Ischemia induced Inflammation in Arteriovenous Malformations.

Jan Rodemerk^1*; Marvin Darkwah Oppong^1*, M.D.; Andreas Junker², M.D.; Cornelius Deuschl³, M.D.; Michael Forsting³, M.D.; Yuan Zhu¹, PhD; Philipp Dammann¹, M.D.; Anne Uerschels¹, M.D.; Ramazan Jabbarli¹, M.D. ; Ulrich Sure¹, M.D. ; Karsten H. Wrede¹, M.D.

*These authors contributed equally

1. Department of Neurosurgery, University Hospital Essen, University Duisburg-Essen, Essen, Germany
2. Institute for Neuropathology, University Hospital Essen, University Duisburg-Essen, Essen, Germany
3. Institute of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, University Duisburg-Essen, Essen, Germany

PMID: 35901719
DOI: 10.3171/2022.4.FOCUS2210

Abstract

Background

Arteriovenous malformations’ (AVMs) pathophysiology of development, growth, and rupture is only partially understood. However, inflammation is known to play an essential role in many vascular diseases. This feasibility study aimed to investigate the expression of enzymes [COX-2 (Cyclooxygenase 2) and NLRP3 (NOD-, LRR- and pyrin domain-containing protein 3)] in the AVM nidus that are essential in their inflammatory pathways and how these influence the pathophysiology of AVMs.

Methods

The study group comprised 21 patients with partly thrombosed arteriovenous malformations. The cohort included 13 ruptured and 8 unruptured AVMs, which had all been treated microsurgically. The formaldehyde-fixed and paraffin-embedded samples were immunohistochemically stained with a monoclonal antibody against COX-2 and NLRP3 (Dako, Santa Clara, CA; Clone: CX-294; ABCAM, Cambridge, MA, ab214185). We correlated MRI and clinical data with immunohistochemistry, using the Trainable Weka Segmentation algorithm for analysis.

Results

The median AVM volume was 2240 mm³. The proportion of NLRP3 positive cells was significantly higher (26.23 % – 83.95 %), compared to COX-2 positive cells (0.25 % – 14.94 %, p < 0.0001). Ruptured AVM had no higher expression of NLRP3 (p = 0.39) or COX-2 (p = 0.44), compared to non-ruptured AVMs. Moreover, no patient characteristics could be reported that showed significant correlations to the enzyme expression.

Conclusion

The inflammatory process in AVMs seems to be mainly associated with ischemic (NLRP3) and not with mechanical (COX-2) driven inflammatory pathways. No direct associations between NLRP3 and COX-2 expression and radiological, standard histopathological, or patient characteristics were found in this cohort.

Introduction

Brain arteriovenous malformations (AVMs) are vascular deformities characterized by an arteriovenous shunt draining blood from arterial feeders via a vascular nidus into a cerebral vein.^{1, 2} AVMs are of particular concern because of the live long risk of hemorrhage. Bleeding from an AVM results in high morbidity (23% -40.5%)^{3, 4} and mortality (12%-66.7%).^5-7  Unruptured AVMs may come to clinical attention by causing other symptoms, such as epilepsy, progressive neurological deficits, headaches, or as an incidental finding in cranial imaging performed for unrelated reasons.^{8, 9} The pathophysiology behind the development of AVMs and their risk of rupture is not fully understood yet.^7,10 Inflammation is known to play a crucial role in different intra- and extracranial vascular diseases.^{11, 12} A better understanding of its role in AVMs might help identifying new cornerstones in the pathophysiology of AVMs and consequently, the development of more specific therapies. It is generally argued that inflammation participates in the growth and rupture of AVMs .¹³ Supporting data include the identification of polymorphism in interleukin-6 and interleukin-1β that have been connected to AVM growth and rupture .^{14, 15} Interleukin-1β is, in turn, part of the priming process of the pyroptosis enzyme NLRP3.¹⁶ Whereas, interleukin-6 is shown to induce the transcription of Cyclooxygenase-2 (COX-2).¹⁷ COX-2 is proven as a critical enzyme in the endovascular inflammatory process .^18-20 The biochemistry of COX-2 is interlinked with NOD-, LRR- and pyrin domain-containing protein 3 (NLRP-3) through a direct and indirect inhibition .¹⁶ NLRP-3 has been linked to autoinflammatory diseases but also more recently to an inflammatory response to cerebral ischemia .^{21, 22} COX-2 is expressed as a direct response to local physical stress in (cerebral) vessels.²³ Considering this, a study on COX-2 expression in AVMs has already been conducted.²⁴ However, only a marginal expression of COX-2 could be detected. This study now aimed to determine the expression and correlation between COX-2 and NLRP3 in the AVM nidus. We aimed to identify if the inflammatory response in the AVM nidus is more likely driven by physical or ischemic factors and how it is connected to its characteristics and clinical course.

Material and Methods

Study Cohort

All cases with brain AVM tissue samples harvested during microsurgical resection between 2015 and 2019 were eligible for this study. The study protocol was authorized by the local ethics advisory board. All patients or their relatives gave written informed consent before inclusion into the study. The study was conducted according to the principles of the Declaration of Helsinki and was Health Insurance Portability and Accountability Act compliant.

Clinical data collection

Demographic, radiographic, and clinical parameters of the patients were prospectively collected and stored in the clinical digital documentation system. Data included radiological images and reports, pathological reports, and clinical data on preexisting conditions, medications, comorbidities, and other risk factors. In addition, a structured telephone interview was performed before the initiation of the study to obtain an up-to-date clinical outcome as well information regarding persisting AVM-associated epilepsy or recurring AVM.

Immunohistochemical Staining

AVM wall sections (1 µm) were deparaffinized and rehydrated through descending alcohol series and cleaned with distilled water. Sections were then immersed in boiling citrate buffer (pH=6.0) in a 700 W Microwave oven for 15 minutes, followed by immersion in 3% hydrogen-peroxide in distilled water to block the activity of the endogenous peroxidase. The tissues were immunohistochemically (IHC) stained with the first antibody (COX-2: 1:50 dilution, DAKO, Clone: CX-294; NLRP3: 1:200 dilution, ABCAM, ab214185). After overnight incubation (>1 hour), sections were washed in PBS, incubated with the biotinylated secondary antibody (1:200 dilution, Vector Laboratories, Burlingame, CA; COX-2: BA-9200; NLRP3: BA-1000), and further supplemented with a Horseradish peroxidase/Strepadividin conjugate. Slides were then incubated with 3,3-diaminobenzidine as substrate for precisely 5 minutes to visualize positively immunostained cells. Finally, all sections were counterstained with hematoxylin. To avoid false-positive results due to the unspecific binding of the secondary antibody, a negative control was used in each staining passage. Furthermore, to ensure that the antibodies detect the specific protein, we used human adrenal parenchyma as positive control tissue for COX-2 and non-small cell lung cancer for NLRP3 (Figure 1). Additional standard histopathologic stainings, such as Hematoxylin and Eosin, Perl’s Prussian blue, and Elastica van Gieson, were collected within the clinical routine. All staining protocols are available on request.

Histological Image Analysis

Histopathologic slides were digitalized using an Aperio Brightfield slide scanner (Leica Microsystems, Wetzlar, Germany) at ×20 and ×40 magnification. For further evaluation and comparability, the same five areas were selected in the COX-2 and NLRP3 slides. using the CaseViewer software package (version 2.3 RTM, https:// www.3dhistech.com/software-downloads). With the FIJI software package semiquantitative analysis was performed (version 2.0.0-rc-69/1.52p, https://imagej.net/Fiji/Downloads) using the microscopy pixel classification machine learning tool Trainable Weka Segmentation (doi:10.1093/bioinformatics/ btx180). Every picture containing at least 16 million pixels was divided into three groups using five variables for each group to train the algorithm to differentiate between positive IHC signal, normal tissue area, and nuclei. To quantify the enzyme expression of COX-2 and NLRP3, we have calculated the coverage of IHC positive cells as a percentage correlative to the whole image area. Every single image evaluated by the algorithm was re-checked for plausibility by an experienced team member. Thus, we established a two-factor analysis and fallback level for our immunohistochemistry. The mean of the positive cell coverage for all five regions was calculated as an estimate of the enzyme expression across the entire arteriovenous malformation.

Radiological data collection and measurements

Four-vessel digital subtraction angiography (DSA) was performed using a Philips Allura angiography suite (Philips Healthcare, Best, The Netherlands) capable of 3-dimensional (3D) rotational DSA. Magnetic resonance imaging was acquired with various MR systems ranging from 1 Tesla up to 3 Tesla magnetic field strength (Siemens, GE, Philips, Toshiba). Pulse sequences additionally included diffusion weighted imaging (DWI) and time of flight (TOF) in all AVM cases. Image evaluation (histology and MRI) was performed by two experienced neurovascular surgeons (KHW, MDO) and an MD student (JR) utilizing the Horos DICOM viewer (version 3.3.5, https://horosproject.org/). To obtain the AVM nidus volume, the maximum diameters of the AVMs were measured in the multiplanar reconstruction of TOF MRI data. In the frontal and sagittal planes, the AVM dimensions were measured horizontally, while in the transversal plane, the dimension was measured vertically. The volume calculation was performed by multiplication of all three values and the factor 0.5.

Statistical Analysis

Statistics were performed using the R and R-Studio software packages (R version 3.6.0, https://cran.r-project.org/bin/windows/base/; R-Studio version 1.2.1335, https://rstu-dio.com/products/rstudio/download/) with these additional packages: GGPlot2, GGExtra, Readxl, and fBasic. For categorical variables proportions were calculated and a 2-sided t-test was used. For continuous variables, median values with interquartile ranges (IQR, between the 25. and 75. percentile) or mean values with standard deviation (SD) were reported for non-normally and normally distributed continuous data. The Spearman correlation was applied for the complete cohort. The significance level α was set to 0.05, and the CI was set to 0.95%.

Results

Study cohort

A total of 21 patients (7 male) with ruptured (n = 8) and unruptured (n = 13) AVMs were included into the study. Nine AVMs were partly embolized before microsurgical extirpation. Age ranged between 14 and 78 years (mean: 37.86, median: 35 SD: ± 17.51) at the time of surgery. There was no mortality during the study period.

In all AVMs, signs of fibrosis and degeneration of the vessel wall structure were present. Moreover, in 81%, elastic fibers were lost in the AVM nidus, and in 52%, iron deposition in the wall structures is reported. Immigration of inflammatory cells were detected in 10% of the cases. The median radiological volume of the AVMs nidus was 2240 mm³ (IQR: 924 – 8372 mm³)

AVM and its association with NLRP3

Expression of NLRP3 in the nidus samples of AVMs varied between 26.23 % and 83.95 % (mean: 55.19 %, median: 53.73%, SD: ±12.57 %, Figure 2). The size of AVMs, in relation to the nidus volume, showed no association with increased or decreased expression of NLRP3 (p = 0.33). Moreover, there was no correlation between pathological characteristics, like loss of elastic fibers (p = 0.86), iron deposition (p = 0.89) or even cellular inflammation signs (p = 0.35). Even patients’ characteristics like age (p = 0.91), body mass index (p = 0.38), hypertension (p = 0.38), ASA intake (p = 0.56), smoking (p = 0.89) or alcohol intake (p = 0.36) showed no connection. Neither radiological intervention for embolization of the AVM (p = 0.16), nor ruptured AVMs (p = 0.39) or clinical severity of the bleeding degrees according to Hunt and Hess (p = 0.33) showed increased NLRP3 expression (Tbl.1).

AVM and its association with COX-2

The COX-2 expression in AVMs ranged between 0.25 % and 14.94 % (mean: 4.71 %, median: 3.65 %, SD: ±3.57 %). Again, no association between AVM volume and increased COX-2 expression could be reported (p = 0.99, Supplementary Fig. 1).

But smoking in general (p = 0.025) and an increased number of pack-years (p = 0.038) showed correlations to higher COX-2 expression. Nevertheless, standard histopathological data, the remaining patients’ characteristics, and radiological interventions or features had no linked association to the COX-2 expression altogether (Tbl. 1).

Relationship between the expressed inflammatory enzymes in AVMs

Overall, COX-2 is expressed at only about 10% of the strength of NLR3 (p < 0.0001). For example, ruptured AVMs expressed on average only 3.76% COX-2, whereas NLRP3 was expressed on average 57.29%. Moreover, patients with diagnosed hypertony expressed a mean of 4.81% COX-2 and 62.01% of NLRP3. In all categories, we thus have an absolute difference between the COX-2 and NLRP3 expression of 50.64 percentage points. A direct correlation between increased NLRP3 and concomitant increased COX-2 expression could not be shown (p = 0.42). This expression scheme and characteristic are found in all study variables (Tbl. 1).

Discussion

This study aimed to assess the influence of inflammatory pathways and their histopathological markers (COX-2 and NLRP) in the development and rupture of AVMs. In line with an earlier publication, we identified a low but consistent expression of COX-2. As a novum, we were able to report a very strong expression of NLRP3 in the AVM nidus. This expression pattern hints to an ischemical rather than a mechanical driven inflammatory pathway in AVMs.

The biochemistry of the inflammatory enzymes COX-2 and NLRP3 are interlinked. NLRP3 shows a variety of activation and regulation factors.¹⁶ Especially in cerebral pathophysiology, ischemia and post-ischemic reorganisation have been identified as the main drivers of this enzyme. In comparison, the activation of COX-2 in the cerebral vessels is most likely due to sheer stress and inflammatory cells.²³ However, a regulatory effect of COX-2 on the protein output of the NLRP3 inflammasome has already been demonstrated.²⁵ Moreover, the catalytic product of COX-2, Prostaglandin E2, inhibits the NLRP3 inflammasome directly.²⁶ This connection leads to the assumption of an anticipatory relationship between these two enzymes.

In their study, Keränen et al. already reported a comparably low COX-2 expression in the AVM nidus.²⁴ Furthermore, they suggested that the pathophysiology of AVMs might be in line with the pathophysiology of cerebral aneurysms leading to small aneurysms in the AVM nidus. These might in turn be the driving force behind a rupture event. However, our results show a very low COX-2 expression, indicating that this enzyme might not be solely responsible for an inflammatory remodelling of the vascular wall.

In our series, the overall enzyme expression of NLRP3 was consistently strong, whereas the COX-2 expression was significantly lower. Given the known impact of inflammation on AVMs, this significantly different expression hints at the noteworthy influence of ischemia on the inflammatory process in AVMs.²⁷ This result contrasts with the findings of histopathological studies of other intracranial vascular pathologies (i.e. aneurysms)¹⁹ where mechanical stress seems to be the main driving force. Ischemia in connection with AVMs has been linked to clinical symptoms induced by these lesions, identifying the size of the nidus as the main factor causing ischemia in the adjacent brain tissue.^{28, 29} However, in our cohort, the expression of NLRP3 in the nidus itself was not correlating with its size. In addition, the consistently high expression of NLRP3 shows that the pyroptosis process in the AVM wall structure is widely activated.¹⁶ Thus, this pyroptosis marker may indicate that a constant remodelling process occurs in all AVMs. Interestingly, we could not demonstrate the anticipated relationship between COX-2 and NLRP3 expression.

Taking these aspects into account, we consider ischemia to be a greater driving force for growth and rupture than COX-2 driven inflammation, because it is strongly expressed in all AVMs regardless of their size. Most likely due to the minor sample size, we did not detect a substantial variation in both enzymes’ expression or could address increased expression to a clinical or radiological characteristic. Nevertheless, our results suggest a relevant role of ischemic rather than mechanical induced inflammatory pathways in the pathophysiology of AVMs, although a direct correlation of the intensity of this inflammation with clinical or radiographic factors was not present in our cohort. After all, the missing correlations between inflammation and rupture risk as well as inflammation and clinical characteristics in AVMs are in line with the literature.^{24, 30}

Limitations

All our samples were from surgically treated patients, which may introduce selection bias to our results that cannot be avoided. Moreover, the small sample size, which remains reasonable in the post ARUBA³¹ era, limits the statistical power of this study. Nevertheless, our results show very consistent enzyme expressions of COX-2 and NLRP3, making the results valuable despite the limitations mentioned above.

Conclusion

The expression rate of the inflammatory enzymes NLRP3 and COX-2 in AVMs showed a significant difference. This study identified a dominance of NLPR3, suggesting a mainly ischemical influence on inflammatory pathways in AVMs. The expression of inflammatory markers showed no correlation to standard histopathological stainings, or to radiological or patients’ characteristics in this cohort. The influence of inflammatory pathways on development and clinical course of AVMs remains unclear. Therefore, we strongly advocate establishing a multicenter registry of histopathological samples and corresponding radiological and clinical data to further investigate AVM pathophysiology in a larger cohort.

Acknowledgments

We thank Janine Szybowicz for the help with patient recruitment and data acquisition.

Sources of Funding

None.

Disclosure

None.

Conflict of Interest

None.

Supplemental Materials Supplementary Figure 1

References

1.            Solomon RA, Connolly ES, Jr. Arteriovenous malformations of the brain. N Engl J Med. 2017;376:1859-1866

2.            Wrede KH, Dammann P, Johst S, Monninghoff C, Schlamann M, Maderwald S, et al. Non-enhanced mr imaging of cerebral arteriovenous malformations at 7 tesla. Eur Radiol. 2016;26:829-839

3.            Fukuda K, Majumdar M, Masoud H, Nguyen T, Honarmand A, Shaibani A, et al. Multicenter assessment of morbidity associated with cerebral arteriovenous malformation hemorrhages. J Neurointerv Surg. 2017;9:664-668

4.            Choi JH, Mast H, Sciacca RR, Hartmann A, Khaw AV, Mohr JP, et al. Clinical outcome after first and recurrent hemorrhage in patients with untreated brain arteriovenous malformation. Stroke. 2006;37:1243-1247

5.            van Beijnum J, Lovelock CE, Cordonnier C, Rothwell PM, Klijn CJ, Al-Shahi Salman R, et al. Outcome after spontaneous and arteriovenous malformation-related intracerebral haemorrhage: Population-based studies. Brain. 2009;132:537-543

6.            Brown RD, Jr., Wiebers DO, Forbes G, O’Fallon WM, Piepgras DG, Marsh WR, et al. The natural history of unruptured intracranial arteriovenous malformations. J Neurosurg. 1988;68:352-357

7.            Fults D, Kelly DL, Jr. Natural history of arteriovenous malformations of the brain: A clinical study. Neurosurgery. 1984;15:658-662

8.            Starke RM, Kano H, Ding D, Lee JY, Mathieu D, Whitesell J, et al. Stereotactic radiosurgery for cerebral arteriovenous malformations: Evaluation of long-term outcomes in a multicenter cohort. J Neurosurg. 2017;126:36-44

9.            von der Brelie C, Simon M, Esche J, Schramm J, Bostrom A. Seizure outcomes in patients with surgically treated cerebral arteriovenous malformations. Neurosurgery. 2015;77:762-768

10.          Rutledge WC, Ko NU, Lawton MT, Kim H. Hemorrhage rates and risk factors in the natural history course of brain arteriovenous malformations. Transl Stroke Res. 2014;5:538-542

11.          Jarvelin P, Wright R, Pekonen H, Keranen S, Rauramaa T, Frosen J. Histopathology of brain avms part i: Microhemorrhages and changes in the nidal vessels. Acta Neurochir (Wien). 2020

12.          Chen Y, Zhu W, Bollen AW, Lawton MT, Barbaro NM, Dowd CF, et al. Evidence of inflammatory cell involvement in brain arteriovenous malformations. Neurosurgery. 2008;62:1340-1349; discussion 1349-1350

13.          Lawton MT, Rutledge WC, Kim H, Stapf C, Whitehead KJ, Li DY, et al. Brain arteriovenous malformations. Nat Rev Dis Primers. 2015;1:15008

14.          Pawlikowska L, Tran MN, Achrol AS, McCulloch CE, Ha C, Lind DL, et al. Polymorphisms in genes involved in inflammatory and angiogenic pathways and the risk of hemorrhagic presentation of brain arteriovenous malformations. Stroke. 2004;35:2294-2300

15.          Kim H, Hysi PG, Pawlikowska L, Poon A, Burchard EG, Zaroff JG, et al. Common variants in interleukin-1-beta gene are associated with intracranial hemorrhage and susceptibility to brain arteriovenous malformation. Cerebrovasc Dis. 2009;27:176-182

16.          Swanson KV, Deng M, Ting JP. The nlrp3 inflammasome: Molecular activation and regulation to therapeutics. Nat Rev Immunol. 2019;19:477-489

17.          Ramsay RG, Ciznadija D, Vanevski M, Mantamadiotis T. Transcriptional regulation of cyclo-oxygenase expression: Three pillars of control. Int J Immunopathol Pharmacol. 2003;16:59-67

18.          Simon LS. Role and regulation of cyclooxygenase-2 during inflammation. Am J Med. 1999;106:37S-42S

19.          Rodemerk J, Junker A, Chen B, Pierscianek D, Dammann P, Darkwah Oppong M, et al. Pathophysiology of intracranial aneurysms: Cox-2 expression, iron deposition in aneurysm wall, and correlation with magnetic resonance imaging. Stroke. 2020;51:2505-2513

20.          Jabbarli R, Rauschenbach L, Dinger TF, Darkwah Oppong M, Rodemerk J, Pierscianek D, et al. In the wall lies the truth: A systematic review of diagnostic markers in intracranial aneurysms. Brain Pathol. 2020;30:437-445

21.          Barrington J, Lemarchand E, Allan SM. A brain in flame; do inflammasomes and pyroptosis influence stroke pathology? Brain Pathol. 2017;27:205-212

22.          Wang H, Zhong D, Chen H, Jin J, Liu Q, Li G. Nlrp3 inflammasome activates interleukin-23/interleukin-17 axis during ischaemia-reperfusion injury in cerebral ischaemia in mice. Life Sci. 2019;227:101-113

23.          Russell-Puleri S, Dela Paz NG, Adams D, Chattopadhyay M, Cancel L, Ebong E, et al. Fluid shear stress induces upregulation of cox-2 and pgi2 release in endothelial cells via a pathway involving pecam-1, pi3k, fak, and p38. Am J Physiol Heart Circ Physiol. 2017;312:H485-H500

24.          Keranen S, Suutarinen S, Mallick R, Laakkonen JP, Guo D, Pawlikowska L, et al. Cyclo-oxygenase 2, a putative mediator of vessel remodeling, is expressed in the brain avm vessels and associates with inflammation. Acta Neurochir (Wien). 2021;163:2503-2514

25.          Hua KF, Chou JC, Ka SM, Tasi YL, Chen A, Wu SH, et al. Cyclooxygenase-2 regulates nlrp3 inflammasome-derived il-1beta production. J Cell Physiol. 2015;230:863-874

26.          Sokolowska M, Chen LY, Liu Y, Martinez-Anton A, Qi HY, Logun C, et al. Prostaglandin e2 inhibits nlrp3 inflammasome activation through ep4 receptor and intracellular cyclic amp in human macrophages. J Immunol. 2015;194:5472-5487

27.          Mouchtouris N, Jabbour PM, Starke RM, Hasan DM, Zanaty M, Theofanis T, et al. Biology of cerebral arteriovenous malformations with a focus on inflammation. J Cereb Blood Flow Metab. 2015;35:167-175

28.          Rangel-Castilla L, Russin JJ, Martinez-Del-Campo E, Soriano-Baron H, Spetzler RF, Nakaji P. Molecular and cellular biology of cerebral arteriovenous malformations: A review of current concepts and future trends in treatment. Neurosurg Focus. 2014;37:E1

29.          Chen CJ, Ding D, Derdeyn CP, Lanzino G, Friedlander RM, Southerland AM, et al. Brain arteriovenous malformations: A review of natural history, pathobiology, and interventions. Neurology. 2020;95:917-927

30.          Wright R, Jarvelin P, Pekonen H, Keranen S, Rauramaa T, Frosen J. Histopathology of brain avms part ii: Inflammation in arteriovenous malformation of the brain. Acta Neurochir (Wien). 2020

31.          Mohr JP, Parides MK, Stapf C, Moquete E, Moy CS, Overbey JR, et al. Medical management with or without interventional therapy for unruptured brain arteriovenous malformations (aruba): A multicentre, non-blinded, randomised trial. Lancet. 2014;383:614-621