C-reactive protein (CRP) is an acute-phase reactant produced almost exclusively in the liver mostly triggered by Il-6. Plasma CRP levels can increase up to 1000-fold in response to microbial infection, trauma, infarction or autoimmune diseases. However, elevated CRP levels are also described in underlying malignancies or premalignancies, or in tissue inflammation associated with tumor growth.
It is still unclear whether the tumor promotes inflammation or is rendered more aggressive by it. Furthermore, high Il-6 levels are known promoters of high-risk prostate cancer (PC). Additionally, high levels of circulating CRP have recently been linked to poor prognosis in various malignancies, including oral squamous cell carcinoma, esophageal SCC, non-small cell as well as small cell lung cancer, melanoma, hepatocellular carcinoma, breast cancer, endometrial cancer, renal cell carcinoma, urothelial carcinoma and castration-resistant prostate cancer. However, the data available on localized PC is still limited.
Therefore we conducted a prospective clinical study to evaluate the association of preoperative CRP levels and high-risk PC in terms of a potential and feasible predictive marker in patients with clinically organ confined and biopsy confirmed PC.
We enrolled 629 men (age 34-85; median/mean 68/66.9 years) scheduled for radical prostatectomy. Patients with a second malignancy, signs of inflammation, autoimmune disease, liver disorders or prior therapies were excluded.
The median/mean preoperative PSA level for all patients was 7.6/11.8 ng/ml (IQR, 5.4–12.6), and the median/mean preoperative CRP level was 1.2/3.1 mg/l (IQR, 0.5–2.8). Preoperative CRP levels were normal (<4 mg/l) in 527 patients (83.8 %) and elevated in 102 (16.2 %). After prostate surgery, pathological examination revealed that there were 219 patients (34.8 %) with locally advanced prostate cancer (≥ pT3a), 49 (8.3 %) with nodal involvement, and 66 (10.5 %) with poorly differentiated cancer (Gleason score ≥ 8).
Poorly differentiated PC tends to be more common in patients with elevated CRP levels (15.5 vs. 9.5%, p=0.08). Analogously, patients with a Gleason score ≥ 8 PC had significantly higher median CRP levels than those with a Gleason score ≤ 7 PC (1.9 vs. 1.2mg/l, p=0.03).
Having shown that the preoperative CRP level was significantly associated only with the pathological Gleason-Score, we applied a multivariate regression analysis to assess its value as an independent predictor of poor PC differentiation and/or advanced disease in patients with clinically localized PC. Surprisingly, this analysis revealed that an elevated CRP level as a continuous cardinal variable was significantly linked with poor tumor differentiation (HR 1.04, 95 % CI 1.00–1.07, p=0.014) but not with advanced disease (HR 0.98, 95 % CI 0.95–1.02, p=0.27). After including CRP as a categorical variable in our multivariate analysis with a 4 mg/l cut-off, an association with tumor differentiation was still indicated, though no longer statistically significant (HR 2.26, 95 % CI 0.98–5.22, p=0.057). But again, a CRP >4 mg/l was no indicator of advanced tumor stage (HR 1.16, 95 % CI 0.70–1.95, p=0.57). Since this multivariate analysis to identify independent predictors of poor tumor differentiation in the surgical specimen could have been biased by a strong correlation between the preoperative biopsy Gleason score and the final Gleason score (rs=0.52, p<0.001, Spearman-Rho), we repeated our analysis excluding the biopsy Gleason score. In contrast to the preoperative PSA level and digital rectal examination findings, the CRP level did now no longer independently predicted poor tumor differentiation, neither as a continuous/cardinal (HR 1.02, 95 % CI 0.995–1.04, p=0.11) nor as a categorical variable (HR 1.51, 95 % CI 0.76–2.99, p=0.24).
CRP is an approved biomarker in patients with metastatic disease in terms of disease prognosis and assessment of aetiopathology. Published molecular factors determining cancer aggressiveness suggested that CRP could be a feasible predictive tool. However, we were able to show that, in patients with clinically localized PC, the serum CRP level does not posses the predictive value as it has shown in other tumor entities or advanced PC. Thus, its inclusion in existing prediction models seems not to be very promising.
T.J. Schnoeller, MD
University of Ulm Medical School, Department of Urology