Introduction

Migraine is a complex and polygenic disorder, featuring different characteristics such as nausea, vomiting, phonophobia, photophobia, and interestingly the cardinal feature i.e., unilateral headache1. International Classification of Headache Disorder 3rd edition (ICHD-3) has classified the disorder into two main clinical subtypes i.e., Migraine with Aura (MA) and Migraine Without Aura (MWA) based on the criteria of presence and absence of aura feature (ICHD-3.org/1-migraine/). Various factors, such as cortical spreading depression (CSD), activation and desensitization of the Trigeminal-vascular system, neurogenic-neuroinflammation, etc., which are collectively responsible for the etiology of migraines, have previously been explored2,3,4,5,6. Factors that are responsible for increasing the risk of migraine are crucial and are broadly categorized into environmental and genetic factors. Former which is responsible for hindering the susceptibility threshold of pain which is set by the latter factor i.e., genetic risk factors1.

Genetic risk factors include genes with changes/variations in the gene sequence with a major type called Single Nucleotide Polymorphism (SNP) or Single Nucleotide Variation (SNV) which are responsible for altering the function of the same. The most recent and updated meta-analysis of the Genome-Wide Association Study (GWAS) data has shown that many genes with modest effects are involved in disease risk7. Other than the advanced GWAS, numerous independent studies have been carried out in various populations and have identified various genes that are responsible for disease risk attribution. Using India as an example of a population, which is part of the Asian ethnic group where migraine disorder is very common8,9,10,11,12, many genes have been studied (Table 1) but the association between these genes and migraine risk has been found to be inconsistent13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36.

Table 1 Multiple gene identified in the population of India.
Full size table

As a result, in the current review study, we first sought to identify the different genes explored in the Indian population before pooling the risk of similar genes to determine the precise association in the same population. The current study is unique in that it is the first of its kind to include all of the risk genes and variants from the Indian population to determine the precise risk.

Method

Literature survey

The presented review aimed to find out the critical genes that increase the risk of migraine and its clinical subtype in the population of India which belongs to the Asian ethnic group. To achieve the aim, we used the approach of a “systematic way of literature survey” which was done from the online database and search engines such as PubMed-NCBI (National Center for Biotechnology Information) (Pubmed.NCBI.nlm.nih.gov), and Google Scholar (Scholar.google.com.tw) respectively. We bypassed exploring other databases and utilized PubMed because of its comprehensive collection of references, which includes MEDLINE (Medical Literature Analysis and Retrieval System Online), life science journals, and electronic books.

Multiple key terms were used in our search strategy, including “Gene polymorphism with migraine in India OR Gene variant with migraine in India AND migraine genes in India OR migraine polymorphism in India”. Only articles published in the English language were evaluated using the linguistic filters as a factor for publication selection. We also tried to exclude study data if unpublished (Research Square/ Researchsquare.com), incomplete, or only partially available. Because partial and missing data are not included in the study, there is no such detrimental effect and we did our utmost to eliminate any undesirable characteristics. The search was completed and studies were included following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (www.prisma-statement.org) (Supplementary Prisma Check List-S1)38.

Inclusion and exclusion features

Concerning the study inclusion criteria, the following inclusion criteria should be met and those include “A case–control or cohort study design must be the prime requirement followed by the “the study must represent the Indian population criteria” for the screening of study”, second the “authors must have investigated/ diagnosed the patients according to the criteria of the International Headache Society (IHS) or ICHD-3″ International Classification of Headache Disorders-3, the authors must have looked at the genetic polymorphisms/variants and must have provided the detailed description about the variant under study with proper reference ID (rsID) or change of nucleotide”, “the genotype frequencies of the investigated variants must be stated unambiguously among migraineurs and controls”, “Hardy–Weinberg Equilibrium (HWE) conditions are required for all experiments”, “studies should provide clear data to calculate the odds ratios (ORs) and the corresponding 95% confidence intervals (CI)”.When necessary information was missing from an article, it was sourced from another publication.

Data extraction

The demographic characteristics of each study were extracted, including the state in which the research was conducted, the number of patients and healthy individuals, any cohort data, the genotypic frequency from both cases and controls, the first authors, the years of publication, and the technique used to determine the genotype? Also features such as gene coding protein, chromosomal location, the function of a protein, and polymorphism ID/SNP-ID/rsID were extracted from the online data where required. If any statistical/numerical data were found missing, previous research/references were analyzed. All the data/features extraction was first done by our two authors (A.S, A.C.P, M.B, & I.S) and then the quality was assessed by the other two authors (A.S, A.C.P, & H.K).

Quality assessment

The quality of research articles is an important factor to consider when performing a meta-analysis, which entails pooling independent research studies to determine the precise result. In light of this, the present research assessed the quality of all previously published studies by utilizing the criteria established by the Newcastle–Ottawa scale (NOS) such as (1): selection of cases and controls that include cases and control definition, and their selection(2): comparability (comparability of cases and controls) and (3): ascertainment of exposures (exposure ascertainment, case, and control ascertainment, and non-response rate). Each section with the correct method is assigned one star (1 point) with an exception in Comparability” section which is with two stars (2 points). Therefore, a study will be disqualified if it receives fewer than 5 stars (< 5) (or 5 points), which is considered to be a good study and can only receive a maximum of 9 stars (Ottawa Hospital Research Institute (ohri.ca) (Supplementary File: SF1-NOS). If any differences in decision-making were noticed concerning article inclusion, data extraction, or quality assessment/ NOS, the third investigator (P.K) investigated and concluded the matter.

Statistical analysis

Genotypic and allelic frequency was first calculated for all studies included in the meta-analysis and then the Chi-square test was used to analyze whether the population is in HWE (p > 0.05 for the population in HWE) or not (p < 0.05). To find out the strength of the association between the variant of interest, and the risk of migraine, the logistic regression utilizing OR (Odds Ratio) model with 95% CI (Confidence Interval) was used. Here odds ratio is defined as the OR > 1 is defined as the odds of exposure among cases being greater than the odds of exposure among controls & OR < 1: the odds of exposure among cases are lower than the odds of exposure among controls. Different genetic models such as allelic (rare allele vs. wild allele), dominant (dominant vs. heterozygote + pure recessive), recessive (recessive vs. dominant + heterozygote), and over-dominant (heterozygote vs. pure dominant + recessive) were used to observe the strength of association (OR) using random: Dersimonian and Laird method or fixed model (Inverse variance method) based on I2(I2 > 75: Random model). I2 is an estimate to define the proportion of inter-study variability attributed to chance rather than heterogeneity.

The publication bias including reporting bias and heterogeneity of the research studies were assessed using Begg's and Egger's tests and χ2 based on Cochran’s Q Test with I-square (I2) tests respectively. Also, we performed the sensitivity analysis to observe the influence of individual studies on the pooled ORs and 95% CIs by the criteria of “exclusion of each study”.All tests were two-sided, and a value < 0.05 was considered statistically significant. The current meta-analysis process, from the choice of statistical tests to the analysis of the findings, was conducted following the Cochrane guidelines (Training.cochrane.org/handbook/current). All the statistical analysis was done with Meta-Genyo online Statistical Analysis System software (MetaGenyo: Meta-Analysis of Genetic Association Studies).

Trial sequential analysis

To reduce the possibility of random error, the current meta-analysis makes use of a method called Trial Sequential Analysis (TSA), which checks to see if the included trials have sufficient numbers of participants. Based on an overall risk of 5% and a relative risk reduction of 20% (with 80% power), the TSA tool (Copenhagen Trial Unit, Denmark) was used to compute the necessary information size for evaluating the validity of a meta-analysis. (TSA—ctu.dk). With specific settings “Set Effect Measure and model” (Effecet Measure: Odds Ratio, Model: Fixed), “Set Zero Event Handling (SZEH)” (Method: Constant, Value: 1.0, Included trials with no events: checked), “Set Confidence Interval (SCI)” (Conventional/ Coverage: 95%). There are, however, two basic possibilities: (1) no further research is needed if the cumulative Z value/curve exceeds the RIS (Required Information Size); (2) additional studies are needed if the Z curve does not surpass the RIS threshold.

Protein–protein interaction

It is critical to identify the most important wiring connection/most connected node/dot in the gene/protein interaction network used to understand disease progression because any change in the peripheral gene will eventually affect the regulation of the core gene/ most connected node39. Therefore, to analyze the most connected gene/node among the studied genes in the respective population, the String v11 (String-db.Org/), a potential protein–protein interaction tool that collects data from several online databases was used. After PPI (Protein–Protein Interaction), the network was built, edited, and analyzed using the Cytoscape tool version 3.9.1, a free and publicly available bioinformatics tool for analyzing and interpreting gene expression profiles, and molecular interaction networks (Cytoscape.org/).

Result

Using the strategy of systematic way of literature survey (Fig. 1), a total of 24 studies was found (Table 1) which explored about 21 genes with 31 variants (Table 1) from 4 different states of India (3 states from north India and 1 state from south India) (Fig. 2) (Paint—Microsoft Apps). Only nine genes have been studied more than two times in the Indian population and thus were found eligible for the meta-analysis and these include six studies which have explored MTHFR gene13, 16,17,18, 22, 23, three studies for ACE (I/D polymorphism)13, 17, 24, LRP1- rs1117211319, 26, 28, PRDM16- rs265189919, 26, TRPM8- rs10166942 and rs1050486121, 26, ESR1 PvuII and XbaI15, 29, 37, DAO- rs10156191, rs205212920, 35 and TNF-α G308A25, 30.

Figure 1
figure 1

Selection of literature according to PRISMA (Preferred Reporting Items for Systematics Reviews and Meta-Analysis) guidelines.

Full size image
Figure 2
figure 2

Indian Map representing different genes explored in 3 different states of India population.

Full size image

Study characteristics

After finding nine genes eligible for meta-analysis, the second step was the inclusion into the meta-analysis of eligible studies which were done based on the NOS (Table 2) and HWE criteria. If the NOS were six or more than it and if the control population were in HWE respectively the study was included After which two variants such as A1298C of MTHFRand rs10166942 of TRPM8 were excluded due to not being found in HWE (Supplementary Tables 2 S1 and 5 S1).

Table 2 Newcastle–Ottawa Scale (NOS) for assessing the quality of non-randomized studies in meta-analyses†
Full size table

Meta-analysis

MTHFR-C677T

In the present analysis, a total of 842 cases and 882 control subjects were included which were found after the inclusion of five studies representing four from the north Indian population13, 16,17,18 and one from the south Indian population22 and exclusion of one study due to not found in HWE23 (Supplementary Table 1 S1). The frequency of the risk allele was 0.195 (n = 329/1684) in contrast to the wild allele 0.804 (n = 1355/1684)within the case group while in a control group, the frequency of the risk allele was 0.168 (n = 297/1764) in contrast to wild allele i.e., 0.831 (n = 1467/1764) in the control group.

To find out the association, a logistic regression model i.e., Odds Ratio associated with a 95% Confidence Interval, p-value < 0.05 were used. The present meta-analysis has shown that there was no significant association between the variant under study (C677T) and the risk of migraine in the Indian population after utilizing different genetic models and these include the allelic model (OR: 1.04 [0.84–1.29], I2 = 53%) (Fig. 3A), recessive model (OR: 1.31 [0.71–2.42], I2 = 38%), dominant model (OR: 1.08 [0.72–1.62], (I2 = 53%), and overdominant model (OR: 1.06 [0.70–1.60], (I2 = 53%). Subgrouping based on the criteria of the ”study conducted in which region of India i.e., such as South India (SI) and North India (NI)”, no significant association was observed (Supplementary Table 11 S1). After sub-grouping based on the clinical subtype i.e., MA and MWA, there was no significant association was found with any genetic models.

Figure 3
figure 3

(A) Forest Plot of MTHFR allele model showing the non-significant association with the risk of overall migraine (B) Symmetrical Funnel Plot representing no publication bias.

Full size image

Egger's test, which is based on the connection between standard error and strength of association (log of OR), was used to examine publication bias across all studies included in the meta-analysis (p-value = 0.35). By placing the most accurate research on top and the least precise studies at the bottom of a scatter plot, we were able to create a "funnel plot" that displays the distribution of accuracy across all investigations. All genetic models resulted in symmetrical funnel plots, indicating that no publication bias existed. (Fig. 3B). The findings of a sensitive analysis performed on all genetic models by systematically removing individual studies showed that the pooled ORs were not significantly altered, confirming the excellent stability of the meta-analysis (Fig. 4).

Figure 4
figure 4

Sensitive plot representing allele model.

Full size image

ACE-I/D

There are 3 studies13, 17, 24 (Supplementary Table 3 S1) that observed the frequency and association of polymorphism in the population of the Indian population. After pooling such independent studies, we found that the overall frequency of risk and wild allele was 0.410 (n = 289/704) and 0.589 (n = 415/704) in the patient group respectively. While in the control population, the frequency of the risk allele was considerably low i.e., 0.343 (n = 289/842) in comparison to the frequency of a minor allele in the patient group (q = 0.410).

The present meta-analysis found a significant association between the selected variant and risk of overall migraine after utilizing the allele (Fig. 5A) and recessive model (OR: 1.37 [1.11–1.69], I2 = 0%) and (OR: 2.05 [1.36–3.11], I2 = 0%) respectively in contrast to dominant and over-dominant model (OR: 1.29 [0.96–1.73], I2 = 0%) and (OR: 0.90 [0.67–1.19], I2 = 24%) respectively. After subgroup based on the clinical subtype i.e., MA and MWA, the variant showed a significant association after utilizing different genetic modes such as allele (OR: 1.41 [1.06–1.88], I2 = 0%), recessive (OR: 2.22 [1.29–3.83], I2 = 0%) in contrast to dominant (OR: 1.32 [0.88–1.98], I2 = 0%) and over-dominant model (OR: 0.90 [0.60–1.34], I2 = 39%) in MA group in compare to MWA where allele and recessive model showed significant association (OR: 1.33 [1.05–1.70], I2 = 42% and OR: 1.94 [1.21–3.11], I2 = 11%) in comparison to dominant and overdominant model (OR: 1.26 [0.89–1.77], I2 = 0% and OR: 0.90 [0.65–1.26], I2 = 0%) respectively. In addition, subgrouping based on the criteria of the ”study conducted in which region of India i.e., such as South India (SI) and North India (NI)”, was not done due to all studies were from north India.

Figure 5
figure 5

(A): Forest Plot of ACE allele model showing the significant association with the risk of overall migraine (B): Symmetrical Funnel Plot representing no publication bias.

Full size image

There was no evidence of publication bias because all funnel plots for genetic models were symmetrical (p-value = 0.68) (Fig. 5B) (Supplementary Table 14 S1). The good stability of the meta-analysis was confirmed by the results of a sensitive study done on all genetic models by carefully removing individual research (Fig. 6).

Figure 6
figure 6

Sensitive plot representing allele model.

Full size image

ESR1

In the present study, we found two variants such as PvuII and XbaI of ESR1 studies in the Indian population by three different research groups14, 29, 37 (Supplementary Tables 6 and 7 S1). Concerning ESR1-PvuII, a significant association has been found where the allele (OR: 1.47 [1.24–1.74], I2 = 0%), dominant (OR: 1.66 [1.30–2.12], I2 = 0%), and recessive model (OR: 1.91 [1.31–2.77], I2 = 0%) significantly increase the risk of migraine. After subgrouping based on clinical type criteria, a significant association was also found in both MA (Allele: OR: 1.72 [1.34–2.20], I2 = 0%, dominant: OR: 2.63 [1.75–3.96], I2 = 9%, overdominant: OR: 1.75 [1.23–2.50], I2 = 0% and recessive: OR: 1.79 [1.07–2.98], I2 = 0%), and MWA (allele: OR: 1.39 [1.15–1.67], I2 = 11%, dominant: OR: 1.43 [1.10–1.87], I2 = 55%), and recessive model: OR: 1.94 [1.30–2.89], I2 = 0%).

Concerning, after critical literature analysis, only two research publications were found29, 37 discussing the impact of XbaI polymorphism on the susceptibility of migraine and its type. The pooled OR of both studies did not show any significant association with the risk of migraine or with the migraine sub-type (Supplementary Table 12 S1). All genetic models resulted in symmetrical funnel plots, indicating that no publication bias existed. The findings of a sensitive analysis performed on all genetic models by systematically removing individual studies showed that the pooled ORs were not significantly altered, confirming the excellent stability of the meta-analysis.

TNF-α G308A

After combining the two studies25, 30 (Supplementary Table 8 S1), there was a significant difference between the genotypic frequency in the patient group (GG: 77.80%, GA: 18.45%, & AA: 3.73%) in comparison to a control group (GG: 79.03%, GA: 18.43%, & AA: 2.53%). The frequency of risk allele (q) was found slightly more (q = 0.129) than the control group (q1 = 0.117).

When comparing the pooled results from the experimental (n = 856) and control (n = 868) groups, the association value was not statistically significant for any genetic model under study such as allelic (OR: 1.12 [0.84–1.51] I2 = 68: random model), dominant (OR: 1.08 [0.77–1.50], I2 = 82%: random model), recessive (OR: 1.54 [0.70–3.39], I2 = 0.0%: fixed model), and over-dominant model (OR: 1.00 [0.70–1.42], I2 = 89%: random model) (Supplementary Table 15 S1). There was no subgrouping analysis in this variant since only one study investigated the clinical subtype25. All genetic models had symmetrical funnel plots, indicating that there was no publication bias. A sensitive investigation of all genetic models was also performed by removing each research one at a time. It was demonstrated that none of the pooled ORs were considerably influenced, indicating the meta-analysis findings' excellent stability.

LRP1- rs11172113

In the present review, we found two studies19, 26 (Supplementary Table 9 S1) representing the north Indian population where the combined frequency of risk allele was less i.e., 0.198 (n = 163/410) in the patient's group in comparison to control group i.e., 0.29 (n = 174/600). To find out the risk using different genetic models, a protective role of variant (allelic model) (Fig. 7A) was observed with an OR of 0.65 [0.50–0.83] (I2 = 44%), dominant (OR: 0.48 [0.35–0.66], I2 = 12%), in contrast to recessive and over-dominant (OR: 1.29 [0.22–7.59], I2 = 91%) and (OR: 0.10 [0.00–5.43], I2 = 88%) respectively where the non-significant association was observed.

Figure 7
figure 7

(A) LRP1 Allele showing the significant protective effect of a rare variant in the Indian population (B) Symmetrical Funnel Plot representing no publication bias.

Full size image

After clinical sub-grouping of migraine, it was observed that allele (OR: 0.54 [0.37–0.78], I2 = 52%), dominant (OR: 0.47 [0.30–0.73], I2 = 0%), and over-dominant (OR: 0.54 [0.34–0.86], I2 = 0%) significantly showed protective role in MA. But, in the case of MWA, only dominant (OR: 0.68 [0.49–0.95], I2 = 58%) and over-dominant model (OR: 0.63 [0.45–0.88], I2 = 0%) showed a protective role in contrast to allele (OR: 0.88 [0.46–1.68], I2 = 82%) and recessive (OR: 1.12 [0.65–1.93], I2 = 74%) where a non-significant association was observed. There was no evidence of publication bias because all genetic models produced symmetrical funnel plots(Fig. 7B). The good stability of the meta-analysis was confirmed by the results of a sensitive study done on all genetic models by carefully deleting individual research(Fig. 8).

Figure 8
figure 8

LRP1 migraine sensitivity plot for allele model.

Full size image

DAO- rs10156191

After combining studies20, 35 (Supplementary Table 10 S1), it was observed that in the patient’s group, the heterozygote (CT: 29.42%) and homozygous recessive (TT: 5.71%) genotypes are slightly greater than the heterozygote (CT: 21.14%) and homozygous recessive (TT: 3.42%) genotype in control’s group. The frequency of the risk allele (q = 0.204) in the patient group was more than the frequency of the risk allele in the control group (q1 = 0.14).

The present meta-analysis provides pieces of evidence that allele (OR: 3.86 [0.37–39.98], I2 = 81%) and recessive model (OR: 1.47 [0.69–3.12], I2 = 52%) showed non-significant association with the risk of migraine in contrast to dominant (OR: 1.69 [1.19–2.42], I2 = 69%) and over-dominant model (OR: 1.62 [1.12–2.34], I2 = 13%) which significantly increase the risk of migraine in Indian population. All genetic models yielded symmetrical funnel plots, eliminating publication bias. A sensitive study on all genetic models carefully removing individual studies proved the meta-analysis's stability.

PRDM16-rs2651899

Concerning PRDM16- rs2651899 (Supplementary Table 4 S1), the frequency of the risk allele in the case group was slightly higher i.e., 0.469 (n = 459/978) compared to risk allele frequency in the control group i.e., 0.471 (n = 330/700). In addition, no significant association was observed in migraine or any clinical subtype (MA and MWA) after utilizing any genetic models (Supplementary Table 13 S1).

Trial sequential analysis

After finding a non-significant association for MTHFR-C677T, DAO- rs10156191, TNF-α G308A, and ESR1-XbaI, the required sample size estimation was done for allele model using TSA. For the MTHFR-C677T, the last point of the Z-curve reached or positioned within the conventional boundary which is considered as a statistically non-significant zone therefore, we cannot conclude that there is any risk association between the variant under study and diseases. Therefore, to achieve power (RIS: 10,616) further studies are required (Fig. 9). Concerning remaining DAO- rs10156191, TNF-α G308A, and ESR1-XbaI, the TSA showed “Boundary RIS is ignored due to little information use”.

Figure 9
figure 9

TSA graph for MTHFR-C677T (allele model) showed a non-significant result with less sample size/ power therefore required more studies to find out the association.

Full size image

Protein–protein interaction

In the present study, we also aimed to find the most connected node in the list of genes studied in the population of India. Therefore, the String database which is a potential PPI tool that collects data from several online databases was used. Concerning the PPI setting, medium confidence (40%-69%), with active interaction sources which include test mining, experiments, databases, co-expression, neighborhood, gene fusion, and co-occurrence were utilized. We found that there were 32 connections/ edges between the processed 16 nodes/protein with an average node degree of four and six expected number of edges, 0.683 of average local clustering coefficient, and a significant PPI enrichment p-value (1.57e-14). It was observed that the highest degree (Degree: 8) was found with TNF-α followed by APOE and SLC6A4 (Degree: 7) (Table 3). String PPI was later edited and presented using Cytoscape tool version 3.9.1, which is an open-source bioinformatics software platform for visualizing molecular interaction networks and integrating them with gene expression profiles and other state data (Fig. 10).

Table 3 Protein–Protein Interaction.
Full size table
Figure 10
figure 10

Protein–Protein interaction where the TNFA shows the highest node.

Full size image

Discussion

Migraine is considered a complex disorder with polygenic inheritance and it has been shown by the most recent and updated meta-analysis of GWAS data has shown that numerous genes contribute to the risk of diseases with small effect sizes7. Other than the advanced GWAS, multiple studies have been conducted in different populations and found different genes. Specifying with the example of population, many genes have also been explored in the Indian population belonging to Asian ethnic groups constituted of different states (Fig. 2). Within the same population, association disparity was found in many genes and risk of migraine. Therefore, the present meta-analysis aimed to find out the precise risk between the different genes that have been explored in the past.

We have found that the variant ”C677T” of MTHFR showed a non-significant association with the risk of overall migraine in the Indian population which supports the independent studies13, 16, 18 in contrast to the positive association found by different independent studies17, 22, 23. Comparing the present pooled result with the most recent meta-analysis which discussed the association of C677T with the risk of migraine, showed a significant risk association40. However their selection of studies was before December 2018 and only two studies from the Indian population were included13, 18 and also missed the inclusion of one more study17. In addition to meta-analysis, we have also estimated the ”Required sample Size” using trial sequential analysis and observed that the Z-curve was unable to cross the required information/ sample size (Fig. 9). Therefore, we cannot conclude that there is any risk association between the variant under study and diseases and thus required more sample size.

Concerning the ACE-I/D polymorphism, the present meta-analysis showed a significant association between the variant of interest and the risk of migraine and both clinical subtypes such as MA and MWA. This present study supports the independent study by Jasrotia and group and Joshi and group13, 17 in contrast non-significant association was observed by Wani and group24. Interestingly all studies13, 17, 24 were from north India, thus disparity between such might be due to different sample sizes i.e., control and case subjects. Evidence from the meta-analysis published in 2016 powered with 7334 patients and 22,990 control showed no relationship between the ACE I/D polymorphism and any migraine but upon subgrouping based on the criteria of ethnicity, they observed a protective effect against migraine with aura and without aura at least in the Turkish population41.

Regarding ESR1 and its variant studied in the Indian population, only PvuII showed a significant association with the risk of migraine including both of its clinical subtypes in contrast to XbaI. which was consistent with Kumar and group29 in contrast to Ghosh and group37. The other intronic variant i.e., PvuII which is separated by the 50 bp from XbaI, two studies discussed the association in the respective populations, and upon combing both studies, the present pooled meta-analysis showed significant association with the risk of migraine including both of its clinical subtypes which supports the result observed by Joshi and group14 in contrast to Kumar and group29.

TNF-alpha is known for its critical role in pro-inflammation and critical regulator of microglial activation which leads to the initiation of neurogenic-neuroinflammation6. But the presence of a functional variant i.e., − 308 G > A leads to elevated plasma level of protein thus hindering the susceptibility of inflammation threshold. In the present study, after analysis of two independent studies25, 30 we did not find any significant association after utilizing different genetic models and risk of migraine which was the opposite of what was observed by the included independent study25, 30. This prime reason for such disparity might be due to the different regions one is from north India25 and the other is from south India30. Therefore, it is very important to conduct more studies in the respective population to find out the precise result. Comparing our meta-analytic data with the pre-existing meta-analysis, the results were found consistent with different studies30, 42,43,44 in contrast to Chen and group44.

Concerning the LRP1- rs11172113, the present review observed a protective role of variant (allelic model) (Fig. 7A) with overall migraine, and such protective effect was found consistent with both of its clinical subtypes. Thus, the present pooled result supports the result that Ghosh and group previously observed26 in contrast to Kaur and group19. Each study was observed in the north Indian population, but such disparity might be due to the low sample size and also the patient group was not in HWE19. Comparing our meta-result with the overall pooled result presented by Siokas and the group where they observed a non-significant association between the rs11172113 and risk of migraine (OR: 1,10 [0.84–1.44], I2: 68%)45.

Strengths and limitations of the present meta-analysis

The prime strength of the present pairwise meta-analysis is the strategy utilized for the literature survey, then the inclusion of searched studies based on the criteria discussed (Section ”Inclusion and exclusion features”), and secondly the use of statistical analysis for finding the risk association between the different risk variants and diseases under consideration. Thirdly, the presentation of pooled summary estimates is considerably simpler to understand. Fourth, we have also found the risk attribution between the selected variants and migraine subtypes (MA and MWA). Fifth, we also presented the protein–protein interaction in an attempt to find out the most connected node in the network of nodes selected from the population under study. Sixth, a precise risk attribution toward the risk of migraine within a specific population i.e., the Indian population has been established. Apart from the strength, the first limitation is that migraine is an extremely heterogeneous condition, as all studies have diagnosed the suspected individual using criteria of ICHD-3 / HIS, but still, there could be misclassification. Second, the present analysis is only limited to clinical subgrouping and no subgroup was done based on gender. Also, there was an incredible disparity in sample size between the studies. Also, the risk of non-significantly associated variants can be modified by different modifier genes which were not explored in the present study. Additionally, the risk of disease can be attributed to the interaction between the markers of the same gene. In addition, concerning with the included studies in the meta-analysis were not enough, therefore for a precise estimate, more studies are required.

Future perspectives

In the present study, we aimed to find out the critical gene or genes that are responsible for the significant risk attribution toward disease susceptibility within a specific population (India) using a high statistical meta-analytical research approach. Different genes such as ACE-I/D, and ESR1-PvuII showed a significant association with the risk of migraine in contrast to LRP1-rs11172113 and MTHFR-C677T, PRDM16-rs2651899, DAO-rs10156191which showed protective and non-significant association in respect to Indian population respectively. We also noticed that there was much disparity in the sample size between studies, specifically the patient’s group was even not found in HWE. Also, the ratio between case and control was not even equal in the different studies, and for a fixed sample size, the chi-square test for independence is most powerful if the number of cases is equal to the number of controls (i.e., 1:1)46. Additionally, we can increase or recruit more controls to boost the study's statistical power if we are unable to find enough cases, but only up to 4 controls for every one case. Given the expense of recruiting them, adding more controls (more than four) might limit the increase in statistical power beyond this ratio47. There have only been a few, sometimes just two, studies exploring specific variants, which makes it necessary for more research to be done to support the risk attribution hypothesis.

Conclusion

In conclusion, this present meta-analysis showed that the ACE-DD variant and ESR1-PvuII showed a significant risk of migraine in the Indian population in contrast to LRP1- rs11172113 which showed a protective role in the respective population.