Author: admin

Study cohort and migraine prevalence

We determined migraine prevalence using EHR ICD codes and self-report Baseline Survey data (Supplementary Data 4). MVP Baseline Survey data was unavailable for 280,957 individuals, and Black and Hispanic women were more likely not to have Baseline Survey data (58 and 44%, respectively) than other strata (e.g., White men; 34%). Based on ICD, lifetime migraine prevalence was 10.4% for White, 13.5% for Black, and 13.9% for Hispanic Veterans. Self-reported migraine lifetime prevalence on the Baseline Survey was 9.3% for White, 12.5% for Black, and 13.2% for Hispanic Veterans. Women had a higher prevalence of migraine than men, with the highest ICD prevalence in Hispanic women (36.9%) and the lowest among White men (8.5%). The GWAS sample cases and control definitions were derived from the combination of EHR and the Baseline Survey (see Methods). There were 118,079 individuals with any history of migraine on either or both EHR or Baseline Survey identified as cases, and 440,255 individuals with no history of migraine on both sources were controls (Supplementary Data 2). While individuals with a history of migraine on any one data source could be counted as cases, individuals with incomplete data (missing Baseline Survey) were excluded from the control classification. Consequently, the GWAS sample proportions (Table 1) do not reflect MVP migraine phenotypic lifetime prevalence but are a function of conservative control definitions to minimize contamination (Supplementary Data 4).

The MVP migraine GWAS sample included 433,010 participants, 87,859 cases, and 345,151 controls from three HARE-derived ancestral backgrounds. Table 1 shows the average age of participants at enrollment and migraine prevalence by HARE-derived ancestry/ethnicity categories and sex. European HARE-derived ancestry was the largest ancestry group (EUR, n = 338,743; 59,975 cases, 278,768 controls), followed by African HARE-derived ancestry (AFR, n = 65,178; 19,358 cases, 45,820 controls), and HARE-derived Hispanic (HIS, n = 29,089; 8,526 cases, 20,563 controls). People of East Asian and South Asian ancestry were not analyzed due to the low number of Asian individuals in MVP. The MVP sample was predominantly male, with 90.4% (n = 391,622) men and 9.6% (n = 41,388) women, consistent with the distribution (10% women) in the U.S. VA population [51]. Most participants (88.1%) were more than 50 years old, with 402,234 individuals in this age group. Average age varied across the groups from 65.24 (SD = 12.87) years for EUR to 56.93 (14.94) years for HIS and was lower for migraine cases than controls in all ancestries (p < 0.001).

Genome-wide significant loci

The multi-ancestry meta-analysis (META_C; cases = 87,859, controls = 345,151) across three HARE-defined categories identified 36 genome-wide significant (GWS; p < 5 × 10⁻⁸) loci when accounting for linkage disequilibrium (r² > 0.1) with 40 lead SNPs corresponding to 188 mapped genes. Quantile-quantile plots showed inflation of test statistic (Supplementary Fig. 3-l; GCl = 1.365), and polygenic effects accounted for 85.6% of this inflation based on the LDSC intercept of 1.0634 (0.009). The META_C findings are summarized in the Manhattan plot in Fig. 1A and Supplementary Data 5, which also reports all loci across all other strata. In total, we identified 23 GWS loci in the EUR strata (EUR_C), one in the HIS strata (HIS_C), and two in the AFR (AFR_C) strata. A meta-analysis of men (META_M; 66,083 cases, 325,539 controls) and women (META_W; 21,776 cases, 19,612 controls) revealed 26 and 2 GWS loci, respectively (Supplementary Data 6). In META_M, 3 loci were novel -replicated, and 14 were novel unreplicated. In META_W, two loci were novel unreplicated. There were no GWS loci in the EUR women, AFR men, and HIS women. QQ and Manhattan plots for all strata are presented in Supplementary Fig. 2(a-l) and Supplementary Fig. 3(a-l), respectively. Collectively, we identified 106 GWS loci across all nine GWAS and four meta-analyses, with some loci exhibiting overlap in genomic position while yielding different lead SNPs for various analytic strata. In total, 49 distinct loci across the different strata were identified (Fig. 1B, Supplementary Data 5).

We identified numerous shared loci across sex and ancestry (see Fig. 1B, Supplementary Data 5, Supplementary Data 6), with 11 loci consistently GWS in EUR_M, EUR_C, META_M, and META_C. These “global” loci can be cross-walked to META_C in Supplementary Data 6. For example, locus global_3 (see META_C_10, Supplementary Fig. 4-f) was novel and unreplicated, with lead SNP rs72712556, while locus global 6 (META_C_14, Supplementary Fig. 4-i) lead SNPs varied by strata (Supplementary Data 6).

To prioritize potentially causal SNPs, we ran fine-mapping on the 23 genome-wide significant loci identified in the European population, using the Sum of Single Effects (SuSiE) model [39]. Credible sets were identified for 21 out of 23 loci (sum of posterior inclusion probability>0.95; Supplementary Data 7). Credible sets contained a median of 9 SNPs (range 1–116). SuSiE identified one locus, EUR_11, with 8 separate credible sets. The presence of multiple credible sets within the same locus indicates allelic heterogeneity and the presence of distinct, statistically independent association signals. Many loci had large credible sets, indicating a diffuse signal that is not well localized. However, one locus (EUR_19), had a single SNP (rs11172113) in the credible set, with PIP = 1.0, indicating strong support for causality. This SNP is in an enhancer region of LRP1, a gene that has been previously associated with migraine [23].

Replication of MVP-migraine genome-wide significant loci

We compared our findings to seven previous migraine studies (from non-overlapping cohorts, available through the GWAS catalog) to classify GWS loci identified in this study as known or novel [21, 23, 25, 26, 28, 30, 40]. GWS loci identified in our study were classified as known if any SNP in the locus around the lead SNP (locus area defined by FUMA’s LD clumping algorithm) was found to be genome-wide significant in a previous migraine study. Otherwise, the GWS loci in our study were classified as novel. Supplementary Data 6 presents all loci, genes, and replication status by strata. Of the 49 GWS loci in the MVP cross-strata analysis (Fig. 1B), 13 had prior associations with migraine, and 36 were new to this study (Fig. 1A; Supplementary Data 6). We used a previous large-scale GWA meta-analysis combining GERA and UKB data [21] to replicate the novel loci. Among the 36 new loci, seven loci contained at least one SNP that was nominally significant (after Bonferroni correction for 36 loci tested: p < 0.05/36). We label these seven loci ‘novel-replicated’. A further 23 loci contained at least one SNP with p < 0.05, but did not remain significant after Bonferroni correction (Supplementary Data 6) [21]. Finally, seven GWS loci were novel to the current study and did not replicate in the GERA-UKBB cohort (all p > 0.05). All seven loci had small (ORs near one) and non-significant effects, with three loci showing concordant effect direction and four showing opposing direction. This suggests that the differences may be due to statistical noise, though other sources of heterogeneity are possible. Within the 36 multi-ancestry meta-analyses (META_C) GWS loci, 12 were known (previous GWAS migraine associations), seven had at least one SNP replicate after Bonferroni correction in the GERA-UKBB cohort, and 17 were novel to this study (Fig. 1A). Locus Zoom plots are provided for each novel-replicated and novel-unreplicated SNPs (Supplementary Fig. 4a-kk) across all strata and are described in more detail below.

Given the potential differences between the study population (Veterans, mainly men), and previous migraine GWAS, we sought to understand which previously identified migraine loci were replicated in our data. We identified all previously identified SNPs associated with migraine in the GWAS catalog and cross-referenced them with the results from the current study (using meta-combined results). There were 180 SNPs associated with migraine in the GWAS catalog which were not overlapping a GWS locus from the current study. Of these 180 SNPs, 24 replicated after Bonferroni correction (p < 0.05/180), and an additional 65 SNPs were nominally significant in the study data (p < 0.05) (Supplementary Data 8). In particular, rs1003194 was Bonferroni-significant in our study (p = 0.0015), and was highlighted in a recent large migraine GWAS, mapped to CALCA/B, and proposed as a target of new migraine therapeutics [23].

Genes and pathways mapped to novel and known loci

In the MVP cross-strata analysis incorporating all individual GWAS and meta-analysis results, we identified 283 genes associated with migraine (see Supplementary Data 9 for 188 genes from the META_C results). Of these, 76 genes mapped to the 13 known loci, 61genes mapped to the seven novel-replicated loci, and 146 genes mapped to the 29 novel-unreplicated loci. Among the 13 known loci were well-documented migraine genes (Supplementary Data 10), including LRP1, TRPM8, PRDM16, ASTN2, and PHACTR1, all found to be disease-associated genes in the DISEASES database migraine gene set [52] and identified in previous migraine GWAS [25, 30]. Among the genes mapped to novel replicated loci were CELF4, CAV2, and FAM167A (Supplementary Data 6 and 10). Notably, seven genes mapped to novel loci had been previously associated with migraine in GWAS, including LINGO2 and HTRA1. In these cases, the loci we identified were novel, but the mapped gene was not novel. We also note that several genes we mapped to known migraine loci had not been previously linked with migraine, including ABCC3 and PARVB (Supplementary Data 10). These differences may reflect differences in gene-mapping strategies.

Functional enrichment analysis of the mapped genes revealed significant association with gene sets from human disease databases, including lipidosis, triglycerides, and obesity (DisGeNet), the mammalian phenotype ontology, including decreasing levels of triglycerides and glucose (MPO), cholesterol metabolic process (GO), and other GWAS traits from UKBB and GWAS catalog, including irritability, neuroticism, and fatty acid levels (Supplementary Data 11, Fig. 2A).

Gene tissue expression and pathway analysis

Tissue-specific enrichment analysis of the multi-ancestry meta-analysis (META_C) using MAGMA revealed significant associations with brain tissues, including the frontal cortex, cortex, anterior cingulate cortex, nucleus accumbens, basal ganglia, and the cerebellar hemisphere (Fig. 2B; Supplementary Data 12). MAGMA tissue results are presented in Supplementary Data 12 for META_C and Supplementary Data 13 for META_EUR, with figures for all strata presented in Supplementary Fig. 5 a-l. Consistent with the MAGMA pathway results, DEPICT (Supplementary Data 14), an alternative method for identifying enriched pathways and gene sets, identified 17 brain regions as significantly enriched, largely consistent with the MAGMA results. Enriched brain regions included the cerebrum/cerebral cortex, parietal lobe, telencephalon, and temporal lobe.

MAGMA identified one significant gene set (at FDR adjusted threshold p = 0.05/19054 gene sets = 2.62 × 10⁻⁶) associated with cytotoxic T lymphocyte function and immune regulation (2.32 × 10⁻⁰⁸, Supplementary Data 15). While no pathway met significance criteria following FDR adjustment (p = 0.05/14465 gene sets = 3.46 × 10⁻⁶), the most enriched DEPICT gene set term was the mammalian phenotype ontology term “increased brain weight” (p = 2.4 × 10⁻⁵; Supplementary Data 14). Notably, DEPICT does not consider the direction of effect.

It is noteworthy that none of the arterial tissues achieved nominal significance in MAGMA (Fig. 2B) and DEPICT (Supplementary Data 14) showed no vascular pathways (including NOTCH signaling subnetworks). This finding contrasts with earlier migraine GWAS [23, 30], which reported strong associations with vascular pathways such as NOTCH.

SNP-based heritability

The SNP-based heritability (h²_SNP) of MIG-MVP was estimated using LDSC. For the liability scale h²_SNP, we used a prevalence of 8.2% for men and 30.1% for women, with a prorated combined lifetime prevalence of 10.0%. Liability scale h2_SNP (Supplementary Data 16) was estimated at h²_SNP = 0.098 (SE = 0.005) for the combined EUR sample, h²_SNP = 0.100 (SE = 0.005) for EUR men, and h2_SNP = 0.155 (SE = 0.031) for EUR women.

Sex-stratified GWAS results

The genetic correlation between EUR_M and EUR_W for migraine was estimated at r_g = 0.93 (SE = 0.07), indicating high trait similarity across sexes. When evaluating sex-stratified GWAS analyses, we observed six GWS loci specific to META_M (Supplementary Data 5), which may contain sex-specific signatures. Locus META_M_18, (Supplementary Fig. 4-cc) is novel to our study but was nominally replicated in a previous migraine GWAS [21]. The lead SNP in this locus falls in an exonic region of the gene AHNAK. Multiple SNPs in this locus have high CADD scores ( > 20), indicating the potential for impact on gene function. Other novel loci specific to men included the unreplicated META_M_5, (Supplementary Fig. 4-y) and associated with KIF3C, RAB10, EPT1, DRC1, MAPRE3, KHK, SNX17 associated with diabetes, and C-reactive protein function. In addition, unreplicated META_M_24, (Supplementary Fig. 4-ee), is associated with LAMP5 and PAK7, both associated with depression.

The women sample was underpowered (cases = 21,776, controls = 19,612), comprising only 9% of the MVP population and revealing only two novel GWS loci. Both loci were novel and unreplicated (Supplementary Data 6). META_W_2 (Supplementary Fig. 4-ff), is associated with LINGO2, a gene previously associated with migraine [21, 23]. META_W_1 (Supplementary Fig. 4-gg) is associated with CMTM1 and CMTM3, genes not previously associated with migraine. In addition, loci associated with migraine in our and previous GWAS trended towards significance in the meta-analysis of women in the MVP (Supplementary Data 5; META_W [e.g., rs11172113, p = 1.42E-06 on LRP1]).

We evaluated the sex-stratified results of four SNPs reported in the GERA-UKB migraine GWAS [53]. Choquet et al. reported rs1047891(CPS1), rs11718509 (PBRM1), and rs10150336 (SLC25A21), rs7858153 (ASTN2) as significantly associated with migraine in women (P  <  5.0  ×  10⁻⁸) but not men. We evaluated these variants within the MVP men and women cohorts. One of these variants trended toward significance in the MVP META_W sample (rs11718509; p = 0.009) and two towards significance in META_M (rs7858153, p = 0.002; rs1047891, p = 0.059).

Ancestry-stratified results

One locus was specific to AFR Women (AFR_W_1, Supplementary Fig. 4-jj) with lead SNP rs2864065 associated with LSAMP linked to neuronal activity within the limbic system and metabolic syndrome and body mass index. A locus in the AFR Combined strata (AFR_C_1, Supplementary Fig. 4-ii) coding for IRX1 that may be involved in the development of the nervous system. A locus in the HIS men strata (HIS_M_1, Supplementary Fig. 4-ii) was associated with seven genes, including RNF4, previously associated with back pain, TNIP2, associated with inflammation, and GRK4 involved in G protein-coupled receptor signaling and vascular regulation. None of the non-EUR lead variants showed nominal significance in any of the EUR strata (all p > 0.05).

Genetic correlations with psychiatric disorders and brain regions

We evaluated LDSC genetic correlations between the EUR cohorts and meta-analysis with summary statistics from the PGC, TBI, and brain structure imaging regions from ENIGMA. Within the psychiatric domain (Supplementary Data 17, Fig. 3A), significant correlations ranged from r_g = 0.27 (SE = 0.05) for Tourette Syndrome to r_g = 0.76 (SE = 0.04) for TBI for the MVP EUR_M sample. Within the brain imaging data (Supplementary Data 17, Fig. 3A), significant correlations ranged from r_g = −0.28 (SE = 0.08) between EUR_W and brainstem volume to r_g = 0.09 (SE = 0.04) in caudate nucleus volume for the EUR_M sample. Significant differences in magnitude between genetic correlations for men and women (Fig. 3A) were observed for MDD (z = 2.34, p = 0.02), PTSD (z = 4.11, p < 0.001), brainstem volume (r_g= 0.20, SE = 0.09), and globus pallidus volume (r_g= 0.26, SE = 0.12).

To evaluate the correlation between specific cerebral regions and migraine while accounting for ICV, we simultaneously estimated the relationship between volumes of specific regions and migraine using GSEM path models. This approach incorporated the correlation between the brain region and ICV and the direct effect of ICV. Most cerebral regions did not exhibit a statistically significant association with migraine after adjusting for ICV (Supplementary Data 18, Supplementary Fig. 6). However, a notable positive association with migraine was observed for the caudate nucleus (β = 0.107, SE = 0.043, p = 0.012), suggesting that larger CN volume may correlate with an increased risk of migraine.

Genomic structural equation modeling

To further explore the shared genetic architecture of migraine and common Veteran migraine comorbidities of TBI, PTSD, and MDD, we modeled the associations simultaneously using a GSEM path model (Fig. 3B) in EUR_C based on LDSC-derived correlations. This approach included both the genetic associations of TBI, PTSD, and MDD with migraine while accounting for the genetic correlations among these traits. Our findings suggest that MDD and PTSD show marginally significant (β = 0.10, SE = 0.06, p = 0.07) and non-significant (β = −0.07, SE = 0.14, p = 0.644) associations with migraine, respectively, while TBI, even after accounting for MDD and PTSD (which are genetically correlated, r_g = 0.64, SE = 0.03), maintains a strong association (β = 0.75, SE = 0.12, p < 0.001). Despite the strong genetic associations between migraine and TBI, PTSD, and MDD (Fig. 3A), only the association with TBI remains influential when modeling these conditions simultaneously.

We conducted EFA and CFA on EUR_C migraine and psychiatric disorders with strong genetic correlations (Fig. 3A). Parallel analysis (Supplementary Fig. 7) indicated the presence of one to two factors, and we evaluated one-, two-, and three-factor EFA factor loadings, comparing the corresponding CFA models for the EUR_C migraine sample. EFA factor loadings for two and three-factor models are presented in Supplementary Data 19A, and CFA model fit statistics in Supplementary Data 20A. The one-factor CFA model provided an adequate fit to the data (df = 14, AIC = 75.818, SRMR = 0.1203), but the two-factor model provided the best fit with a lower AIC and improved SRMR (df = 12, AIC = 43.184, SRMR = 0.0471). The three-factor model introduced an additional not-identified factor with a single trait (PTSD) and only a slight improvement in SRMR but an increase in AIC (df = 11, AIC = 44.032, SRMR = 0.038). Consequently, the two correlated factor model provided the most parsimonious and best-fitting solution. (Fig. 3C) The first latent factor (F1) was associated with migraine and neuropsychiatric disorders (ADHD, PTSD, and TBI), while the second factor was associated with anxiety/stress, mood, and alcohol use disorders (ALCH, MDD, PTSD, and ANX). The two latent factors were correlated at r = 0.61 (SE = 0.06), showing a moderate positive relationship between them. The model explained 53% of the genetic variance in EUR_C migraine. As a sensitivity analysis, we repeated the EFA and CFA using the EUR_M and EUR_W GWAS results, both of which indicated that the two-factor model, with similar factor loadings (Supplementary Data 19B-C), provided the best fit (Supplementary Data 20B-C).

Genetic overlap of migraine with other traits

We conducted an unbiased LDSC screen of genetic correlations with 844 publicly available GWAS using the Complex Trait Genetics Virtual Lab (CTG-VL) [46]. These data include phenotypes from the UK Biobank, GIANT consortium, Psychiatric Genomics Consortium (PGC), FinnGen, and CHARGE, identifying 305 significant genetic correlations with migraine (Bonferroni-corrected p < 5.92 × 10^−5; Supplementary Data 21). Notable correlations were observed between migraine and multisite chronic pain (r_g= 0.71, SE = 0.02), absence of pain (r_g = −0.67, SE = 0.03), back pain (r_g = 0.54, SE = 0.03), headache in the last month (rg = 0.71, SE = 0.04), hip pain (r_g = 0.51, SE = 0.04), as well as medication use including paracetamol (r_g = 0.65, SE = 0.04; r_g = 0.68, SE = 0.04), codeine-acetaminophen (r_g = 0.72, SE = 0.06), and omeprazole (r_g = 0.67, SE = 0.05). Mental health conditions included MDD (r_g = 0.53, SE = 0.03) and ADHD (r_g = 0.47, SE = 0.03). We observed modest but significant associations with triglycerides (r_g = 0.153, SE = 0.033) and circulating calcium (r_g = 0.091, SE = 0.021), as well as a negative correlation with HDL cholesterol (r_g = −0.156, SE = 0.036), while the correlations with total and LDL cholesterol were not significant. Additionally, our results showed positive genetic correlations between migraine and cardiovascular disease, including stroke (r_g = 0.544, SE = 0.124), myocardial infarction (r_g = 0.229, SE = 0.045), and angina (r_g = 0.327, SE = 0.042). Finally, general health indicators including long-standing illness or disability (r_g = 0.52, SE = 0.03), and impairment factors, including the inability to work due to health-related issues (r_g = 0.62, SE = 0.04) and financial difficulties arising from illness (r_g = 0.55, SE = 0.03), were also associated.

We compared the genetic correlation from the unbiased LDSC screen between the MVP migraine cohort and the GERA-UKBB cohort. The genetic correlation between MVP EUR_C and the GERA-UKB migraine meta-analysis [53] was r_g = 0.76 (SE = 0.04). As expected, the MVP EUR_C and GERA-UKB showed similar associations related to headache pain (Fig. 4B; Supplementary Data 21). MVP EUR_C also showed associations with multisite chronic pain (r_g = 0.563, 0.055, p = 8.68 × 10⁻²⁵) and lower back pain (r_g = 0.524, 0.060, p = 2.22 × 10⁻¹⁸) while GERA-UKB did not. We discerned differences unique to the MVP cohort, encompassing socioeconomic factors, musculoskeletal characteristics, and pain-related traits (Fig. 4A and B). Consistent with the genetic correlations with PGC data (Supplementary Data 17), the r_g data for GERA-UKB (Supplementary Data 17), indicates minimal significant correlations with psychiatric conditions from CTG-VL.

Mendelian randomization

Next, we evaluated the potential causality between traits with nine significant genetic correlations with MVP migraine (Fig. 3A; TBI, ADHD, TS, anxiety, PTSD, alcohol dependence, depression, ICV, surface area) using the CAUSE software for MR. We compared null, shared, and causal models using expected log predictive density (ELPD) differences to test for statistical significance, set at p < 0.01. Of the traits tested, none showed evidence of a significant causal relationship on MVP-migraine (Supplementary Data 22), with all p > 0.05. TBI sharing model approached statistical significance when compared to the null model (DELPD = −10.50, SE_DELPD = 4.31, z = −2.45, p = 0.007). Conversely, there was no evidence that MVP migraine caused any of the tested traits (DELPD p < 0.05). The causal model did not provide a significantly better fit than shared polygenicity models, thus indicating no evidence that MVP migraine caused any of the tested traits.

Polygenicity identified in MVP migraine, relative to previous migraine GWAS

We extended our comparison analyses between MVP-migraine and GERA-UKB-migraine genetic architecture using MiXeR, a Gaussian mixture modeling approach to estimate polygenicity. Quantification of the polygenic overlap between MVP-migraine and GERA-UKB-migraine revealed that MVP shared all ~1900 GERA-UKB loci, but there were ~7300 loci predicted to be unique to MVP-migraine (Fig. 4C; Supplementary Data 23, Supplemental Fig. 8). The large number of MVP-migraine unique loci leads us to conclude that MVP migraine is much more polygenic than GERA-UKB despite the high genetic correlation (r_g = 0.76, SE = 0.04). The additional polygenicity of MVP-migraine could reflect features specific to the MVP, such as the predominantly men Veteran population. The polygenicity comparison of MVP migraine Men_C and Women_C was inconclusive due to the low sample size in the MVP women population.

Drug-class and drug-set enrichment analyses

We evaluated each gene associated with any GWS loci for its presence within a comprehensive database of known drug targets (OpenTargets Platform) [54]. We identified 76 drugs that are known to target at least one gene identified in our study based on FUMA prioritization (Supplementary Data 24). Seven genes from loci previously associated with migraine in prior studies were linked to ten drugs, including monoclonal antibody drugs (tanezumab, fasinumab, fulranumab) that target nerve growth factor (NGF) function to mediate pain signaling, as well as metformin hydrochloride (which targets NDUFAF4), a well-established treatment for type 2 diabetes with possible neuroprotective effects and previous unsuccessful clinical trial (NCT02593097) for migraine, in addition to menthol, which targets TRPM8. Six novel loci from our study were linked to several medications associated with established migraine pathways. These include TLR4 antagonists (eritoran, resatorvid), which play a role in neuroinflammation targeting TLR4; p38 mitogen-activated protein kinase inhibitors (losmapimod, doramapimod, neflamapimod) targeting MAPK14; and peroxisome proliferator-activated receptor agonists (bezafibrate, seladelpar, lanifibranor) targeting PPARD. Nine novel loci mapped to additional drugs, including calcium channel modulators used in migraine treatment, GABA analogs used for pain management, and immunomodulators.

19 December 2025

• ECB streamlines how it assesses standardised capital and securitisation operations from January 2026
• Approval time reduced to two weeks, down from the current three months
• Move allows supervisors to focus on more complex assessments
• ECB to check that banks do not overly rely on capital benefits from significant risk transfer securitisations

The European Central Bank (ECB) has streamlined how it assesses banks’ capital and securitisation operations. As of January 2026, banks will be able to receive a faster response from the ECB when they seek to reduce their capital by repurchasing shares or other capital instruments, or to reduce their capital requirements after a significant risk transfer.

The faster processes will apply to standardised operations. Despite the faster process, all global standards and European regulation still apply in full. The normal procedure, which entails a more detailed assessment of the operation and of the risks involved, will still apply to operations ineligible for fast-track processing.

The two new fast-track processes aim to cut approval times to two weeks, down from three months currently. In the broader context of streamlining supervision and making it more efficient and effective, these fast-track processes will save time on routine operations and allow supervisors to focus on more complex assessments.

Capital fast track

EU banking rules require banks to get ECB approval before they buy back shares or repurchase other capital instruments, as these operations reduce banks’ capacity to absorb losses. The ECB therefore checks that these transactions comply with all applicable regulatory requirements. The ECB is streamlining the approval process to become more efficient, while maintaining the same prudential standards as before.

A bank’s plan to repurchase capital instruments other than shares may qualify for the fast-track process if its impact on the bank’s capital ratio is below 100 basis points, and if the bank’s capital is estimated to continue exceeding capital requirements and guidance for at least three years.^[1]

For share buybacks to be eligible for fast-track processing, additional conditions apply: the bank must be scored medium or low-risk in its capital adequacy assessment, it must retain a sufficient portion of its profits, and it must demonstrate the ability to meet its capital requirements and guidance under severe financial stress.

The ECB is also launching a streamlined process for banks to submit their applications to reduce own funds. Banks will immediately know whether their application is complete and in principle eligible for fast-track processing. The application will then be reviewed and assessed by the Joint Supervisory Team, which retains the discretion to request additional information, if needed.

Securitisation fast track

When banks transfer risks through securitisation, EU banking rules provide that capital requirements on the securitised exposures may be reduced only if the ECB acknowledges that the securitisation transfers a significant amount of risk to third parties – this is known as a “significant risk transfer” (SRT).

The SRT fast-track process applies to standardised securitisations, namely when the securitised portfolio is performing, not concentrated, and does not contain more than 20% of leveraged loans; when the impact on the bank’s capital ratios of the significant risk transfer is lower than 25 basis points; and when standardised contractual early termination clauses are used.

To ensure that the SRT process does not lead to undue risk-taking and a weakening of resilience, it will be accompanied by increased scrutiny of micro- and macroprudential risks. In case the use of securitisation raises prudential concerns, the ECB will take adequate measures. More specifically, supervision will focus on complex cases and on bank-level assessments of securitisations activities. The ECB will continue to assess the adequacy of banks’ internal governance, risk management and capital management frameworks, including stress testing, to prevent them from overly relying on the capital benefits generated by SRT securitisations and from elevated rollover risk created by large-scale use of synthetic securitisations. Furthermore, the granular data collected via the fast-track template will contribute to enhanced monitoring of the securitisations market by the ECB in its supervisory and macroprudential functions.

Further details about the securitisation fast track, including the full set of criteria for securitisations to qualify for the fast-track process, are provided in Section 3.3 of the Guide on the notification of significant risk transfer and implicit support for securitisations.

For media queries, please contact François Peyratout, tel.: +49 172 8632 119.

Commenting on today’s ONS retail sales figures, Oliver Vernon-Harcourt, head of retail at Deloitte, said:

“A second consecutive month of decline in retail sales will be a big disappointment for retailers, particularly with Black Friday and early Christmas shopping captured in these figures. It could be that some consumers opted to delay spending decisions until after the Budget, but the rise in volumes across department stores, clothing and footwear, and household goods suggests that the start of the discounting period could have enticed consumers into purchases at the end of the month. Meanwhile, supermarket sales declining for a fourth consecutive month is a sign of the times, with food inflation still having a significant impact on consumer budgets and purchasing decisions.

“A successful December, to round off a subdued Golden Quarter, will be at the top of retailers’ wish lists. The bigger picture of retail in 2025 showed a prolonged pressure from weak consumer confidence and challenging economic conditions. With interest rates and inflation continuing to fall, retailers will hope this can spur a rebound in consumer spending. On the other hand, if economic uncertainty continues, the apprehension among both consumers and retailers may continue in 2026.”

ENDS

– Advertisement –

ISLAMABAD, Dec 17 (APP):The benchmark KSE-100 Index of the Pakistan Stock Exchange (PSX) on Wednesday closed bearish, shedding 133.44 points, a negative change of 0.08 percent, to settle at 170,313.86 points compared to 170,447.30 points on the previous trading day, according to Pakistan Stock Exchange data.

During the session, the ready market witnessed a trading volume of 1,068.512 million shares with a traded value of Rs51.799 billion, against 1,176.638 million shares valuing Rs53.473 billion in the previous session. Market capitalization increased to Rs19.377 trillion from Rs19.370 trillion a day earlier.

Out of 483 active companies in the ready market, 178 advanced, 255 declined, while 50 remained unchanged.

Bank of Punjab topped the volume chart with 90.619 million shares, followed by Hascol Petroleum with 83.922 million shares and TPL Corporation Limited with 53.151 million shares.

The top gainers included Nestle Pakistan Limited, which rose by Rs49.03 to close at Rs8,000.00, and Hafiz Limited, increasing by Rs47.18 to settle at Rs590.07. On the losing side, PIA Holding Company Limited-B declined by Rs130.00 to close at Rs23,070.00, while Unilever Pakistan Foods Limited fell by Rs125.00 to close at Rs29,000.00.

In the futures market, turnover stood at 222.330 million shares with a traded value of Rs12.532 billion, compared to 271.071 million shares worth Rs13.531 billion in the previous session.

Out of 316 futures-market companies, 113 recorded gains, 199 declined, while four remained unchanged. Among futures contracts, BOP-Dec led with 32.678 million shares, followed by PIAHCLA-Dec with 25.874 million shares and SSGC-DecB with 18.075 million shares.

In the futures market, the top gainers included GHNI-DecB, which rose by Rs17.11 to close at Rs839.20, and UBL-DecB, increasing by Rs12.19 to settle at Rs403.13. On the losing side, HUBC-Jan declined by Rs19.51 to close at Rs225.50, while SAZEW-DecB fell by Rs19.50 to close at Rs1,664.03.