Skip to main content
NCBI home page

This is a preprint.

It has not yet been peer reviewed by a journal.

The National Library of Medicine is running a pilot to include preprints that result from research funded by NIH in PMC and PubMed.

medRxiv logoLink to medRxiv
[Preprint]. 2026 Sep 10:2026.09.08.26362541. [Version 1] doi: 10.64898/2026.09.08.26362541

Circulating Inflammatory Markers and Mammographic Density Phenotypes: a Mendelian Randomization Analysis

Austin Hammermeister Suger 1, Tabitha A Harrison 1, Cameron B Haas 1, Hongjie Chen 1, Shaoqi Fan 2, Manjeet K Bolla 3, Qin Wang 3, Joe Dennis 3, Kyriaki Michailidou 4, Alison M Dunning 5, Douglas F Easton 5,3, Melissa C Southey 6,7,8, Peter A Fasching 9, Lothar Haeberle 9, Jennifer Stone 10, Manuela Gago-Dominguez 11, Jose E Castelao 12, Kristan Aronson 13, Rachel A Murphy 14,15, D Gareth Evans 16,17, Sherene Loi 18,19; kConFab Investigators^, Fergus J Couch 20, Siddhartha Yadav 21, Roger L Milne 22,23,6, Christopher A Haiman 24, Christopher G Scott 25, Aaron Norman 26, Esther M John 27,28, Anna Marie Mulligan 29,30, Vessela N Kristensen 31,32, A Heather Eliassen 33,34,35, Irene L Andrulis 36,37, Jonine D Figueroa 38,39,40, Susan Astley 41, Kamila Czene 42, Paul DP Pharoah 43, Antonis C Antoniou 44, Montserrat García-Closas 45,46, Amy Berrington 45,46,3, Gretchen L Gierach 2, Celine M Vachon 26, Rulla M Tamimi 47,35, Sara Lindström 1,48
PMCID: PMC13622778  PMID: 42818447

Abstract

Observational studies have identified associations between circulating inflammatory markers (CIMs) and mammographic density (MD) phenotypes, but whether these associations reflect causal effects remains unclear. We utilized two-sample Mendelian randomization (MR) to assess relationships between 60 CIMs and MD phenotypes in individuals of European genetic ancestry by examining associations between genetically predicted CIM concentrations and three MD phenotypes (dense area, DA: N = 13,965; non-dense area, NDA: N = 14,036; percent density, PMD: N = 17,841). Primary analyses were conducted using the inverse variance weighted MR method. No CIM-MD phenotype associations reached statistical significance after correction for multiple testing. At a nominal significance level of p < 0.05, we observed suggestive evidence for eight associations represented by six unique CIMs (MD phenotype/direction of association): CCL23 (DA/+), CD40 (NDA/−), GDNF (DA/+ & PMD/+), MMP-10 (DA/+ & PMD/+), TRANCE (DA/−), and TWEAK (NDA/+). In multivariable MR analyses accounting for body mass index (BMI), we observed six nominal associations represented by six unique CIMs, although most instruments demonstrated weak conditional instrument strength after inclusion of BMI. Overall, our comprehensive analysis of 60 CIMs provides limited evidence supporting broad causal effects of circulating inflammatory markers contributing to variation in MD phenotypes.

Introduction

Female breast cancer is the most diagnosed cancer among women worldwide, with an estimated 2.3 million incident cases diagnosed in 2022, and remains a major contributor to cancer mortality [1]. Mammographic density (MD), defined by the relative amounts of connective, epithelial, and adipose tissue in the breast, is a well-established risk factor for breast cancer [2–4]. MD phenotypes include mammographic dense area (DA), non-dense area (NDA), and percentage of dense area in the breast (percent mammographic density; PMD). In addition to being a breast cancer risk factor, increased MD (higher PMD or DA) is associated with reduced sensitivity and specificity for mammography screening [5,6]. Characterizing factors associated with MD could help us better understand its influence on breast cancer risk and impact on screening.

Tumor-promoting inflammation is a cancer hallmark [7,8] and circulating inflammatory markers (CIMs) have been associated with breast cancer in molecular and observational epidemiological studies [9–11]. One of the most studied CIMs is C-reactive protein (CRP), which has a functional role in inflammatory processes [12], with studies supporting its involvement in apoptosis, immune cell recruitment, and lipid accumulation [13]. Multiple studies have observed an association between higher concentrations of circulating CRP and risk of breast cancer [14–17], yet many of the identified associations were modest and varied by body mass index (BMI), menopausal status, or other factors. Relationships between other CIMs and breast cancer risk have been studied less extensively and the associations that have been detected are often inconsistent across studies and vary by breast cancer subtype, BMI, menopausal status, menopausal hormone therapy use, or other factors [18–20].

Elucidating the relationships between CIMs and breast cancer risk factors could help disentangle the etiological role of systemic inflammation in breast cancer risk. Studies have identified associations between MD phenotypes and circulating concentrations of interleukin 6 (IL-6) [21–23], interleukin 8 (IL-8) [22,23], tumor necrosis factor receptor 2 (TNFR2) [24], vascular endothelial growth factor (VEGF) [23], and CRP [21,22,24,25]. However, similar to the breast cancer studies, many associations were inconsistent across studies and varied by potential confounding factors [21,25].

The objective of this study was to comprehensively assess associations between genetically predicted CIMs and MD phenotypes using a two-sample Mendelian randomization (MR) approach, as MR can help mitigate some methodological limitations of observational studies, such as reverse causation and uncontrolled confounding. We further performed multivariable MR analyses including genetically-predicted BMI as an exposure, given its strong association with MD phenotypes and involvement in inflammatory processes [4,26]. Previous bidirectional MR studies of CRP and BMI have suggested that BMI is likely to affect CRP, rather than vice versa, and thus BMI is a potential confounder in CRP-MD phenotype relationships [27,28]. In total, we conducted MR analyses of 60 genetically predicted CIM exposures in relation to DA, NDA, and PMD phenotypes.

Methods

MR study design and data sources

The STROBE-MR checklist was used to guide the reporting of our MR analyses [29]. Causal diagrams illustrating the relationships that we investigated using MR and multivariable MR analyses are shown in Figure 1. We used summary statistics from published European genetic ancestry GWAS for BMI (N subjects = 461,460) [30], circulating CRP (N = 575,531) [31] and TNFR2 (N = 21,758) [32], and circulating levels of 58 additional inflammation-related proteins (N = 14,824) for which at least one genome-wide significant (p < 5×10−8) single nucleotide polymorphism (SNP) exists [33] (Table S1), for a total of 60 unique CIMs. GWAS summary statistics were obtained from either the IEU OpenGWAS [34,35], the EBI-NHGRI GWAS Catalog [36], or studyspecific repositories [33], selecting the largest available European genetic ancestry GWAS when multiple were available. For MD phenotypes (DA: N = 13,965; NDA: N = 14,036; PMD: N = 17,841), we used GWAS summary statistics from the Breast Cancer Association Consortium (BCAC) dataset described in Haas et al., 2024 [37] (Table S2). The original GWAS or studies meta-analyzed by those GWAS obtained ethical approval from their organization/institution(s) ethical review boards as described in the cited articles. The GWAS summary statistics used did not contain any individual-level information.

Figure 1. Causal diagrams representing the relationships investigated in Mendelian randomization (MR) and multivariable MR (MVMR) percent mammographic density (PMD), dense mammographic area (DA), and non-dense mammographic area (NDA) analyses with core assumptions indicated.

Figure 1.

Open in a new tab

(A) Causal diagram representing univariable MR analyses for CIM exposures and MD outcomes. Core MR instrumental variable (IV) assumptions (MR IV1-IV3) are highlighted in red. (B) Causal diagram representing CRP MVMR analyses for MD outcomes. Core MVMR IV assumptions (MVMR IV1-IV3) are highlighted in red.

MR assumptions

To be valid, instrumental variables (IVs) used in MR analyses must meet three assumptions (MR IV 1–3): (1) the IVs must be strongly associated with the exposure, (2) the IVs must be independent of all (observed or unobserved) confounders of the exposure-outcome relationship, and (3) the IVs must be associated with the outcome only via the exposure (i.e., independent of the outcome given the exposure; Figure 1A) [38]. The expanded assumptions (multivariable MR IV 1–3) for valid IVs in multivariable MR analyses are that: (1) the IVs must be strongly associated with each exposure given all other exposures in the model, (2) the IVs must be independent of all confounders of any of the relationships between the exposures and the outcome, and (3) the IVs must be independent of the outcome given all exposures (Figure 1B) [38,39].

Statistical analyses

We used the TwoSampleMR R package v0.6.11 [34], custom functions in R v4.3.2 [40], and PLINK 2.0 alpha v6.5 [41,42] to extract and process exposure GWAS summary statistics, and harmonize the exposure and outcome GWAS summary statistics to create MR instruments. For each exposure GWAS, only SNPs with p < 5×10−8 were extracted, then SNPs with linkage disequilibrium (LD) were excluded (r2 = 0.001 and clumping window = 10,000 base-pairs) using the 1000 Genomes Project European genetic ancestry (1000G EUR) data as an LD reference panel [34,43]. For the multivariable MR instrument, if a SNP was not included in both the CIM and BMI GWAS summary statistics, then a proxy SNP with r2 > 0.8 in the 1000G EUR LD reference panel was used [35]. Ambiguous strand information was resolved using allele frequency information when possible. Palindromic SNPs were retained if strand information could be inferred.

When constructing all instruments, we excluded SNPs with a minor allele frequency (MAF) below 5% in the 1000G EUR reference panel [43]. We constructed two instruments for the CRP univariable MR analyses to try to disentangle the influences of CRP from BMI on MD phenotypes. The first instrument (CRP) included SNPs that met the standard inclusion criteria described above. The second instrument (CRP_noBMI) was further restricted to CRP-associated SNPs that did not show an association (p ≥ 0.05) with BMI and were not in LD with a BMI-associated SNP (r2 > 0.1 within 500 kb) in the 1000G EUR LD reference panel [36,43,44]. The number of SNPs in each of the final instruments are shown in Table S3.

For all CIM instruments, we conducted univariable MR and multivariable MR analyses using random-effect inverse-variance weighted (IVW) methods [45]. These methods were implemented in the MendelianRandomization R package v0.10.0 [46]. Primary inference was based on IVW MR analyses, with nominal p-values providing suggestive evidence in the context of multiple testing across CIM-MD phenotype pairs. A total of 61 CIM instruments with at least one valid SNP were tested using the univariable MR method for each of the three MD outcomes. CRP had two instruments for univariable MR analyses as described above. This resulted in 183 (61 × 3) total univariable MR analyses representing 180 (60 × 3) unique CIM-MD phenotype relationships. For multivariable MR, a total of 60 instruments resulted in 180 total multivariable MR analyses for 180 unique CIM-MD associations. As a result, the corresponding Bonferroni-corrected significance thresholds were p < 2.73 × 10−4 (0.05 / 183) and p < 2.78 × 10−4 (0.05 / 180), respectively, for univariable and multivariable IVW MR analyses.

Assessment of MR assumptions and sensitivity analyses

In univariable MR analyses, we calculated F-statistics for each of the instruments to assess the MR IV1 assumption [46]. Although there are no formal tests for the MR IV2 and IV3 assumptions, we used up to seven methods with varying degrees of robustness to help identify potential violations. For instruments with at least one SNP, we calculated MR Wald ratios for each SNP in the instrument. For instruments with more than two SNPs, we further used the MR-Egger [47], MR weighted median (MR median), MR-LASSO [48], and MR constrained maximum-likelihood (MR cML) [49] methods implemented in the MendelianRandomization R package [46]. For instruments with more than 10 SNPs, we used the MR pleiotropy residual sum and outlier (MR-PRESSO) method [50]. The MR-LASSO, MR cML, and MR-PRESSO methods support the identification of outliers and potentially invalid SNPs in the instruments.

To assess the multivariable MR IV1 assumption, we calculated Sanderson Windmeijer conditional F-statistics for each exposure in the multivariable MR analyses. For the CRP multivariable MR analyses, we also incorporated the estimated phenotypic correlation between CRP and BMI in UK Biobank (UKB) participants into the conditional F-statistic calculations. This phenotypic correlation was estimated by Pearson correlation between CRP ln(mg/l) and BMI (kg/m2) at baseline for 466,356 participants with available data. To identify violations of multivariable MR IV2-IV3 and outliers or invalid SNPs, we used multivariable extensions of the MR-Egger, MR weighted median, MR-LASSO, MR-cML, and MR generalized method of moments (MR GMM) methods implemented in the MendelianRandomization R package [34]. For MR cML and multivariable MR cML, the range of potentially invalid IVs was based on the number of potentially invalid IVs detected by MR-LASSO or multivariable MR-LASSO, respectively. Given the strong correlation between BMI and CRP, we used two additional multivariable MR methods for the CRP and BMI (CRP_BMI) analyses: the multivariable MR-PRESSO method implemented in the MR-PRESSO R package [50] and robust Q-statistic minimization method implemented in the multivariable MR (MVMR) R package v0.4.1 [39].

Results

Study population

The SNP-exposure effects estimated in the CIM GWAS populations are expected to be similar to those in the MD GWAS population, as all individuals were of inferred European genetic ancestry. None of the BCAC studies included in the MD GWASs were included in any of the CIM GWASs, and any sample overlap is deemed minimal (Tables S1 & S2). The estimated Pearson correlation between CRP ln(mg/l) and BMI (kg/m2) at baseline in UKB participants (N = 466,356) was r = 0.43.

Univariable MR analysis

All instruments used in univariable MR analyses had F-statistics > 10, suggesting strong instruments (Table S3). All the SNPs included in the CIM instruments were either associated with that CIM or were in high LD (r2 > 0.8) with a SNP associated with that CIM in the GWAS Catalog or a previous publication [36,44].

In univariable random-effects IVW MR analyses, we observed nominally statistically significant (p < 0.05) associations for CCL23, GDNF, MMP-10, and TRANCE with DA, CD40 and TWEAK with NDA, and GDNF and MMP-10 with PMD (Table 1, Figure 2, & Table S4). For DA, the effect estimates for CCL23 and TRANCE were directionally consistent across multiple MR sensitivity analyses (Figure 2, Tables S4-S9). Similarly, the effect estimate for TWEAK with NDA was generally consistent across sensitivity analyses(Figure 2, Tables S4 & S5). The effect estimates for GDNF and MMP-10 were directionally consistent across DA and PMD analyses (Table S4). The instruments for GDNF, MMP-10, and CD40 each included a single SNP, precluding additional multi-SNP sensitivity analyses.

Table 1.

Nominal associations between genetically predicted circulating inflammatory markers and mammographic density phenotypes in IVW univariable and multivariable MR analyses.

Exposure1 Outcome2 Univariable MR Multivariable MR
N SNP 95% CI p-value N SNP 95% CI p-value
CCL23 DA 3 0.15 0.01, 0.28 0.039 275 −0.11 −0.21, −0.01 0.032
GDNF DA 1 0.99 0.35, 1.63 0.003 272 0.10 −0.12, 0.32 0.387
MMP-10 DA 1 0.73 0.27, 1.18 0.002 272 −0.04 −0.29, 0.21 0.769
TRANCE DA 5 −0.34 −0.60, −0.08 0.010 277 0.17 −0.10, 0.43 0.216
CD40 NDA 1 −0.78 1.57, −0.0005 0.049 272 −0.10 −0.51, 0.31 0.631
TWEAK NDA 4 0.36 0.01, 0.71 0.042 273 −0.36 −0.83, 0.11 0.132
CRP3 PMD 59 0.14 −0.03, 0.31 0.098 281 0.13 0.0003, 0.26 0.049
GDNF PMD 1 0.51 0.04, 0.99 0.034 272 0.13 −0.04, 0.30 0.137
MMP-10 PMD 1 0.60 0.26, 0.93 <0.001 272 0.05 −0.18, 0.27 0.678
Open in a new tab
1

Exposure units were ln(mg/l) for CRP and rank-based inverse normalized log2 (pg/ml) for all other CIMs

2

Outcome units were sqrt(cm2) for DA and NDA and sqrt(%) for PMD

3

The results shown here are for the CRP_noBMI instrument in univariable MR and CRP_BMI instrument for multivariable MR

Figure 2. Mendelian randomization (MR) sensitivity analysis results for CIMs with significant MR IVW results for any of the mammographic density phenotypes.

Figure 2.

Open in a new tab

CIM instruments with at least one significant (p < 0.05) association with an MD phenotype are listed on the y-axes. MR estimates and 95% confidence intervals for PMD, DA, and NDA are on the x-axes. MR methods are indicated by the point and error bar color, with the Egger, median, LASSO, IVW, and cML MR methods plotted from top to bottom for each CIM. Open circles represent MR estimates with p ≥ 0.05. Instruments for CD40, GDNF, and MMP-10 only included a single SNP and no additional sensitivity analyses could be performed. Instruments not constructed using summary statistics from the large GWAS of 91 CIMs are indicated with an asterisk.

Across remaining CIMs, univariable IVW effect estimates were generally consistent with those from MR sensitivity analyses (Tables S4-S9). Exceptions were CCL28-NDA, CRP-PMD, and FGF19-PMD, which had significant MR-Egger intercepts, indicating possible horizontal pleiotropy (Table S6). The CRP instrument also had significant MR-PRESSO global test p-values in the NDA and PMD analyses, further indicating possible horizontal pleiotropy for those associations (Table S10). We observed limited evidence of horizontal pleiotropy across most univariable MR analyses. No other CIM instruments were nominally associated with any MD outcome in IVW analyses (all p ≥ 0.05). MR-LASSO generally flagged few (0 to 24) SNPs as potentially invalid across instruments (Table S8). MR-PRESSO identified three potential outlier SNPs (rs28429148 and rs4817984) for the CRP instrument in the NDA analysis, two potential outlier SNPs (rs2393794 and rs4817984) for the CRP_noBMI instrument in the NDA analysis, and one outlier SNP (rs28429148) for the CRP instrument in the PMD analysis (Table S10).

Multivariable MR analysis

In multivariable MR analyses taking BMI into account, conditional F-statistics for all exposures except BMI and CRP were F < 10, indicating the potential for weak instrument bias (Table S3). In agreement with our previous work [37], we observed strong inverse associations between BMI and PMD (−0.81 sqrt(%) per 1 kg/m2 increase in BMI; 95% CI: −1.02, −0.61; p = 1.6 × 10−15) and DA (−0.35 sqrt(cm2) per 1 kg/m2 increase in BMI; 95% CI: −0.60, −0.10; p = 0.006), and a positive association between increased BMI and NDA (1.34 sqrt(cm2) per 1 kg/m2 increase in BMI; 95% CI: 0.95, 1.73; p = 1.1 × 10−11) in the CRP_BMI multivariable MR analysis. The estimated associations between BMI and MD outcomes were generally consistent across multivariable IVW MR models and similar to those observed in the CRP-BMI multivariable MR analysis (Table S4). We observed nominally significant association between CRP and PMD (0.131 sqrt(%) per 1 ln(mg/l) increase in CRP; 95% CI: 0.00, 0.26; p = 0.05), but no evidence of association for DA (p = 0.24) or NDA (p = 0.06) (Figure 3 & Table S4). We also observed nominally statistically significant associations for CCL23 with DA, TRAIL with NDA, and CCL20, OPG, and TRANCE with PMD, with no evidence of horizontal pleiotropy based on multivariable MR-Egger intercepts (Table S6).

Figure 3. Multivariable Mendelian randomization (MVMR) sensitivity analysis results for CIMs with significant MVMR IVW results for any of the mammographic density phenotypes.

Figure 3.

Open in a new tab

CIM exposures in MVMR instruments with at least one significant (p < 0.05) association with an MD phenotype are listed on the y-axes. MR estimates and 95% confidence intervals for PMD, DA, and NDA are on the x-axes. MVMR methods are indicated by the point and error bar color, with the Egger, median, LASSO, IVW, and cML MVMR methods plotted from top to bottom for each CIM. Open circles represent MR estimates with p ≥ 0.05. The CRP exposure has additional results from the Q-statistic minimization (MVMR Q-stat min) method listed second from the top. Instruments not constructed using summary statistics from the large GWAS of 91 CIMs are indicated with an asterisk.

Discussion

The purpose of this study was to investigate the relationships between genetically predicted CIMs and MD phenotypes using an MR approach. In this comprehensive MR analysis of 60 CIMs, we found limited evidence supporting broad causal effects of circulating inflammatory markers on mammographic density phenotypes. Although several nominal associations were observed, none survived correction for multiple testing. Overall, the consistency of largely null findings across the CIMs evaluated suggests that circulating inflammatory markers are unlikely to have broad or strong causal effects on MD phenotypes. The nominal associations identified from multi-SNP instruments in univariable MR analyses (CCL23, TRANCE, and TWEAK) were highly consistent across all of the MR methods we utilized. With the exception of CRP, none of these CIMs have previously been reported to be associated with MD phenotypes, but all have previously been linked to breast cancer biology, including pathways involving immune activation, apoptosis, and tumor progression [14–17,51–59]. However, because none of these observed associations were significant after multiple testing correction, these findings should be viewed as hypothesis-generating and require replication in independent datasets. Notably, CRP was the only CIM with previously reported observational associations with MD phenotypes for which we observed nominal associations in the present study. We did not identify associations (all p ≥ 0.05) between genetically predicted concentrations of IL-6, IL-8, TNFR2, or VEGF and any MD phenotype, suggesting that some previously reported observational associations may reflect confounding, reverse causation, or differences between circulating and local tissue-specific inflammatory processes [22–24].

The association between CCL23 and DA was the only association observed in both the univariable and multivariable MR analyses. However, the direction of the observed CCL23-DA association changed from positive in univariable MR to negative in multivariable MR analyses, further reducing confidence that this association reflects a robust causal effect. In addition, the observed negative association for PMD and TRANCE in the multivariable MR analysis conflicts with the observed positive association for DA and TRANCE in the univariable MR analysis (Figures 2 & 3). For all CIM instruments with the exception of CRP, conditional F statistics in the multivariable MR analyses indicated weak instruments. In contrast to univariable MR where weak instruments tend to bias estimates towards the null, the direction of bias from weak instruments in multivariable MR is more challenging to predict (39)39. Thus, these multivariable MR analyses should be interpreted cautiously and considered exploratory.

Our results for CRP were inconsistent across MR methods. In univariable IVW analyses, we observed no associations between CRP and MD phenotypes, whereas in multivariable MR IVW analyses including BMI, we observed a nominal positive association between CRP concentrations and PMD. The magnitude of the CRP-PMD association was generally consistent across the multivariable MR analyses. Overall, we found that inclusion of BMI in multivariable MR analyses reduced conditional instrument strength for most CIMs, highlighting the difficulty of disentangling adiposity-related pathways from systemic inflammatory processes and/or theoretical challenges with multivariable MR when the number of genome-wide significant SNPs varies widely between exposures of interest. Given the strong association between BMI and both CRP and MD phenotypes [4,26–28], it is possible that our multivariable MR results better reflect the underlying relationship between CRP and MD phenotypes, especially as the GWAS summary statistics used to construct CRP instruments were not adjusted for BMI. Indeed, it has been proposed that multivariable MR approaches may produce less biased estimates in the absence of confounder-adjusted exposure or outcome GWAS summary statistics, regardless of the bias correction approaches used for univariable MR [60]. Notably, the CRP-PMD association reached nominal significance only in the MVMR analysis, whereas the standard CRP instrument was null and the CRP_noBMI instrument produced a positive but non-significant estimate, suggesting that the estimated direct effect of CRP may be sensitive to how BMI-related genetic variation is accounted for. Further, it is also important to note that previous bidirectional MR studies only estimated the linear effects between BMI and CRP [27,28]. Non-linear relationships between BMI and CRP are possible [17], and could make modeling the BMI-CRP relationship in multivariable MR analyses more complex.

While tumor-promoting inflammation is a cancer hallmark [7,8], there is some evidence to suggest that CIMs may not be good proxies for tissue-specific inflammatory markers. An MR study of the relationships between 66 CIMs and 30 adult cancers (including breast cancer) in a European genetic ancestry population found limited evidence that CIM concentrations were associated with cancer [61]. While both local and systemic immune responses are relevant to breast cancer risk [62], associations between inflammatory marker concentrations and MD measures have generally been stronger in studies that measured levels in breast tissue [22,23] than in studies that measured CIM levels [21,24,25], suggesting that local inflammation may be more directly relevant to MD. Assessing the relationships between tissue-specific inflammatory markers and MD would therefore be informative given the potential relevance of the breast microenvironment to MD and breast cancer [7,8,22,23]. However, we are not aware of studies that provide genetic instruments for breast-specific inflammation markers and were therefore unable to assess these associations here.

Our study has several strengths. To our knowledge, this is the largest and most comprehensive MR analysis of the relationships between CIMs and MD phenotypes conducted to date. We applied multiple MR methods with varying degrees of robustness to violations of MR assumptions and used both univariable and multivariable MR approaches to help mitigate the impact of BMI on CRP estimates.

Our study also has some limitations. First, all GWAS were conducted using European genetic ancestry individuals, limiting the generalizability of results to other populations. Given the global importance of breast cancer, examining these relationships in more genetically diverse populations would be informative. In addition, MR results only represent causal effect estimates when all assumptions are satisfied [38,39,47]. The presence of pleiotropy or invalid SNPs may violate the assumptions required for causal inference. In the case of our multivariable analyses, we observed the potential for weak instrument bias in the instruments for all CIMs except CRP. While we only observed direct evidence of pleiotropy in our sensitivity analyses for a small number of CIMs, we observed evidence of heterogeneity in a larger number of MR analyses. In addition, several of the identified associations were based on single-SNP instruments, limiting our ability to assess potential horizontal pleiotropy through standard MR sensitivity analyses. Finally, the GWAS for all CIMs except CRP had modest sample sizes (N < 55,000), and larger GWAS may provide more informative IVs.

Overall, this study provides limited evidence supporting broad causal effects of circulating inflammatory markers on MD phenotypes. These findings suggest that previously reported observational associations between CIMs and MD may not reflect true causal effects, with the possible exception of CRP, for which we observed a small positive association with PMD in multivariable MR analyses. A small number of markers may warrant follow-up. Replication in independent datasets and evaluation using larger GWAS will help determine whether any of the nominally significant signals identified here represent robust causal associations.

Supplementary Material

Supporting Information

Table S1: GWAS summary statistic sources and information.

Table S2: Sample sizes and information about BCAC studies included in outcome MD phenotype meta-analysis GWAS summary statistics.

Table S3: MR and MVMR instrument F-statistics and SNPs.

Table S4: IVW MR and MVMR results.

Table S5: Single SNP MR wald ratio results.

Table S6: MR- and MVMR-Egger results.

Table S7: MR and MVMR median method results.

Table S8: MR- and MVMR-LASSO results.

Table S9: MR- and MVMR-cML results.

Table S10: MR- and MVMR-PRESSO results.

Table S11: MVMR GMM estimator results.

Table S12: MVMR Q-statistic minimization method results.

Acknowledgements

Funding acknowledgements

This work was supported by CA244670. A.H.S. was supported by grant number T32CA009168 from the NCI, NIH. The study’s contents are solely the responsibility of the authors and do not necessarily represent the official views of the NCI, NIH. This research was supported [in part] by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH author(s) are considered Works of the United States Government. The findings and conclusions presented in this paper are those of the author(s) and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services.

BCAC acknowledgements

We thank all the individuals who took part in these studies and all the researchers, clinicians, technicians, and administrative staff who have enabled this work to be carried out.

ABCFS thanks Maggie Angelakos, Judi Maskiell, and Gillian Dite; BBCS thanks Eileen Williams, Elaine Ryder-Mills, and Kara Sargus.

BCEES thanks Lin Fritschi, Jane Heyworth, and BreastScreen Western Australia.

The BREOGAN study would not have been possible without the contributions of the following: Manuela Gago-Dominguez, Jose Esteban Castelao, Angel Carracedo, Victor Muñoz Garzón, Alejandro Novo Domínguez, Maria Elena Martinez, Sara Miranda Ponte, Carmen Redondo Marey, Maite Peña Fernández, Manuel Enguix Castelo, Maria Torres, Manuel Calaza (BREOGAN), José Antúnez, Máximo Fraga, and the staff of the Department of Pathology and Biobank of the University Hospital Complex of Santiago-CHUS, Instituto de Investigación Sanitaria de Santiago, IDIS, Xerencia de Xestion Integrada de Santiago-SERGAS; Joaquín González-Carreró and the staff of the Department of Pathology and Biobank of University Hospital Complex of Vigo, Instituto de Investigacion Biomedica Galicia Sur, SERGAS, Vigo, Spain.

CBCS thanks study participants, co-investigators, collaborators, staff of the Canadian Breast Cancer Study, and project coordinators Agnes Lai and Celine Morissette.

FHRISK and PROCAS thank NIHR for funding.

kConFab/AOCS wish to thank Heather Thorne, Eveline Niedermayr, all the kConFab research nurses and staff, the heads and staff of the Family Cancer Clinics, and the Clinical Follow-Up Study [which has received funding from the NHMRC, the National Breast Cancer Foundation, Cancer Australia, and the National Institute of Health (USA)] for their contributions to this resource and the many families who contribute to kConFab.

The MCCS was made possible by the contribution of many people, including the original investigators, the teams that recruited the participants and continue working on follow-up and the many thousands of Melbourne residents who continue to participate in the study.

We thank the coordinators, the research staff, and especially the MMHS participants for their continued collaboration on research studies in breast cancer.

The following are NBCS Collaborators: Kristine K. Sahlberg (PhD), Anne-Lise Børresen-Dale (Prof. Em.), Lars Ottestad (MD), Rolf Kåresen (Prof. Em.), Dr. Ellen Schlichting (MD), Marit Muri Holmen (MD), Toril Sauer (MD), Vilde Haakensen (MD), Olav Engebråten (MD), Bjørn Naume (MD), Alexander Fosså (MD), Cecile E. Kiserud (MD), Kristin V. Reinertsen (MD), Åslaug Helland (MD), Margit Riis (MD), Jürgen Geisler (MD), OSBREAC, and Grethe I. Grenaker Alnæs (MSc).

For NHS and NHS2, the study protocol was approved by the Institutional Review Boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health and those of participating registries as required. We would like to thank the participants and staff of the NHS and NHS2 for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, and WY. The authors assume full responsibility for analyses and interpretation of these data.

The OFBCR thanks Gord Glendon, Nayana Weerasooriya, Teresa Selander and the Sinai Health Biospecimen Repository.

PBCS thanks Louise Brinton, Mark Sherman, Neonila Szeszenia-Dabrowska, Beata Peplonska, Witold Zatonski, Pei Chao, and Michael Stagner.

SASBAC thanks the Swedish Medical Research Counsel.

We thank the SEARCH and EPIC teams.

UKBGS thanks Breast Cancer Now and the Institute of Cancer Research for support and funding of the Generations Study and the study participants, study staff, and the doctors, nurses, and other health care providers and health information sources who have contributed to the study. We acknowledge NHS funding to the Royal Marsden/ICR NIHR Biomedical Research Center.

BCAC funding acknowledgements

BCAC is funded by the European Union’s Horizon 2020 Research and Innovation Program (Grant Numbers 634935 and 633784 for BRIDGES and B-CAST, respectively) and the PERSPECTIVE I&I Project, funded by the Government of Canada through Genome Canada and the Canadian Institutes of Health Research, the Ministère de l’Économie et de l’Innovation du Québec through Genome Québec, and the Quebec Breast Cancer Foundation. The EU Horizon 2020 Research and Innovation Program funding source had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Additional funding for BCAC is provided by Cancer Research UK Grant PPRPGM-Nov20/100002 and via the Confluence Project, which is funded with intramural funds from the National Cancer Institute Intramural Research Program, National Institutes of Health.

Genotyping of the OncoArray was funded by the NIH Grant U19 CA148065 and Cancer Research UK Grant C1287/A16563 and the PERSPECTIVE Project supported by the Government of Canada through Genome Canada and the Canadian Institutes of Health Research (Grant GPH-129344), the Ministère de l’Économie, Science et Innovation du Québec through Genome Québec and the PSRSIIRI-701 Grant, and the Quebec Breast Cancer Foundation. Funding for iCOGS came from: the European Community’s Seventh Framework Program under Grant Agreement n° 223175 (HEALTH-F2–2009-223175) (COGS), Cancer Research UK (C1287/A10118, C1287/A10710, C12292/A11174, C1281/A12014, C5047/A8384, C5047/A15007, C5047/A10692, C8197/A16565), the National Institutes of Health (CA128978) and Post-Cancer GWAS Initiative (1U19 CA148537, 1U19 CA148065, and 1U19 CA148112—the GAME-ON Initiative), the Department of Defense (W81XWH-10–1-0341), the Canadian Institutes of Health Research (CIHR) for the CIHR Team in Familial Risks of Breast Cancer, and Komen Foundation for the Cure, the Breast Cancer Research Foundation, and the Ovarian Cancer Research Fund.

The Australian Breast Cancer Family Study (ABCFS) was supported by Grant UM1 CA164920 from the National Cancer Institute (USA). The content of this manuscript does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the Breast Cancer Family Registry (BCFR), nor does mention of trade names, commercial products, or organizations imply endorsement by the USA Government or the BCFR. The ABCFS was also supported by the National Health and Medical Research Council of Australia, the New South Wales Cancer Council, the Victorian Health Promotion Foundation (Australia), and the Victorian Breast Cancer Research Consortium. J.L.H. is a National Health and Medical Research Council (NHMRC) Senior Principal Research Fellow. M.C.S. is a NHMRC Senior Research Fellow.

The work of the BBCC was partly funded by ELAN-Fond of the University Hospital of Erlangen.

The BCEES was funded by the National Health and Medical Research Council, Australia and the Cancer Council Western Australia and acknowledges funding from the National Breast Cancer Foundation (JS).

The BREast Oncology GAlician Network (BREOGAN) is funded by Acción Estratégica de Salud del Instituto de Salud Carlos III FIS PI12/02125/Cofinanciado and FEDER PI17/00918/Cofinanciado FEDER; Acción Estratégica de Salud del Instituto de Salud Carlos III FIS Intrasalud (PI13/01136); Programa Grupos Emergentes, Cancer Genetics Unit, Instituto de Investigacion Biomedica Galicia Sur. Xerencia de Xestion Integrada de Vigo-SERGAS, Instituto de Salud Carlos III, Spain; Grant 10CSA012E, Consellería de Industria Programa Sectorial de Investigación Aplicada, PEME I+D e I+D Suma del Plan Gallego de Investigación, Desarrollo e Innovación Tecnológica de la Consellería de Industria de la Xunta de Galicia, Spain; Grant EC11–192. Fomento de la Investigación Clínica Independiente, Ministerio de Sanidad, Servicios Sociales e Igualdad, Spain; and Grant FEDER-Innterconecta. Ministerio de Economia y Competitividad, Xunta de Galicia, Spain.

CBCS is funded by the Canadian Cancer Society (Grant # 313404) and the Canadian Institutes of Health Research.

FHRISK and PROCAS are funded from NIHR grant PGfAR 0707–10031. DGE, AH and WGN are supported by the NIHR Manchester Biomedical Research Centre (IS-BRC-1215–20007).

kConFab is supported by a Grant from the National Breast Cancer Foundation and previously by the National Health and Medical Research Council (NHMRC), the Queensland Cancer Fund, the Cancer Councils of New South Wales, Victoria, Tasmania and South Australia, and the Cancer Foundation of Western Australia. Financial support for the AOCS was provided by the United States Army Medical Research and Materiel Command (DAMD17–01-1–0729), Cancer Council Victoria, Queensland Cancer Fund, Cancer Council New South Wales, Cancer Council South Australia, The Cancer Foundation of Western Australia, Cancer Council Tasmania, and the National Health and Medical Research Council of Australia (NHMRC; 400413, 400281, 199600). G.C.T. and P.W. are supported by the NHMRC. RB was a Cancer Institute NSW Clinical Research Fellow.

MCBCS was supported by the NIH Grants R35CA253187, R01CA192393, R01CA116167, and R01CA176785, a NIH Specialized Program of Research Excellence (SPORE) in Breast Cancer (P50CA116201), and the Breast Cancer Research Foundation.

The Melbourne Collaborative Cohort Study (MCCS) Cohort Recruitment was funded by VicHealth and Cancer Council Victoria. The MCCS was further augmented by Australian National Health and Medical Research Council Grants 209057, 396414, and 1074383 and by infrastructure provided by Cancer Council Victoria. Cases and their vital status were ascertained through the Victorian Cancer Registry.

The MEC was supported by NIH Grants CA63464, CA54281, CA098758, CA132839, and CA164973.

The MMHS study was supported by NIH Grants CA97396, CA128931, CA116201, CA140286, and CA177150.

The NBCS has received funding from the K.G. Jebsen Center for Breast Cancer Research; the Research Council of Norway Grant 193387/V50 (to A.-L. Børresen-Dale and V.N. Kristensen) and Grant 193387/H10 (to A.-L. Børresen-Dale and V.N. Kristensen), South Eastern Norway Health Authority (Grant 39346 to A.-L. Børresen-Dale), and the Norwegian Cancer Society (to A.-L. Børresen-Dale and V.N. Kristensen).

The Northern California Breast Cancer Family Registry (NC-BCFR) and Ontario Familial Breast Cancer Registry (OFBCR) were supported by Grant U01CA164920 from the USA National Cancer Institute of the National Institutes of Health. The content of this manuscript does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the Breast Cancer Family Registry (BCFR), nor does mention of trade names, commercial products, or organizations imply endorsement by the USA Government or the BCFR.

The NHS was supported by NIH Grants P01 CA87969, UM1 CA186107, R01CA131332, and U19 CA148065. The NHS2 was supported by NIH Grants UM1 CA176726, R01 CA124865, and U19 CA148065.

The PBCS was funded by the Intramural Research Program of the National Cancer Institute, National Institutes of Health, Department of Health and Human Services, USA.

The SASBAC study was supported by funding from the Agency for Science, Technology, and Research of Singapore (A*STAR), the US National Institute of Health (NIH) and the Susan G. Komen Breast Cancer Foundation.

SEARCH is funded by Cancer Research UK (C490/A10124, C490/A16561) and supported by the UK National Institute for Health Research Biomedical Research Center at the University of Cambridge. The University of Cambridge has received salary support for PMDPP from the NHS in the East of England through the Clinical Academic Reserve.

The UKBGS is funded by Breast Cancer Now and the Institute of Cancer Research (ICR), London. ICR acknowledges NHS funding to the NIHR Biomedical Research Center.

UK Biobank acknowledgement

We are grateful to the UK Biobank participants who have generously agreed to provide a broad range of information for health-related research. This research was conducted using the UK Biobank Resource under Application 95770.

Footnotes

Disclosure of Potential Conflicts of Interest

P.A.F. conducts research funded by Amgen, Novartis and Pfizer and received Honoraria from Roche, Novartis and Pfizer. R.A.M. is a consultant for Pharmavite. S.Lo. receives research funding to institution from Novartis, Bristol Myers Squibb, Astra Zeneca/Daiichi Sankyo, Roche-Genentech, MSD, Pfizer, Gilead Sciences, Nektar Therapeutics and Eli Lilly and has acted as consultant to Roche-Genentech, MSD, Gilead Sciences, Astra Zeneca/Daiichi Sankyo, Bristol Myers Squibb, Novartis, Eli Lilly, Amaroq Therapeutics, Mersana Therapeutics, Domain Therapeutics, BioNTech, Bicycle Therapeutics, Exact Sciences, Menari Asia-Pacific, SAGA Diagnostics, Adanate. The other authors declare no conflicts of interest.

Data Availability Statement

We used publicly available GWAS summary statistics from individual study, GWAS Catalog, and IEU OpenGWAS project databases for circulating inflammatory markers and BMI as cited. The BCAC data used to generate the mammographic density GWAS summary statistics used in this study are available via application to the Data Access and Co-ordination Committee (BCAC@medschl.cam.ac.uk).

Code Availability Statement

Code used in these analyses are publicly available in our project GitHub repository (https://github.com/UW-Epidemiology/MD_CIM_MR).

References

  • 1.Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: A Cancer Journal for Clinicians. 2024;74: 229–263. doi: 10.3322/caac.21834 [DOI] [PubMed] [Google Scholar]
  • 2.Boyd NF, Guo H, Martin LJ, Sun L, Stone J, Fishell E, et al. Mammographic Density and the Risk and Detection of Breast Cancer. In: The New England Journal of Medicine [Internet]. Massachusetts Medical Society; 18 Jan 2007. [cited 14 June 2024]. doi: 10.1056/NEJMoa062790 [DOI] [PubMed] [Google Scholar]
  • 3.Bond-Smith D, Stone J. Methodological Challenges and Updated Findings from a Meta-analysis of the Association between Mammographic Density and Breast Cancer. Cancer Epidemiol Biomarkers Prev. 2019;28: 22–31. doi: 10.1158/1055-9965.EPI-17-1175 [DOI] [PubMed] [Google Scholar]
  • 4.Bodewes FTH, van Asselt AA, Dorrius MD, Greuter MJW, de Bock GH. Mammographic breast density and the risk of breast cancer: A systematic review and meta-analysis. Breast. 2022;66: 62–68. doi: 10.1016/j.breast.2022.09.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Strandberg R, Illipse M, Czene K, Hall P, Humphreys K. Influence of mammographic density and compressed breast thickness on true mammographic sensitivity: a cohort study. Scientific Reports. 2023;13: 14194. doi: 10.1038/s41598-023-41356-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Force UPST, Nicholson WK, Silverstein M, Wong JB, Barry MJ, Chelmow D, et al. Screening for Breast Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2024;331: 1918–1930. doi: 10.1001/jama.2024.5534 [DOI] [PubMed] [Google Scholar]
  • 7.Hanahan D. Hallmarks of cancer—Then and now, and beyond. Cell. 2026;189: 2254–2277. doi: 10.1016/j.cell.2025.12.049 [DOI] [PubMed] [Google Scholar]
  • 8.Hanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov. 2022;12: 31–46. doi: 10.1158/2159-8290.CD-21-1059 [DOI] [PubMed] [Google Scholar]
  • 9.Jiang X, Shapiro DJ. The immune system and inflammation in breast cancer. Molecular and cellular endocrinology. 2013;382: 673. doi: 10.1016/j.mce.2013.06.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Gunter MJ, Wang T, Cushman M, Xue X, Wassertheil-Smoller S, Strickler HD, et al. Circulating Adipokines and Inflammatory Markers and Postmenopausal Breast Cancer Risk. J Natl Cancer Inst. 2015;107: djv169. doi: 10.1093/jnci/djv169 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kim E-S, Kim SY, Moon A. C-Reactive Protein Signaling Pathways in Tumor Progression. Biomol Ther (Seoul). 2023;31: 473–483. doi: 10.4062/biomolther.2023.132 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Sproston NR, Ashworth JJ. Role of C-Reactive Protein at Sites of Inflammation and Infection. Front Immunol. 2018;9: 754. doi: 10.3389/fimmu.2018.00754 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Salazar J, Martínez MS, Chávez-Castillo M, Núñez V, Añez R, Torres Y, et al. C-Reactive Protein: An In-Depth Look into Structure, Function, and Regulation. International Scholarly Research Notices. 2014;2014: 653045. doi: 10.1155/2014/653045 [DOI] [Google Scholar]
  • 14.Wang J, Lee I-M, Tworoger SS, Buring JE, Ridker PM, Rosner B, et al. Plasma C - reactive Protein and Risk of Breast Cancer in two Prospective Studies and a Meta-analysis. Cancer Epidemiol Biomarkers Prev. 2015;24: 1199–1206. doi: 10.1158/1055-9965.EPI-15-0187 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Price TR, Friedenreich CM, Robson PJ, Li H, Brenner DR. High-sensitivity C-reactive protein, hemoglobin A1c and breast cancer risk: a nested case–control study from Alberta’s Tomorrow Project cohort. Cancer Causes Control. 2020;31: 1057–1068. doi: 10.1007/s10552-020-01329-6 [DOI] [PubMed] [Google Scholar]
  • 16.Jiang R, Shen J, Wang X, Chen S, Wu S, Cai H. Association between body mass index combined with high-sensitivity C-reactive protein and the risk of postmenopausal breast cancer: A prospective cohort study. Mol Clin Oncol. 2024;21: 64. doi: 10.3892/mco.2024.2762 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Zhang Y, Song M, Yang Z, Huang X, Lin Y, Yang H. Healthy lifestyles, systemic inflammation and breast cancer risk: a mediation analysis. BMC Cancer. 2024;24: 208. doi: 10.1186/s12885-024-11931-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Kehm RD, McDonald JA, Fenton SE, Kavanaugh-Lynch M, Leung KA, McKenzie KE, et al. Inflammatory Biomarkers and Breast Cancer Risk: A Systematic Review of the Evidence and Future Potential for Intervention Research. Int J Environ Res Public Health. 2020;17: 5445. doi: 10.3390/ijerph17155445 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Fontvieille E, His M, Biessy C, Navionis A-S, Torres-Mejía G, Ángeles-Llerenas A, et al. Inflammatory biomarkers and risk of breast cancer among young women in Latin America: a case-control study. BMC Cancer. 2022;22: 877. doi: 10.1186/s12885-022-09975-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Peng J, Madduri S, Clontz AD, Stewart DA. Clinical trial-identified inflammatory biomarkers in breast and pancreatic cancers. Front Endocrinol (Lausanne). 2023;14: 1106520. doi: 10.3389/fendo.2023.1106520 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Reeves KW, Weissfeld JL, Modugno F, Diergaarde B. Circulating levels of inflammatory markers and mammographic density among postmenopausal women. Breast cancer research and treatment. 2010;127: 555. doi: 10.1007/s10549-010-1249-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Hanna M, Dumas I, Orain M, Jacob S, Têtu B, Sanschagrin F, et al. Association between expression of inflammatory markers in normal breast tissue and mammographic density among premenopausal and postmenopausal women. Menopause. 2017;24: 524. doi: 10.1097/GME.0000000000000794 [DOI] [PubMed] [Google Scholar]
  • 23.Abrahamsson A, Rzepecka A, Romu T, Borga M, Leinhard OD, Lundberg P, et al. Dense breast tissue in postmenopausal women is associated with a pro-inflammatory microenvironment in vivo. Oncoimmunology. 2016;5: e1229723. doi: 10.1080/2162402X.2016.1229723 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Fang CY, Egleston BL, Byrne C, Bohr GS, Pathak HB, Godwin AK, et al. Inflammation and Breast Density among Female Chinese Immigrants: Exploring Variations across Neighborhoods. Cancer Causes Control. 2019;30: 1113–1126. doi: 10.1007/s10552-019-01206-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Makboon K, Gold EB, Harvey DJ, Butler LM, Habel LA. Association between high-sensitivity C-reactive protein (hsCRP) and change in mammographic density over time in the SWAN mammographic density subcohort. Cancer causes & control : CCC. 2015;26: 431. doi: 10.1007/s10552-015-0522-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Choi J, Joseph L, Pilote L. Obesity and C-reactive protein in various populations: a systematic review and meta-analysis. Obesity Reviews. 2013;14: 232–244. doi: 10.1111/obr.12003 [DOI] [PubMed] [Google Scholar]
  • 27.Welsh P, Polisecki E, Robertson M, Jahn S, Buckley BM, de Craen AJM, et al. Unraveling the Directional Link between Adiposity and Inflammation: A Bidirectional Mendelian Randomization Approach. J Clin Endocrinol Metab. 2010;95: 93–99. doi: 10.1210/jc.2009-1064 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Timpson NJ, Nordestgaard BG, Harbord RM, Zacho J, Frayling TM, Tybjærg-Hansen A, et al. C-reactive protein levels and body mass index: Elucidating direction of causation through reciprocal Mendelian randomization. Int J Obes (Lond). 2011;35: 300–308. doi: 10.1038/ijo.2010.137 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Skrivankova VW, Richmond RC, Woolf BAR, Yarmolinsky J, Davies NM, Swanson SA, et al. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement. JAMA. 2021;326: 1614–1621. doi: 10.1001/jama.2021.18236 [DOI] [PubMed] [Google Scholar]
  • 30.MRC IEU UK Biobank GWAS pipeline version 2. In: University of Bristol; [Internet]. [cited 6 Mar 2025]. Available: https://research-information.bris.ac.uk/en/datasets/mrc-ieu-uk-biobank-gwas-pipeline-version-2 [Google Scholar]
  • 31.Said S, Pazoki R, Karhunen V, Võsa U, Ligthart S, Bodinier B, et al. Genetic analysis of over half a million people characterises C-reactive protein loci. Nature Communications. 2022;13: 2198. doi: 10.1038/s41467-022-29650-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Folkersen L, Gustafsson S, Wang Q, Hansen DH, Hedman ÅK, Schork A, et al. Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals. Nat Metab. 2020;2: 1135–1148. doi: 10.1038/s42255-020-00287-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Zhao JH, Stacey D, Eriksson N, Macdonald-Dunlop E, Hedman ÅK, Kalnapenkis A, et al. Genetics of circulating inflammatory proteins identifies drivers of immune-mediated disease risk and therapeutic targets. Nat Immunol. 2023;24: 1540–1551. doi: 10.1038/s41590-023-01588-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife. 2018;7: e34408. doi: 10.7554/eLife.34408 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Elsworth B, Lyon M, Alexander T, Liu Y, Matthews P, Hallett J, et al. The MRC IEU OpenGWAS data infrastructure. bioRxiv. 2020; 2020.08.10.244293. doi: 10.1101/2020.08.10.244293 [DOI] [Google Scholar]
  • 36.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 Res. 2019;47: D1005–D1012. doi: 10.1093/nar/gky1120 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Haas CB, Chen H, Harrison T, Fan S, Gago-Dominguez M, Castelao JE, et al. Disentangling the relationships of body mass index and circulating sex hormone concentrations in mammographic density using Mendelian randomization. Breast Cancer Res Treat. 2024;206: 295–305. doi: 10.1007/s10549-024-07306-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Sanderson E, Smith GD, Windmeijer F, Bowden J. An examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings. International Journal of Epidemiology. 2018;48: 713. doi: 10.1093/ije/dyy262 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Sanderson E, Spiller W, Bowden J. Testing and correcting for weak and pleiotropic instruments in two sample multivariable Mendelian randomization. Statistics in Medicine. 2021;40: 5434. doi: 10.1002/sim.9133 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2023. Available: https://www.R-project.org/ [Google Scholar]
  • 41.Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MAR, Bender D, et al. PLINK: A Tool Set for Whole-Genome Association and Population-Based Linkage Analyses. Am J Hum Genet. 2007;81: 559–575. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience. 2015;4: s13742–015-0047–8. doi: 10.1186/s13742-015-0047-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.The 1000 Genomes Project Consortium, Corresponding authors, Auton A, Abecasis GR, Steering committee, Altshuler DM, et al. A global reference for human genetic variation. Nature. 2015;526: 68–74. doi: 10.1038/nature15393 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Machiela MJ, Chanock SJ. LDlink: a web-based application for exploring population-specific haplotype structure and linking correlated alleles of possible functional variants. Bioinformatics. 2015;31: 3555–3557. doi: 10.1093/bioinformatics/btv402 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Burgess S, Thompson SG. Multivariable Mendelian Randomization: The Use of Pleiotropic Genetic Variants to Estimate Causal Effects. American Journal of Epidemiology. 2015;181: 251. doi: 10.1093/aje/kwu283 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Patel A, Ye T, Xue H, Lin Z, Xu S, Woolf B, et al. MendelianRandomization v0.9.0: updates to an R package for performing Mendelian randomization analyses using summarized data. Wellcome Open Res. 2023;8: 449. doi: 10.12688/wellcomeopenres.19995.2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Burgess S, Thompson SG. Interpreting findings from Mendelian randomization using the MR-Egger method. European Journal of Epidemiology. 2017;32: 377. doi: 10.1007/s10654-017-0255-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Rees JMB, Wood AM, Dudbridge F, Burgess S. Robust methods in Mendelian randomization via penalization of heterogeneous causal estimates. PLoS ONE. 2019;14: e0222362. doi: 10.1371/journal.pone.0222362 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.H X, X S, W P. Constrained maximum likelihood-based Mendelian randomization robust to both correlated and uncorrelated pleiotropic effects. American journal of human genetics. 2021;108. doi: 10.1016/j.ajhg.2021.05.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Verbanck M, Chen C-Y, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50: 693–698. doi: 10.1038/s41588-018-0099-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Piskór BM, Przylipiak A, Dąbrowska E, Sidorkiewicz I, Niczyporuk M, Szmitkowski M, et al. Plasma Level of MMP-10 May Be a Prognostic Marker in Early Stages of Breast Cancer. J Clin Med. 2020;9: 4122. doi: 10.3390/jcm9124122 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Shin SY, Lee DH, Lee J, Choi C, Kim J-Y, Nam J-S, et al. C-C motif chemokine receptor 1 (CCR1) is a target of the EGF-AKT-mTOR-STAT3 signaling axis in breast cancer cells. Oncotarget. 2017;8: 94591–94605. doi: 10.18632/oncotarget.21813 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Zhao X, Li Y, Feng Y, Lv S. Elevated CCL20 expression was associated with poor prognosis for breast cancer. J Chin Med Assoc. 2025;88: 469–480. doi: 10.1097/JCMA.0000000000001233 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Lam C, Lanchoney O, Maddipatla V, Markosyan N, Joshi N, Hu X, et al. Agonistic CD40 elicits CD8+ T-cell-dependent primary responses and CD4+ T-cell-dependent long-term immunity in breast cancer. npj Breast Cancer. 2026;12: 25. doi: 10.1038/s41523-025-00889-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Fielder GC, Yang TW-S, Razdan M, Li Y, Lu J, Perry JK, et al. The GDNF Family: A Role in Cancer? Neoplasia. 2017;20: 99–117. doi: 10.1016/j.neo.2017.10.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Sim N, Carter J-M, Deka K, Tan BKT, Sim Y, Tan S-M, et al. TWEAK/Fn14 signalling driven super-enhancer reprogramming promotes pro-metastatic metabolic rewiring in triple-negative breast cancer. Nat Commun. 2024;15: 5638. doi: 10.1038/s41467-024-50071-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Geerts D, Cusick JK, Connelly L. Editorial: The Tumor Necrosis Factor Superfamily: An Increasing Role in Breast Cancer. Front Oncol. 2020;10: 622588. doi: 10.3389/fonc.2020.622588 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Radhi JH, El-Hagrasy AMA, Almosawi SH, Alhashel A, Butler AE. The Role of Osteoprotegerin in Breast Cancer: Genetic Variations, Tumorigenic Pathways, and Therapeutic Potential. Cancers (Basel). 2025;17: 337. doi: 10.3390/cancers17030337 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Rahman M, Davis SR, Pumphrey JG, Bao J, Nau MM, Meltzer PS, et al. TRAIL induces apoptosis in triple-negative breast cancer cells with a mesenchymal phenotype. Breast Cancer Res Treat. 2009;113: 217–230. doi: 10.1007/s10549-008-9924-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Wang P, Lin Z, Pan W. Unbiased causal inference with Mendelian randomization and covariate-adjusted GWAS data. HGG Adv. 2025;6: 100412. doi: 10.1016/j.xhgg.2025.100412 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Yarmolinsky J, Robinson JW, Mariosa D, Karhunen V, Huang J, Dimou N, et al. Association between circulating inflammatory markers and adult cancer risk: a Mendelian randomization analysis. eBioMedicine. 2024;100: 104991. doi: 10.1016/j.ebiom.2024.104991 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Gerashchenko T, Frolova A, Patysheva M, Fedorov A, Stakheyeva M, Denisov E, et al. Breast Cancer Immune Landscape: Interplay Between Systemic and Local Immunity. Advanced Biology. 2024;8: 2400140. doi: 10.1002/adbi.202400140 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

We used publicly available GWAS summary statistics from individual study, GWAS Catalog, and IEU OpenGWAS project databases for circulating inflammatory markers and BMI as cited. The BCAC data used to generate the mammographic density GWAS summary statistics used in this study are available via application to the Data Access and Co-ordination Committee (BCAC@medschl.cam.ac.uk).

Articles from medRxiv are provided here courtesy of Cold Spring Harbor Laboratory Preprints

RESOURCES