Cervical cancer systemic inflammation score: a novel predictor of prognosis

Inflammation contributes to development and progression in a variety of cancers, including cervical cancer. We developed a novel cervical cancer systemic inflammation score (CCSIS) based on the preoperative platelet-to-lymphocyte ratio (PLR) and serum albumin levels. A retrospective analysis of clinical data from 795 patients with operable cervical cancer was then conducted to investigate the prognostic value of CCSIS and its association with the patients' clinicopathological features, overall survival (OS), and disease-free survival (DFS). CCSIS was predictive of OS and DFS. High CCSIS was correlated with more advanced FIGO stages, poor tumor differentiation, and the presence of PLN and LVSI. Both albumin levels and the PLR were independent prognostic indicators for operable cervical cancer. The use of the CCSIS could improve risk stratification and traditional clinicopathological analysis in cervical cancer.


INTRODUCTION
Emerging evidence suggests that inflammation plays a key role in the initiation and progression of various cancers, including cervical cancer. Cancer-related inflammation can influence tumor cell proliferation, invasion, metastasis, cell survival, and angiogenesis [1,2]. Accordingly, inflammation-based prognostic indicators, such as the modified Glasgow Prognostic Score (mGPS) [3,4], platelet to lymphocyte ratio (PLR) [5,6], neutrophil to lymphocyte ratio (NLR) [5][6][7], albumin, and C-reactive protein (CRP) [8,9] are prognostic factors for cervical cancer as well as other cancers. These cost-effective biomarkers are used routinely in clinical settings and could be used to provide information regarding patient outcomes. However, there are few studies investigating the value of these biomarkers as independent indicators in cervical cancer. Additionally, to the best of our knowledge, integration of individual biomarkers to enhance the outcome prediction for cervical cancer has not yet been studied.
In this study, the prognostic values of PLR and albumin were investigated. We also developed a novel prognostic score, the cervical cancer systemic inflammation score (CCSIS), based on preoperative serum albumin levels and PLR. The correlation between CCSIS and clinicopathological parameters and the prognostic value of CCSIS in cervical cancer patients were investigated.

Patient characteristics
795 patients with clinically and histologically confirmed cervical cancer were included in the present analysis. These patients were diagnosed at a mean age of 49.5 ± 10.7 years, with a mean follow-up period of 62.3 ± 26.7 months. By the last follow-up, 123 patients had relapsed, and 98 patients had died. 105 (13.1%) of these
Kaplan-Meier curves showed that high CCSIS scores were associated with shorter OS and DFS (p < 0.001 for both) (Figures 7 and 8). The cumulative 3-year OS rates were 84.8%, 90.4%, and 96.2% for patients with CCSIS scores of 2, 1, and 0, respectively; the 5-year OS rates in corresponding subgroups were 81.9%, 88.9% and 94.3%, respectively. The differences in cumulative OS among the three subgroups were statistically significant (p < 0.001 for both). Furthermore, the cumulative 3-year DFS rates were 76.4%, 87.2%, and 93.9% for patients with CCSIS scores of 2, 1, and 0, respectively; 5-year DFS rates in the corresponding subgroups were 75.8%, 85.3%, and 91.6 %. The differences among the three subgroups for cumulative DFS were also statistically significant (p < 0.001 for both).
CCSIS was predictive of prognosis as measured by both OS and DFS in a univariate analysis (p < 0.05 for both) ( Table 1). In multivariate analysis, both CCSIS and tumor differentiation (p < 0.05 for all) were independent predictors of OS and DFS in cervical cancer patients (Tables 2 and 3). Furthermore, the presence of LVSI was associated with higher risk of recurrence and increased FIGO stage.

Correlations between PLR, albumin, CCSIS, and other clinical parameters of cervical cancer
Clinicopathologic characteristics were compared between patients grouped by PLR and serum albumin as shown in Table 4. Increased PLR and decreased serum     albumin were both associated with advanced FIGO stage (p = 0.040 and p < 0.001, respectively) and poor tumor differentiation (p = 0.010 and p < 0.001, respectively). Additionally, elevated PLR was associated with the presence of PLN (p = 0.002). In contrast, decreased serum albumin was associated with the presence of LVSI (p < 0.001). As shown in Table 5, high CCSIS score was correlated with advanced FIGO stage (p < 0.001), poor tumor differentiation (p < 0.001), and the presence of PLN (p = 0.002) and LVSI (p = 0.010).

DISCUSSION
The clinicopathological characteristics and prognoses of 795 cervical cancer patients were investigated in this retrospective study. Univariate Cox proportional hazard analysis revealed that PLR, serum albumin, FIGO stage, tumor differentiation, PLN, and LVSI, but not NLR, were associated with OS and DFS. Additionally, increased PLR and decreased serum albumin before surgery were independent predictors of shorter OS and DFS in a multivariate analysis. To find a more objective marker, the cervical cancer systemic inflammation score (CCSIS), an integrated indicator based on PLR and serum albumin, was created. Kaplan-Meier curves demonstrated that high CCSIS scores were associated with shorter OS and DFS. Furthermore, increased CCSIS was associated with advanced FIGO stage, poor tumor differentiation, and the presence of PLN and LVSI.
Platelets and lymphocytes are related to immune surveillance [10]. Therefore, PLR, a combined index of platelets and lymphocyte counts, has been investigated as a prognostic factor in various cancers [11]. Recently, a meta-analysis including 12,754 patients demonstrated that high PLR was associated with shorter OS in various solid tumors [12]. The prognostic value of PLR in cervical cancer is unclear. Mesut et al. found that PLR may provide useful information regarding cervical carcinoma invasion in a study of 75 patients with pre-invasive and 30 patients with invasive disease [13]. On the contrary, Wang et al. suggested that pretreatment PLR did not predict prognosis in cervical cancer patients. However, that study only included 111 FIGO stage IB2-IIB patients who had received neoadjuvant chemotherapy and underwent radical hysterectomy [14]. In a study of 460 FIGO stage I or II cervical cancer patients, Yu et al. found that NLR, but not PLR, was an independent prognostic marker for PFS, but not for OS [6]. However, unlike phosphoglycerate dehydrogenase [15], PLR is a cheap and easily available biomarker that may be an independent predictor of poor prognosis in patients with cervical cancer.
Serum albumin is a negative acute phase protein. It has been integrated with other markers to create new prognostic markers, such as mGPS [4]. Decreased preoperative albumin is associated with reduced survival and poor prognosis in other cancer patients [16][17][18]. However, few studies have investigated the relationship between preoperative albumin and patient survival in cervical cancer. In a study of 238 patients who underwent chemoradiotherapy instead of operations, mGPS was correlated with advanced stages in cervical cancer patients and was an independent prognostic indicator for OS and PFS [19]. Consistent with previous studies, we showed that decreased albumin was associated with poor OS and DFS in operable cervical cancer patients. So far, no studies have explored the links between PLR and serum albumin in cervical cancer. Here, we demonstrated for the first time that decreased albumin predicted shorter OS and DFS. By creating a new score that combines PLR and albumin, we stratified patients into three groups. This approach was particularly successful in predicting differences among the three groups in OS and DFS in multivariate models. Moreover, subgroup analysis supported the validity of this stratification method. Therefore, PLR and albumin were useful prognostic biomarkers that, when combined, provided additional risk stratification for cervical cancer patients. Patients with high CCSIS scores, which were correlated with aggressive cancer biology phenotypes such as advanced FIGO stage, poor tumor differentiation, and the presence of PLN and LVSI, might benefit from aggressive therapeutic regimens such as radiotherapy and/or chemotherapy, or even neoadjuvant chemotherapy followed by surgery [20,21].
These results reflect complex interactions between systematic inflammation and tumor progression. Both albumin and lymphocyte levels are indicative of host immunity. Albumin, platelets and lymphocytes play different roles in systematic inflammation which leads to tumor progression. Synthesized predominantly in the liver [22], albumin is a safe and immunogenic protein. Decreased preoperative albumin may indicate a malnourished state [23] and sustained systemic inflammation [24,25], suggesting a suppressed immune system. Platelets are the major serum source of vascular endothelial growth factor (VEGF), which induces angiogenesis and endothelial proliferation [3,[26][27][28]. Angiogenesis in turn is critical for the growth and metastasis of tumors [29]. T lymphocytes are critical adaptive immune cells, which are commonly divided into two subsets based on the expression of CD4 and CD8 receptors [30]. On one hand, decreased lymphocyte levels can cause immune suppression, which can produce inflammatory cytokines in the tumor microenvironment [31]. On the other hand, according to Ramello et al, after co-incubation with CD4+ or CD8+ tumorinduced senescent T cells, monocytes produce more pro-inflammatory cytokines (TNF, IL-1β and IL-6) and angiogenic factors (MMP-9, VEGF-A and IL-8) than those co-cultured with controls [32]. Thus, T cells may play key role in angiogenesis and impact tumor progression and metastasis. Overall, PLR may reflect the balance between tumor angiogenesis and host immunity. Increased PLR indicates a predominance of pro-tumor angiogenesis responses and is associated with poor oncologic outcomes. Increased PLR together with decreased albumin, which indicates hypoimmunity in patients, may lead to cancer. The present study supports this theory.
Currently, this is the largest study of the prognostic roles of preoperative PLR and albumin in cervical cancer. It is also the first study to propose a potential prognostic role for CCSIS for this disease. However, the present study does have some limitations. First, it is difficult to control for potential confounding factors in retrospective studies, and the results might differ in other populations; future studies replicating these findings in diverse patient groups are needed. Second, the present study was conducted at a single institution, and multicenter studies of the markers used here would strengthen our conclusions. Finally, only patients who underwent surgery were included, and the results may not apply to patients without surgical indications, including stages IIB, III and IV.
In conclusion, this study highlights the potential role of PLR and albumin as prognostic factors for OS and DFS in early-stage cervical cancer patients. CCSIS, a new biomarker integrating PLR with serum albumin levels, was evaluated. CCSIS served as an independent prognostic factor and might aid in the identification of cervical cancer patients and in managing their treatment.

Patients
The current analysis included 795 patients with cervical cancer who received treatment at the First Affiliated Hospital of Wenzhou Medical University (Zhejiang province, China) between May 2005 and December 2012. Only patients with IA1-IIA2 stage cervical cancer who had received radical hysterectomy with additional chemotherapy or radiotherapy at the hospital were included. Patients with any history of immunological, hematological, or liver disease, other cancers, adolescents ( < 18 years old), and patients who took medications that could cause hematological disease or liver disease were excluded. This study was approved by the Hospital Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University. As this was a retrospective study, written consents were waived, and verbal consents were obtained via telephone.

Data collection and definitions of NLR, PLR and LMR
Albumin levels and preoperative blood cell counts, which included platelet, absolute lymphocyte, neutrophil, and monocyte levels, from one week prior to operation were collected through the electronic medical record system (EMRS). NLR, PLR, and lymphocyte to monocyte ratio (LMR) were calculated as absolute neutrophil count divided by absolute lymphocyte count, absolute platelet count divided by absolute lymphocyte count, and absolute lymphocyte count divided by absolute monocyte count, respectively. Patient clinical and pathological characteristics, such as age at diagnosis, parturition number, FIGO stage, pathological types, tumor differentiation, vascular lymph node invasion (LVSI), and pelvic lymph node metastasis (PLN) were also collected.

Follow-up and prognosis evaluation
Out-patient medical records, telephone consultations, and social security death indices were used to calculate overall survival (OS) and disease free survival (DFS). Follow-ups were conducted every 3 months for the first 2 years, 6 months for the next 3 years, and annually in the following years. Gynecological examinations, cervical cytology, trans-vaginal ultrasound scanning, and computed tomography (CT) and/or magnetic resonance imaging (MRI) were conducted during follow-up evaluations. OS was defined as the time from surgery to the time of death regardless of cause. DFS was calculated from the date of surgery to the date of recurrence.

Statistical analysis
Continuous variables with normal distributions are expressed as mean ± standard deviation (SD) and were compared using standard t-tests. Wilcoxon's rank-sum test or the Kruskal-Wallis test were used for continuous variables with non-normal distributions. Additionally, the Chi-square test or Fisher's exact test were applied to analyze categorical variables. Survival analyses were performed using Kaplan-Meier survival curves, and significant differences between groups were compared by the log-rank test. Cox proportional hazard regression was used to identify associations between outcomes and variables. Variables identified as statistically significant in the univariate Cox regression analysis were then used in multivariate analysis with backward stepwise selection. Threshold values for NLR, PLR, LMR, and serum albumin were selected using receiver operating characteristic (ROC) curve analysis.
NLR, PLR, LMR and serum albumin biomarkers were evaluated together with traditional clinicopathological variables in univariate and multivariate analyses. PLR and serum albumin were identified as independent prognostic indicators for OS and DFS. The CCSIS combined PLR and serum albumin levels and was assessed in a multivariate analysis with traditional clinicopathological variables. All p values were 2-sided, and a p value < 0.05 was considered statistically significant. Statistical analysis was performed using the Statistical Package for Social Sciences version 21.0 (SPSS Inc., Chicago, IL, USA).