The Elevated Pre-Treatment C-Reactive Protein Predicts Poor Prognosis in Patients with Locally Advanced Rectal Cancer Treated with Neo-Adjuvant Radiochemotherapy

The aim of the present study was to investigate the association of the pre-treatment C-reactive protein (CRP) plasma level with survival outcomes in a cohort of 423 consecutive patients with locally advanced rectal cancer treated with neo-adjuvant radiochemotherapy followed by surgical resection. To evaluate the prognostic value of the CRP level for clinical endpoints recurrence-free survival (RFS), local-regional control (LC), metastases-free survival (MFS), and overall survival (OS), uni- and multivariate Cox regression analyses were applied, and survival rates were calculated using Kaplan–Meier analysis. The median follow-up time was 73 months. In univariate analyses, the pre-treatment CRP level was a significant predictor of RFS (hazard ratio (HR) 1.015, 95% CI 1.006–1.023; p < 0.001), LC (HR 1.015, 95% CI 1.004–1.027; p = 0.009), MFS (HR 1.014, 95% CI 1.004–1.023; p = 0.004), and OS (HR 1.016, 95% CI 1.007–1.024; p < 0.001). Additionally, univariate analysis identified the MRI circumferential resection margin (mrCRM) and pre-treatment carcinoembryonic antigen (CEA) as significant predictor of RFS (HR 2.082, 95% CI 1.106–3.919; p = 0.023 and HR 1.005, 95% CI 1.002–1.008; p < 0.001). Univariate analysis also revealed a significant association of the mrCRM (HR 2.089, 95% CI 1.052–4.147; p = 0.035) and CEA (HR 1.006, 95% CI 1.003–1.008; p < 0.001) with MFS. Age and CEA were prognostic factors for OS (HR 1.039, 95% CI 1.013–1.066; p = 0.003 and HR 1.005, 95% CI 1.002–1.008; p < 0.001). In multivariate analysis that included parameters with a p-level < 0.20 in univariate analysis, the pre-treatment CRP remained a significant prognostic factor for RFS (HR 1.013, 95%CI 1.001–1.025; p = 0.036), LC (HR 1.014, 95% CI 1.001–1.027; p = 0.031), and MFS (HR 1.013, 95% CI 1.000–1.027; p = 0.046). The results support the hypothesis that an elevated pre-treatment CRP level is a predictor of poor outcome. If confirmed by additional studies, this easily measurable biomarker could contribute to the identification of patients who might be candidates for more aggressive local or systemic treatment approaches or the administration of anti-inflammatory drugs.


Introduction
Colorectal cancer (CRC) is one of the most commonly occurring cancers worldwide with a remarkable increase in incidence among adults younger than 50 years [1,2]. Approximately 35% derived from studies analyzing relatively small numbers of patients. Toiyama et al. evaluated the prognostic role of the pre-treatment plasma CRP level in a cohort of 84 rectal cancer patients and demonstrated that an elevated CRP level predicted poorer OS and early recurrence in patients with locally advanced rectal cancer treated with nRCT followed by total mesorectal excision (TME) [24]. In contrast, Buijsen et al., who investigated various pre-treatment biomarkers as predictive factors for tumor response after nRCT in rectal cancer patients, did not detect a significant association between the pre-treatment CRP level and tumor response after nRCT [25].
The aim of the present study was to investigate the association of the pre-treatment CRP level with recurrence-free survival (RFS), loco-regional control (LC), metastases-free survival (MFS), and overall survival (OS) in order to validate and further clarify the prognostic significance of the CRP level in a large European cohort of locally advanced rectal cancer patients treated with nRCT followed by definitive surgery.

Materials and Methods
A total of 540 consecutive cases with histologically confirmed locally advanced primary rectal cancer referred for nRCT from 2004-2015 at the Department of Therapeutic Radiology and Oncology were evaluated retrospectively. Patients with metastatic disease (n = 52) or secondary malignancies (n = 60), patients without surgery after nRCT (n = 4), as well as those receiving neo-adjuvant radiation without concurrent chemotherapy (n = 1) were excluded. The remaining 423 received nRCT and subsequent surgical resection and were included in further analysis (Appendix A).
Demographic characteristics as well as clinical, pathological, and laboratory data were obtained from paper and electronic medical records. The distance to the anal verge was measured by means of either colonoscopy or rigid proctoscopy and additional digital rectal examination. The preoperative depth of tumor invasion was assessed by pelvic MRI, endorectal ultrasound and/or CT scan. Staging was performed according to the edition of the Union for International Cancer Control (UICC) classification valid at the time of rectal cancer diagnosis (6th edition for cancer diagnosis from 2004-2009 and 7th edition for 2010-2015) [26,27]. The used versions did not differ in terms of the primary tumor stage.
Treatment decisions were based on the recommendations of the multidisciplinary tumor board. Radiotherapy was performed with high-energy photons (6 or 18 MeV) up to a total dose of 45-46 Gy in 23-25 fractions of 1.8 or 2 Gy/fraction (5 days/week) prescribed to the 95% isodose. The radiation method consisted of either three-dimensional conformal radiotherapy delivered in a four-field technique or intensity-modulated radiotherapy (IMRT), including volumetric modulated arc therapy (VMAT). All patients received concurrent fluoropyrimidine-based chemotherapy that was performed as a continuous intravenous infusion with 5-fluoruracil (1000 mg/m 2 ) during the first and last week of radiotherapy or as oral administration of capecitabine (1650 mg/m 2 daily) on each day of radiation treatment. After a median interval of 7 weeks, TME or abdominoperineal rectum resection (APR) was performed. Adjuvant 5-fluorouracil based chemotherapy was administrated in 224 patients (53%).
The CRP were obtained as part of a routine clinical evaluation prior to the start of nRCT and analyzed using the standard clinical testing methodology. Additional parameters of interest assessed in the present study included age, sex, smoking status, body mass index (BMI), clinical T-stage (cT), clinical lymph node involvement, and tumor grade.
Follow-up examinations were performed in regular intervals (3 months intervals in years 1-3, 6 months intervals in years 4-5, and 12 months intervals in years 6-10 after diagnosis) and included clinical examination, evaluation of CEA and carbohydrate antigen 19-9 (CA 19-9), radiological assessment (liver scan or ultrasound and chest X-ray every 6 months within the first three years), and colonoscopy every 2 years.
The study complied with all the relevant national regulations, institutional policies, and in accordance with the tenets of the Declaration of Helsinki, and has been approved by the review board of the Medical University Graz (approval number: EK 28-046 ex 15/16, approval date 21 October 2015). As this is a retrospective non-intervening study, informed consent from the study participants was not required.
The primary endpoint, RFS, was calculated from the start of treatment to the date of locoregional or distant tumor recurrence. OS was defined as the time from the start of treatment to the date of death of any cause, LC was calculated from initiation of treatment to the development of local recurrence and/or regional lymph node metastases, and MFS was defined as the time from the start of treatment to the development of distant metastases. The clinical endpoints were calculated using the Kaplan-Meier method and log-rank tests were applied for statistical comparisons between Kaplan-Meier curves. Non-parametric tests (Mann-Whitney U Test, Kruskal-Wallis Test, and Spearman correlation) were applied to evaluate the association between the pre-treatment CRP level and clinico-pathological parameters. Univariate Cox proportion analysis was performed to determine the influence of an elevated CRP level and other clinico-pathological and laboratory factors such as age at diagnosis, sex, BMI, smoking status, tumor grade, tumor stage, nodal involvement, LDH, CEA, CA19-9 on OS, LC and RFS. Hazard ratios (HRs) estimated from the Cox proportion analysis were reported as relative risks with corresponding 95% confidence intervals (CIs). Multivariate Cox proportion analyses were conducted to evaluate the impact of potential confounders and included parameters with a p-value < 0.20 in univariate analyses. To evaluate the relationship between baseline clinico-pathological and laboratory parameters and tumor response after nRCT, logistic regression analysis was applied. Receiver operating characteristic (ROC) curve analysis was performed to estimate the optimal CRP cutoff value as reported previously [28,29]. The optimal cutoff value was determined as the point on the ROC curve that maximizes the Youden Index.
All statistical analyses were performed using the Statistical Package for Social Sciences version 26.0 (SPSS Inc., Chicago, IL, USA). A two-sided p < 0.05 was considered statistically significant.
The pre-treatment tumor stage was a significant predictor of complete response (p = 0.007), and none of the remaining baseline parameters were significantly associated with a complete response after nRCT. Furthermore, pre-treatment tumor site (p = 0.007), tumor stage (p = 0.027), CEA (p = 0.005), and CA 19-9 (p = 0.043) were significantly associated with a good response.
Using ROC curve analysis, a pre-treatment CRP cutoff value of 3.8 mg/L was determined to be optimal for both PFS and LC. Overall, there were 254 patients (60%) with a pre-treament plasma CRP level ≤3.8 mg/L and 169 patients (40%) with a CRP level >3.8 mg/L. In the Kaplan-Meier analysis, decreased PFS (p = 0.006) and LC (p = 0.001) was detected for patients with a pre-treatment CRP level >3.8 mg/L ( Figure 1A,B).
For OS, ROC curve analysis showed that a CRP level of 4.3 mg/L was optimal to discriminate patient survival and death. There were 272 patients (64.3%) with a CRP level

Discussion
The establishment of easily determinable and objective predictive factors for the outcome is critical for improving treatment strategies. In the present study, we examined the prognostic impact of the pre-treatment plasma CRP level in patients with locally advanced rectal cancer treated with nRCT and detected a significant relationship between an elevated pre-treatment CRP level and a decreased RFS, LC, MFS, and OS in univariate and multivariate analysis. These data suggest that a high pre-treatment CRP level may identify patients with poor loco-regional control, a high risk for developing distant metastases, and a poor overall survival outcome.
There is strong evidence for the role of inflammation in cancer development, progression, and metastasization [12,30]. Most solid malignant tumors trigger an intrinsic inflammatory response that builds up a protumorigenic microenvironment [14]. Various oncogenes have been shown to induce a transcriptional program that leads to remodeling of the tumor microenvironment through the recruitment of leukocytes and expression of tumor-promoting chemokines and cytokines [31,32]. Furthermore, solid malignancies become oxygen-and nutrient-deprived, resulting in necrotic cell death and in an inflammatory response promoting angiogenesis and the synthesis of additional cytokines [33]. Cytokines such as interleukin (IL)-6, tumor necrosis factor (TNF)-a, transforming  Figure 1C).

Discussion
The establishment of easily determinable and objective predictive factors for the outcome is critical for improving treatment strategies. In the present study, we examined the prognostic impact of the pre-treatment plasma CRP level in patients with locally advanced rectal cancer treated with nRCT and detected a significant relationship between an elevated pre-treatment CRP level and a decreased RFS, LC, MFS, and OS in univariate and multivariate analysis. These data suggest that a high pre-treatment CRP level may identify patients with poor loco-regional control, a high risk for developing distant metastases, and a poor overall survival outcome.
There is strong evidence for the role of inflammation in cancer development, progression, and metastasization [12,30]. Most solid malignant tumors trigger an intrinsic inflammatory response that builds up a protumorigenic microenvironment [14]. Various oncogenes have been shown to induce a transcriptional program that leads to remodeling of the tumor microenvironment through the recruitment of leukocytes and expression of tumor-promoting chemokines and cytokines [31,32]. Furthermore, solid malignancies become oxygen-and nutrient-deprived, resulting in necrotic cell death and in an inflammatory response promoting angiogenesis and the synthesis of additional cytokines [33]. Cytokines such as interleukin (IL)-6, tumor necrosis factor (TNF)-a, transforming growth factor (TGF)-ß, and epithelial growth factor receptor (EGFR) ligands control the immune and inflammatory milieu and also have direct effects on cancer cell growth and survival through the activation of various downstream effectors [14,33].
There is rising evidence supporting the prognostic role of the plasma CRP level in different cancer entities, including localized and metastatic CRC [39][40][41][42][43][44][45][46]. In addition, pre-therapeutic indices of systemic inflammation based on CRP and albumin have been found to provide prognostic information in CRC patients. However, only a few studies have investigated the prognostic significance of the plasma CRP level in rectal cancer. Kim et al. analyzed the prognostic role of the pre-operative CRP level determined in 125 patients after completion of nRCT and detected a significant association of an elevated pre-operative CRP level with DFS and CSS [47]. In contrast, Giessen et al., who also evaluated the prognostic role of the pre-operative CRP level in rectal cancer patients, did not detect a significant relationship between the CRP level and outcome [48].
In patients with locally advanced rectal cancer treated with nRCT, data on the prognostic significance of the pre-treatment plasma CRP level are very limited. Buijsen et al. investigated various pre-treatment biomarkers, including the plasma CRP level, as predictive factors for tumor response after nRCT, but did not detect a significant relationship between the pre-treatment CRP level and complete or good response after nRCT [25]. However, data on the relationship between the pre-treatment CRP level and the survival outcome are not provided. To our knowledge, to date, only 1 series has examined the prognostic role of the pre-treatment plasma CRP for survival outcomes in rectal cancer patients treated with nRCT. Toiyama et al. analyzed the prognostic impact of the pre-treatment CRP level in a cohort of 84 patients treated with nRCT and subsequent TME and identified an elevated pre-treatment CRP level as a significant prognostic factor for poor OS and DFS [24].
In the present study, univariate analysis identified a CRM > 1mm as a significant prognostic factor for improved RFS and MFS, but did not reach significance in multivariate analysis. Our data are in line with the results of Nikberg et al. who also detected that the CRM is not an independent factor for RFS [49]. Furthermore, pre-treatment clinical parameters such as tumor stage or tumor site were associated with tumor response; however, a significant relationship with survival outcomes was not detected. Similar findings have been obtained by Toiyama et al. who also were unable to detect a significant association between pre-treatment clinical stage and disease-free survival or OS [24]. In addition, we did not detect a significant relationship between the pre-treatment nodal stage and outcome. This finding might probably be explained by the limited accuracy achieved with current imaging techniques. For instance, Brouwer et al. reported nodal understaging in 24% of rectal cancer patients staged as cN-and overstaging in 44% of patients staged as cN+ that may result in patients' under-or overtreatment [50]. Thus, the identification of additional prognostic factors determined at the baseline is particularly important for improvements in the treatment decision-making process.
In our study, the pre-treatment plasma CRP level was identified as a significant prognostic factor in rectal cancer patients treated with nRCT. There is evidence that CRP itself exerts pro-inflammatory and tumor-promoting effects [51]. It has been shown that CRP may induce the expression of various molecules involved in cell-cell and cell-matrix adhesion as well as the expression of matrix metalloproteinases (MMPs) that are essential for the degradation of the basement membrane and extracellular matrix, a key step for tumor growth and metastasization [52,53]. Additionally, CRP may promote tumor growth by protecting tumor cells from apoptosis [54].
CRP is a significant marker of inflammation that is known to represent a critical component of tumor development and progression. Several agents affecting the CRP level have been described. Cyclooxygenase inhibitors, platelet aggregation inhibitors, lipid-lowering agents, antioxidants, and angiotensin-converting enzyme inhibitors may reduce serum levels of CRP [55]. Anti-inflammatory treatment leading to the reduction of the CRP level has been proposed to represent an effective therapeutic strategy for cancer. Previous studies have shown a significantly lower incidence of colon cancer in individuals taking aspirin and non-steroidal anti-inflammatory drugs (NSAIDs) [56,57]. Because inflammatory factors contribute to the initiation and progression of various types of cancer, NSAIDs are suggested to not only reduce the incidence but also prevent the progression of cancer by suppressing various inflammatory pathways [58]. Furthermore, NSAIDs may inhibit cancer progression via the induction of tumor cell apoptosis, protection, and repair of DNA damage, and suppression of platelet activity [59]. Statin use has also been shown in several preclinical and clinical studies to demonstrate anticancer properties in several preclinical and clinical studies [60]. Furthermore, in rectal cancer, statin use was predictive of a pathologically complete response in nRCT [61].
Worse nutritional status is associated with increased morbidity, mortality, and prolonged hospitalization and may copy cancer-related processes. In unresectable or relapsed colorectal cancer, a strong correlation between malnutrition and overall survival has been reported [62]. Furthermore, the loss of muscle mass during chemotherapy has been associated with lower rates of survival in patients with metastatic CRC [63]. In our cohort, no association between BMI and survival could be observed.
To the best of our knowledge, the present study currently represents the largest one analyzing the prognostic role of the pre-treatment CRP level in patients with locally advanced rectal cancer treated with nRCT. Major strengths include the well-defined, homogenously treated study cohort and the relatively long follow-up time.
Nevertheless, some limitations of the present study have to be taken into account. Due to the retrospective study design, a selection bias cannot be fully excluded. Furthermore, CRP is a nonspecific marker of inflammation and might be influenced by several conditions such as bacterial or viral infection, inflammatory diseases, connective tissue disorders, severe stress, and medical treatments. In view of the lack of a standardized cut-off value, ROC curve analysis was applied to determine optimal cut-off levels. For OS, the cut-off value determined using ROC analysis was different when compared to the cut-off value for the remaining endpoints. This finding might be explained by the fact that OS is a non-cancer-specific endpoint that is influenced by several other conditions or diseases. However, future investigations are necessary for the validation of the cut-off levels presented in our study.
Beside CRP, other inflammation-related parameters, such as white blood cell count, neutrophil/lymphocyte ratio, or erythrocyte sedimentation rate, are potential biomarkers for outcome in rectal cancer patients [64,65]. Further studies are required to establish whether one distinct inflammatory marker or a combination of several markers yields the highest predictive power in this patient group.
In view of the limitations of this study, our results have to be interpreted cautiously until validated in additional prospective large-scale studies before firm conclusions about the role of the pre-treatment CRP level for prognosis in patients with locally advanced rectal cancer can be drawn. Nonetheless, even considering these limitations, our data indicate that a high plasma CRP level is an independent prognostic factor for RFS, LC, MFS, and OS in locally advanced rectal cancer patients. If confirmed by additional studies, determination of the pre-treatment CRP level could provide additional prognostic information and contribute to the identification of patients who might be candidates for a more aggressive local or systemic treatment approach or the administration of anti-inflammatory drugs.

Conclusions
An increased pre-treatment plasma CRP level seems to represent an independent prognostic factor for RFS, LC, MFS, and OS in rectal cancer patients treated with nRCT. If the present findings are replicated in future studies, determination of the pre-treatment plasma CRP level may help to obtain a more precise individual risk profile and contribute to the tailored treatment of rectal cancer patients.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Table A1. Inclusion and exclusion criteria for patients who were referred for neoadjuvant radiochemotherapy and retrospective analysis at our department.

Inclusion Criteria Exclusion Criteria
Neoadjuvant radiochemotherapy with subsequent surgical resection Metastatic disease (n = 52) Secondary malignancies (n = 60) No surgical resection (n = 4) No concurrent chemotherapy (n = 1) Between 2004-2015 records of 540 patients with histological confirmed rectal cancer were identified. 423 patients received neoadjuvant radiochemotherapy and subsequent surgical resection and were included in further analysis of this study.