Introduction
A recent paper titled “Genome-wide Association Study of Long COVID” (Lammi et al., 2023) has been published on the preprint server medRxiv. This groundbreaking study conducted a genome-wide association study (GWAS) on long COVID, analyzing data from approximately 6.5 thousand long COVID cases and from over 1 million population controls across 24 studies spanning 16 countries. It stands as the first and largest GWAS of long COVID to date. The study identified one independent association signal represented by a single nucleotide polymorphism (SNP) known as rs9367106, which exhibited an odds ratio of 1.65. This SNP was found to be mapped to the gene FOXP4. This remarkable discovery suggests that individuals infected with SARS-CoV-2 who also carry the rs9367106 SNP have a 1.65-fold increased likelihood of experiencing long COVID (Lammi et al., 2023).
The identified long COVID marker, rs9367106, gains robust support due to its potential functional impact on the expression of FOXP4. In their study, the authors used a proxy SNP (rs12660421) closely linked to rs9367106 to represent the significant association between rs9367106 and elevated FOXP4 expression in lung tissue according to data from the Genotype-Tissue Expression database (GTEx) (GTEx-Consortium, 2020). Furthermore, single-cell expression data from lung tissues provided additional confirmation of the potential roles of FOXP4 in immune and alveolar cells within the lung. These pieces of evidence align coherently with our understanding, given that the lung serves as the primary target for SARS-CoV-2 infection. The presence of severe inflammation in the lung and the subsequent chronic lung damage resulting from it would be likely to contribute significantly to the development of long COVID.
However, because FOXP4 has been reported to be linked to lung cancer (Yang et al., 2015), it remains uncertain whether the long COVID risk SNP is also connected to lung cancer. The present study endeavors to explore the possible relationship between rs9367106 and long COVID, severe COVID-19, and lung cancer. Furthermore, it aims to ascertain whether the association between rs9367106 and both long COVID and severe COVID-19 is influenced by its connection to lung cancer.
Materials and methods
Summary statistics of long COVID GWAS
Lammi et al. (2023) have generously provided download links for the summary statistics of four distinct long COVID GWASs. These GWASs were conducted with slightly varying case-control designs, with the criterion of retaining only SNPs having a minor allele frequency of ≥1%. The specifics of these four long COVID GWASs, along with their respective download links, are outlined below, with only the long COVID GWAS that compared strict cases of long COVID after test-verified SARS-CoV-2 infection to general population controls showing only one independent genome-wide significant hit (P < 5.0 × 10−8):
Strict cases of long COVID after test-verified SARS-CoV-2 infection (n = 3,018) vs. general population controls (n = 994,582), with its download link provided as (https://my.locuszoom.org/gwas/192226/?token=09a18cf9138243db9cdf79ff6930fdf8).
Broad long COVID cases identified as infected by any SARS-CoV-2 virus (n = 6,450) vs. general population controls (n = 1,093,995), with its download link provided as (https://my.locuszoom.org/gwas/826733/?token=c7274597af504bf3811de6d742921bc8).
Strict long COVID cases defined (n = 2,975) vs. strict controls restricted to individuals who were infected by SARS-CoV-2 but were not diagnosed with long COVID (n = 37,935): with its download link provided as (https://my.locuszoom.org/gwas/793752/?token=0dc986619af14b6e8a564c580d3220b4).
Broad long COVID cases defined (n = 6,407) vs. strict controls as defined in (n = 46,208): with its download link provided as (https://my.locuszoom.org/gwas/91854/?token=723e672edf13478e817ca44b56c0c068).
Summary statistics of severe COVID-19 GWAS from the host genetics initiative (HGI)
Our study also incorporated five severe COVID-19 GWASs, encompassing mixed populations and ancestry-specific populations. These severe COVID-19 GWASs are accessible for download from the HGI website (COVID-19 GWAS meta-analyses round 7, release date: 8 April 2022). These GWASs investigated severe respiratory-confirmed COVID-19 in comparison to the general population across mixed and ancestry-specific subpopulations, including European (EUR), East Asian (EAS), South Asian (SAS), and African (AFR) populations. Only SNPs that met the criteria of a minor allele frequency (MAF) ≥ 1% and an imputation score > 0.6 were selected for subsequent analysis.
The severe COVID-19 GWASs involved in this study feature the following sample sizes and associated download links:
A2_ALL for mixed populations (cases = 18,152; controls = 1,145,546), with the following download link: https://storage.googleapis.com/covid19-hg-public/freeze_7/results/20220403/main/sumstats/COVID19_HGI_A2_ALL_leave_23andme_20220403_GRCh37.tsv.gz.
A2_AFR for populations with African ancestry (cases = 628; controls = 12,568), with the following download link: https://storage.googleapis.com/covid19-hg-public/freeze_7/results/20220403/pop_spec/sumstats/COVID19_HGI_A2_ALL_afr_leave23andme_20220403_GRCh37.tsv.gz.
A2_EAS for populations with East Asian ancestry (cases = 794; controls = 4,862), with the following download link: https://storage.googleapis.com/covid19-hg-public/freeze_7/results/20220403/pop_spec/sumstats/COVID19_HGI_A2_ALL_eas_leave23andme_20220403_GRCh37.tsv.gz.
A2_EUR for populations with European ancestry (cases = 13,769; controls = 1,072,442), with the following download link: https://storage.googleapis.com/covid19-hg-public/freeze_7/results/20220403/pop_spec/sumstats/COVID19_HGI_A2_ALL_eur_leave23andme_20220403_GRCh37.tsv.gz.
A2_SAS for populations with South Asian ancestry (cases = 980; controls = 47,696), with the following download link: https://storage.googleapis.com/covid19-hg-public/freeze_7/results/20220403/pop_spec/sumstats/COVID19_HGI_A2_ALL_sas_leave23andme_20220403_GRCh37.tsv.gz.
Summary statistics of self-reported lung cancer GWAS from UK Biobank (UKB)
The self-reported lung cancer GWAS data from UK Biobank was obtained to perform a comparative analysis of association signals around the FOXP4 locus with both the long COVID GWAS of mixed ancestries and the severe COVID-19 GWAS of European ancestry. The self-reported lung cancer GWAS was generously provided by Neale’s lab (GWAS round 2: http://www.nealelab.is/uk-biobank), and the GWAS dataset consists of 360,938 controls and 203 cases, which is publicly accessible via the following link:
Analysis procedures and related scripts
Following the acquisition of these long COVID GWAS summary statistics, a custom Perl script was employed to merge the association signals of each SNP by rsID across the four long COVID GWAS summary statistics. This procedure resulted in a final table where each row is specific to a single SNP, with corresponding columns housing the summary statistics extracted from the four GWASs. To reduce file size, this merged table was subsequently compressed into a “gz” file format using 7zip. It was then uploaded to the workspace of the freely accessible online software SAS OnDemand for Academics. Within this platform, the integrated data underwent comprehensive analysis alongside summary statistics from severe COVID-19 GWASs, which were downloaded from the HGI website.
To facilitate the analysis and visualization of association signals related to both long COVID and severe COVID-19 around the FOXP4 gene, a custom SAS script leveraging the SAS package COVID19_GWAS_Analyzer was developed. This script automates the extraction of FOXP4 SNPs from the aforementioned GWASs and visualizes their association signals in a localized Manhattan plot.
The COVID19_GWAS_Analyzer package is openly accessible on GitHub (https://github.com/chengzhongshan/COVID19_GWAS_Analyzer), where the corresponding SAS codes for extracting FOXP4 SNPs from these GWASs and generating the local Manhattan plot can be found at this link (https://github.com/chengzhongshan/COVID19_GWAS_Analyzer/tree/main/LongCOVID_data_and_scripts/Evaluate_FOXP4_SNPs_with_both_long_COVID_and_severe_COVID.sas).
Details for the analysis process were divided into four distinct steps. First, the merged long COVID GWASs were screened for association signals of SNPs within the genomic region chr6:41000000–42000000 (hg38), where the FOXP4 gene resides. Second, the severe COVID-19 GWAS summary statistics for mixed populations and ancestry-stratified populations were downloaded and queried for SNPs in the same genomic region as FOXP4. Third, the association signals of these SNPs from both long COVID GWASs and severe COVID-19 GWASs were merged using SNP rsIDs. The resulting table was employed to visualize the association signals of these SNPs in a local Manhattan plot via the COVID19_GWAS_Analyzer. Finally, we employed a similar approach to integrate GWAS signals related to self-reported lung cancer around the FOXP4 loci with those of long COVID and severe COVID-19 within the European population. We conducted a specific examination of the association between the target SNP, rs9367106, and self-reported lung cancer. Additionally, we identified 20 SNPs displaying suggestive association signals (P < 5 × 10−4) in the vicinity of the FOXP4 locus in relation to self-reported lung cancer. These selected SNPs underwent expression quantitative trait locus (eQTL) analysis using the GTEx API (https://gtexportal.org/api/v2/redoc#tag/GTEx-Portal-API-Info), and COVID19_GWAS_Analyzer implemented a SAS Macro to facilitate the automation of the eQTL analysis.
All the analyses described above were executed within the SAS On Demand for Academics. Additionally, the customized Perl script used for merging the association signals of each SNP by rsID across the four long COVID GWAS summary statistics is available on GitHub (https://github.com/chengzhongshan/COVID19_GWAS_Analyzer/tree/main/LongCOVID_data_and_scripts/MergeBigFiles.PL).
Results
rs9367106 of FOXP4 associated with both long COVID and severe COVID-19
The FOXP4 variant, rs9367106, was strongly associated with both long COVID and severe COVID-19 outcomes. Notably, the authors reported a significant association of this SNP with long COVID, as depicted in Figure 1. To have a deep understanding of this finding, we merged long COVID GWAS data with the most recent severe COVID-19 GWAS data shared by the HGI (refer to Materials and Methods). Our analysis affirmed that, across a diverse range of populations, including a mixed population and four major populations stratified by ancestry (European—EUR, African—AFR, East Asian—EAS, and South Asian—SAS), rs9367106 was consistently identified as a genome-wide significant hit (P < 5.0 × 10−8) in the GWAS of severe COVID-19. This association also held true for the mixed population as well as the South Asian—SAS population, with a somewhat weaker association observed in the East Asian—EAS population, nearing genome-wide significance, as illustrated in Figure 1. Furthermore, rs9367106 exhibited a strong linkage disequilibrium with the top SNP associated with severe COVID-19, rs2496644 (with an R2 value of 0.88 in the European population). Interestingly, rs2496644 demonstrated similar association signals across four GWAS datasets, encompassing the long COVID GWAS and the severe COVID-19 GWAS conducted in mixed populations, as well as in EAS and SAS populations. These findings align with the understanding that severe COVID-19 patients are more likely to develop long COVID.
FIGURE 1
The long COVID and severe COVID-19 risk SNP rs9367106 close to the FOXP4 gene does not associate with lung cancer
To explore the potential relationship between rs9367106 and lung cancer, we conducted an analysis using the UK Biobank self-reported lung cancer GWAS dataset, which primarily comprises individuals of European ancestry. Additionally, we compared the association signals around the FOXP4 gene in the self-reported lung cancer GWAS with those from the severe COVID-19 GWAS of European samples and the long COVID GWAS, which includes individuals of mixed ancestries. As illustrated in Figure 2, the association peaks near rs9367106 observed in both the long COVID GWAS and the severe COVID-19 GWAS were not present in the self-reported lung cancer GWAS. This discrepancy can be attributed mainly to two specific SNPs, along with others in close proximity, which showed no significant association with lung cancer (with Ps > 0.05). However, it is noteworthy that 20 SNPs located within the gene body of FOXP4 displayed suggestive associations with lung cancer (Ps < 5 × 10−4). Interestingly, these SNPs did not achieve nominal significance in the two COVID-related GWAS datasets. This observation strongly suggests that different SNPs may underlie the potential associations of FOXP4 with lung cancer, as well as with long COVID and severe COVID-19. In summary, our findings indicate that FOXP4 is indeed implicated in lung cancer, but the long COVID and severe COVID-19 risk SNP rs9367106 does not exhibit an association with lung cancer.
FIGURE 2
Expression quantitative trait locus (eQTL) analysis for two COVID risk SNPs and 20 SNPs of FOXP4 that are potentially predisposing to lung cancer
To further substantiate our hypothesis regarding the regulatory influence of these two COVID risk SNPs, rs9367106 and rs2496644, along with the 20 SNPs potentially associated with lung cancer, on FOXP4 expression, we conducted an extensive eQTL analysis across 49 GTEx tissues. This analysis involved correlating the genotypes of each SNP with FOXP4 expression in each GTEx tissue. Given the absence of rs9367106 in GTEx, we focused on its highly linked SNP, rs12660421. Our findings revealed that both COVID risk SNPs, rs2496644 and rs12660421, exhibited robust eQTL effects on FOXP4 expression in the lung and brain hippocampus tissues. Notably, the lung tissue displayed the most pronounced association between FOXP4 expression and rs12660421 (as illustrated in Figure 3). In contrast, these 20 SNPs potentially associated with lung cancer did not exhibit eQTL effects in the lung and brain hippocampus tissues. However, these SNPs displayed consistent eQTL associations across various other tissues, including the thyroid, left ventricle of the heart, esophagus mucosa, and cervical region of the spinal cord. In conclusion, the observation of these SNPs displaying distinct eQTL patterns in different tissues suggests that diverse genetic elements within FOXP4 specifically contribute to COVID or lung cancer by influencing FOXP4 expression across different tissues.
FIGURE 3
Discussion
The long COVID GWAS has illuminated a compelling connection between a genetic marker proximal to the FOXP4 gene and the incidence of long COVID. The potential involvement of FOXP4 in the development of long COVID finds support through multiple layers of evidence, including expression quantitative trait locus analysis, single-cell gene expression profiling, and epigenetic investigations. However, two noteworthy limitations merit discussion.
First, distinguishing whether the identified SNP or FOXP4 itself is more linked to severe COVID-19 or long COVID posed a challenge due to the distinct clinical profiles of two COVID phenotypes. Lammi et al. (2023) attempted to address this by employing Mendelian randomization (Davis et al., 2023) to explore the causal relationship between COVID-19 hospitalization and long COVID by excluding overlapping samples used in both GWASs. However, since these samples originated from the same source (i.e., HGI), potential confounding factors might have a similar prevalence in these shared samples. This limitation is underscored by reports of FOXP4 variants being associated with lung cancer (Yang et al., 2015). In our analysis, we identified different FOXP4 regulatory variants that were associated with long COVID/severe COVID-19 or lung cancer but not all of these traits at the same time. One possible explanation for these findings is that other underlying factors, instead of lung cancer, could influence the association between FOXP4 variants and both severe COVID-19 and long COVID. Further association tests considering additional covariates may clarify these associations.
Second, SNP associations with long COVID varied among populations, emphasizing the need for confirmation through further investigation. The SNP (rs9367106) displayed varying minor allele frequencies across populations, with the lowest in Europeans and higher frequencies in Asian populations. It showed nominal significance in only a few sub-cohorts used for the meta-analyses of long COVID GWAS (Lammi et al., 2023), indicating the required validation in diverse populations.
It is essential to emphasize that the relationship between long COVID and other COVID phenotypes is intricately intertwined. Long COVID represents a broad phenotype encompassing persistent post-COVID symptoms, which endure for more than 6 months following SARS-CoV-2 infection (Davis et al., 2023). Researchers are still actively working to define the key symptoms that can accurately characterize long COVID. The current long COVID GWAS did not encompass all the key long COVID symptoms (Davis et al., 2023) from affected patients. Future GWAS efforts may focus on more cohesive long COVID phenotypes or sub-phenotypes.
Despite limitations, Lammi et al. (2023)’s findings represent a significant breakthrough. With the complexity of long COVID symptoms, future genetic studies are expected to uncover more associations related to this condition.
Statements
Author contributions
Y-SL: Investigation, Writing–review and editing. KZ: Conceptualization, Investigation, Resources, Writing–original draft, Writing–review and editing. Z-SC: Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Project administration, Software, Supervision, Validation, Visualization, Writing–original draft, Writing–review and editing.
Funding
The author(s) declare financial support was received for the research, authorship, and/or publication of this article. The current study is supported by the Science and Technology Supporting Plan of the Scientific and Technological Department of Guizhou Province, China [2020]4Y163.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
DaviesN. M.HolmesM. V.Davey SmithG. (2018). Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ362, k601. 10.1136/bmj.k601
2
DavisH. E.McCorkellL.VogelJ. M.TopolE. J. (2023). Long COVID: major findings, mechanisms and recommendations. Nat. Rev. Microbiol.21 (3), 133–146. 10.1038/s41579-022-00846-2
3
Gtex-Consortium (2020). The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science369 (6509), 1318–1330. 10.1126/science.aaz1776
4
LammiV.NakanishiT.JonessS. E.AndrewsP. J.KarjalainenJ.CortésB. (2023). Genome-wide association study of long COVID. medRxiv. 2023.06.29.23292056.
5
ThaweethaiT.JolleyS. E.KarlsonE. W.LevitanE. B.LevyB.McComseyG. A.et al (2023). Development of a definition of postacute sequelae of SARS-CoV-2 infection. JAMA329 (22), 1934–1946. 10.1001/jama.2023.8823
6
YangT.LiH.ThakurA.ChenT.XueJ.LiD.et al (2015). FOXP4 modulates tumor growth and independently associates with miR-138 in non-small cell lung cancer cells. Tumour Biol.36 (10), 8185–8191. 10.1007/s13277-015-3498-8
Summary
Keywords
long COVID, severe COVID-19, SNP, FOXP4, lung cancer, rs9367106, GWAS
Citation
Luo Y-S, Zhang K and Cheng Z-S (2023) Absence of Association between a Long COVID and Severe COVID-19 Risk Variant of FOXP4 and Lung Cancer. Front. Genet. 14:1258829. doi: 10.3389/fgene.2023.1258829
Received
14 July 2023
Accepted
06 October 2023
Published
26 October 2023
Volume
14 - 2023
Edited by
Silviu-Alin Bacanu, Virginia Commonwealth University, United States
Reviewed by
Magdeldin Elgizouli, University of Zurich, Switzerland
Updates
Copyright
© 2023 Luo, Zhang and Cheng.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Ke Zhang, kezhang@gmc.edu.cn; Zhong-Shan Cheng, cheng.zhong.shan@gmail.com
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.