HLA-DQA1*04:01 is related to a higher multiple sclerosis lesion load on T2/Flair MRI sequences

ABSTRACT Background: The genetic predisposition to multiple sclerosis (MS) is associated with HLA alleles, especially HLA-DRB1*15:01. Objective: To identify associations between findings in magnetic resonance imaging (MRI) and genetic features in a Brazilian cohort of patients with MS. Methods: We retrospectively studied data from 95 consecutive patients with MS. Two independent observers who were blinded to the clinical data identified black holes and enhanced lesions on T1 MRI sequences, and counted and measured contrast-enhanced lesions on T2 and Flair (fluid attenuation inversion recovery) sequences. Cases were classified according to lesion size, number, and volume. The HLA-DRB1, HLA-DQB1, and HLA-DQA1 alleles, and the rs4774, rs3087456, rs6897932, rs731236, and rs1033182 single nucleotide polymorphisms were identified by polymerase chain reaction amplification with sequence-specific primers using the One Lambda Inc. Kit, Canoga Park, CA, USA. Results: Patients with the HLA-DQA1*04:01 allele had lesion load (adjusted for age, sex, and MS duration) above median compared with patients with other HLA-DQA1 alleles (p=0.02). There were no differences among all the other HLA alleles and single nucleotide polymorphisms and lesion load. Conclusions: The correlation of the HLA-DQA1*04:01 allele with a higher lesion load on T2/Flair MRI sequences suggests that the presence of this allele is associated with the risk of greater MS severity.

Recently, the International Multiple Sclerosis Consortium published a large metanalysis demonstrating the importance of HLA-DRB1*15:01, HLA-DQA1, and HLA-DQB1 interaction and its role in peripheral immune cells and microglia susceptibilities in MS patients 9 .
Baranzini et al. 10 found 242 single nucleotide polymorphisms (SNPs) related to MS susceptibility, including 65 SNPs in the major histocompatibility complex of chromosome 6p21. 3. Another work suggests that the polymorphisms CIITA -168AA, CIITA +1614GG, and CIITA +1614 GC are associated with a better clinical course of MS in Brazilian patients with the disease 11 .
In this retrospective study, we searched for associations among the HLA-DRB1, HLA-DQA1, and HLA-DQB1 haplotypes and the following SNPs: rs4774 and rs3087456 (CIITA gene), rs6897932 (IL7R gene), rs731236 (VDR gene), and rs1033182 (ESR gene) and MRI features, mainly lesion load (LL), number of black holes (black lesions on T1 MRI) (BH), and enhanced lesions (EL) in a cohort of 95 Brazilian patients with MS.

Patients
We retrospectively analyzed data from 95 patients (60 women and 35 men) with MS diagnosed on clinical and laboratory bases who were followed as outpatients and during periods of eventual hospitalization during the last 15 years at the Hospital Universitário Clementino Fraga Filho/Universidade Federal do Rio de Janeiro (HUCFF-UFRJ). All subjects met the 2017 McDonald criteria for the diagnosis of MS 12 . According to disease progression, MS was classified as relapsing-remitting (RR), primarily progressive (PP), and secondarily progressive (SP).
We did not include patients older than 71 years old (at the time of MRI) because of the usual hyperintensities from the natural process of aging that could be interpreted as MS lesion load.
The National Council for Ethics in Research approved this study (no. 1265), and written informed consent was obtained from all participants. A single MRI examination of the skull and whole spine (neuroaxis) was chosen for comparison with the clinical situation at a random moment in MS evolution for each patient. We also recorded disease duration, the interval between MS symptom onset and MRI examination, the clinical situation, and the relationship to genetic characteristics.
Clinical evaluations were performed by the team of neurologists at HUCFF-UFRJ, which was blinded to the MRI findings, using Kurtzke's Expanded State Disability Scale 13 .

Magnetic resonance imaging evaluation
MRI examinations were performed in a 1.5-T scanner (Magneton Avanto; Siemens, Munich, Germany) with a 12-channel head coil using a conventional protocol ( Table 1).
The presence, size, and location of hyperintense lesions on T2/Flair ( fluid attenuation inversion recovery) sequences were determined. The number of BH and enhanced lesion (EL) were counted. Following the modified 2017 McDonald criteria, lesion locations were recorded as periventricular, justacortical (subcortical/cortical), posterior fossa, and spinal cord 11 . Two observers with 25 and 10 years of experience who were blinded to patient information counted and measured the lesions visually/manually, without the use of an automatic tool. Any disagreement was resolved by consensus.
The lesion load (LL) was estimated by multiplying the number of lesions by their respective estimated average volumes and summing the results. The LL was also calculated separately according to the McDonald criteria locations. All the LL comparisons among groups were adjusted for age, sex, and illness duration.
The median adjusted lesion load (mLL) was 19.8 mL (in the whole cohort), and we considered this value as the threshold to compare different genetic features groups.

DNA typing
DNA was extracted from blood samples collected on filter paper using the organic method and quantified by spectrophotometry at 260/280 nm. The alleles HLA-DRB1, HLA-DQB1, and HLA-DQA1 and SNPs rs4774, rs3087456, rs6897932, rs731236, and rs1033182 were identified by polymerase chain reaction amplification with sequence-specific primers using the One Lambda Inc. Kit (Canoga Park, CA, USA) according to the manufacturer's recommendations. Then, capillary electrophoresis was performed using an ABI PRISM ® 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA),

Statistical analysis
Due to the non-normal distribution of LL data within groups, we used the median rather than mean for comparing different genetic features (HLA type and its alleles and SNPs).
Patient information was entered into a Microsoft Excel ® (Redmond, WA, USA) database and later exported to the Statistical Package for the Social Sciences (SPSS ver. 14.0, IBM, Armond, NY, USA). Proportional data were compared using the chi-squared test (Fisher or Yates, as needed). Interval and ratio data were submitted to analysis of variance with a comparison of multiple groups according to Tamhane's statistics, as the variance between groups was not homogeneous. P value was considered significant if <0.05

Associations of demographic and clinical characteristics with clinical multiple sclerosis categories
Of the 95 patients analyzed, 73 had RR, 9 had PP, and 13 had SP MS. Patient characteristics and timing of MRI examination are shown in Table 2. The mean age was significantly greater in the RR group than in the PP group (p=0.02). Male sex predominated in the PP group relative to the RR and SP groups (chi-square=5.3, p=0.01) The mean age at disease onset was significantly greater in the PP group than in the RR group (p=0.02). The average disease duration and age at the time of MRI examination did not differ significantly among groups.

Associations of magnetic resonance imaging findings with clinical multiple sclerosis categories
MRI features and parameters are shown according to MS progression type in Table 3. The mean LL in the posterior fossa was significantly greater in the SP group than in the RR group (p<0.05); no significant difference was found in the LL or number of lesions in any other region of the brain. The mean number of BHs was significantly greater in the SP group than in the RR group (p<0.02), and the mean number of ELs was significantly greater in the RR group than in the PP group (p<0.04).

Associations of lesion load and genetic features
We compared the mLL with the SNPs rs3087456, rs4774, rs6897932, rs731236, and rs1033182, considering the three possible genetic variations (wild type homozygous, heterozygous, and polymorphic homozygous), and found no significant correlation (Table 4). These SNPs were the only ones available for this study, given the scarcity of resources.

DISCUSSION
The significance of a high mLL in patients who have the HLA-DQA1*04:01 allele may suggest a possible susceptibility to high disease severity. MRI criteria are widely used for the diagnosis and monitoring of MS, but they are constantly evolving. For example, the 2017 modifications to the MRI criteria changed the dissemination in space concept 12 .
An increasing number of studies have examined genetic associations with the LL, lesion shape, and topological lesion distribution in patients with MS. Gouraud et al. 14 identified 31 significant genetic variations related to MS lesion topology on MRI. They combined with genetic risk score in MS activity and progression. Kalincik et al. 15 found similar results in another study. Patients with MS carrying the susceptibility allele HLA-DRB1*15:01 had a greater brain lesion volume than non-carriers 7 .
In Brazilian patients, a population characterized by ethnic admixture, HLA-DRB1*15:01, has been shown to confer MS susceptibility based on clinical features 16 . Additionally, genetic predictors of MS susceptibility, disease activity, and severity have been identified in two other studies of Brazilian patients 11,17 .
In this study, we highlighted the role of the HLA-DQA1 gene in MS susceptibility. We did not find in the literature a specific relationship between the allele 04:01 and MS susceptibility or severity.
The role of DRB1* and DQA1* molecules in susceptibility to experimental autoimmune encephalomyelitis have been demonstrated 23 . We found no statistical significance between MRI features (LL, number of BH, and EL) and other HLA haplotypes, especially HLA-DRB1, which is most frequently reported in association with MS severity on MRI. However, there are controversial reports in the literature.
In 2003, Zivadinov et al. 24 reported a significant relationship of HLA-B7 with the LL and number of BHs. In 2007 and 2009, Zivadinov et al. 25,26 reported correlations of HLA-DRB1*15:01 and HLA-DRB1*12 with a larger number of BHs and smaller cerebral volumes, but not with the LL. In 2009, Okuda et al. 27 reported a correlation between high LLs and the HLA-DRB1*15:01 allele. Hooper-van Veen et al. 28 described associations of the CD28, IFNGR2, and IL1B-511 genes with a larger number of BHs, but not with the LL. In contrast, Schreiber et al. 29 reported that they found no significant correlation between the MS LL and HLA genes.
Recently, Lysandropoulos et al. 30 reported greater clinical severity and more lesions in patients with HLA-A*2. However, their results for the HLA-DRB1, HLA-DQB1, and HLA-B*08 alleles were inconclusive. In 2020, Lysandropoulos et al. 31 confirmed these findings in a slightly larger group of patients, with a longer clinical and imaging follow-up.
We found no relationship between MRI features of MS severity and SNPs, specifically rs3087456, rs4774, rs6897932, rs731236, and rs1033182. Sombekke et al. 32 and Baranzini et al. 10 found no relationship between HLA-DRB1*15:01 and the LL. However, the latter found correlations of the LL and brain volume with multiple SNPs (but they did not examine any SNP examined in the present study).
In a genetic study, the peculiarities of the population of interest can sometimes explain the differences in the results. We studied a Brazilian cohort, and the diversity of our findings could be related to this feature.
This study has some limitations. First, clinical and imaging data were not obtained over prolonged MS disease courses. We randomly selected a single time point representing each patient' s illness, which was infrequently the time of the last MRI examination. This random selection was made to mitigate selection bias. Second, the analysis of the LL was done manually rather than automated an in recent publications. We chose the manual technique (old method) because it allows simultaneous evaluation of the brain in three different regions justacortical, periventricular and posterior fossa, optic nerve, and spinal cord. Automatic segmentation methods require separate analyses and have limitations in spliting the central nervous system. This limitation was mitigated by independent evaluation by two experienced observers blinded to patient clinical data. Imaging companies need to develop a reliable method to do this automatically.
In conclusion, in this analysis of MRI features in patients with MS, we found a significant association between a high LL and the presence of the HLA-DQA1*04:01 allele, which may represent a genetic susceptibility or predisposition. This specific allele has been associated with many different autoimmune diseases and MS.
Future structure-function studies are needed to uncover the specific mechanisms by which DQA1*04:01 or other haplotypes may cause these neuroradiological findings.