ABSTRACT
Biological behavior of Cerebral Cavernous Malformation (CCM) is still controversial without clear-cut signature for biological mechanistic explanation of lesion aggressiveness. There is plenty evidence implicating dysregulated inflammatory and immune responses in vascular malformation pathogenesis, including CCM. In the present study, we evaluated the predictive capacity of the SNPs VDRrs7975232, VDRrs731236, VDRrs11568820 as well as expanded the analysis of PTPN2rs72872125 and FCGR2Ars1801274 in relation to the aggressive behavior of CCM and its implications in biological processes. This was a single-site prospective observational cohort study with 103 patients enrolled, 42 had close follow-up visits for a period of 4 years, focused on 2 main aspects of the disease: (1) symptomatic event that composed both intracranial bleeding or epilepsy and (2) precocity of symptoms. We report a novel observation that the PTPN2rs72872125 CT and the VDRrs7975232 CC genotype were independently associated with an asymptomatic phenotype. Additionally, PTPN2rs72872125 CC genotype and serum level of GM-CSF could predict a diagnostic association with symptomatic phenotype in CCM patients, while the FCGR2Ars1801274 GG genotype could predict a symptomatic event during follow-up. The study also found a correlation between VDRrs731236 AA and VDRrs11568820 CC genotype to the time to first symptomatic event. In summary, this study provides valuable insights into the genetic markers that could potentially impact the development and advancement of CCM.
INTRODUCTION
Cerebral cavernous malformation (CCM) are vascular malformations consisting of capillary-like channels with a single layer of endothelium and no intervening nervous tissue1,2. CCM are among the most common vascular malformation of the central nervous system (CNS) affecting 0.5%–1% of the population, with magnetic resonance image (MRI) as the gold standard for the diagnosis and the lesions evaluation should include the hemosiderin-sensitive techniques as the susceptibility-weighted imaging (SWI)3,4. Intracerebral hemorrhage (ICH) and seizure, are the main clinical manifestations, followed by neurological disability without evidence of bleeding. However, many patients remain asymptomatic throughout their life and it is not known what factors may predict aggressive manifestations in individual cases4,5.
Usually, single cavernomas are detected in patients affected by sporadic forms and often associated with a developmental venous anomaly. Familial CCM is mostly associated with the occurrence of multiple lesions that might increase in number and size with aging 6. The familial form of the disease is linked to mutations on specific genes, (CCM1/KRIT1, CCM2/MGC4607 and CCM3/PDCD10), which encode distinct proteins involved in the endothelial cell junction function and in the interaction with cytoskeletal proteins. The most frequent and well-studied of the CCM genes is the CCM1/KRIT1, with mutational analyses showing a Hispanic-American ancestral haplotype7,8. CCMs are characterized by an incomplete disease penetrance, being 80% to the CCM1 form, near 100% to the CCM2 and 60% to the CCM3 form9,10. Furthermore, the discovery of PIK3CA mutations in CCM lesions provides important new insights into the genetic factors that contribute to the variable expression and progression of the disease phenotype 11.
Biological behavior of CCM is still controversial without clear-cut signature for biological mechanistic explanation of lesion aggressiveness. There is plenty evidence implicating dysregulated inflammatory and immune responses in vascular malformation pathogenesis, including CCM. Several dysregulated pathways have been confirmed in the transcriptome of CCM lesions, which are out of balance having an impact on the permeability of the blood-brain barrier and the stability of endothelial tight junctions 12. Notably, inflammatory and immune cells such as monocytes, macrophages, B and T cells, are present in human CCM lesions as well as in mouse models of CCM13–17. Recently, our group have demonstrated an adaptive immune-cellular reaction to CCM within CD20+ and CD68+ in the pericavernous tissue of an aggressive pharmaco-resistant epilepsy patient18. Moreover, the severity of the disease has been linked to genetic polymorphisms within inflammatory and immune response genes 19.
The significance of certain genetic variants in the development of CCM has been established, underscoring the need for further investigation into their specific roles. Tang et al 2017 showed allele T for SNP in TLR4 (rs10759930) gene is associated with increased CCM lesion number 20. Some studies have also identified genetic polymorphisms in genes related to oxidative stress with a significant impact on inter-individual variability in CCM disease onset and severity21. Of particular interest, in a previous study, our group demonstrated the association of FCGR2Ars1801274 GG and PTPN2rs72872125 CT genotype with a symptomatic profile of CCM patients in a smaller cohort 22.
In the present study, we evaluated the predictive capacity of the SNPs VDRrs7975232, VDRrs731236, VDRrs11568820 as well as expanded the analysis of PTPN2rs72872125 and FCGR2Ars1801274 in relation to the aggressive behavior of CCM and its implications in biological processes. We also looked for plasmatic inflammatory cytokines expressed in the patients, verifying a pattern of heterogeneity of plasmatic expression and any correlation with the genetic variation identified with different clinical phenotypes of CCM. Using a multi-step Bayesian approach, we thought to build a biomarker that could predict a diagnostic and a prognostic aggressive phenotype of CCM.
METHODS
Study Design and Population
This was a single-site prospective observational cohort study. Qualified researcher on human subjects conducted this study having been approved by the National Council for Ethics in Research (CAAE 69409617.9.0000.5258). Blood samples were collected between October 2017 to March 2023 and informed consent was obtained. Strobe criteria were used to report the findings of this study 63.
The epidemiological data were defined at each clinical assessment. Patients were stratified as multiform when they harbor multiples CCM on SWI or Gecho or as isolated form when they had a single lesion at the CNS. Both definitions were based on MRI study with SWI or Gecho and the absence of venous development anomalies in the multiple cases in proximity.
CCM clinical presentation were identified as symptomatic (hemorrhage or seizure) and asymptomatic. Intracerebral hemorrhage was defined based on the consensus of Cavernous Angiomas with Symptomatic Hemorrhage (CASH) 50. The solely epileptic patients were defined as not having achieved previous hemorrhagic consensus and whom the MRI evaluation depicted no hint of hemosiderin beyond the usual smooth and regular deposit around a CCM lesion.
In this study, we focused in 2 mains aspects of the disease: (1) symptomatic event that composed both intracranial bleeding or epilepsy and (2) precocity of symptoms. We considered that these presentations may reflect an aggressive phenotype of the disease. We also designed six sub-group analysis in order to check for correlation - 60 years or older, female, familiar form, pure epileptic without signs of lesion bleeding, pure symptomatic bleeding and infratentorial lesion. Among the 103 consecutive cases, 42 had closely follow-up visit for a period of 4 years. The remaining patients either did not have a clinical follow-up visit or underwent surgical resection.
Sample collection and processing
Blood samples were collected with EDTA and processed by centrifugation at 720 × g and 4 °C for 5 min to separate the plasma. The plasma supernatants were immediately stored at -80°C until further processing. The Genomic DNA (gDNA) extraction was performed using the PureLink Genomic DNA Mini Kit according to the manufacturer’s recommendations (ThermoFisher Scientific, Waltham, Massachusetts, USA). The quality of gDNA was determined by NanoDrop 2000 (ThermoFisher Scientific) followed by quantification using the Qubit dsDNA HS Assay Kit (ThermoFisher Scientific) and Qubit Fluorometer3.0 (Thermo Fisher Scientific).
Plasma isolation and Inflammatory Modulators Assessment
Proinflammatory cytokine and chemokine levels in plasma were measured using a multiparametric immunoassay based on XMap-labeled magnetic microbeads (Luminex Corp – Austin, TX, USA). A human ProcartaPlex ™ Panel (Invitrogen – Waltham, MA, USA) was used to analyze a set of 18 cytokines and chemokines (IL-1β, IL-2, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12p70, IL-13, IL-17A, IL-18, IL-21, IL-22, IL-23, IL-27, GMCSF and INF-γ). The samples were measured according to the manufacturer’s instructions and simultaneously to avoid potential batch, as described previously, using a BioPlex MAGPIX system (Biorad -Hercules, CA, USA) 64,65. Cyto/chemokine expression was measured in duplicate, and the levels of inflammatory modulators were analyzed using Xponent v. 3.0 software (Luminexcorp) and expressed in pg/ml.
Analysis of polymorphism of PTPN2rs72872125, VDRrs7975232, VDRrs731236, VDRrs11568820 and FCGR2Ars1801274 genes
The SNPs investigated in this study are detailed in Table 1. The SNPs were genotyped by allelic discrimination performed in QuantStudio™ 3 Real-Time PCR System (ThermoFisher) using TaqMan SNP genotyping assays (ThermoFisher). The characteristics of the PTPN2rs72872125, VDRrs7975232, VDRrs731236, VDRrs11568820 and FCGR2Ars1801274 genes were obtained from the SNP bank of the National Center of Biotechnology Information - NCBI (http://www.ncbi.nlm.nih.gov/). The probes for each SNP were produced by Applied Biosystems ™ rs72872125 (C 98019281_10), rs7975232 (C 28977635_10), rs731236 (C 2404008_10), rs11568820 (C 2880808_10) and rs1801274 (9077561_20). As the assay was designed and standardized to work with the same thermal cycles, a single protocol was applied for both genes. Briefly, PCR was performed with a 25 uL reaction mixture containing 10 ng DNA, TaqMan®_Universal PCR Master Mix (1X), Probe TaqMan® _Gene Expression Assay (1X), and DNAse free water for the final volume. The Real Time PCR conditions were: initially 60°C for 30 seconds and then 95°C for 10 min, and subsequently 40 cycles of amplification (95 °C for 15 seconds and 60°C for 1 min), and then 60°C for 30 seconds. The five selected SNPs were amplified on separate plates.
In silico Structural Prediction of FCGR2Ars1801274 Protein and Phylogenetic Analysis
Since the FCGR2Ars1801274 SNP is the only with amino acid change, we performed a structural analysis using bioinformatic tools. FCGR2Ars1801274 transcript structure was predict by directly modifications of the wild-type FCGR2A using FASTA sequence obtained on NCBI database 66. The new FASTA file was submitted to the Chimera software (developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from NIH P41-GM103311) for overall structural analysis 67.
The evolutionary history was inferred using the MEGA-X software using the maximum likelihood method and Jones-Taylor-Thornton (JTT) matrix-based model. We first gathered the FCGR2A wild-type sequence from Homo sapiens and 19 other mammals to study the evolutionary conservation of this protein. Structural alignment was performed using the muscle algorithm that allows multiple sequencing alignment with high accuracy and high throughput 68.
Statistical Analysis
Statistical analyses were performed with STATA 13 (StataCorp LP, TX, USA) and GraphPad Prism 9.0.0 MacOS (GraphPad Software, San Diego, California USA, www.graphpad.com). Categorical variables were expressed as n (%) and continuous data were given as mean and standard deviation (SD). A 2-tailed 2-sample t test or Mann-Whitney test was applied for continuous variables while the association between two categorical variables was measured by Pearson Chi-square (χ2) test when appropriate.
Allelic and genotypic frequencies were calculated for patient and control subjects via direct gene counting. Genotypic distributions in Hardy–Weinberg equilibrium (HWE) were evaluated by two-tailed χ2-test. Linkage disequilibrium (LD) were analyzed using: Linkage Disequilibrium Calculator (https://grch37.ensembl.org/Homo_sapiens/Tools/LD), LDlink on https://ldlink.nih.gov/ and Heatmap was plotted by https://www.bioinformatics.com.cn/en.The degree of LD between SNPs is represented by R2 and D’.
Genetic polymorphisms were evaluated through a univariate and multivariable binary logistic regression adjusting for sex, age and familiar form when adequate in matter of phenotype (symptomatic and asymptomatic). Plasma cytokines were also evaluated between genotypes and phenotypes (symptomatic and asymptomatic). Values over 2 times de standard deviation were excluded from analysis due to possible bias results.
Genetic polymorphism and plasma cytokines verified as statistically different between groups (both diagnostic and prognostic) were combined through a canonical discriminant function analysis in order to build a biomarker-model of symptomatic profile. All the possible combinations showing significant association in symptomatic individuals were build. Receiver Operating Characteristic (ROC) curves were generated along with a computed area under the curve (AUC) for each combination individually and ROC curves were compared to identify if any of them were statically superior then the others. The best model-biomarker to differentiate symptomatic and asymptomatic patients in both diagnostic and prognostic scenario was selected according to the Akaike Information Criteria (AIC), representing the best fit parsimonious model to the data with the fewest number of predictors. The optimal cutoff point was generated from ROC curves utilizing the Youden index method. Mann-Whitney test was then used to verify the difference in values generated using the modeled equation among symptomatic and asymptomatic patients and a logistic regression was used to control for age and sex. A p value <0.05 was considered significant.
To balance baseline covariates between patients with and without symptoms during follow-up we used a propensity score matching (PSM) strategy. The strategy involved 1:1 pairing and nearest-neighbor methods. After PSM, the distribution of gender, familial form, and age was balanced between the groups and the biomarker analysis was remade.
Finally, A Kaplan-Meyer survivor analysis was carried out among patients who presented different polymorphism combinations. Hazard ratios were calculated with 95% confidence intervals through a log-rank test. Failure events were determined according to if there had been symptomatic presentation at some point in the patient’s lifetime
Results
Demographic and CCM lesions characteristics
Out of 103 CCM patients enrolled in the study, 44 are multifocal/familial CCMs, 70 patients presented with symptomatic phenotype, which 48 presented a symptomatic hemorrhage secondary to their CCM, and 22 presented with seizures and no history or signs of CCM bleeding on MRI. The median age of enrolment was 45.6, within a mean age of 41.6 in the symptomatic group and 53.7 in epileptic sub-group. There were no significant differences in terms of sex, age and form (sporadic vs familiar/multifocal) between the outcome groups (Table 2).
Association of PTPN2rs72872125 and VDRrs7975232 to a Symptomatic Phenotype and Plasmatic Cytokine Levels
The distribution of the genotypic frequencies of the PTPN2rs72872125, VDRrs7975232, VDRrs731236, VDRrs11568820 and FCGR2Ars1801274 in symptomatic and asymptomatic patients are shown in Table 3. All genotypic distributions were in Hardy-Weinberg equilibrium. Haplotype analysis indicated linkage disequilibrium between VDRrs7975232 and VDRrs731236 (D’>0.99), corroborating what has already been found in other populations (Figure 1A and Supplementary material). While the other combinations segregate independently VDRrs11568820 and VDRrs731236 (D’>0.04), VDRrs11568820 and VDR rs7975232 (D’>0.317) (Figure 1A and Supplementary material). A higher frequency of the PTPN2rs72872125 CT genotype (OR 0.34, 95% CI 0.11-0.99, p = 0.04) and the VDRrs7975232 CC genotype (OR 0.06, 95% CI 0.006-0.612, p = 0.017) was observed in asymptomatic phenotype in the age, familiar and female adjusted multivariable analysis, when compared with symptomatic patients (Table 3). The ROC analysis (Figure 1B) revealed that patients with PTPN2rs72872125 CT genotype had a modest Area Under de Curve (AUC 0.420, SE 0.05 CI 95% 0.334-0.506) and VDRrs7975232 CC individuals (Figure 1C) had a pour accuracy (AUC 0.439, SE 0.03 CI 95% 0.370-0.508).
Association of PTPN2 and VDR to a Symptomatic Phenotype, Linkage Disequilibrium Map and Plasmatic Cytokine Expression. (A) Haplotype map showing the 3 SNPs analyzed and the range of R2 value, suggesting that VDRrs7975232 and VDRrs731236 are in linkage disequilibrium (D’>0.99), while the other combinations segregate independently VDRrs11568820 and VDRrs731236 (D’>0.04), VDRrs11568820 and VDR rs7975232 (D’>0.317). (B) ROC analysis evidencing the performance of PTPN2rs72872125 CT genotype (AUC 0.420) and (C) VDRrs7975232 CC (AUC 0.439) to distinguish symptomatic patients. (D) Plasmatic cytokine expression between PTPN2rs72872125 groups showing the PTPN2rs72872125 CT genotype have higher plasma level of IL-10 (p = 0.0146), (E) low levels of IL-18 (p = 0.0450) and (F) low levels of IFN-γ (p = 0.0310) (Figure 1F), while individuals with at least one VDRrs7975232 C allele had (G) low plasma levels of IL-27 (p = 0.0055) and (H) low plasma levels of IL-23 (p =0.0034).
We also hypothesized that these genetic variants might influence the plasmatic inflammation signature of the CCM patients. We found that patients who harbor the PTPN2rs72872125 CT genotype showed a higher plasma level of IL-10 (p = 0.0146) (Figure 1D) and low levels of IL-18 (p = 0.0450) (Figure 1E) and IFN-γ (p = 0.0310) (Figure 1F), while individuals with VDRrs7975232 at least one C allele had low plasma levels of IL-27 (p = 0.0055) and IL-23 (p =0.0034) (Figure 1G-H). For other SNPs no significant difference in the levels of immune markers was observed. We did not observe any significant differences in the levels of these plasma molecules in relation to patient sex and form (sporadic/solitary or familial/multifocal).
Performance of Diagnostic Biomarker
In order to build a diagnostic biomarker of CCM activity that could improve our previously published weighted biomarker formula, we tested new combinations of genetic and cytokines factors. First, the previously published diagnostic weighted biomarker was again confirmed to distinguish symptomatic and asymptomatic patients with 14.2% sensitivity and 97.4% specificity (AUC 0.641 SE 0.06 CI 95% 0.512 – 0.770, p =0.03), worse than the range reported previously. A similar canonical discriminant analysis approach was then implemented to determine if a weighted combination of different SNPs and cytokines could improve the diagnostic association with symptomatic phenotype. The best weighted-biomarker included the PTPN2rs72872125 CC genotype and GMCSF plasma levels as formulated -0,89*(GMCSF)+0,41*(PTPN2rs72872125CC) (AUC 0.663, SE 0.06 CI 95% 0.534 – 0.792, p = 0.01) with a specificity and sensitivity 85.7% and 41.3%, respectively (Figure 2A). This formula outperformed all the other possible formula (Table 1 Supplement) that had a statistically significant value. The median weighted combination value was 2.65-times increased (p =0.01) in symptomatic patients (median estimated value -0.36) compared with asymptomatic individuals (median estimated value -0.97) (Figure 2B). The formula performed fairly in the sub-group analysis of hemorrhagic patients (AUC 0.661, SE 0.07, CI 95% 0.517 – 0.806) but had a good accuracy in familiar form of the disease (AUC 0.774, SE 0.09, CI 95% 0.595 – 0.953, p = 0.01) and patients with infratentorial lesions (AUC 0.804, SE 0.08, CI 95% 0.642 – 0.966, p = 0.001) (Figure 2 C-D). Female symptomatic patients, pure epileptic patients and elderly patients were not statically significant.
Performance of Diagnostic Biomarker. (A) ROC curve of the best weighted-biomarker included PTPN2rs72872125 CC genotype and GMCSF plasma levels as formulated -0,89*(GMCSF)+0,41*(PTPN2rs72872125CC) (AUC 0.663, SE 0.06 CI 95% 0.534 – 0.792, p = 0.01) with a specificity and sensitivity 85.7% and 41.3%, respectively. (B) The median weighted combination value was 2.65-times increased (p =0.01) in symptomatic patients (median estimated value -0.36) compared with asymptomatic individuals (median estimated value -0.97). Sub-group analysis demonstrating that the formula performed fairly in the sub-group analysis of (C) symptomatic bleeding patients (AUC 0.661, SE 0.07, CI 95% 0.517 – 0.806) but had a good accuracy in (D) familiar form of the disease (AUC 0.774, SE 0.09, CI 95% 0.595 – 0.953, p = 0.01) and (E) patients with infratentorial lesions (AUC 0.804, SE 0.08, CI 95% 0.642 – 0.966, p = 0.001).
Validation Cohort with Symptomatic Event during Follow-up: Building a Prognostic Biomarker
We then tested if the newly weighted-diagnostic biomarker could act as a prognostic biomarker in a sub-group analysis of 42 patients that were prospectively followed-up after initial blood collection. 4 patients experienced a symptomatic event (1 lesion growth, 2 symptomatic hemorrhage and 1 newly onset epilepsy without signs of bleeding) while the others 38 stayed asymptomatic during the next 4 years. The weighted-biomarker had a fair accuracy (AUC 0.644, SE 0.078, CI 95% 0.491-0.976) in distinguishing these patients (Figure 3A). In order to verify if other combination of genetic variants and cytokines could act as a prognostic biomarker, we first tested the 5 SNPs in the sub-group of 42 patients that were closely followed-up. We found an independent statistically significant association of the FCGR2Ars1801274 GG genotype (OR 16.10, 95% CI 1.32 – 195.52, p = 0.029) in the age and familiar adjusted multivariable analysis. This variant had an excellent accuracy (AUC 0.796, SE 0.12, CI 95% 0.631-0.897) in distinguishing these individuals (Figure 3B). We further tested the same gene in a propensity matched subgroup. The presence of the FCGR2Ars1801274 GG genotype could predict a symptomatic event with high accuracy (AUC 0.875, SE 0.12, CI 95% 0.473-0.996) (Figure 3C). We also tested if this mutation would maintain its accuracy power when comparing symptomatic patients during follow-up and individuals who have never experienced any symptoms. We also found that FCGR2Ars1801274 GG genotype could predict a symptomatic event with high accuracy (AUC 0.829, SE 0.14, CI 95% 0.519 - 0.956) (Figure 3D).
Performance of Prognostic Biomarker and In Silico Structural Analysis of Transcript FCGR2Ars180127. (A) The weighted-biomarker had a fair accuracy (AUC 0.644, SE 0.078, CI 95% 0.491-0.976) as prognostic biomarker while FCGR2Ars1801274 GG (B) genotype had an excellent accuracy (AUC 0.796, SE 0.12, CI 95% 0.631-0.897) in distinguishing these patients. (C) ROC curve evidencing the performance of the FCGR2Ars1801274 GG genotype after PSM analysis (AUC 0.875) and (D) in a sub-group of pure asymptomatic patients (AUC 0.829). (E) In silico structural analysis prediction reveals that the FCGR2Ars1801274 leads to a substitution of a histidine for an arginine at position 131 leading to a predicted conformational structure slightly different than the wild-type. (F) Structural alignment predicts that the mutation c.500A>G (p.His131Arg) occurred in domain conserved in higher primates, within 30% of the species sharing this site.
In Silico Structural Analysis of Transcript FCGR2Ars1801274
In silico structural analysis prediction reveals that the FCGR2Ars1801274 leads to a substitution of a histidine for an arginine at position 131 (Figure 3E). The structural alignment predicts that the variation c.500A>G (p.His131Arg) occurred in domain that appears to be conserved mainly in higher primates, within 30% of the species sharing this site (Figure 3F).
VDR variants relates to Precocity of symptoms
We tested if any of the VDR genetic variants could relates to a precocity of symptoms in the CCM cohort. Patients with the VDRrs731236 AA genotype tend to present their symptoms late in their life compared to those who did not have (HR 0.52, SE 0.14, CI 95% 0.299-0.909 p = 0.020) (Figure 4A). Also, individuals that harbor the VDRrs11568820 CC genotype tend to have symptoms earlier (HR 1.58, SE 0.46, CI 95% 1.092-2.282 p = 0.016 (Figure 3B).
(A) VDRrs731236 AA genotype tend to have symptoms later in life compared to those who did not have (HR 0.52, p = 0.020). (B) Individuals that harbor the VDRrs11568820 CC genotype tend to have symptoms earlier (HR 1.58, p = 0.016).
Discussion
Neuroinflammation is increasingly a focus of research in symptomatic events in CCM patients 23. In the present study, we evaluated a large and phenotypically well characterized CCM cohort and provided some key information about the genetic influence on the behavior of this unique neurovascular disease. Here, we have found an individual higher frequency of PTPN2rs72872125 CT and the VDRrs7975232 CC genotypes in asymptomatic patients, associated with changes in cytokine levels, suggesting a possible protective role. Our main new observation is that the combination of a balanced formula using the PTPN2rs72872125 CC genotype and serum level of GM-CSF could predict a diagnostic association with symptomatic phenotype in CCM patients, while the FCGR2Ars1801274 GG genotype showed the best accuracy in predicting a symptomatic event in the next years, possibly functioning as prognostic genetic biomarker. In previous studies, we reported the association of this same variants with an aggressive phenotype of cerebral cavernous malformation but in a smaller sample22. Furthermore, another exciting discovery is that individuals with the VDRrs731236 AA and VDRrs11568820 CC genotypes may experience an earlier onset of symptoms in the CCM cohort.
To date, there have been few published studies on SNPs in the PTPN2 gene, and among them, a weak association with certain diseases or different outcomes was found. However, it is important to highlight that in combination with other variants they have been shown to possibly increase the susceptibility of chronic inflammatory disorders, including rheumatoid arthritis, type 1 diabetes and celiac disease24–28. Furthermore, there have been research findings indicating the presence of epistasis between PTPN2 and VDR gene 29,30. Interestingly, patients who inherit the PTPN2rs72872125 CT genotype have high levels of IL-10 when compared with CC genotype (Figure 1). IL-10 is an anti-inflammatory molecule, with a well-established function in restraining and regulating both acute and chronic inflammatory processes 3132. Lyne et al. (2019) demonstrated that IL-10 is an important molecule present in diagnostic CASH biomarker and the levels were decreased in cases of individuals who had experienced symptomatic CCM hemorrhage in the prior year 33. Furthermore, the CT genotype was associated with low levels of IL-18 and INF-γ (Figure 1), which are increased in patients with epilepsy and CCM hemorrhagic phenotype 34–36. The Canonical values derived from the best weighted formula combination with GM-CSF were 2,65X higher in patients who suffered a subsequent symptomatic event (Figure 2). Like this, the a weighted combination of PTPN2rs72872125 SNP and GM-CSF levels is a potential diagnostic genetic biomarker for the symptomatic phenotype in CCM patients.
We had also provided more evidence that the FCGR2Ars1801274 GG genotype could act as a prognostic biomarker of the disease. The FCGR2A gene is located on chromosome 1q23 and consists of seven exons, encoding a member of a family of Fcγ receptors for immunoglobulin G (IgG). Through it is expression in immune system cells such as macrophages, dendritic cells and neutrophils, it is possible to link cellular and humoral immunity37,38. The FCGR2A rs1801274 variant leads to a substitution of a histidine (A allele) for an arginine (G allele) at position 131, also known as H131R. This polymorphism is capable to increase the binding affinity of FCGR2A to IgG, resulting in activation of the FCGR2A signaling pathway and upregulation of IgG2-dependent phagocytosis 39,40. The GG genotype has been associated with several autoimmune diseases, such as Systemic Lupus Erythematosus, Type 1 Diabetes Mellitus, Crohn’s Disease and others, based on its relationship with the release and stimulation of responsive inflammatory processes41. Protein structure prediction using bioinformatic tools is an important approach to understand the FCGR2A variants. In this study we also performed an in-silico analysis that demonstrated that the H131R position is strongly conserved among higher primates, which may be related to its clinical importance.
Lyne et al 2019 showed an upregulation of FCGR2B gene in CASH transcriptomic, shedding light on the importance of Fcγ receptors in CCM disease33. FcγRII receptors mediates the C-reactive protein (CRP)-induced changes in endothelial function and inflammatory response 42,43. A higher binding avidity of CRP to FcγRIIa on immune cells was identified for allotype FcγRIIa-R131 compared to other genotypes 44. The variant homozygous genotype (GG) was able to increase the expression of ICAM-1 and E-selectin in HUVEC (Human Umbilical Vein Endothelial Cells) and the levels of tPA, MCP-1, and IL-6 secreted45. In addition, the G allele resulted in a significant defect in endothelium dependent vasodilatation and reduced NO activity during endothelial cell stimulation in patients with hypercholesterolaemia, corroborating with what has already been found in models of CCM 46,47. This data could potentially impact our comprehension of the pathophysiology of CCM disease as well as consequences with regard to the interpretation of prognostic biomarker for fallow-up of patients.
Growing evidence suggests that vitamin D signaling role in cavernous malformation behavior 48. Peripheral plasma vitamin D has been shown to reflect the severity of CCM disease 49,50. Indeed, cholecalciferol (vitamin D3), was shown to decrease CCM lesion burden in a murine model of CCM, and to inhibit ROCK activity, known to affect CCM development22,47. The effects of vitamin D on the immune system are accomplished by binding to the nuclear Vitamin D Receptor (VDR). Some single nucleotide polymorphisms (SNPs) in genes involved in vitamin D singling were reported to have association with vitamin D deficiency 51. The VDRrs731236 also known as TaqI, is a synonymous variant, while VDRrs7975232 (ApaI) and VDRrs11568820 are located in the 3′ and 5′-untranslated regions of the gene respectively. These SNPs do not alter the amino acid sequence in the VDR protein, but they can exert influence on mRNA stability and gene transcription 52. Especially in relation to VDRrs7975232, the literature does not make clear its role in relation to vitamin D levels, on the other hand we showed that patients who carried the CC genotype have decreased levels of proinflammatory cytokines (Figure 1). Corroborating with a possible protective role of the VDRrs7975232CC genotype, Jiang et al (2015) showed that A allele slightly increased the risk of temporal lobe epilepsy in children 53.
Age at the first symptom is also a fundamental information during the counselling of CCM patients. Distinguishing individuals with higher chances of precocity of symptoms that could influence their productive life is of great importance to the CCM community. Our study provided the correlation between VDRrs731236 AA and VDRrs11568820 CC genotypes to the time to first symptomatic event (Figure 4). However, we could not demonstrate a straight correlation between the VDRrs11568820 CC and VDRrs731236 AA genotype presence and the blood concentration of any of the analyzed cytokines. The relationship between VDR polymorphisms and susceptibility to autoimmunity diseases have been conducted in different settings, while the results obtained so far are conflicting54–56. We hypothesized that VDR polymorphism may affect the way proinflammatory cells react to the CCM in an autoimmunity fashion leading to molecular level disruption. To the best of our knowledge, this is the first study to investigate the associations between VDR polymorphisms and cerebral cavernous malformations.
Previously reports demonstrated that CCM1/KRIT1 regulates vascular permeability through interaction with CCM2/MGC4607 to stabilize endothelial cell–cell junctions, and together they suppress RhoA activity and, thus, activation of the RhoA effector ROCK57. Mutation on these genes lead to augment of actin stress fiber formation and increased permeability of CCMs58–60. Moreli et al (2007) have shown that BXL-628, a VDR agonist, prevented RhoA activation and inhibits RhoA/Rho Kinase signaling in rat and human bladder61. Since vitamin D is a natural agonist of VDR and animal models have suggested that fasudil and high doses of atorvastatin promotes significant inhibition of CCM lesional development and hemorrhage through the RhoA/Rho Kinase rational, it is reasonable to suppose that the prognostic ability of low levels of serum vitamin D is related to these VD/VDR interactions and its genetic variants.
Our study pioneers a novel classification denoted as the “aggressive behavior” paradigm within cerebral cavernous malformations (CCM), incorporating both hemorrhage and epilepsy as integral components of its symptomatic criteria. This redefined paradigm marks a transformative departure, acknowledging the critical role of epilepsy alongside hemorrhage in defining the clinical spectrum of CCM. Unlike a diagnostic biomarker, which detects the presence of a specific medical condition, a prognostic biomarker assesses the likelihood of progression or stability of medical conditions 62. These biomarkers contribute to a more precise and expedited diagnosis, as well as improved patient follow-up. In summary, our study provides new pieces of evidence for possible genetic biomarkers that may influence the behavior of cerebral cavernous malformations. We report a novel observation that the PTPN2rs72872125 CT and the VDRrs7975232 CC genotypes were independently associated with an asymptomatic phenotype. Additionally, PTPN2rs72872125 CC genotype and serum level of GM-CSF could predict a diagnostic association with symptomatic phenotype in CCM patients, while the FCGR2Ars1801274 GG genotype could predict a symptomatic event during follow-up. The study also found a correlation between VDRrs731236 AA and VDRrs11568820 CC genotype to the time to first symptomatic event. Overall, this study provided valuable information on the genetic factors that may influence the development and progression of CCM. Moreover, beyond the conventional genetic biomarker purview, our investigation postulates a novel etiological framework. We posit an inflammatory cascade precipitating a self-non-self-interaction, instigating an autoimmune response against CCM. This theoretical construct delineates the changes of CCM from a benign anatomical variant—traditionally construed as a common disease—into an aggressive lesion, now occupying the realm of a rare affliction—the aggressive behavior of CCM. This theoretical construct challenges extant paradigms, accentuating the potential role of immune-mediated mechanisms in the clinical trajectory of CCM, thereby offering prospects for targeted immunomodulatory interventions. In essence, this study augments our comprehension of genetic determinants influencing the advancement of CCM while advancing a transformative hypothesis concerning the disease’s underlying inflammatory dynamics. These insights herald a paradigm shift in conceptualizing the clinical spectrum of CCM, fostering an enriched landscape for future investigative pursuits and potential therapeutic interventions aimed at modulating the intricate inflammatory cascades underpinning the aggressive behavior of CCM.
Data Availability
All data produced are available online at
Supplemental material
Table Supplementary 1 (S1)
Sources of Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Brazilian National Council for Scientific and Technological Development (CNPq Number 440779/2016-2), Senator Romario Faria parliamentary amendment (no. 37990007 EIND), financial support of Coordination for the Improvement of Higher Education Personnel (CAPES Number 88887.130752/2016-00), FAPERJ E-26/210.657/2021, E-26/210.273/2018 and E-26/201.040/2021 and Chamada Pública MCTI/FINEP/CT-INFRA-PROINFRA 02/2014 – Equipamentos Multiusuários – Ref. n° 0097/2016. Thanks to Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) and Casa Hunter for the financial help to this Project.
Disclosures
The authors declare no competing interests
Acknowledgements
The authors thank the Cavernoma Alliance Brazil Research Institute – Aliança Cavernoma Brasil for the logistic assistance.
References
