To our knowledge, this is the largest population-based cohort study to date examining associations between systemic inflammation and breast cancer survival and the first to evaluate SAA as a prognostic marker for breast cancer. We observed significant associations between reduced overall survival and elevated concentrations of the inflammatory biomarkers SAA and CRP (measured approximately 31 months after diagnosis and 24 months after enrollment). Similarly, we observed a significant association between elevated CRP and reduced disease-free survival and a borderline association between elevated SAA and reduced disease-free survival. All associations were independent of disease stage, BMI, self-reported cardiovascular events, estradiol concentrations, smoking, and physical activity and did not seem to be modified by ER status, PR status, or disease stage.
Results from smaller studies16–18
suggest that an association between CRP and survival may be present only in groups of patients with metastatic disease. However, the results from our population-based study of nonmetastatic breast cancer (stage 0 to IIIA) indicate that this association is also present in patients with less advanced disease. The discrepancy between our work and that of Al Murri et al18
may be attributable to the timing of biomarker measurement, which was approximately 31 months after treatment in our study but before treatment in the study by Al Murri et al. Our study reflects longer term survival among breast cancer patients and is unlikely to be impacted by tumor burden.
The associations of SAA and CRP with disease-free survival were weaker in magnitude than the associations for SAA and CRP with overall survival, suggesting that these markers may be more closely related to overall survival than breast cancer. However, the CIs for these associations indicate that their magnitudes may be similar. Further investigation is needed to characterize these associations and get more precise risk estimates.
SAA and CRP are regulated by related cytokines, and the amplitude of the inflammation-related increase in SAA is similar to, or slightly greater than, that of CRP.24
However, the physiologic level of circulating SAA is approximately 10-fold greater than that of CRP, suggesting that SAA may be more useful for detecting slight elevations in systemic or chronic inflammation.24
SAA is superior to CRP for detecting inflammation associated with severe burns, viral infections, kidney transplantation, Crohn's disease, and ulcerative colitis.24
The mechanism by which chronic inflammation is related to breast cancer prognosis is unclear. Chronic inflammation may promote carcinogenesis through complex processes such as polarization of M2 tumor-associated macrophages via cytokines and the subsequent production of tumor growth factors or promotion of angiogenesis.13,25
In addition, inflammatory status is correlated with several prognostic factors such as body fatness, physical activity, and cardiovascular comorbidity, which may affect prognosis through alternate mechanisms. It is also possible that inflammation is a response to the presence of undetected cancer cells, rather than being solely a contributor to tumor promotion.
CRP and SAA were associated with a self-reported history of myocardial infarction and history of heart failure in the HEAL study,23
and these histories seem to partially confound the relationship between these inflammatory markers and survival. However, the associations of SAA and CRP with overall survival remained strong after adjustment for these factors. In addition, our analyses restricted to patients not reporting a history cardiovascular events and focused on noncardiovascular deaths suggest that the associations of SAA and CRP with overall survival are not attributable to cardiovascular disease.
To our knowledge, this is the first large, population-based study of post-treatment inflammatory markers and breast cancer survival. Our study measured biomarkers of inflammation at a later time point (approximately 31 months after diagnosis) and thus concerns longer term survival. Accordingly, we did not assess the relationship between inflammation and survival in participants who were not disease free at the time of biomarker measurement (n = 46). However, the timing minimizes the effect of treatment on inflammatory markers, resulting in measurements that reflect long-term inflammatory status. It is possible that acute inflammatory conditions at the time of blood collection caused elevated CRP and SAA concentrations that do not reflect long-term inflammatory status. However, this bias will be nondifferential (ie, similar among those with and without events) and bias risk estimates toward a null effect, unless such acute conditions are also associated with the event. Aromatase inhibitors were not yet available when HEAL patients were diagnosed, and data on HER-2/neu status were not routinely collected in SEER; therefore, we could not assess their effects on the associations observed in this study. Furthermore, power to assess associations within and compare associations between ER and PR subgroups was limited as a result of the small number of events in each group. We acknowledge that multiple statistical tests were conducted in this analysis, emphasizing the importance of follow-up studies to confirm these results.
In summary, this study suggests that systemic inflammation, as measured by circulating CRP and SAA, may be an important long-term prognostic factor for breast cancer, even after adjustment for age, stage, BMI, and cardiovascular events. The strong associations between overall survival and these inflammatory biomarkers require further investigation, as do the effects of reductions in inflammatory markers on breast cancer recurrence and survival (either using medications, such as nonsteroidal anti-inflammatory drugs/statins, or lifestyle changes, such as weight loss in overweight/obese patients).