular extracellular molecule. It can leak into the interstitial fluid¹⁹ in pathologic conditions when the BBB is disrupted and, therefore, requires leakage correction. By contrast, due to dOHb being contained in red cells, it remains intravascular regardless of the state of the BBB. This capacity may be a particularly useful advantage when the BBB is disrupted, provided that there is no active bleeding.²⁰ Furthermore, the analysis of the signal generated by GBCA can be affected during a second pass through the circulation, which will not occur with THx-dOHb unless there is a right-to-left cardiopulmonary shunt. On the contrary, THx-dOHb is cleared within 1 blood circulation time following pulmonary reoxygenation, enabling repeat measures as required. THx-dOHb may also be a suitable alternative contrast for patients with renal insufficiency who are at risk of nephrogenic systemic fibrosis if exposed to GBCA or at risk of worsening of renal function if exposed to iodinated contrast used for CTP.

During the past decade, there has been a drive to develop noninvasive techniques that are capable of identifying vascular dysfunction in individuals who are at risk. Arterial spin-labeling is a noninvasive MR imaging technique that provides quantitative measures of CBF. However, this technique has sensitivity to transit times and long collateral flow pathways associated with advanced cerebrovascular disease.^21⇓-23 Currently, the only clinical standard for assessing perfusion metrics in patients with cerebrovascular steno-occlusive disease is DSC imaging with GBCA. However, future work comparing CBF measured via Thx-dOHb with arterial spin-labeling may be beneficial.

Bolus Profile Comparison

The stimulus profile we used in this study was chosen to be consistent with its antecedent studies. The arterial input function profile of THx-dOHb differs from that of GBCA by having 2 susceptibility oxygen desaturations (Fig 1). Due to the strong paramagnetic properties of GBCA and the rapid IV injection, GBCA causes a very sharp decrease in the BOLD signal. In contrast, the rate of dOHb formation is limited by the rate of deoxygenation in the lungs. This process is one of breath-by-breath dilution of the oxygen in the lungs, over about 5–7 breaths taking about 20 seconds. Thus, the THx-dOHb performed a double 60-second hypoxic exposure targeting partial pressure of oxygen (PO₂) of 40 mmHg to improve the CNR. Previous studies have determined that the CNR increased marginally with repeat hypoxic stimuli. Although the susceptibility profiles of GBCA and Thx-dOHb differ, the arterial input functions can be used for both agents for the calculation of the respective perfusion metrics.^7,11 Nevertheless, further changes in the hypoxic exposure profile may be considered for implementing strategies to shorten the time to reach target hypoxia.

Limitations

The tolerance of hypoxia in humans has been well-established. In formal studies, participants maintained a PO₂ at about 50 mm Hg for >20 minutes.^24⇓-26 In another study, a PO₂ of 55 mm Hg was maintained for 8 hours.²⁷ Hypoxia is experienced continuously by the millions who live at high altitudes, especially those with severe lung and heart disease, and intermittently by hundreds of millions throughout the world who have multiple episodes of obstructive sleep apnea nightly. Indeed, extended periods of intermittent hypoxia have even shown benefits of preconditioning against myocardial infarction,^28,29 heart failure,³⁰ hypertension,^30,31 and cognitive decline.³²

There is likely a range of sensitivity to hypoxia in humans and a range of its tolerability. In this study, 2 of 12 patients did not tolerate the hypoxia and aborted their study. Both participants who aborted the study did so during the second hypoxic exposure. It is possible that the low tolerance for hypoxia was related to the severity of the perfusion limitation, though there are no indications of this effect in people with severe steno-occlusive disease living at high altitudes or with concomitant severe pulmonary disease or severe, prolonged obstructive sleep apnea.

Another limitation of THx-dOHb is the issue of CBF increase in response to hypoxia, which could confound the measurement of perfusion metrics. Ellingsen et al³³ had reported that a normocapnic reduction of PO₂ to 65 mm Hg resulted in an increase in carotid artery blood flow velocity over a period of about 10 minutes with a halftime of about 4 minutes. More recently, Harris et al²⁴ used pulsed arterial spin-labeling to measure the delay and time constant of a CBF increase in response to acute hypoxia. They found a delay to onset of flow change of about 3 minutes with a time constant of about 4.5 minutes. The duration of our hypoxic stimuli was only 60 seconds, repeated twice. However, the precise kinetics of CBF responses to acute severe hypoxia have not been rigorously documented.³⁴ If acute hyperemic responses do occur, reducing the extent of hypoxia or shortening the hypoxic interval may be feasible. These measures would also increase the tolerability of hypoxia.^7,35,36 A further limitation may occur in areas of misery perfusion in which blood flow is low and the oxygen extraction fraction is high, resulting in a high baseline concentration of dOHb. With the initiation of THx-dOHb, the arterial oxygen delivery is further reduced, resulting in an exaggerated fall in the BOLD signal, perhaps marking the area of misery perfusion.

The presence of steno-occlusive disease or cerebral ischemia may itself result in dispersion of the bolus of contrast³⁷ and difficulty in accurately applying the input function. There are numerous ways to analyze perfusion data including, but not limited to, adopting a mono-/biexponential model from pharmacokinetics,³⁸ or using a vascular model³⁹ or a singular value decomposition.⁴⁰ However, these deconvolution techniques do not account for dispersion. For that, a complex model for the vasculature is required.⁴¹ When there is no correction for contrast dispersion, caution should be used in interpreting the calculated MTT, rCBF, and rCBV metrics from either contrast. Similarly, if such peripheral late dispersion does exist, the deconvolution model would be equally offset with both GBCA and THx-dOHb profiles.