Local genetic correlations between systemic sclerosis and common cancer types

Figures Abstract Introduction The incidence of cancer, encompassing all major types, is significantly higher among patients with systemic sclerosis (SSc) compared to the general population. While previous studies have assessed the global genetic relationships between SSc and various cancers and found no evidence of causality or pleiotropy, these methods average effects across the genome and thus ignore potential opposing directional effects at different loci, thereby missing locus-specific associations. To address this gap, we assessed fine-scaled genetic overlap between SSc and commonly associated cancers, namely breast and lung cancers, as well as hematologic malignancies, by applying local genetic correlation analyses. Methods We assessed the genetic relationship between SSc and cancers that frequently co-occur with SSc: breast cancer and its’ four main molecular subtypes: HER2-enriched-like, luminal A and B-like, and triple-negative breast cancers, as well as lung cancer, lymphocytic leukemia, and non-Hodgkin's lymphoma. Published genome-wide association study statistics were obtained from the GWAS catalog and The Breast Cancer Association Consortium, and data were standardized, quality control filtered, and preprocessed. Global and local genetic correlations were evaluated using linkage disequilibrium score regression and Local Analysis of [co]Variant Annotation software. Gene-based cross-trait meta-analysis, colocalization and fine-mapping approaches were used to assess evidence of pleiotropy across regions identified by local genetic correlation analyses. We implemented visualization of the local genetic correlations and functional enrichment analyses of pleiotropic genes from these local correlations. Results We did not detect any significant global genetic correlation between SSc and the analyzed cancer subtypes. However, we identified 23 significant local bivariate correlations; 21 were with different molecular subtypes of breast cancer. However, only the locus shared between SSc and lung cancer showed strong evidence of pleiotropy. Genes within loci shared with lung cancer were involved in cell communication and signaling, extracellular matrix remodelling and skin morphogenesis. Citation: Patasova K, Che WI, Westerlind H, Diaz-Gallo LM, Holmqvist M (2026) Local genetic correlations between systemic sclerosis and common cancer types. PLoS One 21(5): e0350006. https://doi.org/10.1371/journal.pone.0350006 Editor: Elingarami Sauli, Nelson Mandela African Institute of Science and Technology, TANZANIA, UNITED REPUBLIC OF Received: September 26, 2024; Accepted: May 7, 2026; Published: May 27, 2026 Copyright: © 2026 Patasova et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: All GWAS used in the study analyses were extracted from GWAS catalog and Breast cancer consortium. Statistical analyses code published on Zenodo https://doi.org/10.5281/zenodo.18633624. Funding: The study was funded by Swedish Research Council (grant 2020-02445). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Introduction Systemic sclerosis (SSc) is an immune-mediated disease primarily affecting women. It is characterized by vasculopathy and abnormally high extracellular matrix deposition in skin and other organs, leading to fibrosis [1,2]. Similar to other immune-mediated disorders, patients with SSc are at greater risk of developing cancer compared to the general population. In patients with SSc, cancer remains the leading cause of death unrelated to SSc, accounting for over a third of fatalities [3]. The types of cancer that are more common in SSc than in the general population differ across populations. Still, the risk of lung cancer, breast cancer, and hematological malignancies is consistently reported as elevated in SSc [4]. Additionally, evidence points to an intriguing temporal relationship between the onset of SSc and cancer development. In particular, there seems to be a biphasic incidence of cancer; the first peak of cancer coincides with SSc diagnosis, occurring within five years before or after SSc onset and the second more than ten years later. Interestingly, breast carcinomas tend to develop during the first peak, while gastrointestinal tract cancers predominate at the second peak [5]. The first cancer incidence peak coinciding with SSc diagnosis suggests that some cancer types are part of a paraneoplastic phenomenon, whereby autoimmunity is triggered by a protective antitumor response against malignant cells in a genetically susceptible host [6–8]. Cancer risk also appears to be mediated by SSc-associated autoantibodies, such as anti-RNA polymerase III (anti-POL3) and anti-topoisomerase (anti-Scl-70), which have been associated with increased cancer risk in these patients [9,10]. The temporal clustering and SSc autoantibodies linked to cancer indicate that there are shared and diverse molecular mechanisms involved in SSc-related cancer. However, our understanding of the molecular mechanisms underlying the pathological link is limited. The few Mendelian Randomization (MR) studies systematically examining the genetic overlap between SSc and cancer, have not found evidence supporting causal or pleiotropic associations. For example, Peng et al. reported no causal relationship between SSc and lung cancer [11]. Similarly, Xu and colleagues found no causality between SSc and various breast cancer subtypes in European and East-Asian cohorts [12]. MR analyses examining whether genetically predicted predisposition to immune-mediated diseases could predict risk of non-Hodgkin's lymphoma, did not support causal or pleiotropic associations with SSc [13]. Additionally, one of the few MR studies on SSc and cancer [13], purposefully excluded the Major Histocompatibility Complex (MHC) region due to its complexity. The MHC region, spanning ~3.5–5 Mb on chromosome 6, contains numerous immune-related genes, such as human leukocyte antigen (HLA) alleles, which are involved in both autoimmunity [14] and cancer [15]. By ignoring this region, these studies may miss key pleiotropic signals and biologically pertinent genetic mechanisms shared between SSc and cancer. Overall, MR and genetic correlation methods estimate average effects across the entire genome, which can obscure important signals by averaging out locus-specific associations and masking opposing directional effects at different loci [16], Therefore, methods that assess genetic relationships at a global scale are unable to identify specific regions that may drive associations between SSc and various cancers. To address this gap, our study applied local genetic correlation analyses, examining the genetic overlap of specific loci between SSc and the most common concomitant cancer subtypes, namely breast and lung cancers and hematological malignancies, including non-Hodgkin's lymphoma and lymphocytic leukemia. By focusing on fine-scale genetic overlap, we sought to identify genomic regions of shared susceptibility and shed light on biological pathways shared between SSc and cancer. Materials and methods Data pre-processing Genetic markers associated with SSc were obtained from the largest SSc meta-GWAS [17], which comprised 9,846 patients with SSc and 18,333 healthy controls from 14 cohorts, primarily from Western and Northern European countries and the USA. Full details are outlined elsewhere [17], but in brief, cases were required to fulfill the 1980 American College of Rheumatology [18] or LeRoy and Medsger criteria [19]. Different versions of Illumina and Affymetrix arrays were used for genotyping, and following quality-control procedures, 4,723,365 SNPs were analyzed [17]. We included genetic data on breast [20] and lung cancers [21], lymphocytic leukemia, and non-Hodgkin's lymphoma [22]. Given the high heterogeneity of breast tumors [23], which significantly impacts cancer behavior, we investigate the genetic correlations between SSc and four main molecular subtypes of breast cancer, HER2-enriched-like, luminal A and B-like, including HER2-negative subtype, and triple-negative breast cancers. These GWAS were focused on individuals of European descent and were obtained from the GWAS catalog [24] and Breast Cancer Association Consortium [25]. GWAS on breast, lung cancers, and lymphocytic leukemia/non-Hodgkin's lymphoma were respectively performed using OncoArray [20], custom genome-wide Illumina [21], and different versions of Axiom GWAS arrays [22]. Data standardization and quality control were performed using Bioconductor MungeSumstats R package [26] in R version 4.3.2 [27]. Specifically, summary statistics were aligned to the reference genome (1000 genomes Phase2 Reference Genome Sequence hs37d55; ‘BSgenome.Hsapiens.1000genomes.hs37d5’ R package [28]), and alleles were flipped to match the reference file (‘SNPlocs.Hsapiens.dbSNP155.GRCh37’ R package [29]). SNPs absent from the reference file and variants with betas and p-values equal to zero were excluded. Where necessary, standard errors were calculated from betas and p-values using impute_se command. The pipeline of the study analyses is shown in S1 Fig. Table 1 presents information on SSc and cancer GWAS summary statistics after standardization and quality control steps. The column “Traits (Reference)” lists the names of analyzed traits with a study reference. The total number of genome-wide association study (GWAS) participants, cases and controls is displayed in the fields “Total sample size,” “N cases,” and “N controls,” respectively. “N markers in GWAS” and “N markers after QC” are the number of genetic variants in GWAS summary statistics and after quality control (QC), respectively. Statistical analyses Global genetic correlations To calculate the total genetic correlation between SSc and breast cancer, lung cancer, and hematological malignancies, we used an approximation method implemented in the linkage disequilibrium score regression (LDSC) software where LD scores are generated from the 1000 Genomes Project Phase 3 European population panel [30–32]. In LDSC analyses, summary statistics from GWAS were additionally processed with HapMap Project Phase 3 SNPs [33] using munge_sumstats.py script [34] Specifically, variants with MAF < 1% and SNPs within the extended MHC locus (chromosome 6: 20,000,000–40,000,000) were excluded. Genetic correlations with p-value < 0.05 were considered statistically significant. Local genetic correlations Specific loci contribution was assessed through pairwise local genetic correlations across numerous loci, using the Local Analysis of [co]Variant Annotation (LAVA) software [16]. LAVA was included as one of the main methods in our study due to its robust capabilities in assessing local genetic correlations. Traditional genetic correlation analyses methods such as LDSC regression often focus on global correlations, which can obscure important local variations. LAVA addresses this limitation by analyzing genetic correlations at specific genomic loci, providing a more comprehensive understanding of the genetic architecture underlying complex traits and providing gene candidates for future functional studies [16]. Moreover, the complex MHC region, which is very important for autoimmune diseases, can be included in LAVA analyses, whereas in global correlation analyses, it is often excluded. LAVA analyses were performed on 2,495 loci, which were generated by Werme et al. by partitioning the whole genome into equal-sized semi-independent LD blocks with at least 2,500 SNPs and LD between SNPs less than 0.25 [16]. To control for potential sample overlap between GWASs, LAVA analyses included a sample overlap file generated using LDSC [30,31]. To identify relevant loci, associations between individual diseases and 2,495 loci, including the MHC region, were evaluated using a univariate genetic correlation test and loci that showed evidence of significant local heritability (p-value < 2.0 x10-05 (0.05/2,495)) for both traits in the SSc-cancer pairs were advanced for bivariate local genetic correlation analyses. Local bivariate genetic correlations were considered statistically significant if their Bonferroni-adjusted p-values were < 0.05. Bonferroni-adjusted p-values were calculated using p.adjust function in R [27] based on the number of loci that passed univariate tests performed for each SSc-cancer pair. Following the assessment of bivariate local genetic correlations, regional SNPs-level associations within loci shared between SSc and cancers were visualized using topr package [35] in R [27]. The replication of significant bivariate genetic correlations was performed using publicly available SSc GWAS summary statistics published by FinnGen [36]. Similarly, Bonferroni-adjusted p-values were used to determine the significance of local genetic correlations. Gene-based pleiotropy cross-trait meta-analysis To investigate gene-level evidence of pleiotropy across regions shared with cancer, we performed cross-trait gene-set association analysis using GCPBayes [37,38]. GCPBayes aggregated SNP-level association signals into gene-level Bayes factors and estimated posterior probabilities of a gene (group) or SNPs assigned to a gene (within group) being truly associated with both traits. Hierarchical Bayesian model with hierarchical spike (HS) and slab priors facilitated variable selection by distinguishing relevant from irrelevant SNPs while accounting for LD and gene size [37,38]. To assess cross-trait associations at a gene/group level, SNPs were annotated to genes using protein-coding gene coordinates. LD clumping was performed to mitigate redundancy in LD blocks [37,38]. The clumping procedure was performed on SNPs with p-value ≤ 0.99 using r² threshold of 0.6 and a 10 Mb window to identify independent SNPs. Genes were considered strongly associated with both traits if they satisfied criteria of log10Bayes Factor (BF) > 1, Local Bayesian False Discovery Rate (lBFDR) < 0.05 and met pleiotropy criteria derived from both the posterior median and the credible interval. Significant results produced by the analyses using the Dirac spike function were confirmed by running Gibbs sampler with HS. Colocalization We performed Bayesian colocalization analyses to determine whether genetic signals at pleiotropic loci were driven by a common causal variant shared between SSc and each cancer type. Bayesian colocalization analyses, performed in coloc R package [39], calculated posterior probabilities for 5 different hypotheses [40]: H0: No association with either trait; H1: Association with SSc only; H2: Association with cancer only: H3: Association with both traits, but different causal variants: H4: Association with both traits, same causal variant (colocalization). Posterior probabilities (PP.H0–PP.H4) were calculated for each pair of traits and regions, separately, with PP.H4 ≥ 0.8 indicating the presence of a shared causal SNP, whereas high PP.H3 favoured pleiotropy without shared causality. Fine mapping We applied SuSiE (Sum of Single Effects) regression for fine-mapping within regions showing evidence of pleiotropy [40]. SuSiE identified credible sets of variants likely to contain causal SNPs by modelling multiple effects simultaneously [40]. For each region, the top hits and their posterior inclusion probabilities (PIPs) were reported [40]. Functional exploration of significant bivariate genomic correlations We investigated possible functional implications of pleiotropic genes identified in GCPBayes analyses in functional enrichment analyses using gprofiler2 R package [41]. The gprofiler2 R package employed data from various sources, including Gene Ontology (GO) [42] and Kyoto Encyclopedia of Genes and Genomes (KEGG) [43]. Biological pathways and functions with FDR p-value < 0.05 were considered statistically significant. Results of functional analyses were visualized using clusterProfiler [44] and enrichplot [45] R packages. Results Global genetic correlations Global genetic correlations between SSc and breast and lung cancers, lymphocytic leukemia and non-Hodgkin's lymphoma, including different molecular subtypes, were assessed in LDSC-regression. We did not find evidence supporting global genetic correlations between SSc and the studied cancer subtypes (Table 2). Local genetic correlations LAVA analyses were applied to breast cancer and its molecular subtypes, lung cancer, lymphocytic leukemia, and non-Hodgkin’s lymphoma. Fig 1A displays the number of loci that displayed significant local heritability for both SSc and each studied cancer. Panel A displays the number of loci that passed univariate test and showed significant local heritability for both SSc and the studied cancers. Panel B shows Bonferroni-significant local bivariate genetic correlations. We detected twenty-three significant local genetic correlations passing the threshold for statistical significance (Table 3; Fig 1B); 21 of them were with different molecular subtypes of breast cancer, including HER2-enriched-like, luminal A-like, luminal B-like, luminal B HER2-negative-like, and triple-negative breast cancers. Specifically, SSc displayed strong negative genetic correlations with HER2-enriched-like breast cancer across two loci on chromosome 6 and 12, respectively (chr6:106053916–107309327, genetic correlation coefficient (rg)=−0.8; chr12:34340496–38108284, rg = −0.8). Similarly, SSc and luminal A-like breast cancer also shared a locus on chromosome 6 (chr6:134313644–135544083) with a positive genetic correlation of 0.8 and a locus on chromosome 4 that showed a negative genetic correlation of −0.8. Positive local genetic correlations ranging from 0.5–0.8 were detected between SSc and luminal B-like breast cancer across five loci on chromosomes 1, 7, 12, and 14. In addition to these, another common locus was mapped to chromosome 5 (chr5:65606871–67096191), exhibiting a negative genetic correlation of −0.5. There were four loci shared between SSc and luminal B HER2-negative-like breast cancer, of which only one displayed a negative genetic correlation (chr4: 38984839−39820061; rg = −0.8), and the remaining three local genetic correlations on chromosomes 17 and 22 varied between 0.6 and 0.8, respectively. We also identified three positive genetic correlations between SSc and triple-negative breast cancer across three distinct loci on chromosomes 7, 17, and 20; these genetic correlations ranged between 0.6–0.8. Four negative genetic correlations were also observed across loci on chromosomes 4, 7, 13, and 18, ranging between −0.8 and −0.4. In contrast to numerous common loci identified between SSc and breast cancer, lung cancer and lymphocytic leukemia each shared a single locus with SSc mapped to chromosomes 6 and 20, accordingly (rg = 0.4 for lung cancer and rg = 0.6 for lymphocytic leukemia). We did not detect significant local genetic correlations between SSc and non-Hodgkin's lymphoma. Most of these loci did not contain SNPs associated at genome-wide significance threshold (p < 5 x 10-08) for SSc or breast cancer. For instance, genomic locus chr6:106053916–107309327 carried genetic markers associated with SSc at genome-wide significance but this was not the case for HER2 enriched like breast cancer (Fig 2A). By contrast, the locus chr18:41425455–42974165 harbored SNPs that showed genome wide-significant SNP associations with triple-negative breast cancers but not SSc (Fig 2B). The locus chr6:30715007–31106493 carried SNPs associated at genome-wide significance level with SSc and lung cancer, mirrored by local genetic correlation between these two traits (Table 3, Fig 2C). Genes were mapped to the loci shared between SSc and breast cancer subtypes (panels A & B) and lung cancer (panel C). Strongest genetic associations within the loci are marked with SNP rsid names. Genetic association probabilities are plotted as –log10 (p-value) on the y-axis against chromosomes on x-axis. Gene-based cross-trait meta-analysis Of the 21 genomic regions flagged by LAVA genetic correlation analyses, GCPBayes found evidence of gene-level pleiotropy in two. In the locus shared with the lung cancer (6:30715007–31106493), ten genes, (PSORS1C2, C6orf15, LINC02570, DDR1-DT, MUC21, CDSN, DDR1, PSORS1C1, HCG20, and MUC22), were classified as pleiotropic under GCPBayes confidence interval and median-based criteria and confirmed by HS-based analyses (S2 Table). In contrast, the lymphocytic leukaemia locus (20:44072211–45673603) exhibited a more focused association signal, with only one gene, SLC12A5, implicated based on median criteria. Theta (0.6) for this gene indicated a moderate shared signal (S1 Table). Loci shared with breast cancers, such as HER2-enriched-like, Luminal A-like, Luminal B-like, Luminal B HER2-negative-like, and triple-negative subtypes lacked detectable pleiotropic effects under the applied GCPBayes framework, with significant regional associations driven mostly breast cancer-specific signals (PPA phenotype > 0.5) (S2 Table) Colocalization In the 6p locus (6:30715007–31106493), identified by GCPBayes analyses as containing pleiotropic genes, colocalization analyses strongly supported distinct causal variants associated with SSc and lung cancer (PP.H3 ≈ 1.0), while the probability of a shared causal variant (PP.H4) was zero (S2 Fig, Panel A) Fine-mapping Across 6p locus (6:30715007–31106493), fine-mapping identified multiple SNPs with PP.H3 = 1.0 (e.g., rs886422 and rs9263631). Given all of these SNPs had PP.H4 close to 0, SuSiE findings were congruent with coloc analyses which supported strong pleiotropy with distinct causal SNPs for SSc and lung cancer (S2 Fig, Panel B). Functional analyses Gene-set enrichment analyses showed that pleiotropic genes mapped to the 6p locus (6:30715007–31106493), which was shared between SSc and lung cancer, participated in biological processes related to cellular communication, including negative regulation of cell adhesion and extracellular matrix organization among other. Collagen-activated signalling and tyrosine kinase activity pathways were also implicated. Additionally, pleiotropic genes were involved in developmental programs and morphogenesis, as highlighted both by GO and KEGG analyses, as well as wound healing (Fig 3, S3 Table). Gene-set enrichment analyses used Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases. Discussion In this study, we sought to examine the genetic relationship between SSc and cancer by applying genetic correlation analyses to GWAS data on SSc and related cancers. We did not find evidence supporting global genetic correlations between SSc and the nine evaluated cancers comprising breast cancer, including HER2-enriched-like, luminal A-like, luminal B-like, luminal B HER2-negative-like and triple-negative molecular groups, lung cancer, lymphocytic leukemia, and non-Hodgkin's lymphoma. Nonetheless, we detected twenty-three significant local genetic correlations between SSc and seven types of cancers; twenty-one were shared with different breast cancer molecular subtypes. Subsequent gene-based cross-trait meta-analysis found sufficient evidence for pleiotropy only at the locus shared with lung cancer, providing gene candidates whose utility can be further explored in functional and clinical studies. Nonetheless, we were not able to replicate observed local genetic correlation using data from FinnGen cohort. Given that FinnGen represents genetically distinct population shaped by founder bottlenecks – combined with a small number of SSc cases in the replication cohort, statistical power was limited in validation analyses. However, currently, no other larger GWAS on SSc are available. Although published research indicates that SSc and cancer may share several common mechanisms, previous studies found minimal genetic correlations between SSc and different cancer types. Nonetheless, it has been demonstrated that local genetic correlations may be present without significant genome-wide correlation estimates or diverge from global correlations [46], therefore providing a motivation for exploring genetic pleiotropy on the local scale. In our study, we did not observe any significant global correlations between SSc and the studied cancer types, consistent with previous investigations that used various bioinformatic approaches to assess the genetic relationships. A recently published meta-analysis aggregating SSc GWAS data from European cohorts and Biobank of Japan measured genetic correlations between SSc and 60 different traits, including several cancers [47]. This meta-analysis reported non-significant global correlations for breast cancer (0.17) and hematological malignancies (0.45) [47]. In a MR study, Lu and colleagues [48] examined a causal link between different immune-mediated diseases and breast cancer, finding an OR of 1.00 for SSc. The authors concluded that there was no causal association between SSc and increased risk of developing breast cancer, supporting that some shared environmental and genetic factors could explain previously observed associations [48]. Similarly, MR analyses evaluating the relationship between SSc and lung cancer did not support causality between these traits [11]. Additional stratified analyses did not identify significant associations with lung adenocarcinoma, squamous cell, or small-cell subtypes [11]. Our study identified several loci with significant local genetic correlations between SSc and various subtypes of breast cancer. These local genetic correlations were both positive and negative, suggesting a complex genetic relationship between SSc and cancer. Broadly, positive genetic correlations between SSc and breast cancer at observed loci suggested that genetic variants in a particular region increased the risk for both diseases and contributed to their development. Conversely, negative genetic correlations implied that genetic markers in a particular locus may have opposite effects on SSc and analysed cancers. Negative genetic correlations could hint at different biological mechanisms or pathways influenced by the same genetic region in opposite directions. While numerous loci showed significant local genetic correlations between SSc and various breast cancer subtypes, no consistent evidence of pleiotropy emerged from these local correlations. These discrepancies could be explained by several factors. For instance, trait causality or mediated pleiotropy, where a trait causally influences another trait rather than because genetic variants directly affect both diseases or horizontal pleiotropy [49]. Genetic correlations between traits can arise from pleiotropy or from linkage disequilibrium alone [50]. Additionally, observed negative genetic correlations could indicate antagonistic pleiotropy, in which SNPs within a specific locus increase the risk of one disease while decreasing the risk of another [51]. GCPBayes methodology employed Bayesian sparse group sampling that aggregated SNP-level evidence of horizontal pleiotropy at gene level, which assumed co-directionality of the SNP effects [37,38]. Conversely, divergent SNP associations diluted evidence of pleiotropy, resulting in low posterior probabilities of shared effects [37,38]. Both, LAVA or GCBayes did not distinguish between vertical and horizontal pleiotropy [16,37,52,53]. LAVA detected regions with genetic correlations, but could not establish whether those arose from pleiotropy or coincidental linkage disequilibrium [16]. This limitation stems from LAVA focus on the statistical associations without a mechanistic insight. The inclusion of expression data could clarify pathways underlying local genetic correlations, providing mechanistic explanations and generating a testable hypothesis for experimental validation. Overall, both genetic and environmental factors are postulated to influence the association between SSc and breast cancer. Published epidemiological research indicates that breast cancers are typically diagnosed before or shortly after diagnosis of SSc, occurring within a 3–5 years window [5]. Interestingly, mutations in POLR3A gene and co-occurring anti-RNA polymerase III autoantibodies were linked to emergence of breast cancer closely to SSc diagnosis [8,54], which supports the role of genetics in paraneoplastic relationship with breast cancer, By contrast, cancer surfacing after diagnosis of SSc would point to potential influence from immunosuppressive medications, such as cyclophosphamide, mofetil and azathioprine, several of which were implicated in cancer [54–58]. Additionally, the temporal interplay between SSc and breast cancer can be a by-product of intense cancer screening of newly diagnosed patients with SSc. It is also important to acknowledge that although genetic correlations may provide novel insights into shared genetics between SSc and cancer, stratifying patients with SSc by their autoantibody profile, comparing the genetic background of patients with SSc that develop malignancies with those that did not and performing functional studies are necessary to validate the biological relevance of these loci. Experimental validation can help confirm whether the identified genetic variants directly influence the risk of SSc-related cancer. In conjunction with several types of breast cancer, our study also revealed local genetic correlations with lung cancer and lymphocytic leukemia. The locus common to SSc and lymphocytic leukemia (chr20:44072211–45673603) showed inconsistent evidence of pleiotropy, while the 6p locus (chr6:30715007–31106493), which was shared between SSc and lung cancer displayed robust evidence of gene-level pleiotropy with ten genes. These genes suggest shared mechanisms between SSc and lung cancer, pertaining to extracellular matrix (ECM) remodeling, collagen signaling and aberrant tyrosine-kinase-mediated cell adhesion. ECM remodeling was markedly altered in both SSc and lung cancer, underscoring ECM’s role in fibrosis and tumor progression [59,60]. SSc and SSc-related interstitial lung disease featured aberrant deposition of collagen and other ECM proteins leading to progressive fibrosis of organs, including lungs [60,61]. In lung cancer, ECM with altered collagens supported tumor microenvironment by facilitating tumorigenesis, invasion and metastasis [62]. Gene enrichment for pathways related to peptidyl-tyrosine autophosphorylation and cell adhesion provided additional context for mechanistic overlap between SSc and lung cancer. Tyrosine phosphorylation was a key regulator of multiple cellular processes disrupted in SSc, namely, autoimmunity, fibrosis, and vasculopathy [63]. Dysregulation of tyrosine kinases also influences cell proliferation, adhesion, migration and metastasis in lung cancer [64]. Genes involved in epithelial function were among pleiotropic candidates associated with SSc and lung cancer. These genes maintained mucosal barriers and regulated intercellular signaling [65] and were implicated in inflammation-related oncogenic pathways [66,67]; their overexpression, has been linked to lung adenocarcinoma growth, invasion, and metastasis [65,68]. Colocalization and fine-mapping showed that the locus that was shared between SSc and lung cancer harbored distinct causal SNPs underlying these traits. These findings highlight that local genetic correlation and gene-based pleiotropy do not necessarily reflect shared causal variants, as LD and polygenicity can generate regional signals without true colocalization. We could not successfully replicate local genetic correlations using data from FinnGen cohort. Low number of SSc cases and unique genetic background of the FinnGen cohort might have contributed to non-significant local genetic correlations and overall lower statistical power because genetic variants that are common in Europeans can be rare or absent in Finish population. Isolated populations like Finland often accumulate deleterious mutations that occur at a lower frequencies in other Europeans [69]. Sufficient differences in the genetic profiles of these populations could diminish power to detect associations and lead to failed replication attempts. Other limitations of our study included small GWAS samples, such as in the case of hematological malignancies, which decreased overall statistical power. Additionally, GWAS on cancer included close to 10 million markers whereas the largest to date SSc GWAS covered only 4 million variants, contributing to modest marker overlap. Moreover, common susceptibility variants identified by GWAS confer incremental increases in the disease risk, explaining only small proportion of overall trait heritability. Missing heritability is attributed to rare markers and structural genetic variation [70] that can be explored in future genome-sequencing studies. It has been also shown, that local genetic correlation estimation methods, such as LAVA, generally have lower power to detect genetic correlation between binary traits compared to continuous traits [71]. In addition to potential power limitations, we analyzed traits derived from individuals of European descent potentially limiting generalizability of our findings [72,73]. Future investigations should focus on exploration of genetic correlations between SSc and cancer in non-European populations. Importantly, different disease subsets characterized by specific autoantibody profiles have been identified in SSc [74]. We analyzed published GWAS on SSc that did not include data on specific autoantibodies; analyses stratified by autoantibody profiles and other clinical characters of patients with SSc would help better understand pathoetiological mechanisms underlying cancer development in SSc. Although, our analyses were controlled for potential sample overlap between SSc and analyzed cancer studies, there was still a possibility that cancer patients included in SSc meta-GWAS biased the results of correlation analyses. Even though we identified several loci that could be interesting candidates for future research, most of these loci carried genes with known role in cancer but not SSc. It is worth mentioning that observed local genetic correlations do not necessarily indicate true pleiotropy. In particular, misclassification or complex linkage-disequilibrium could lead to detection of spurious local pleiotropic signals [52]. One of the strengths of this study was rigorous quality control applied to the GWAS summary data. GWAS datasets included extensive information on genetic markers across genome, providing a rich resource for exploring genetic correlations. By contrast, case-case studies on cancer in SSc would require large samples which are less feasible given rarity of SSc. Our analyses also considered high heterogeneity of breast cancer, assessing genetic correlations with its’ most common molecular subtypes and revealing heterogeneous regional signals, that would have been missed otherwise. To comprehensively examine genetic relationship between SSc and cancer, we applied both global and local genetic correlation methods, gene-based cross-trait meta-analysis, colocalization and fine-mapping analyses. In contrast to LDSC, which required removal of the MHC due to its complexity and inflation of estimates, LAVA allowed us to assess genetic correlations within the MHC region, which was a significant risk factor associated with SSc [72]. Integration of multiple published GWAS datasets enhanced the ability to carry out more granular genetic correlation analyses, providing detailed insights into genetic relationships at different scales. The identification of shared genetic loci can expand our understanding of etiological basis underpinning a longstanding association between SSc and lung cancer, and inform the risk assessment and screening strategies for cancer in patients with SSc. The shared genetic loci identified in our study could be potentially utilized in earlier cancer detection or used in personalized treatment strategies. Genes residing within common loci can be explored as potential biomarkers and targeted by specific therapies, which would optimize treatment efficacy and attenuate adverse effects Conclusions To conclude, we report twenty-three loci associated with SSc and numerous breast cancers subtypes as well as lung cancer and lymphocytic leukemia. These findings provide new insight into the pathological link between SSc and the commonly associated cancer types from a genetic perspective. The identified local pleiotropy between SSc and the studied cancer types needs further replications in more powered studies. Supporting information S1 Fig. Schematic illustration of study statistical analyses. https://doi.org/10.1371/journal.pone.0350006.s001 (TIF) S2 Fig. Results of colocalization and fine-mapping analyses. Panel A displays summary table of hypothesis posterior probabilities (PP.H0–PP.H4) for colocalization (first row) and SuSiE signal pairs (subsequent rows). First two columns provide representative lead SNPs for systemic sclerosis (SSc) and lung cancer signals, respectively. The first row shows the colocalization result for the entire region, while subsequent rows show SuSiE signal comparisons. Panel B shows sensitivity analysis for the SuSiE analyses (SSc signal vs lung cancer signal). Manhattan-style plots of −log₁₀(p) for SSc (top) and lung cancer (bottom) across the locus; blue points indicate SNPs in the SuSiE credible set for the selected signals. Prior probabilities (top) and posterior probabilities (bottom) for hypotheses H0–H4 as the shared-causal prior (p₁₂) varies. The posterior curve shows H3 ≈ 1 across all prior settings, suggesting strong evidence for two distinct causal variants rather than colocalization (H4). https://doi.org/10.1371/journal.pone.0350006.s002 (TIF) S1 Table. Gene-level pleiotropic associations identified by GCPBayes analyses. Columns list cancer that was analyzed with SSc (“Trait paired with SSc “), genomic regions (“loci”) and gene within that region (“Gene”). Theta indicates overall probability of gene being associated with at least one of the traits. DS-based and HS-based columns provide evidence of pleiotropy based on credible intervals (CI) and median criteria under the Dirichlet Spike (DS) and Hierarchical Spike (HS) priors. https://doi.org/10.1371/journal.pone.0350006.s003 (DOCX) S2 Table. Gene-level associations with SSc and cancers. Columns list cancer that was analyzed with SSc (“Trait paired with SSc “), genomic regions (“loci”) and gene within that region (“Gene”). The base-10 logarithm of the Bayes Factor (log10BF), shows the strength of evidence for pleiotropy between SSc and cancer at a given locus with higher values suggesting stronger evidence. Local Bayesian False Discovery Rate (lBFDR) provides the probability that the observed association is a false positive. “Theta” provides overall probability of gene being associated with at least one of the traits, while “PPA SSc” and " PPA phenotype” list posterior probabilities with SSc and cancer, respectively. https://doi.org/10.1371/journal.pone.0350006.s004 (DOCX) S3 Table. Gene-set enrichment results. The source of gene list and pathways incorporated in the analyses are listed in the columns “Dataset” and “Pathway.” Field “Gene ratio” represents the proportion analyzed genes mapped to a particular pathways, while “Background ratio” shows the proportion of background genes associated with the same pathways. Column “Gene” includes genes associated with a particular pathway. “Fold enrichment” compares these two ratios and assesses the strength of enrichment. Columns “p-value” and “FDR p-value” list statistical significance of the enrichment and multiple testing adjusted statistical significance (false discovery rate). https://doi.org/10.1371/journal.pone.0350006.s005 (DOCX) Acknowledgments We want to acknowledge all participants and investigators from the studies used in our analyses, and particularly those from the FinnGen study. References - 1. Gabrielli A, Avvedimento EV, Krieg T. Scleroderma. N Engl J Med. 2009;360:1989–2003. - 2. Peoples C, Medsger TA Jr, Lucas M, Rosario BL, Feghali-Bostwick CA. Gender differences in systemic sclerosis: relationship to clinical features, serologic status and outcomes. J Scleroderma Relat Disord. 2016;1(2):177–240. pmid:29242839 - 3. Hao Y, Hudson M, Baron M, Carreira P, Stevens W, Rabusa C, et al. Early Mortality in a Multinational Systemic Sclerosis Inception Cohort. Arthritis Rheumatol. 2017;69(5):1067–77. pmid:28029745 - 4. Nevskaya T, Chandran S, Roos AM, Pasarikovski CR, Kron A, Chau C, et al. Epidemiology of Cancer in Systemic Sclerosis—Systematic Review and Meta-Analysis of Cancer Incidence, Predictors and Mortality*. OJRA. 2013;03(04):231–45. - 5. Partouche L, Goulabchand R, Maria ATJ, Rivière S, Jorgensen C, Rigau V, et al. Biphasic Temporal Relationship between Cancers and Systemic Sclerosis: A Clinical Series from Montpellier University Hospital and Review of the Literature. J Clin Med. 2020;9(3):853. pmid:32245039 - 6. Shah AA, Casciola-Rosen L. Cancer and scleroderma: a paraneoplastic disease with implications for malignancy screening. Curr Opin Rheumatol. 2015;27(6):563–70. pmid:26352736 - 7. Shah AA, Rosen A. Cancer and systemic sclerosis: novel insights into pathogenesis and clinical implications. Curr Opin Rheumatol. 2011;23(6):530–5. pmid:21825998 - 8. Maria ATJ, Partouche L, Goulabchand R, Rivière S, Rozier P, Bourgier C, et al. Intriguing Relationships Between Cancer and Systemic Sclerosis: Role of the Immune System and Other Contributors. Front Immunol. 2019;9:3112. pmid:30687318 - 9. Watad A, McGonagle D, Bragazzi NL, Tiosano S, Comaneshter D, Shoenfeld Y, et al. Autoantibody status in systemic sclerosis patients defines both cancer risk and survival with ANA negativity in cases with concomitant cancer having a worse survival. Oncoimmunology. 2019;8(6):e1588084. pmid:31069155 - 10. Igusa T, Hummers LK, Visvanathan K, Richardson C, Wigley FM, Casciola-Rosen L. Autoantibodies and scleroderma phenotype define cancer risk subgroups. Ann Rheum Dis. 2018;77:1179–86. - 11. Peng H, Wu X, Wen Y, Li C, Lin J, Li J, et al. Association between systemic sclerosis and risk of lung cancer: results from a pool of cohort studies and Mendelian randomization analysis. Autoimmun Rev. 2020;19(10):102633. pmid:32801043 - 12. Xu Y-C, Wang J-X, Chu Y-R, Qian H, Wang H-Y, Wang F. Diseases of the musculoskeletal system and connective tissue and risk of breast cancer: Mendelian randomization study in European and East Asian populations. Front Oncol. 2023;13:1170119. pmid:37182186 - 13. Shi X, Wallach JD, Ma X, Rogne T. Autoimmune Diseases and Risk of Non-Hodgkin Lymphoma: A Mendelian Randomisation Study. Cancer Med. 2024;13:e70327. - 14. Hanson AL, Sahhar J, Ngian G-S, Roddy J, Walker J, Stevens W, et al. Contribution of HLA and KIR Alleles to Systemic Sclerosis Susceptibility and Immunological and Clinical Disease Subtypes. Front Genet. 2022;13:913196. pmid:35754823 - 15. Carabaño M, Liu T, Liu A, Lee JC, Cheng L, Wang L-J, et al. Association of HLA-A, B, and C alleles and cancer susceptibility in 179 solid malignancies. Am J Transl Res. 2023;15(9):5642–52. pmid:37854217 - 16. Werme J, van der Sluis S, Posthuma D, de Leeuw CA. An integrated framework for local genetic correlation analysis. Nat Genet. 2022;54(3):274–82. pmid:35288712 - 17. López-Isac E, Acosta-Herrera M, Kerick M, Assassi S, Satpathy AT, Granja J, et al. GWAS for systemic sclerosis identifies multiple risk loci and highlights fibrotic and vasculopathy pathways. Nat Commun. 2019;10(1):4955. pmid:31672989 - 18. Preliminary criteria for the classification of systemic sclerosis (scleroderma). Subcommittee for scleroderma criteria of the American Rheumatism Association Diagnostic and Therapeutic Criteria Committee. Arthritis Rheum. 1980;23(5):581–90. pmid:7378088 - 19. LeRoy EC, Medsger TA Jr. Criteria for the classification of early systemic sclerosis. J Rheumatol. 2001;28(7):1573–6. pmid:11469464 - 20. Zhang H, Ahearn TU, Lecarpentier J, Barnes D, Beesley J, Qi G, et al. Genome-wide association study identifies 32 novel breast cancer susceptibility loci from overall and subtype-specific analyses. Nat Genet. 2020;52(6):572–81. pmid:32424353 - 21. McKay JD, Hung RJ, Han Y, Zong X, Carreras-Torres R, Christiani DC, et al. Large-scale association analysis identifies new lung cancer susceptibility loci and heterogeneity in genetic susceptibility across histological subtypes. Nat Genet. 2017;49(7):1126–32. pmid:28604730 - 22. Rashkin SR, Graff RE, Kachuri L, Thai KK, Alexeeff SE, Blatchins MA, et al. Pan-cancer study detects genetic risk variants and shared genetic basis in two large cohorts. Nat Commun. 2020;11(1):4423. pmid:32887889 - 23. Guo L, Kong D, Liu J, Zhan L, Luo L, Zheng W, et al. Breast cancer heterogeneity and its implication in personalized precision therapy. Exp Hematol Oncol. 2023;12(1):3. pmid:36624542 - 24. Buniello A, MacArthur JAL, Cerezo M, Harris LW, Hayhurst J, Malangone C, et al. The NHGRI-EBI GWAS Catalog of Published Genome-Wide Association Studies, Targeted Arrays and Summary Statistics 2019. Nucleic Acids Research. 2019;47:D1005–12. - 25. Breast Cancer Association Consortium. Commonly studied single-nucleotide polymorphisms and breast cancer: results from the Breast Cancer Association Consortium. JNCI J Natl Cancer Inst. 2006;98:1382–96. - 26. Murphy AE, Schilder BM, Skene NG. MungeSumstats: a Bioconductor package for the standardization and quality control of many GWAS summary statistics. Bioinformatics. 2021;37(23):4593–6. pmid:34601555 - 27. R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. 2023. - 28. Gehring J. BSgenome.Hsapiens.1000genomes.hs37d5: 1000genomes reference genome sequence (hs37d5). Bioconductor. - 29. Pages H. BSgenome: Software infrastructure for efficient representation of full genomes and their SNPs. Bioconductor. http://bioconductor.org/packages/BSgenome/ 2025. Accessed 2025 March 5. - 30. Bulik-Sullivan BK, Loh P-R, Finucane HK, Ripke S, Yang J, Schizophrenia Working Group of the Psychiatric Genomics Consortium, et al. LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet. 2015;47(3):291–5. pmid:25642630 - 31. Bulik-Sullivan B, Finucane HK, Anttila V, Gusev A, Day FR, Loh P-R, et al. An atlas of genetic correlations across human diseases and traits. Nat Genet. 2015;47(11):1236–41. pmid:26414676 - 32. 1000 Genomes Project Consortium, Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74. pmid:26432245 - 33. International HapMap 3 Consortium, Altshuler DM, Gibbs RA, Peltonen L, Altshuler DM, Gibbs RA, et al. Integrating common and rare genetic variation in diverse human populations. Nature. 2010;467(7311):52–8. pmid:20811451 - 34. ldsc/munge_sumstats.py at master · bulik/ldsc. GitHub. https://github.com/bulik/ldsc/blob/master/munge_sumstats.py Accessed 2024 September 11. - 35. Juliusdottir T. topr: an R package for viewing and annotating genetic association results. BMC Bioinformatics. 2023;24(1):268. pmid:37380954 - 36. Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613(7944):508–18. pmid:36653562 - 37. Asgari Y, Sugier P-E, Baghfalaki T, Lucotte E, Karimi M, Sedki M, et al. GCPBayes pipeline: a tool for exploring pleiotropy at the gene level. NAR Genomics Bioinforma. 2023;5. - 38. Baghfalaki T, Sugier P-E, Truong T, Pettitt AN, Mengersen K, Liquet B. Bayesian meta-analysis models for cross cancer genomic investigation of pleiotropic effects using group structure. Stat Med. 2021;40(6):1498–518. pmid:33368447 - 39. Giambartolomei C, Vukcevic D, Schadt EE, Franke L, Hingorani AD, Wallace C, et al. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genet. 2014;10(5):e1004383. pmid:24830394 - 40. Wallace C. A more accurate method for colocalisation analysis allowing for multiple causal variants. PLoS Genet. 2021;17(9):e1009440. pmid:34587156 - 41. Kolberg L, Raudvere U, Kuzmin I, Vilo J, Peterson H. gprofiler2 -- an R package for gene list functional enrichment analysis and namespace conversion toolset g:Profiler. F1000Res. 2020;9:ELIXIR-709. pmid:33564394 - 42. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. Nat Genet. 2000;25:25–9. - 43. Kanehisa M, Sato Y, Furumichi M, Morishima K, Tanabe M. New approach for understanding genome variations in KEGG. Nucleic Acids Res. 2019;47(D1):D590–5. pmid:30321428 - 44. Yu G, Wang L-G, Han Y, He Q-Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–7. pmid:22455463 - 45. Yu G. Enrichplot: visualization of functional enrichment result. 2021. - 46. Reynolds RH, Wagen AZ, Lona-Durazo F, Scholz SW, Shoai M, Hardy J, et al. Local genetic correlations exist among neurodegenerative and neuropsychiatric diseases. NPJ Parkinsons Dis. 2023;9(1):70. pmid:37117178 - 47. Ishikawa Y, Tanaka N, Asano Y, Kodera M, Shirai Y, Akahoshi M, et al. GWAS for systemic sclerosis identifies six novel susceptibility loci including one in the Fcγ receptor region. Nat Commun. 2024;15(1):319. pmid:38296975 - 48. Lu Y, Ma L. Investigation of the causal relationship between breast cancer and autoimmune diseases: A bidirectional mendelian randomization study. Medicine (Baltimore). 2023;102(34):e34612. pmid:37653788 - 49. Ballard JL, O’Connor LJ. Shared components of heritability across genetically correlated traits. Am J Hum Genet. 2022;109(6):989–1006. pmid:35477001 - 50. Chebib J, Guillaume F. Pleiotropy or linkage? Their relative contributions to the genetic correlation of quantitative traits and detection by multitrait GWA studies. Genetics. 2021;219(4):iyab159. pmid:34849850 - 51. Lockwood C, Vo AS, Bellafard H, Carter AJR. More evidence for widespread antagonistic pleiotropy in polymorphic disease alleles. Front Genet. 2024;15:1404516. pmid:38952711 - 52. van Rheenen W, Peyrot WJ, Schork AJ, Lee SH, Wray NR. Genetic correlations of polygenic disease traits: from theory to practice. Nat Rev Genet. 2019;20(10):567–81. pmid:31171865 - 53. Gerring ZF, Thorp JG, Gamazon ER, Derks EM. A Local Genetic Correlation Analysis Provides Biological Insights Into the Shared Genetic Architecture of Psychiatric and Substance Use Phenotypes. Biol Psychiatry. 2022;92(7):583–91. pmid:35525699 - 54. Lepri G, Catalano M, Bellando-Randone S, Pillozzi S, Giommoni E, Giorgione R. Systemic Sclerosis Association with Malignancy. Clin Rev Allergy Immunol. 2022;63:398–416. - 55. Bruni C, Furst DE. The burning question: To use or not to use cyclophosphamide in systemic sclerosis. Eur J Rheumatol. 2020;7(Suppl 3):S237–41. pmid:33164737 - 56. Robson R, Cecka JM, Opelz G, Budde M, Sacks S. Prospective registry-based observational cohort study of the long-term risk of malignancies in renal transplant patients treated with mycophenolate mofetil. Am J Transplant. 2005;5(12):2954–60. pmid:16303010 - 57. Bichari W, Bartiromo M, Mohey H, Afiani A, Burnot A, Maillard N, et al. Significant risk factors for occurrence of cancer after renal transplantation: a single center cohort study of 1265 cases. Transplant Proc. 2009;41(2):672–3. pmid:19328953 - 58. Sanders ML, Karnes JH, Denny JC, Roden DM, Ikizler TA, Birdwell KA. Clinical and Genetic Factors Associated with Cutaneous Squamous Cell Carcinoma in Kidney and Heart Transplant Recipients. Transplant Direct. 2015;1(4):e13. pmid:26146661 - 59. Song K, Yu Z, Zu X, Li G, Hu Z, Xue Y. Collagen Remodeling along Cancer Progression Providing a Novel Opportunity for Cancer Diagnosis and Treatment. Int J Mol Sci. 2022;23(18):10509. pmid:36142424 - 60. Leask A, Naik A, Stratton RJ. Back to the future: targeting the extracellular matrix to treat systemic sclerosis. Nat Rev Rheumatol. 2023;19(11):713–23. pmid:37789119 - 61. Mouawad JE, Feghali-Bostwick C. The Molecular Mechanisms of Systemic Sclerosis-Associated Lung Fibrosis. Int J Mol Sci. 2023;24(3):2963. pmid:36769282 - 62. Parker AL, Cox TR. The Role of the ECM in Lung Cancer Dormancy and Outgrowth. Front Oncol. 2020;10:1766. pmid:33014869 - 63. Sacchetti C, Bottini N. Protein Tyrosine Phosphatases in Systemic Sclerosis: Potential Pathogenic Players and Therapeutic Targets. Curr Rheumatol Rep. 2017;19(5):28. pmid:28397126 - 64. Elkamhawy A, Lu Q, Nada H, Woo J, Quan G, Lee K. The Journey of DDR1 and DDR2 Kinase Inhibitors as Rising Stars in the Fight Against Cancer. Int J Mol Sci. 2021;22(12):6535. pmid:34207360 - 65. Li X, Chen Y, Lan R, Liu P, Xiong K, Teng H, et al. Transmembrane mucins in lung adenocarcinoma: understanding of current molecular mechanisms and clinical applications. Cell Death Discov. 2025;11(1):163. pmid:40210618 - 66. Johnson JL, Najor NA, Green KJ. Desmosomes: regulators of cellular signaling and adhesion in epidermal health and disease. Cold Spring Harb Perspect Med. 2014;4(11):a015297. pmid:25368015 - 67. Zeng H, Pan W, Xia Q, Shu H, Ji Z. Diagnostic, prognostic, and immunological roles of CDSN in ovarian cancer. Heliyon. 2024;10:e33357. - 68. Li M, Li H, Yuan T, Liu Z, Li Y, Tan Y, et al. MUC21: a new target for tumor treatment. Front Oncol. 2024;14:1410761. pmid:38933439 - 69. Kääriäinen H, Muilu J, Perola M, Kristiansson K. Genetics in an isolated population like Finland: a different basis for genomic medicine? J Community Genet. 2017;8(4):319–26. pmid:28730583 - 70. Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ, et al. Finding the missing heritability of complex diseases. Nature. 2009;461(7265):747–53. pmid:19812666 - 71. Zhang C, Zhang Y, Zhang Y, Zhao H. Benchmarking of local genetic correlation estimation methods using summary statistics from genome-wide association studies. Brief Bioinform. 2023;24(6):bbad407. pmid:37974509 - 72. Ishikawa Y, Terao C. Genetics of systemic sclerosis. J Scleroderma Relat Disord. 2020;5(3):192–201. pmid:35382527 - 73. Mustak MS, Das R, Mishra A, Nizamuddin S. Editorial: Individual and population-specific variation in cancer susceptibility, prevalence, and disease presentation. Front Genet. 2023;14:1287610. pmid:37928240 - 74. Fiorentino DF, Casciola-Rosen L. Autoantibodies and Cancer Association: the Case of Systemic Sclerosis and Dermatomyositis. Clin Rev Allergy Immunol. 2022;63(3):330–41. pmid:35593962