We recently described the CCSVI/VM phenotype associated with MS, which might participate in iron accumulation in the brain, a feature known to be present in several neurodegenerative diseases, including MS [34
We designed a locus-specific CGH array in order to explore the occurrence of CNVs in the HLA region in 15 patients with the peculiar association of CCSVI/VM and MS phenotype.
CNVs are very abundant in the human genome, being involved in genetic variation among populations in a similar fashion to SNPs. Innovatively, this strategy was adopted instead of SNP-based arrays, as the latter tend to suffer from false positive and false negative results, in addition to the fact that no algorithm able to unequivocally detect genomic variants of less than about 30 Kb seems to be currently available.
However, CGH CNV detection also has inherent limitations, such as probe density, and failure of hybridisation due to the presence of SNPs in the probed region. In addition, it is widely recommended that each novel CNV detected by CGH be validated by an alternative method (for example RealTime PCR or qPCR). Nevertheless, being aware of the limit of confidence of CNV identification by CGH, the advantage of searching for CNVs by this technique was the possibility of directly correlating known, validated CNVs with a potential function, which can be related to imbalance either of specific gene(s) or of genomic regulatory regions. In fact, the CGH approach allows fine mapping of the identified CNVs in non-genic regions, which may be correlated to gene regulatory functions, epigenetic changes or other non-coding functions. Indeed, the potential impact on the expression regulation of many genes by genomic "perturbation" of one (or more) specific genomic region(s) is intriguing, as it tentatively implicates it in a variety of different complex phenotypes. This is the reason why this genetic model applies very well to polygenic or multi-factorial diseases [37
As expected, we detected a high number (234) of known CNVs, thereby confirming that the HLA region is very rich in structural variations, in addition to its known high polymorphism in terms of SNPs.
Analysing the distribution of the polymorphic CNVs identified in patients, we observed a peak of CNV numbers within the HLA region. Outside this specific region however, the number of CNVs per patients remained high, though with variable distribution.
We also genotyped the HLA-DRB1 region in our patients. Statistical analysis failed to show any correlation between the presence of the HLA-DRB1*15 allele and VM-related clinical signs. It is well known that SNPs or complex polymorphism density is haplotype-dependent within the human Major Histocompatibility Complex (MHC) [38
However, in order to link the CNV profile and the HLA haplotype, phasing both polymorphism types is mandatory. In addition, no studies have yet fully characterised either CNVs (array-based) or SNPs in the whole genomic region of the HLA locus. For this reason, although the frequency of HLA-DRB1*15 in our patients' cohort seems to be lower than the one reported in classic MS, it is not possible to conclude whether the VM phenotype has a distinct association with the HLA haplotype, especially when one considers our small patient number. This seems to be a major goal for future studies.
Interestingly, while no statistically significant association was found between CNV type or distribution and patient phenotype, the overall number of CNVs showed a significant correlation with the number of stenosing malformations demonstrated by venography in the extracranial segments of the cerebrospinal veins. However, the major shortcoming of this pilot study is the dimension of the sample, which should be expanded in the future to strengthen the as yet unconfirmed significance of our findings. The small number of patients also affected further sub-analysis, and both the extragenic and the intragenic component of the CNVs were not found to be associated to the phenotype VM. Moreover, no correlations were discovered between CNVs and either the number of anomalous Doppler haemodynamic parameters or the VHISS.
Nonetheless, the phenotype studied here, correlated with the CNVs, is strongly associated to MS (OR 43, p < 0.0001) [16
]; we speculate that the presence of VM may contribute to the increase in iron accumulation in MS as a pathogenic component of the disease [36
]. The hampered cerebral venous return consequent to extracranial venous malformations is peculiar to MS, and was not found in a miscellany of patients affected by other neurodegenerative disorders characterized by iron stores, such as Parkinson's, Alzheimer's, and amyotrophic lateral sclerosis [16
]. Theoretically, venous haemodynamic overload may facilitate local microbleeding following damage to vein or venule walls, becoming a distinctive mechanism of iron deposition in MS, as we recently demonstrated [36
Taken as a whole, our data, though preliminary, suggest that the number of polymorphic CNVs in the HLA region did correlate with the number of VMs in our patient cohort. Since HLA is the only region consistently associated with the disease, further studies are certainly needed to discriminate whether the CNV findings are specific for MS patients with venous malformations or, instead, are similarly observed in the general MS population, regardless of the presence of vascular malformations.
In addition, the region studied contains 211 known genes. Using a functional bioinformatics tool, we identified many genes interacting in both neurodegenerative and angiogenesis circuits. Notably, HSPA1L, HSP1A and HSP1B and the HLA-DQ2 gene network in both pathways. Heat-shock proteins (HSPs) represent a group of regulatory proteins involved in a variety of processes, including immunity and angiogenesis [39
]. In particular HSPA1L expression is modulated by ETS1 transcription factor and by SP100, a nuclear autoimmune antigen. Interestingly, genes negatively regulated by ETS1 and up-regulated by SP100, such as HSPA1L, have anti-migratory or anti-angiogenic properties [41
MS possesses a recognised major heritable component, since its susceptibility is associated with the MHC class II region, especially the HLA-DRB5*0101-HLA-DRB1*1501-HLA-DQA1*0102-HLA-DQB1*0602 haplotypes, which dominate genetic contribution to MS susceptibility [42
]. Interestingly, HLA-DQA2 is known to be involved in pro-inflammatory CD4(+) T-cell-mediated autoimmune diseases such as MS and type 1 diabetes [43
]. CD4 is also a very well known inhibitor of tumour angiogenesis [44
], thus supporting a link between the two pathways. The interpretation of the pathway interaction is obviously complex, but it does suggest biological and functional links among these genes as well as, intriguingly, between angiogenesis and immunity.