Circulating inflammatory and immune response proteins and endometrial cancer risk: a nested case-control study and Mendelian randomization analyses

Abstract

Background: Inflammation and immune dysregulation are hypothesized contributors to endometrial carcinogenesis; however, the precise underlying mechanisms remain unclear.

Methods: We measured pre-diagnostically 152 plasma protein biomarkers in 624 endometrial cancer case-control pairs nested within the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Odds ratios (ORs) were estimated using conditional logistic regression, accounting for confounding and multiple comparisons. Proteins considered as associated with endometrial cancer risk were further tested in a two-sample Mendelian randomization (MR) analysis using summary data from the UK Biobank (n = 52,363) and the Endometrial Cancer Association Consortium (12,270 cases and 46,126 controls).

Findings: In the EPIC nested case-control study, IL-6 [OR per NPX (doubling of concentration) = 1.28 (95% confidence interval (CI) 1.03-1.57)], HGF [1.48 (1.06-2.07)], PIK3AP1 [1.22 (1.00-1.50)] and CLEC4G [1.52 (1.00-2.32)] were positively associated; HSD11B1 [0.67 (0.49-0.91)], SCF [0.68 (0.49-0.94)], and CCL25 [0.80 (0.65-0.99)] were inversely associated with endometrial cancer risk; all estimates had multiple comparisons adjusted P-value > 0.05. In complementary MR analysis, IL-6 [OR per inverse-rank normalized NPX = 1.19 (95% CI 1.04-1.36)] and HSD11B1 [0.91 (0.84-0.99)] were associated with endometrial cancer risk.

Interpretation: Altered IL-6 signalling and reduced glucocorticoid activity via HSD11B1 might play important roles in endometrial carcinogenesis.

Funding: Funding for IIG_FULL_2021_008 was obtained from Wereld Kanker Onderzoek Fonds (WKOF), as part of the World Cancer Research Fund International grant programme; Funding for INCA_15849 was obtained from Institut National du Cancer (INCa).

Keywords: Endometrial cancer; HSD11B1; Interleukin-6; Mendelian randomisation; Proteomics.

Fig. 1

The association between plasma protein levels and endometrial cancer risk in nested case-control study, estimated from (a) univariable regression, plotted against Q-values; (b) minimally adjusted model¹, plotted against Q-values; (c) fully adjusted model², plotted against P-values; (d) fully adjusted model² plus C-peptide, plotted against P-values. ¹Minimally adjusted model included physical activity, smoking status, age at menarche, parity, ever use of oral contraceptives, and ever use of hormone replacement therapy as covariates; matching factors (i.e., study centre, age, menopausal status, fasting status, and time of blood collection) were accounted using conditional logistic regression. ²Fully adjusted model included all variables from minimally adjusted model plus BMI.

Fig. 2

(a) Composition of the 44 protein clusters, including 31 clusters of correlated proteins and 13 individual proteins. The heat map represents the Pearson correlation between each cluster centre (i.e., the 1st principal component) and their respective proteins; (b) Odds ratio (OR) and 95% confidence interval (CI) for the association between each cluster and endometrial cancer risk, estimated from the primary analysis model including body mass index, physical activity, smoking status, age at menarche, parity, ever use of oral contraceptives, and ever use of hormone replacement therapy as covariates. Clusters are ranked by the proportion out of 1000 bootstrap samples that were selected by LASSO regression.

Fig. 3

The association between genetically instrumented plasma protein levels and endometrial cancer risk in Mendelian randomization analyses. ¹The OR (95% CI) estimated from MR analyses are not comparable with those estimated from the nested case-control study, as the units used for the GWAS are different. The UKB-PPP GWAS for plasma proteins were measured in inverse-rank normalized NPX. The CHARGE + UK Biobank GWAS for CRP was measured in natural log-transformed CRP. ²No cis variant for IL-6 and HSD11B1 reached the P < 5 × 10⁻⁸ threshold. These estimates were obtained using trans variants selected using linkage disequilibrium clumping at r < 0.001. ³CRP-weighted IL6R results were obtained using GWAS data for CRP (CHARGE + UKB). The estimates have been inverted to correspond to endometrial cancer risk for decrease in CRP, a proxy measure for increase in plasma IL-6R.