Cerebral Amyloid Angiopathy‐Related Inflammation Spectrum Disorders: Introduction of a Novel Concept and Diagnostic Criteria

Cerebrovascular amyloid-β deposition within small cortical arteries and in those in the overlying subarachnoid space, known as cerebral amyloid angiopathy (CAA), often spells trouble.1, 2 Sporadic CAA is primarily a slowly progressive, non-inflammatory, age-related condition, typically known as a cause of lobar intracerebral hemorrhage in some patients and a contributor to cognitive impairment.2 The cerebrovascular amyloid deposition seems to increase the brittleness of affected vessels, and multiple small, and occasionally larger, hemorrhages can ensue. Radiologically, the consequent hemosiderin deposits appear as cortical or cortical–subcortical cerebral microbleeds on gradient-echo or susceptibility-weighted magnetic resonance imaging (MRI), forming the backbone of the Boston criteria for diagnosing CAA during life.2, 3 Over the past decade, however, both the clinical and MRI spectrum of CAA has broadened, revealing a heterogeneous group of disease phenotypes and pathophysiological variants.2 Among these, 2 subtypes stand out for their subtlety and potential to be overlooked: presentations with amyloid spells4 and CAA with inflammation.5 CAA-related transient focal neurological episodes (TFNEs), often referred to as "amyloid spells," typically occur in the context of acute convexity subarachnoid hemorrhage (cSAH).4 While their precise pathophysiology remains puzzling, with (rather elusive) mechanisms such as cortical spreading depression implicated, these episodes are widely believed to result from the rupture of a leptomeningeal arteriole severely affected by CAA in the subarachnoid space.6 In fact, a lot of the "action" in CAA appears to occur superficially. Cortical superficial siderosis, the chronic sequela of prior acute cSAH episodes on blood-sensitive MRI sequences (with a distinctive curvilinear deposition of blood-breakdown products along the gyral cortical surface, within the superficial cortical layers, the subarachnoid space, or both) is the most specific indicator of advanced, clinically significant CAA.7 It is a core feature of the Boston criteria v2.08 and also serves as the strongest independent prognostic marker for future CAA-related bleeding.9 Thus, severe leptomeningeal CAA likely drives brain bleeding in CAA, with cortical superficial siderosis reflecting multiple spontaneous, self-limiting rupture events and CAA-TFNEs representing the clinical correlate when such rupture events happen in eloquent brain regions. What ultimately triggers the rupture of CAA-laden vessels remains an unresolved question. Occasionally an autoinflammatory response is ignited against amyloid-β deposits in vessels affected by sporadic CAA, giving rise to CAA variants with distinct clinical, radiological, and neuropathological features.10, 11 This condition, often referred to as CAA-related inflammation (CAA-RI), encompasses both frank vasculitis and perivascular inflammatory changes associated with CAA.12 The clinical presentation of the typical CAA-RI syndrome is that of acute or subacute cognitive decline, personality changes, confusion, alterations of consciousness, seizures, or new and persistent headaches—symptoms that are more inflammatory in nature than hemorrhagic. Acute focal and progressive neurologic deficits are also frequently observed. On MRI, CAA-RI is typically characterized by numerous cortical and subcortical microbleeds, less commonly cortical superficial siderosis, and focal, multifocal, or diffuse subcortical white matter vasogenic edema.5 A firm diagnosis is essential, as immunosuppressive therapy is required to control the disease effectively.5 In fact, CAA-RI, remains the only potentially reversible form of CAA. Serendipitously, a paper published in this issue of the Annals of Neurology suggests a potential link between certain CAA-related TFNEs and features of CAA-RI.13 The study by Sellimi et al.,13 a retrospective case series from 2 European stroke centers, examined 6 patients with CAA who experienced unremitting TFNEs lasting at least 2 months. All patients underwent high-resolution post-contrast MRI, and 2 also had vessel wall imaging. These imaging studies were prompted by the unusually persistent nature of their symptoms, as CAA-related TFNEs typically fade away within days to weeks. Prolonged, unremitting TFNEs pose significant clinical challenges: for patients, they cause substantial discomfort, and for clinicians, they raise suspicion of occult cSAH. As a result, sequential imaging, including contrast-enhanced MRI and vessel wall imaging, is often performed to rule out alternative diagnoses. In this case series, the authors identified diverse enhancement patterns in the leptomeningeal, parenchymal, and vessel wall regions, with enhancements closely aligning with symptomatic areas in 5 patients.13 In 3 cases, TFNE symptoms and enhancement resolved, including 2 patients treated with corticosteroids, suggesting a possible inflammatory mechanism contributing to symptom persistence.13 However, one could argue that the mechanism may also involve local bleeding or micro-infarction. Of note, leptomeningeal enhancement extended beyond the locations of cSAH (observed in 3 patients) and diffusion weighted imaging (DWI) lesions (seen in 3 others),13 further supporting a primary inflammatory or vasculitic process rather than an epiphenomenon. Additionally, vessel wall and parenchymal enhancement, key observations, were not readily explained by acute bleeding or micro-infarction. More precise understanding how blood–brain barrier dysfunction interacts with inflammatory and hemorrhagic processes in CAA could provide key insights into its pathophysiology and guide targeted interventions. This study,13 like any small case series, carries inherent limitations, including a high risk of selection bias, variability in imaging protocols, and the lack of pathological confirmation in these patients. The authors acknowledge these shortcomings thoughtfully and in detail. Notwithstanding these limitations, their findings are both novel and hypothesis generating. Although not yet replicated or validated, the phenomenon of occult inflammation in ostensibly "sporadic CAA" patients aligns with observations from my clinical experience and that of other CAA specialists (personal communication).14-16 There are several ways to put these findings into perspective. If we focus on CAA-related TFNEs, this case series provides an alternative pathophysiological mechanism, raising the possibility that some CAA-TFNEs are triggered by localized inflammation (as indicated by leptomeningeal, parenchymal enhancement, etc.) that may culminate in superficial bleeding. This could explain why, in some CAA-related TFNEs (traditionally thought to be caused by spreading depolarizations or electrical activity triggered by acute cSAH), only cortical superficial siderosis occurs without acute bleeding, or why the anatomical site of acute superficial bleeding or cortical superficial siderosis does not always align with the TFNE semiology.4 The implications for sporadic CAA-related bleeding more broadly are also intriguing. After all, CAA-related TFNEs, due to their close link with superficial bleeding, represent one of the most aggressive CAA phenotypes, with a future bleeding risk exceeding 25% per year. This study touches on the central question of CAA research: why do some CAA patients develop bleeding into or around the brain, while others never do? By supporting the theory that inflammation may contribute to CAA-related bleeding, these findings open avenues for potential anti-inflammatory treatment strategies to prevent such events. There is a third, and likely more practical and clinically grounded, perspective to approach these findings that I would like to focus on: the lens of CAA-RI's expanding spectrum.1 The authors suggest that their observations may indicate an unrecognized role of inflammation in sporadic CAA, proposing a possible continuum of inflammation within sporadic CAA, with CAA-RI representing the extreme end of this spectrum. While this hypothesis is compelling—and the boundary between what is perceived as sporadic CAA (i.e., without inflammation) and CAA-RI is indeed becoming increasingly blurred—I would argue that the cases presented in this series fall within the spectrum of CAA-RI, albeit in a somewhat atypical (or less frequently recognized) form. After all, the defining characteristic of CAA-RI is the addition of an inflammatory component, to varying degrees, to sporadic CAA. The patients in this case series,13 like others increasingly encountered in CAA clinical practice, do not meet the current diagnostic criteria for CAA-RI.17 In fact, TFNEs are not included in these criteria, which highlights the essence of the problem: the current diagnostic criteria are narrow in scope, and in fact, limited in clinical utility.17 The existing criteria were developed and validated to facilitate non-invasive diagnosis by distinguishing patients with a typical CAA-RI clinical syndrome and characteristic confluent, asymmetric white matter hyperintensities (consistent with vasogenic edema) on fluid-attenuated inversion recovery (FLAIR) from those with pathologically confirmed non-inflammatory sporadic CAA.17 However, in real-world clinical settings, these criteria exhibit limited sensitivity, partly due to the stringent requirement for confluent vasogenic edema to even consider a diagnosis of CAA-RI. Notably, up to one-third of patients with biopsy-proven CAA-RI lack such typical lesions.18 It is now increasingly recognized that CAA-RI without subcortical vasogenic edema—but instead with prominent meningovascular involvement evident on MRI—may be more common than previously thought and likely underdiagnosed.18, 19 For example, a recent multicenter study in France comparing patients with CAA-RI (104 pathologically proven or probable cases) to those with biopsy-positive primary angiitis of the central nervous system (a key differential diagnosis, n = 52) demonstrated that superficial hemorrhagic and leptomeningeal MRI features—including convexity subarachnoid hemorrhage, cortical superficial siderosis, and leptomeningeal contrast enhancement—augmented diagnostic accuracy beyond the current criteria.20 A recent case series highlighted an additional meningovascular MRI feature of CAA-RI: sulcal hyperintensities,21 with or without associated gyral swelling, presumably representing early autoinflammatory-induced effusion. The study included 4 patients with CAA-RI who, similar to cases I have encountered, initially lacked the asymmetric white matter hyperintensities traditionally considered essential for diagnosis. Instead, these patients presented with varying degrees of sulcal hyperintensities and, occasionally, subtle leptomeningeal enhancement.21 Over time, 3 patients developed the characteristic asymmetric white matter hyperintensities, and 2 had biopsy-proven CAA-RI, confirming the diagnosis. Of note, sulcal hyperintensities ("effusion") and gyral swelling, are observed, and are well accepted imaging manifestations, in amyloid-related imaging abnormalities (ARIA), essentially an iatrogenic form of CAA-RI induced by anti-amyloid-β monoclonal antibody treatments in patients with Alzheimer's disease.22 This underscores the need for revising diagnostic criteria to encompass the broadened spectrum of CAA-RI clinical and radiologic presentations, including cases with atypical features and emerging imaging features.1, 18, 21 CAA-related TFNEs with evidence of inflammation might, at first glance, seem distinct from what we understand by the traditional "CAA-RI syndrome".17 However, this distinction merely reflects the narrow clinical and radiological parameters of the current criteria, which fail to capture the full spectrum of CAA-RI.1 This highlights the need to redefine and expand the understanding of the spectrum of inflammation in CAA. Recently, a pragmatic diagnostic framework has been operationalized—one that I and other colleagues have adopted in contemporary clinical practice when suspecting CAA- RI.1 This approach, summarized in Table 1, often deviates from the previously established criteria, integrating the most up-to-date imaging markers and recognizing TFNEs as a potential clinical manifestation of CAA-RI. To address this fundamental gap, I propose the unifying term "CAA-RI spectrum disorders" to encompass less typical symptoms, including unremitting CAA-related TFNEs and minimally symptomatic cases.23 This term also includes presentations lacking typical subcortical vasogenic edema but demonstrating evidence of purely meningovascular inflammation (as outlined in Table 1). The rationale for including CAA-related TFNEs presentations with evidence of inflammation under the CAA-RI spectrum disorders umbrella is their potentially treatable and reversible nature using corticosteroids. However, caution is warranted to avoid over-treatment of all CAA-TFNEs, particularly in non-specialist settings. Employing the framework in Table 1 and seeking expert opinion is strongly recommended. When to suspect a possible inflammatory element in the pathophysiology of CAA-related TFNEs? As a simple guide, in patients with multiple recurrent not remitting CAA-related TFNEs of different types and/or recurrent cSAH over a short period (eg, 1–2 months), underlying CAA-RI spectrum disorders should be considered as a potential cause. Probable CAA-Related Inflammation Spectrum Disorders Diagnosis can be made when all 5 of the following criteria are met: At least 1 of the following MRI findings: Supporting Features of CAA-Related Inflammation Spectrum Disorders Diagnosis The presence of 1 or more of the following features may further support the diagnosis: - Multiple cortical–subcortical cerebral microbleeds (especially ≥10) - Presence of cortical superficial siderosis or convexity subarachnoid hemorrhage - Colocalization of hemorrhagic lesions within areas or in the vicinity of vasogenic edema - Evidence of non-hemorrhagic sporadic CAA markers (as defined in Boston criteria v2.0) - History of previous lobar ICH and/or prior diagnosis of probable CAA per Boston criteria - Absence of parenchymal contrast enhancement (not due to acute ischemia) - Multiple small DWI hyperintense lesions - Absence of arterial abnormalities suggestive of large to medium vessel vasculitis on CTA/MRA - Apolipoprotein E (APOE) e4 (especially e4/e4) - CSF profile with decreased β-amyloid42 levels (with variable tau levels) Definite CAA-Related Inflammation A definite diagnosis can be made in the appropriate clinical context when brain biopsya shows: The role of neuroinflammation in sporadic CAA remains an area of active investigation,24 with prospective studies urgently needed to unravel its complexities. In the meantime, these preliminary observations13 lay a foundation for future research and support the introduction of the CAA-RI spectrum disorders concept. This framework not only encourages critical discussions within the field but also offers a pragmatic tool for clinical application, particularly given the need for timely diagnosis and the judicious use of immunosuppressive therapy in selected cases to reduce morbidity and mitigate the risk of further vascular damage and future bleeding. However, multidisciplinary evaluation and consultation with CAA experts are paramount to ensure appropriate management and to avoid unnecessary or potentially harmful use of immunosuppressive treatments in patients with suspected CAA-RI spectrum disorders. None. Dr A. Charidimou prepared the manuscript. None. Not applicable.