Fibromuscular dysplasia (FMD) is defined as a non-atheromatous, non-inflammatory arteriopathy involving medium and small-sized arteries [1]. The core of this definition is focused on systemic arteries and includes cerebro-afferent arteries. In fact, the involvement of cerebrovascular extracranial arteries in FMD is frequent and similar in proportion to the involvement of renal arteries [2,3]. These data come from international registries and refer mainly to the two FMD subtypes distinguished in a recent expert consensus [1], i.e., multifocal and focal FMD. In addition, FMD is a multisite disease. In the Assessment of Renal and Cervical Artery Dysplasia (ARCADIA) registry [4], the prevalence of multisite involvement was 48% in the whole cohort of patients with FMD and 54% in the subgroup with cerebrovascular presentation. One of the main strengths of this study is the systematic assessment of multisite involvement.

Carotid web is partially out of this scheme, although, already several decades ago, it was described as belonging to the neuroimaging spectrum of FMD. Over time, the evidence derived from past studies in which diagnosis was made by catheter angiography with histopathological confirmation has been progressively forgotten or neglected. Recently, however, interest in carotid web as a possible cause of ischemic stroke and as part of the neuroimaging spectrum of FMD has regained strength. However, beyond spontaneous arterial dissections, which have a higher prevalence in the cerebrovascular vessels compared to other arterial sites, the role of FMD as a cause of ischemic stroke remains a debated issue [1,2-4]. Another potential reason for the current skepticism about the association of carotid web with FMD is the international binary radiological classification of FMD, which does not explicitly consider it, unlike what happened in the past.

The aim of this review is to summarize current knowledge about carotid web as a manifestation of FMD and its potential role in the pathophysiology of ischemic stroke. When addressing these aims, the neuroradiological investigation of carotid web deserves attention and may help to explain some pathophysiological mechanisms of web-related stroke.

The “carotid web” is a shelf-like lesion along the posterior wall of the internal carotid artery (ICA) bulb and an under-recognized cause of stroke in the young. It has been known by several names in the past literature: web-like formation, diaphragm, web-like septum, septal FMD, thrombotic carotid megabulb, spur, carotid bulb atypical FMD, shelf, pseudo-valvular fold, etc.

The carotid web is a thin, membranous layer of proliferative intimal tissue that originates from the arterial wall and extends into the vessel lumen. It is regarded as a subtype of intimal FMD based on pathological classifications [4-6]. Reports of carotid web have predominantly appeared in case studies and retrospective series focusing on patients who have experienced first-time or recurrent ischemic strokes and transient ischemic attacks (TIAs). While its incidence is thought to be low, it is also likely underdiagnosed in clinical practice due to its subtle presentation. The role of the carotid web in increasing ischemic stroke risk remains an active area of investigation, and its optimal treatment strategies are yet to be determined. The different levels of the definition come from the different levels of investigations (anatomical studies, imaging studies using different techniques, histopathological studies), which are associated with different rates of prevalence.

Connett and Lansche [7] first described a case of fibromuscular hyperplasia of the ICA in a 34-year-old woman in 1965 with recurrent transient aphasia and right hemiparesis. The arteriographic findings were the involvement of both ICAs, with left ICA occlusion preceded by irregular narrowing, and right ICA focal stenosis followed by long extracranial aneurysm. Three years later, Rainer et al. [8] described a case of positional cerebral ischemia due to involvement of the left ICA. The patient was surgically treated, and at the carotid opening through a vertical anterior arteriotomy, a firm, discrete ridge traversing the posterior wall of the origin of the ICA was observed. Analysis of the surgical sample of the artery disclosed abundant subintimal and medial fibrosis, and a diagnosis of FMD was made. In a retrospective angiographic series from a single center published in 1973, four cases of carotid “web formation” were reported as incidental findings [9]. In 1979, the description of a carotid septum with superimposed thrombus and ischemic stroke was published for the first time [10]. Comparison with reported cases of the more common “string of beads” lesion of FMD suggested to the authors that the risk of development of ischemic neurological problems is higher for this type of fibromuscular lesion, which may justify surgical correction in symptomatic patients.

Currently, the prevalence of ipsilateral carotid web in patients who underwent a supratentorial ischemic stroke is proposed to be about 1.2% [11]. Most studies suggest that it is more frequent in men than in women [11-13], but conflicting data exist. In all available studies [11-13], unilateral carotid webs are more frequent than bilateral ones. In the first reports, the number of Black or Afro- Caribbean individuals with a carotid web was greater than that of individuals of other ethnicities. However, no large-scale epidemiological research looking for prevalence of carotid web according to ethnic origin has been performed [14].

The main site of the carotid web is the posterior wall of the carotid bulb, but other more distal locations in the ICA are possible (Figure 1).

The pathogenesis of the carotid web remains unclear. It has been proposed that the carotid web is a congenital abnormality arising from genetic factors or a vascular injury with a hormonal trigger [15]. Some evidence suggests that the carotid web in young women may be associated with the use of oral contraceptives, which can induce hyperplasia of the intimal artery layer [16]. Additionally, early angiographic classifications of FMD categorized the carotid web as a pattern of atypical FMD [17]. An older theory posits that the carotid web could be a consequence of atherosclerosis [18]. Alternatively, another hypothesis suggests that the carotid web represents a developmental anomaly of the brachiocephalic system [19]. This theory is supported by pathological findings in a 42-year-old patient, where the strip-like structure at the bifurcation of the left common carotid artery was found to consist of myofibrillary elastic tissue [19]. The authors concluded that the web might result from incomplete degeneration of the third aortic arch [19]. However, this hypothesis has been proposed on a single case, and it might raise some criticism. In fact, the extracranial arteries do not arise from the longitudinal fusion of primitive arteries, and it is difficult to consider the carotid web as a sort of “aberrant fenestration.” In addition, no other arterial anomalies have been associated with carotid web in the reported cases, and it is not easy to propose the embryonic stage at which a supposed unfusion or regression of a very short arterial segment could have occurred.

Considering the carotid web within the spectrum of FMD patterns, it is possible that its pathogenesis, from the genetic and molecular side, is included in the FMD pathogenesis, where genetic risk factors were recently demonstrated in a locus of the PHACTR1 gene, which is involved in several vascular diseases [20]. One of the proposed molecular consequences is related to the fact that the phosphatase and actin-binding protein encoded by PHACTR1 regulates both actin fibers and cell motility, ultimately affecting the cellular organization in vascular smooth muscle cells. To our knowledge, no genetic association studies have been performed on patients with carotid web.

FMD is primarily characterized by beaded (multifocal) or focal lesions in medium- or small-sized arteries. However, the clinical phenotype of FMD has recently been broadened to include arterial dissection, aneurysm formation, and arterial tortuosity [1]. FMD most commonly involves the renal and extracranial carotid and vertebral arteries, although nearly all arterial beds can be affected, and multivessel involvement is frequently observed. Notably, 80%-90% of patients diagnosed with FMD are women [1-3]. Currently, FMD is predominantly identified through radiographic imaging. With the advent of percutaneous revascularization, surgical bypass procedures have become infrequent, and obtaining histological samples has become rare. The typical arteriographic characteristics include multiple sites of stenosis and dilation, often referred to as a “string of beads,” as well as tubular and focal stenoses. Medial fibroplasia is most commonly associated with the “string of beads” pattern [1,3].

The first classifications of neurovascular patterns in FMD are based on digital subtraction angiography (DSA). In particular, in 1968, Houser and Baker [21] described the typical characteristic angiographic finding of multifocal and focal FMD with circular spastic contractions and tubular stenosis, respectively, narrowing the lumen in the midportion of the ICA. They raised an interesting point, i.e., the frequent involvement of the ICA origin. Houser and colleagues [22], three years later, added a detailed description of “atypical” FMD but without mentioning the carotid web. It should be remembered, however, that at that time, the first pathology-proven case of carotid web within the umbrella of FMD had not yet been described. A few years later, Osborn and Anderson [17] systematically divided the FMD pattern of cerebrovascular involvement into three subtypes and related them to the pathological findings. In the atypical subtype, two examples were provided, i.e., an outpouching pattern and carotid web. The main features of the two forms are the non-circumferential involvement of the arterial wall and the focal involvement. Atypical FMD was the rarest subtype, accounting for 4% of patients [17].

In the following decades, the progressive decrease of DSA indication for diagnostic purposes and the simultaneous decrease of surgical treatment of FMD resulted in a substantial loss of detail. This leads to simplification in imaging classifications of FMD, based largely on non-invasive radiological imaging and neuroimaging techniques, such as computed tomography angiography (CTA) and magnetic resonance angiography (MRA). This transition between different diagnostic modalities occurred in a period immediately following the first descriptions of carotid web and gradually reduced the possibilities of histopathological confirmation as FMD, contributing to its uncertain classification. Both the French/Belgian consensus [23], published in 2012, and the American consensus [15] proposed two subtypes, multifocal and (uni)focal FMD, not considering atypical forms and carotid web. These classifications [15,23] formally maintained a link with the histopathological subtype of the disease, while in the 2019 consensus, 1 which unified the definitions, terminology, and intentions between the American and European study groups, the link with histopathology is explicitly lost. In fact, one of the main issues emerging from this consensus [1] is that the histopathological classification of FMD is no longer applicable in modern clinical practice. According with this classification, FMD can manifest in two distinct angiographic patterns: (1) focal FMD, which can affect any segment of the artery, and (2) multifocal FMD, characterized by alternating stenotic and dilated regions (often referred to as a “string of beads”), typically seen in the mid to distal sections of the artery. Two different clinical and imaging phenotypes are proposed, multifocal and focal, potentially representing different diseases [24]. Despite the poor subsequent consideration, the first descriptions of carotid web as FMD were pathologyproven [25]. Rainer et al. [8] provided a detailed description of the gross and microscopic pathology (hematoxylin and eosin/trichromic/elastic stain preparation), finding abundant subintimal and medial fibrosis and an adventitia free of fibrous or inflammatory reaction. In fact, the pathological descriptions of carotid webs are in agreement with the histopathological criteria of intimal FMD. In isolated cases, a longitudinal section showed vascular intimal hyperplasia accompanied by fibrosis and myxoid degeneration that protruded into the lumen to form a valve [5]. In a review of 21 cases of carotid web, all samples had abnormalities in the intimal artery layer [26]. This is the main reason why carotid web has been proposed as a subtype of intimal FMD, more rarely reported than the medial subtype. A histopathological series of nine patients [27] showed that the gross appearance is a significant focal thickening of the arterial wall arising from the intima and protruding into the lumen. The common histological feature of the carotid web is eccentric intimal hyperplasia, characterized by spindle cells (consistent with migrated vascular smooth muscle cells) intermingled within an abundant loose extracellular matrix. Notably, three out of nine patients exhibited a recent thrombus adherent to the carotid web, while one patient displayed a remote thrombus embedded in the intimal hyperplasia. Additionally, six out of nine patients showed focal scarcity of medial vascular smooth muscle cells, replaced by substantial mucoid extracellular matrix accumulation beneath the carotid web.

A single retrospective study, albeit with several limitations, sought to evaluate the prevalence of systemic involvement in patients with carotid web [28]. The study’s main limitation was its retrospective design and the lack of a systematic investigation for FMD. The researchers selected 66 patients (74% women) through neck CTA, with 47 out of 66 (71%) undergoing additional DSA, which evaluated 47 carotid arteries ipsilateral to the stroke and 10 contralateral carotids. Renal artery catheter-based angiography was performed on 16 patients, examining 32 arteries. Using this methodology, classic FMD changes in the ICA were identified in only 6 of the 66 patients (9%) on the ipsilateral side. Additionally, classic FMD changes were observed in just one patient across two renal arteries. The limitations of this study are evident, and the result can hardly answer conclusively the question of the relation between carotid web and FMD. Another consideration derives from the fact that the phenotypic spectrum of FMD is not yet fully explored. It is possible that the histopathological information is in itself identifying the belonging of the carotid web to the spectrum of manifestations of FMD, but it is not sufficient to completely delineate the clinical and imaging phenotype of this condition. In addition, subtle arterial changes may be missed with non-dedicated investigations, and the main target was probably the classical “string of beads” appearance, i.e., well-evident multifocal FMD, missing focal and atypical lesions.

Interestingly, a recent study involving patients with carotid and vertebral artery dissection [29] highlighted the association between spontaneous cervical artery dissection and carotid web, diagnosed using vessel-wall magnetic resonance imaging (VWMRI). In fact, the rate of carotid web was significantly different between the carotid dissection group, vertebral dissection group, and healthy controls (19 [57.6%] vs. 5 [20%] vs. 36 [21.8%], respectively; P<0.001). The odds ratios (OR) for carotid web were 5.41 (95% confidence interval [CI]: 1.634-17.973) and 4.81 (95% CI: 2.176-10.651) for patients with carotid and vertebral dissection, respectively. Furthermore, 16 out of 19 patients with carotid dissection and carotid web had occurrence of dissection in the C1 segment of the ICA, with a mean distance of 1.91± 1.71 cm from the carotid web. One of the relevant points in this study is the use of VW-MRI, which is useful in the diagnostic assessment of both dissection and carotid web, providing accurate clues of morphology and location. In addition, a subset of 10 patients also had a VW-MRI in the follow-up, confirming the unchanged persistence of the carotid web, resolving the possibility of a dissective origin of the vessel abnormality. The diagnosis of carotid web was performed by applying to MRI the same features already validated on CTA (thin intraluminal filling defect along the posterior wall of the carotid bulb confirmed in oblique sagittal and axial images) and adding a slight enhancement of the web on contrast-enhanced imaging sequence (CE CUBE T1-weighted). Notably, in a subset of patients, the carotid web is not visible on thin-slice contrast-enhanced MRA (CE-MRA) images, but only in VW-MRI images. The increased transverse wall shear stress might be a shared feature between carotid web and ICA dissection, as an effect in the former and as a risk factor in the latter [29].

An example of the association of carotid web with focal FMD is illustrated in Figure 2.

As previously highlighted, the histopathological diagnostic criteria of FMD [5,6] perfectly include the pathological appearance of carotid web into the FMD spectrum as an intimal subtype. In fact, the multifocal subtype is mainly, but non exclusively, associated with the medial involvement. The following transition from invasive to non-invasive diagnostic techniques produced a progressive simplification in the FMD classification, missing some subcategories. In fact, there are several clinical and radiological phenotypes of FMD—e.g., the multifocal and focal subtypes, which are strongly different [24]. In particular, the absence of a clear association of carotid web with other localizations of FMD [28] is fragile, referring only to multifocal subtype and having been investigated retrospectively on diagnostic studies performed for other reasons in an urgent setting. In addition, the focal FMD subtype is more complex to diagnose, and it could be missed in non-dedicated studies. Until now, there is no report derived from a systematic, prospective, preplanned screening for multisite FMD in the literature in patients with carotid web. Hopefully, the ongoing prospective registries may help to solve this gap. In the meantime, a critical appraisal of the old FMD classifications and the need to update the current one have recently been advised [30].

The prevalence of carotid web in non-selected consecutive patients with ischemic stroke has been investigated in a few studies with cohorts of different sizes [12,31-34]. The main data are summarized in Supplementary Table 1.

Mei et al. [35] investigated the prevalence of carotid webs in a large hospital-based cohort in the United States, finding 24/1,467 patients (1.6%) with carotid webs (mean age: 63 years; 62.5% female; 50% African American); among them, 12/24 patients (50%) had a history of ipsilateral stroke (8/12 cryptogenic), and 4/24 patients (16.7%) had recurrent strokes. In studies that have introduced selection criteria for some subgroups of patients with ischemic stroke—for example, etiology, age, as well as occlusive pattern [11,32,36-40]—the data are of greater interest and are illustrated in Supplementary Table 2.

A systematic review assessing the prevalence of carotid web in supratentorial embolic stroke of undetermined source (ESUS) [13] reported a total prevalence of 9.58% (95% CI: 5.62%-15.85%). The majority of patients were women (76.5%±22.3%) and of African descent (58%±39%). Compared to patients without ischemic stroke, a significant association was observed with ipsilateral carotid web, with a risk ratio of 2.74 (95% CI: 2.14-3.51), whereas no significant association was found with contralateral webs (risk ratio of 1.50, 95% CI: 0.94-2.40). In a meta-analysis by Mac Grory et al. [11] involving 3,192 patients, the pooled prevalence of ipsilateral carotid web in cryptogenic stroke patients under 60 years old was 13% (95% CI: 7%-22%; I²=66.1%). The relative risk (RR) of ipsilateral versus contralateral carotid web in all patients was 2.5 (95% CI: 1.5-4.2, P=0.0009), while in cryptogenic stroke patients under 60 years old, the RR increased to 3.0 (95% CI: 1.6-5.8, P=0.0011). Few studies have evaluated the risk of recurrent stroke in patients with a first stroke related to carotid web. One such study analyzed cohorts of large vessel occlusion (LVO) stroke from the MR CLEAN (Multicenter Randomized Clinical trial of Endovascular treatment for Acute ischemic stroke in the Netherlands) trial (2010-2014) and MR CLEAN Registry (2014-2017) [41] to assess the rate of symptomatic carotid web and the risk of recurrent stroke over two years. Only 168 patients without carotid web had an extended follow-up. In the total cohort, 30 out of 3,439 patients (0.87%) with baseline CTA had a symptomatic carotid web, while 6 out of 3,438 patients (0.17%) had an asymptomatic carotid web. Patients with symptomatic carotid web were younger (median age: 57 years [interquartile range: 46-66 years] vs. 66 years [interquartile range: 56-77 years]; P=0.01) and more frequently women (73% vs. 40%; P=0.001). Among these patients, 3 out of 30 (10%) had a previous stroke (2 of these in the same vascular territory), and 24 out of 30 (80%) had no other identifiable cause for the index stroke. The remaining 6 patients (20%) had atrial fibrillation. During the 2-year follow-up period, 5 out of 30 patients (17%) with a carotid web ipsilateral to the index stroke experienced a recurrent stroke, compared to 5 out of 168 patients (3%) without carotid web (hazard ratio: 5.0, 95% CI: 1.4-17.3; adjusted hazard ratio: 4.9, 95% CI: 1.4-18.1). Most recurrent strokes in the carotid web group (4 out of 5) were ipsilateral to the index stroke, yielding a hazard ratio of 9.9 (95% CI: 1.8-54.2; adjusted hazard ratio: 8.1, 95% CI: 1.4-46.8).

The pathophysiological mechanisms behind strokes associated with carotid webs remain unclear. It is proposed that the presence of a carotid web leads to thrombogenic hemodynamic changes downstream, which may facilitate thrombus formation and subsequent embolization, resulting in stroke [42]. This theory is corroborated by histological analysis of thrombi found in the intracranial arteries of patients with LVO strokes and carotid webs [33]. These thrombi were found to have a mixed composition, consisting of fibrin, platelets, and red blood cells (RBCs), with RBCs accounting for between 6% and 46% of the thrombus volume. Moreover, the thrombi exhibited a consistent leukocytic inflammatory response, while neither cholesterol clefts nor atheromatous features were identified.

DSA stands as the most effective imaging modality for visualizing hemodynamic changes downstream of a carotid web. Contrast pooling observed in the venous phase of DSA serves as a marker for prothrombogenic flow stasis, and can be seen distally to the web, even when the stenosis is relatively mild [31]. This phenomenon has been quantitatively assessed using computational fluid dynamics alongside regional contrast density measurements from sequential contrast injections, generating time-density curves and calculating parameters such as time to peak (TTP), relative TTP, maximal upslope, and angiographic full width at half maximum [43,44]. The intraluminal shelf-like projection at the site of the web results in accelerated flow proximally, but as the lumen widens distally, it creates a highly unstable, separated flow pattern with a large recirculation zone and potential stagnation region. This is reflected by a larger total area under the curve. The altered hemodynamics caused by the carotid web likely correlate with the contrast pooling observed in the venous phase of DSA at the level of the ICA bulb [44] (Figure 3).

In line with the hypothesized stroke mechanism in individuals with carotid webs, neuroimaging findings of ischemic lesions align with an embolic origin [45]. Symptomatic carotid webs consistently presented with cortical infarctions, supporting the proposed thromboembolic pathway. The majority of these incidents involved LVO embolic strokes that affected both cortical regions and deep structures across various vascular territories. About one-third of patients experienced infarctions in one or more cortical subdivisions, while another third had infarcts spanning two or more subdivisions. Borderzone infarctions, typically located at the junctions of major vascular territories, were observed in 10% of patients, all of whom had associated LVO and a cortical infarction pattern.

The primary imaging modalities for diagnosing carotid webs include DSA, CTA, MRI, and ultrasound (US). Imaging efforts predominantly target the carotid bulb, which is defined as the region extending 10 mm both proximally and distally from the bifurcation [45]. Initial diagnosis is often made using any available non-invasive imaging technique, though this should be confirmed with a complementary method, such as US or CTA. On US, the carotid web appears as a thin, isoechoic or hypoechoic structure arising from the vessel wall and protruding into the lumen, occasionally accompanied by peripheral atherosclerosis. It is commonly observed as a “cliff-like” stenosis when viewed longitudinally. Color Doppler imaging may reveal the presence of a flow disturbance or eddy at the junction between the carotid web and the vessel wall. Techniques such as three-dimensional (3D) US, contrast-enhanced ultrasonography, and intravascular US are less commonly utilized, and their standardized use remains limited. An example of US appearance of a carotid web is provided in Figure 4.

CTA offers several advantages, including high resolution, rapid scanning times, and the ability to perform multiplanar reconstructions. The characteristic appearance of a carotid web on CTA is a hypodense linear septum, visible in both the transverse axial and sagittal planes. In clinical practice, diagnosis of a carotid web via CTA typically requires imaging in at least these two planes. Multiplanar reconstructions enhance the ability to differentiate carotid webs from other conditions, such as carotid dissection and aneurysms [31,37,38]. The inter-rater reliability for CTA in diagnosing carotid webs has been reported to range between 0.78 and 0.93. Earlier studies have made a distinction between carotid webs and small protruding lesions, which appear less prominent on CTA compared to carotid webs. The presence of a septum at the axial CTA view is essential for confirming the diagnosis of a carotid web. If the septum is not visible on the axial images, the lesion is classified as a small protruding lesion. In a sub-study from MR-CLEAN, these small protruding lesions were found to be equally distributed between symptomatic and asymptomatic sides, suggesting that they are not linked to ischemic stroke [31]. On CTA, a carotid web is identified as an intraluminal shelf-like projection within the lumen of the carotid bifurcation, exhibiting three distinct features. Proposed diagnostic criteria for identifying a carotid web on CTA include [46]: (1) “a filling-defect“ (shelf-like protrusion); (2) “a thin line or a septum,” dividing the lumen in axial plane; and (3) “double-lumen sign,” similar to dissection.

On MRI, the carotid web typically appears as a thin, film-like structure originating from the posterior wall of the carotid artery and protruding into the lumen, usually directed upward and inward [47-50]. Using two-dimensional (2D) fast spin echo (FSE) and 2D cine FSE sequences, MRI can reveal thickening of the vessel wall associated with the carotid web, accompanied by increased signal enhancement. A distinct pulsatile expansion of the unilateral vessel wall is observed, differing from the homogeneous expansion seen in a healthy carotid artery. VW-MRI typically shows a film-like protrusion that is isointense on T1-weighted images, slightly hyperintense on fat-suppressed T2-weighted images, and demonstrates significant enhancement. Additionally, MRA can help differentiate thrombus from the carotid web itself while offering detailed insight into the morphology and dynamics of the vessel wall [47-50].

The location and morphology of the carotid web are clearly visible on 3D CE-MRA, with axial source images displaying a false bifurcation in the lumen [50]. Multi-contrast FSE images reveal vessel wall thickening and signal enhancement around the web. Both 3D CE MRA and multi-contrast FSE highlight slowed blood flow distal to the web, as well as blood pooling on one side after contrast injection [50]. CineFSE, a cardiac phase-resolved imaging technique, demonstrates atypical pulsatility of the carotid wall near the web. These imaging features strongly suggest impaired hemodynamics in the region surrounding the carotid web, which may serve as a substrate for thrombus formation [50].

VW-MRI diagnostic criteria for carotid web have also been proposed [46]: (1) “thickness” (carotid vessel wall thickness), (2) “projection” from the posterior wall, (3) ”value sign,” (4) “double lumen sign” (meaning double-lumen or multi-lumen appearance in axial plane), and (5) “contrast stasis.”

In contrast to CTA, VW-MRI not only offers valuable hemodynamic insights but also delivers comprehensive information regarding the arterial wall and lesions, facilitating differential diagnosis from atherosclerotic plaques and arterial dissection. VW-MRI provides a significant benefit in the diagnosis of arterial dissection, identifying an intimal flap in 42% of dissections versus 16% of luminal techniques (CTA, MRA, or DSA) [48]. Similarly, a study on carotid web prevalence in patients with carotid and vertebral dissection showed a three-times increase in detection rate of carotid web using VW-MRI versus MRA [29].

An example of MR-based imaging of carotid web is provided in Figures 5 and 6.

On DSA, the carotid web typically appears as a linear filling defect along the carotid artery wall or as a shelf-like filling defect located in the carotid bulb. During the late venous phase, contrast agent retention is commonly observed at the distal end of the carotid web. In comparison to atherosclerosis, the degree of stenosis associated with the carotid web is generally less pronounced on DSA [49]. Given that the carotid web is predominantly located on the posterior wall of the ICA, and considering the posterolateral positioning of the ICA, misdiagnosis can occur if DSA is performed using only two standard projections—posterior- anterior and lateral. However, since DSA is typically not the first-choice imaging modality for vascular assessments, this limitation is less significant in clinical practice. Few studies on small cohorts of patients have investigated the diagnostic concordance across techniques. A monocenter study [46] proposed the inter-rater agreement for diagnosis (based on a 256-slice CT scanner or 64-slice CT scanner): CTA (n=55), κ=0.88, P<0.0001; DSA (n=28), κ=0,86, P<0.0001; US (n=33), κ=0.553, P=0.001.

In addition, the inter-modality agreement for diagnosis was as follows: CTA and DSA, κ=0.92, P<0.0001; US and CTA, κ=0.633, P=0.001.

The diagnostic performance of various CTA and MRI features for identifying carotid webs has been analyzed, revealing complementary roles for both imaging modalities in diagnosis [46]. Key signs, such as the “double lumen sign” and “a line or septum” on CTA, along with the “valve sign” and “contrast stasis” observed on VW-MRI, are considered critical for diagnosing carotid webs. VW-MRI, in particular, offers superior diagnostic performance in visualizing the “projection” feature compared to the “filling defect” seen on CTA in patients with carotid webs [50].

Imaging studies provide valuable morphological insights into the carotid web [14]. There is considerable variability in the orientation, length, thickness, and presence or absence of a thickened base of the web. Typically, carotid webs have a shelf-like contour with a smooth, straight outline; however, when an intraluminal thrombus is superimposed on the web, it can complicate imaging. In such cases, follow-up studies may be necessary to clarify the diagnosis [40].

The primary differential diagnoses for a carotid web include arterial dissection, non-calcified atherosclerotic plaques, and intraluminal thrombus. To differentiate these conditions, multiple imaging techniques used at baseline and during follow-up may be essential, particularly for identifying superimposed thrombus. It is also important to note that carotid webs may coexist with atherosclerosis [51]. An example of ulcerated plaque is provided in Supplementary Figure 1.

Mac Grory et al. [14] identified several key challenges in the radiological detection of carotid webs (Supplementary Table 3).

In general, CTA is the primary diagnostic technique in symptomatic patients due to its extensive use in the diagnostic workup of acute stroke, while in asymptomatic patients, diagnosis or misdiagnosis is more frequently made using US techniques, which provide access for confirmatory neuroradiological diagnostics and/or for differential diagnosis, usually with CTA. CE-MRA and VW-MRI are not usually employed in an emergent setting for a time-dependent disease, but they represent useful techniques for the differential diagnosis. All the morphological characteristics of symptomatic carotid webs, proposed as markers of future stroke risk, need to be prospectively validated before being considered in a global, imaging-based risk score. Another issue is the heterogeneity of diagnostic behaviors and reading/reporting in clinical practice, claiming for an agreed and systematic approach. Most of the available studies on diagnostic techniques are retrospective and come from acute stroke management settings, which prompted DSA after CTA, so these are not dedicated studies on carotid web, but often retrospective diagnoses made by an external reader. At the moment, the most reasonable suggestion is to use at least two different techniques to confirm the diagnosis of carotid web (e.g., CTA and MRA, or US and CTA), leaving DSA for unsolved questions and implementing the use of VW-MRI for the differential diagnosis. However, DSA is the technique with the best time resolution, giving reliable information on the hemodynamics of carotid webs.

The treatment of symptomatic carotid web (i.e., the carotid web implicated as embolic source in ischemic stroke) is a matter of discussion. If several observational data proposed a high risk of recurrent stroke in medically treated patients (in particular on single antiplatelet therapy), these data have limitations, and there is no strong evidence favoring non-medical treatment (carotid endarterectomy and carotid stenting) [52-55]. In fact, the observational data come from case series, usually retrospective, and strongly depend on individual treatment choices. Conversely, carotid endarterectomy was performed in isolated cases in order to obtain a sample of the arterial lesion in cases where the diagnosis was controversial [54]. With growing attention and knowledge about the carotid web and its diagnosis, carotid stenting has emerged as the preferred treatment over medical therapy, due to its favorable safety profile [53]. With these limitations, a comparison between medically treated patients and non-medically treated patients retrieved a small sample of heterogeneous patients (138 vs. 151) [55]. The aggregate data indicate that carotid revascularization is both effective and safe for preventing recurrent ischemic stroke, whereas medical therapy alone may be less advantageous. In interventionally treated patients, no recurrent strokes were reported, while 26.8% of those treated medically experienced recurrent cerebral ischemia. However, the high event rate in the medical group over a follow-up period of 255 months is notably elevated and does not align with real-world experiences. The low rate of procedural complications in the interventional group is likely influenced by factors such as the younger age of the patients, the absence of extensive or irregular atheromatous plaques, no intraplaque hemorrhage, and the non-inflammatory nature of carotid webs. Carotid stenting is technically less complex than stenting for atherosclerotic stenosis. However, surgery offers the added advantage of providing a tissue sample for further analysis. In the systematic review of Zhang et al. [52], the follow-up was differently reported in the subgroups of treatment, ranging from 14 to 100 months in strokefree medically treated patients and from 3 to 144 months in interventionally treated patients, but it is not clear if patients with a recurrent stroke stopped the follow-up. They reported a 56% rate of recurrent stroke in medically managed patients with a median time of 12 months and no recurrent stroke in interventionally treated patients. In a retrospective series of interventionally treated patients, only this subgroup was followed up for 12 months without reporting events [54]. In a more recent systematic review [55], no recurrent ischemic events were observed over a follow-up range of 3-60 months in the interventional group. Conversely, the recurrence stroke rate was 26.8% in the medical group, over a follow-up of 2-55 months. While several caveats could be raised regarding these data, they may not be sufficient to conclusively support stenting over medical therapy in an evidence-based clinical context. Nonetheless, vascular surgery guidelines have given weight to the evidence, offering a grade IIb recommendation (level of evidence C) for stenting versus medical therapy in symptomatic carotid webs [56].

Meanwhile, several prospective observational registries are ongoing to further investigate carotid webs. One such study, the CAROWEB (CAROtid WEB) registry, is a nationwide multicenter real-life registry collecting clinical, radiological, and therapeutic data from patients presenting with symptomatic carotid webs (ClinicalTrials.gov ID: NCT04431609). Initiated in June 2019, the CAROWEB registry aims to prospectively include patients with carotid webs responsible for at least one ischemic stroke or TIA, in addition to retrospective cases dating back to January 2015 [57]. In a recent update to the registry [58,59], data from 202 patients were included. The majority of these patients were symptomatic for ischemic stroke (91.6%), with a high prevalence of intracranial LVO (71.8%) and ipsilateral chronic cerebral infarction (CCI) (47.5%). The mean age of patients was 50.8±12.2 years, with 62.9% being female and 47.5% Caucasian. Over half of the patients (57.3%) were treated with endovascular thrombectomy (EVT), though the carotid web was not identified during EVT in 30 out of 85 patients (35.3%). Secondary prevention was predominantly invasive, with 55.6% of patients undergoing procedures such as stenting (n=80) or surgery (n=30). Multivariable analysis revealed that the invasive therapeutic approach was associated with the presence of ipsilateral CCI (OR: 4.24, 95% CI: 1.27-14.2, P=0.019) and inversely associated with risk factors (OR: 0.47, 95% CI: 0.24-0.91, P=0.025) as well as admission National Institutes of Health Stroke Scale (NIHSS) score (OR: 0.93, 95% CI: 0.89-0.97, P=0.001). These findings suggest that invasive treatment is more likely in patients with a history of ipsilateral CCI, while risk factors and NIHSS scores may play a role in therapeutic decision-making. Another unresolved issue is the need for treatment for an asymptomatic, incidentally found carotid web. Some neuroradiological studies proposed imaging differences between asymptomatic and symptomatic carotid webs, suggesting that some morphological features may account for a thrombogenic pattern of carotid webs. The length, area, and volume of both symptomatic and asymptomatic carotid webs were compared on CTA images, finding that patients with a thinner carotid web were more likely to have symptoms [59]. Another proposed parameter useful to distinguish symptomatic from asymptomatic webs is the web-to-bulb ratio [60].

In a recent study [58], specific carotid web characteristics were found to correlate with the symptomatic status, if compared with contralateral asymptomatic webs (the study enrolled only patients with symptomatic carotid webs), as summarized in Table 1. These findings emphasize that web morphology plays a key role in predicting stroke risk in patients with carotid webs, but prospective studies are needed to validate these findings.

If the definition of the etiological role of carotid web in the case of ischemic stroke and the subsequent treatment is a matter of debate, the identification of an asymptomatic carotid web (e.g., incidental finding or on the contralateral ICA compared to the stroke) raises questions that are more difficult to answer definitively with the currently available information. The main issues concern the follow-up in terms of necessity, timing, and the most appropriate technique(s), as well as the indication for or against antithrombotic therapy.

Regarding the first issue, considering the carotid web as an expression of FMD, with or without other affected arteries, the monitoring and follow-up recommendations suggested for FMD by the aforementioned international consensus [1] should be applied. They suggest an annual imaging control in all major vascular sites. The best investigation technique for follow-up is not clearly defined. However, in the absence of other indications for performing CTA or CE-MRA, it seems reasonable to suggest monitoring the carotid web at neurovascular centers experienced in FMD, preferring the use of US techniques and reserving the use of a different diagnostic technique for cases where changes are identified compared to a previous control. Moreover, the presence of atherosclerotic overlap near the carotid web is possible and described in the literature, and this could also see the use of a non-invasive technique like US as the first step in follow-up [61-63]. Warning signs may include the identification of spontaneous echocontrast in the pouch of the carotid web and/or incomplete filling of the pouch with the suspicion of hyper- or hypoechoic apposition at the bottom of the pouch. The morphological features of asymptomatic carotid webs do not have a validated role in predicting the risk of stroke. Then, the conservative suggestion is to perform a regular follow-up of all patients with asymptomatic carotid web.

Regarding treatment, while morphological criteria have been proposed to differentiate symptomatic carotid webs from asymptomatic ones [59,60], there are no prospective evaluations that have associated this morphological variation, identified in cross-sectional studies, with the absence of stroke in follow-up. Therefore, unless the carotid web is incidental in a patient with a stroke, for whom the choice of medical therapy is not influenced by the presence of the carotid web, there is no support for initiating antithrombotic therapy in an asymptomatic patient for primary prevention. Nevertheless, a web-based survey performed by the Society of NeuroInterventional Surgery found that medical management with mono- or dual-antiplatelet therapy is the first-line treatment for asymptomatic carotid web for most of the respondents [64]. Future studies on antithrombotic treatment and the ongoing registries can help to overcome this gap.

Carotid web was proposed several decades ago as an atypical presentation of intimal FMD, and this claim was supported by pathological demonstration. It has not been described ab initio as associated with other subtypes of FMD, but it is not excluded that multiple subtypes can be combined in the same patient, as is known for the multifocal and focal subtypes. The transition from pathology and DSA to non-invasive neuroimaging left some information on the ground. For diagnosis, the concordance among techniques and readers needs to be assessed in prospective cohorts, and similar considerations hold true for the natural history and treatment. Carotid web is a convincing cause of stroke, and thrombi have been documented in the web pouch, but the long-term risk has not been assessed and the optimal treatment has not been determined, although in clinical practice, the proposal of carotid stenting is quite widespread. Conversely, it is not well defined how to deal with an asymptomatic carotid web, but it may be a reason for extended vascular screening for FMD, which may represent an opportunity for prevention.

Prospective data are needed, and observational registries are ongoing in the field of FMD. A more important neurovascular contribution to these registries might be useful.