Prognostic role of pre-diagnostic circulating inflammatory biomarkers in breast cancer survival: evidence from the EPIC cohort study

Abstract

Background: Inflammation influences tumour progression and cancer prognosis, but its role preceding breast cancer (BC) and its prognostic implications remain inconclusive.

Methods: We studied pre-diagnostic plasma inflammatory biomarkers in 1538 women with BC from the EPIC study. Cox proportional hazards models assessed their relationship with all-cause and BC-specific mortality, adjusting for tumour characteristics and lifestyle factors.

Results: Over a 7-year follow-up after diagnosis, 229 women died, 163 from BC. Elevated IL-6 levels were associated with increased all-cause mortality risk (HR_1-SD 1.25, 95% CI 1.07-1.47). Among postmenopausal, IL-6 was associated with higher all-cause (HR_1-SD 1.41, 95% CI 1.18-1.69) and BC-specific mortality (HR_1-SD 1.31, 95% CI 1.03-1.66), (P_{Heterogeneity (pre/postmenopausal)} < 0.05 for both), while IL-10 and TNFα were associated with all-cause mortality only (HR_1-SD 1.19, 95% CI 1.02-1.40 and HR_1-SD 1.28, 95% CI 1.06-1.56). Among ER+PR+, IL-10 was associated with all-cause and BC-specific mortality (HR_1-SD 1.35, 95% CI 1.10-1.65 and HR_1-SD 1.42 95% CI 1.08-1.86), while TNF-α was associated with all-cause mortality in HER2- (HR_1-SD 1.31, 95% CI 1.07-1.61). An inflammatory score predicted higher all-cause mortality, especially in postmenopausal women (HR_1-SD 1.30, 95% CI 1.07-1.58).

Conclusions: Higher pre-diagnosis IL-6 levels suggest poorer long-term survival among BC survivors. In postmenopausal survivors, elevated IL-6, IL-10, and TNFα and inflammatory scores seem to predict all-cause mortality.

Fig. 1. Spearman correlations of log-transformed inflammatory biomarkers adjusted for laboratory batch and age at blood collection.

*Asterisks indicate non-statistically significant correlations (p > 0.05).

Fig. 2. Associations between inflammatory biomarkers (per 1-SD increment) and all-cause and breast cancer-specific mortality among breast cancer cases with available inflammatory biomarkers in EPIC.

All individual biomarkers and scores have been assessed on a continuous scale, per 1-SD increase (unless for IL-13, dichotomised based on values above and below the LOQ). Models are stratified by country, menopausal status at diagnosis and stage of tumour (metastatic, non-metastatic, unknown), and adjusted for age at diagnosis, laboratory batch, fasting status at blood collection, education level, physical activity, body mass index, alcohol consumption, smoking status and intensity, ever use of hormones for menopause at diagnosis, cancer grade, and tumour receptor status (ER, PR, HER2). The Inflammatory scores are derived from log-transformed biomarker concentrations standardised using the mean and standard deviation of our population (z-score) of 1538 BC cases; the z-scores for adiponectin were multiplied by −1 to account for its anti-inflammatory effect. These z-scores are summed to generate an overall inflammatory score for each individual. The Inflammatory score-10 is composed of the following biomarkers: IL6, IL8, IL10, IL17D, IL-1RA, IFN-γ, TNFα, CRP, Leptin, and Adiponectin; the Inflammatory score-8 includes the same components as the Inflammatory score-10, except for the adipokines Leptin and Adiponectin. BC breast cancer, HR hazard ratio, CI confidence interval, Inflam.score Inflammatory score, SD standard deviation, IL interleukin, IL-1RA interleukin-1 receptor antagonist, IFN‐γ interferon‐gamma, TNF-α tumour necrosis factor alpha, CRP C-reactive protein.

Fig. 3. Associations between inflammatory biomarkers (per 1-SD increment) and all-cause and breast cancer-specific mortality according to menopausal status at diagnosis.

Individual biomarkers and scores have been assessed on a continuous scale, per 1-SD increase. Models are stratified by country and stage of tumour (metastatic, non-metastatic, unknown), and adjusted for age at diagnosis, laboratory batch, fasting status at blood collection, education level, physical activity, body mass index, alcohol consumption, smoking status and intensity, ever use of hormones for menopause at diagnosis, cancer grade, and tumour receptor status (ER, PR, HER2). The Inflammatory scores are derived from log-transformed biomarker concentrations standardised using the mean and standard deviation of our population (z-score) of 1538 BC cases; the z-scores for adiponectin were multiplied by -1 to account for its anti-inflammatory effect. These z-scores are summed to generate an overall inflammatory score for each individual. The Inflammatory score-10 is composed of the following biomarkers: IL6, IL8, IL10, IL17D, IL-1RA, IFN-γ, TNFα, CRP, Leptin, and Adiponectin; the Inflammatory score-8 includes the same components as the Inflammatory score-10, except for the adipokines Leptin and Adiponectin. BC breast cancer, HR hazard ratio, CI confidence interval, Inflam.score Inflammatory score, SD standard deviation, IL interleukin, TNF-α tumour necrosis factor alpha, Het Heterogeneity.