Association of Carotid Artery Disease with Collateralization and Infarct Growth in Patients with Acute Middle Cerebral Artery Occlusion

DISCUSSION

Because collaterals are considered an important prognostic factor in large vessel occlusion, factors influencing the collateral status are of high interest.^1,2 There are heterogeneous data on the correlation of CAD with the intracranial collateral supply.^{11⇓⇓⇓⇓-16} Until now, the pathophysiological role of CAD in collateralization was unclear. According to our results, ipsilateral CAD is associated with good collateralization. However, there is no rank correlation of the stenosis degree with the collateral grading, ie, patients with stroke with a concomitant 20% ICA stenosis do not necessarily have better collaterals compared with patients with a 10% stenosis.

Furthermore, we could demonstrate a positive association of ipsilateral CAD with infarct growth independent of collateralization. This means that patients with CAD and poor collaterals show less infarct progression than do patients without CAD.

Hence, our data, with the largest cohort (n = 709), confirms the study results of Rebello et al¹⁴ (n = 122), Pienimäki et al¹¹ (n = 247), Hassler et al¹² (n = 281), and Guglielmi et al¹³ (n = 666), all of whom reported an association between good collateral status and carotid stenosis in their studies.

However, the study designs including patient inclusion criteria have noteworthy differences and potential bias. All 4 studies supporting our results included terminal ICA occlusions, consequently impacting the circle of Willis flow by obstructing a potential blood pathway from the posterior communicating artery to the A1 segment of the anterior cerebral artery or vice versa. Depending on anatomic variation of the circle of Willis, this could have a relevant impact.²³ In the cases of Pienimäki et al¹¹ and Guglielmi et al¹³, it is also unclear how the stenosis degree on these particular patient groups was assessed on CTA, as the pseudo-occlusion of the ICA are frequently observed in patients with terminal ICA occlusions. Rebello et al¹⁴ solved this issue by using DSA, and Hassler et al¹² also mention that they validated their results with DSA.

Furthermore, Hassler et al¹² and Guglielmi et al¹³ included M2 occlusions, while also using the Tan score for collateral status evaluation. This potentially increased the number of patients with good collateral supply, as they considered Tan 2 as good collateral status. Rebello et al¹⁴ also included M2 occlusions while using the Souza score, which is a modified version of the Tan score. Nevertheless, there was also a potential increase of patients with better collaterals, as dichotomized between a Souza score of 0–1 (absent/poor collaterals) and a Souza score of ≥2 (moderate/good collaterals).

We tried to homogenize our cohort by solely including intracranial M1 occlusions and only considering Tan 3 as good collaterals, similarly to Pienimäki et al¹¹, who considered Souza 3 and 4 as very good collaterals and statistically conjoined them due to a small number of patients having a Souza score of 4.^{11⇓⇓-14,31}

Contrary results were reported by Dankbaar et al¹⁵ and Sobczyk et al.¹⁶ Dankbaar et al¹⁵ analyzed 188 patients with M1 occlusion from the prospective Dutch Acute Stroke Trial and found no correlation between stenosis degree and collateralization. However, only 18 patients had a stenosis of ≥70%. As pointed out by Pienimäki et al¹¹, this is a relatively small number of patients compared with their study of 51 patients (approximately 20%) with ≥75% stenosis. In our cohort, 118 patients (approximately 17%) presented with carotid stenosis of ≥70%. Therefore, there is a higher likelihood of bias in the data of Danbkaar et al.^11,15 This correlation is also supported by the results of Guglielmi et al¹³, who report a slightly better collateralization in patients with severe carotid stenosis (>70%) compared with moderate stenosis (51%–70%).

However, Sobczyk et al¹⁶ had a completely different study design and did not investigate patients with LVO. They included 58 patients with different stenosis degrees from their database and primarily investigated cerebrovascular reactivity. They concluded that collateralization varies greatly and cannot be predicted solely based on the degree of carotid stenosis.

As pointed out initially, we could demonstrate a positive association of ipsilateral CAD with infarct growth independent of collateralization. Patients with CAD and poor collaterals still had a lower rate of infarct progression than did patients without CAD. To our knowledge, this is the first report of a correlation between CAD and infarct growth.

Interestingly, while patients with ipsilateral CAD had a greater baseline infarct size, infarct growth was smaller compared with patients without ipsilateral CAD. In the end, follow-up ASPECTS was similar in patients with and without CAD in our cohort. To our knowledge, this particular finding has not been reported previously, and it stands in contrast to the study results by Deng et al³², who, in their retrospective study of 158 patients who underwent thrombectomy, demonstrated that greater baseline ASPECTS was associated with less infarct growth. However, they did not differentiate between patients with CAD.

The pathophysiological correlation between CAD and collateralization/infarct growth is not yet fully understood. One possible explanation for the lower baseline ASPECTS could be that patients with initially activated collaterals due to CAD have impaired reserve capacities.^5,16 However, in theory, the chronic hypoxic state caused by CAD may have preconditioned neurons to hypoxia, which ultimately prolonged their resilience in the event of an acute intracranial large vessel occlusion.³³ This hypothesis is supported by an animal experiment by Choi et al³⁴, who experimented on rats with either bilateral common carotid artery ligation or sham operation that were exposed to MCA occlusion and reperfusion. Conclusively, rats with bilateral common carotid artery ligation had smaller infarcts, less DNA-damaged cells, increased cellular defense mechanisms, and evidence of extracellular matrix remodeling. Another possible explanation for the association of CAD and collateralization is that a chronic cerebral hypoxic state stimulates arteriogenesis and angiogenesis, as suggested by Busch et al¹⁸ and Kahn et al¹⁷, based on animal experimental studies. Similar results were reported by Ohtaki et al¹⁹ and Hai et al²⁰, who observed angiogenesis in chronic cerebral hypoperfusion due to an upregulation of vascular endothelial growth factor in rat models.

Therefore, in view of clinical practice, our retrospective data suggest that patients with large ischemic core and tandem lesions should not be excluded from endovascular treatments, as infarct growth seems to be less present in patients with CAD. This is notable, considering that patients with tandem occlusions are typically more technically challenging and take longer to recanalize than do those with isolated MCA occlusions.³⁵

This finding is of particular interest, as the 6 randomized controlled trials on thrombectomy in patients with large ischemic core were in favor of thrombectomy. Only 3 of the 6 large core trials included cervical tandem occlusions.^{36⇓⇓⇓⇓-41} Further studies and post hoc analyses are necessary to understand the relevance of tandem lesions.

In contrast to other studies, we also analyzed the contralateral ICA for the presence of CAD and its relationship with collateral status and infarct growth. According to our results, contralateral CAD is associated with infarct growth, but it is not an independent predictor when taken together with ipsilateral CAD. Regarding this topic, Maus et al⁴² retrospectively analyzed 197 patients with tandem occlusion who were undergoing endovascular therapy. Even though they did not explicitly investigate infarct progression, their results suggest an adverse effect of contralateral stenosis on clinical outcome. They reported that the presence of a contralateral carotid stenosis of >50% was associated with a worse clinical outcome.⁴² However, further studies are necessary.

The strengths of our study are its large cohort size with 709 patients and its assessment of not only ipsilateral but also contralateral ICA regarding CAD. In addition, to our knowledge, this is the first report of a correlation between CAD and infarct growth. We potentially avoided multiple pitfalls by only focusing on intracranial M1 occlusions. By excluding terminal ICA occlusions, we prevented the potential impairment of collateral recruitment through the circle of Willis and avoided issues with stenosis assessment with pseudo-occlusion. By excluding M2 and distal occlusions, we made the assessment of collateral more distinct.

However, this study has some limitations. The retrospective design of our study may have led to selection bias. Our study was limited to patients with M1 segment occlusions and therefore may not be applied to other occlusion locations. Another bias may have resulted from the use of conventional (single phase) CTA for the collateral assessment, as it is static and only displays a single momentum in the arterial phase. However, multiphase or dynamic CTA was not available for every patient. Excluding patients without multiphase or dynamic CTA would have led to a selection bias. Moreover, nonenhanced CT and CTA images were acquired using various CT machines within our regional stroke network. Even though the image quality was similar overall, it may have resulted in a bias. In addition, CT or MR imaging was used for follow-up imaging, which may have led to a bias in ASPECTS interpretation.⁴³ Moreover, we did not distinguish between different onset times to imaging. As such, this may have an impact, especially on the initial ASPECTS, as reported by Potreck et al.⁴⁴ Similarly, the exact time of follow-up imaging was not recorded in our institutional thrombectomy registry. It is usually done 1 day after thrombectomy, but this time interval might have an impact on the assessment of the postinterventional infarct size.