The use of QMRA becomes of increasing importance to study ischemic stroke in the context of intracranial atherosclerosis [24]. TOF MRA allows an accurate assessment of the severity of narrowed or occluded intracranial vessels [24]. While NOVA QMRA allows a more comprehensive evaluation of the cerebral flow and hemodynamics (collateralization) in ischemic stroke through non-invasive measurement of VFR [14], it is a highly specific software available in selected neurovascular centers. On the other hand, TOF MRA is accepted as a standard imaging technique in cerebrovascular examination and widely accessible even in regional hospitals. Both imaging modalities are greatly beneficial in the evaluation of large vessel steno-occlusive disease, but, to our knowledge, no record of comparison between NOVA QMRA and MRA TOF is found in recent literature, neither in clinical settings nor with healthy subjects.

This study is the first on in which the correlation between SIR and VFR was evaluated in a cohort of patients presenting with symptomatic unilateral ICA occlusion. The comparison was performed between the M1- and P2-segments ipsi- and contralateral to the occlusion.

The analysis showed a significant correlation between VFR measured via NOVA QMRA and SIR in TOF MRA. The strength of this correlation is moderate in M1- and P2-segments ipsilateral to the occluded ICA; while, the contralateral vessels show strong correlations with R-values greater than 0.7.

Subgroup analysis (acute, subacute and chronic groups) confirmed the significant correlation regardless of time of execution of MR after symptoms. Although this statement holds generally true, further analysis between ipsi- and contralateral M1- and P2-segments in the different subgroups shows few exceptions that need to be discussed (Table 3).

When considering the vessels specifically, in the subgroup of patients with acute ICA occlusions, we observed a weaker correlation between VFR and TOF SIR in the contralateral M1- and P2-segment. The weaker correlation in those specific arteries could be explained by the impaired intracranial hemodynamic following the first occurrence of the symptomatic occlusion [25].

In the subgroup of patient with subacute ICA occlusions, we could not find significant p values in the ipsi- and contralateral M1-segments. This result is most likely due to a disproportionate number of tandem occlusions (3/7, respectively, 3/8) in this subgroup.

The strongest correlation between NOVA VFR and TOF SIR was found in the vessels of patients with chronic ICA occlusion. Probably because this group had the greatest chance to undergo a long-term adaptation and see the re-emergence of a hemodynamic equilibrium after the ICA occlusion [25, 26] (Table 2).

Another important finding regards the difference in hemodynamic behavior between ipsi- and contralateral M1 and P2 in the whole cohort. While M1 VFR and SIR tend to be higher in the contralateral side to the occlusion, the opposite is true for P2 VFR and SIR (Table 3).

The higher VFR and TOF SIR in ipsilateral P2-segments support the concept of previous studies that showed that cerebral perfusion of the ipsilateral hemisphere may be rescued by activation of leptomeningeal collaterals [27].

The present comparative analysis was designed to assess whether TOF SIR could give a rapid qualitative information about flow/perfusion in patients with symptomatic ICA occlusion. TOF SIR does correlate with VFRs of the ipsi- and contralateral M1- and P2-segments, although the correlation is higher on the side contralateral to ICA occlusion.

Although able to show a significant correlation between VFR and TOF SIR in M1- and P2-segments the obtained insights need to be interpreted in the framework of the retrospective nature and small sample size of this study.

Another limiting factor for the accuracy of the correlation between VFR and TOF SIR is the impact of arterial geometry on the SI in TOF MRA imaging. Contrast between vessels and stationary tissue is produced by flow related enhancement, a phenomenon that is highly dependent on perpendicular out of scan plane movement of blood. Arteries staying in the scan plane or even presenting with a backwards flow experience a spin saturation and therefore loose flow related enhancement faster, thus displaying a lower SI than arteries that move out of the scan plane. Furthermore, this study did not investigate further the added value of NOVA-QMRA compared to other advanced imaging modalities which study cerebral perfusion in patients presenting intracranial large vessel occlusions. Indeed this aspect should represent the focus for future research to further validate the use of NOVA in clinical settings. Lastly, in our study population, only a very limited number of patients were identified as having Circle of Willis variants; therefore, no analysis differentiating between a normal and a Circle of Willis showing anatomical variations was possible.

Nonetheless, TOF SIR sequences could be valuable for a rapid and qualitative assessment of the flow in M1- and P2-segments in patients with unilateral ICA occlusions; while, NOVA QMRA remains the only technique allowing precise quantitative measurement in ml/min. Figures 1 and 2 display this relationship: Fig. 1 shows symmetrical results of TOF SIR and NOVA VFR in bilateral M1-segments (Fig. 1); whereas, Fig. 3 shows a clear asymmetry of the SIRs in the two M1-segments and consequently different NOVA VFRs (Fig. 4). Higher VFR and TOF signal in ipsilateral P2-segments suggest an activation of the leptomeningeal collateral pattern. The results support the use of TOF SIR as a qualitative surrogate of NOVA-VFR. These findings may promote the referral of patients with symptomatic ICA occlusion to tertiary centers for further flow hemodynamic evaluation.

Graphs show the difference between mean SIR and VFR according to examined vessel and side of occluded ICA. ICA: internal carotid artery; SIR: signal intensity ratio; VFR: volume flow rate