Impact of the Preoperative Controlling Nutritional Status (CONUT) Score on the Survival after Curative Surgery for Colorectal Cancer

Background Recently, the preoperative immune-nutritional status has been reported to correlate with the survival rate in patients with colorectal cancer (CRC). However, there have been no reports on the relationship between the controlling nutritional status (CONUT) score and the clinical outcome after curative surgery for CRC. We herein evaluated the prognostic significance of the CONUT score in patients with CRC, and then compared the accuracy of the CONUT score and the prognostic nutritional index (PNI) as a predictor of survival. Methods We retrospectively reviewed a database of 204 patients who underwent curative surgery for Stage II/III CRC. Patients were divided into two groups according to the CONUT score and the PNI. Results The five-year cancer-specific survival (CSS) rate was significantly higher at 92.7% in the low CONUT group, compared to a rate of 81.0% in the high CONUT group (p=0.0016). The five-year CSS was 71.2% in the low PNI group and 92.3% in the high PNI group, which showed a significant difference (p=0.0155). A multivariate analysis showed that lymph node metastasis and the CONUT score were independent risk factors for CSS. Conclusion This study suggested that the CONUT score is a strong independent predictor of the survival among CRC patients.


Introduction
Colorectal cancer (CRC) is the third most common cancer in the world [1].
Although the surgical procedures and chemotherapy for CRC have improved, the clinical outcome of CRC is still poor, as one-third of the patients who undergo curative resection die within five years after surgery [2]. Therefore, it is necessary to identify biomarkers that can predict the prognosis and individualize the therapy based on the stratification of risks. Many studies about the potential prognostic factors for CRC have been carried out, and the preoperative immune-nutritional status has been reported to correlate with the survival for CRC [3][4][5][6][7].
Recently, the prognostic nutritional index (PNI), which was calculated from the serum albumin concentration and the total peripheral lymphocyte count, has been used to predict the risk of postoperative complications [8], and it has also been reported to correlate with the survival in CRC patients [3,9]. Regarding the PNI, we consider that the serum albumin concentration tends to be excessively emphasized. However, the serum albumin concentration has been reported to be easily influenced by not only the nutritional status, but also by changes in the body fluid volume, such as those due to the dehydration/fluid retention status and inflammation caused by chronic disease [10,11]. Therefore, this study focused on evaluating the Controlling Nutritional Status (CONUT) score [10].
The CONUT score is an index calculated from the following factors; the serum albumin concentration, the total peripheral lymphocyte count and total cholesterol concentration. Total cholesterol concentration has also been reported to correlate with the progression of cancer [11]. A more accurate evaluation can be obtain by reducing the importance of the serum albumin concentration and adding the total cholesterol concentration to the evaluation criteria [12]. Although the prognostic significance of the PNI has been reported in numerous previous reports, there have been no reports on the relationship between the CONUT score and the clinical outcome after curative surgery for CRC.
The aim of this retrospective study is to determine whether the preoperative CONUT score could be a useful predictor of the survival in patients with CRC, and to compare the accuracy of the CONUT score and the PNI as a predictor of the survival rate of such patients.

Patients
We retrospectively reviewed a database of 204 patients who underwent curative surgery for Stage II/III CRC at the Department of Surgical Oncology, Osaka City University, Japan between April 2004 and December 2009. We performed a retrospective review of 204 patients with Stage II or III CRC. We obtained written informed consent from the patients for participation and the study protocol was approved by the ethics committee of Osaka City University. Our investigation was conducted according to the principles expressed in the Declaration of Helsinki. The resected specimens were assessed using The International Union Against Cancer (UICC) staging classification of colorectal cancer [13]. All patients were followed up until April 2012 or until their deaths.
The indications for undergoing adjuvant chemotherapy included patients with Stage III or high-risk Stage II disease. T4 tumors, lymphatic vessel invasion, blood vessel invasion, highgrade histology, presentation with obstruction/perforation and inadequate lymph node sampling were defined as high-risk Stage II disease. The decision of whether or not the patients should undergo adjuvant chemotherapy was determined by the surgeons with the patients' consent. The doctors judged the indication for chemotherapy, for example, the criteria, performance status, general condition, age and patient's wishes.
There were no patients treated with neoadjuvant chemotherapy in this study.

Methods
The preoperative blood samples were obtained within two weeks before the operation. The CONUT score was calculated using the serum albumin concentration, peripheral lymphocyte count and the total cholesterol concentration, as described in Table 1 [12].
We used the continuous variable Controlling Nutritional Status (CONUT) as the test variable and cancer-specific survival as the state variable. An investigation of the cut-off value for the CONUT score using the receiver operating characteristic (ROC) curve showed the most appropriate cut-off value for the CONUT score to be 3 (AUC; 0.624, 95%CI: 0.476-0.771, p = 0.076, the sensitivity was 0.5263 and the specificity was 0.7622.). We indicated the ROC curve on Fig 1. Therefore, we set 3 as the cut-off value for the CONUT score in this study and classified the patients into high CONUT (3) and low CONUT (2) groups.
The following formula was used to calculate PNI: 10 × serum albumin concentration (g/dL) + 0.005 × total peripheral lymphocyte count (per mm 3 ) [8]. As with previous reports in which a low PNI (<40) was reported to be a prognostic factor for CRC [3], we also set 40 as the cutoff value of the PNI in the present study. The patients were divided into two groups; a low PNI (<40) group and a high PNI ( 40) group.
Regarding the relapse-free survival, relapse was noted as cancer relapse and deaths of all causes was treated as relapse. The survival times were measured from the date of the operation to the date of being lost to follow-up, the date of relapse, the date of death of all causes or April 30, 2012, whichever occurred first.
As to cancer-specific survival, deaths noted to be caused by colorectal cancer were treated as deaths, and other deaths were regarded as censored events. The survival times were measured from the date of operation to the date of being lost to follow-up, the date of death or April 30, 2012, whichever occurred first.
Differences between the groups were analyzed using the χ 2 test and the Wilcoxon signed rank test. The duration of survival was calculated according to the Kaplan-Mayer method. Differences in the survival curves were assessed with the log-rank test. A multivariate analysis of the clinicopathological factors for survival was performed using a Cox proportional hazard model. Statistical significance was set at a value of p<0.05. The JMP 11 software program (SAS Institute, Cary, NC, USA) was used to analyze the data.

Clinical characteristics
The patient characteristics are shown in Table 2. Fifty-four patients were classified into the high CONUT group and one hundred and fifty patients were classified into the low CONUT group, based on the cut-off value of 3. One hundred and seventy-seven patients were classified into the high PNI group and twenty-seven patients were classified into the low PNI group, based on the cut-off value of 40. All patients with a low PNI were included in the high CONUT group (Table 3). There were no operation-related deaths or hospitalization deaths in this study. Regarding complications, the number of cases with more than Clavien-Dindo classification 2 complications was 40 (19.6%). Specifically, 10 patients had anastomotic leakage, 23 patients had infectious complications and 19 patients had other complications.
The primary tumor was resected in all patients. In cases of recurrent disease, some tumors were removed surgically, and one patient chose chemotherapy. A total of 24 patients (51.1%) received resection for recurrent tumors out of 47 patients with recurrent disease. The receiver operating characteristic (ROC) curve for the controlling nutritional status (CONUT). We used the continuous variable Controlling Nutritional Status (CONUT) as the test variable and 5-year cancer-specific survival as the state variable. An investigation of the cut-off value for the CONUT score using the receiver operating characteristic (ROC) curve showed the most appropriate cut-off value for the CONUT score to be 3 (AUC; 0.624, 95%CI: 0.476-0.771, p = 0.076, the sensitivity was 0.5263 and the specificity was 0.7622.). We indicated the ROC curve on Fig 1. Therefore, we set 3 as the cut-off value for the CONUT score in this study and classified the patients into high CONUT (3) and low CONUT (2) groups.

The correlations between the CONUT/PNI and the clinicopathological factors
The CONUT score had a significant relationship with the age (p = 0.0016) and tumor location (p = 0.0168). The PNI had significant relationship with the age (p = 0.0001), tumor location (p = 0.0224), and adjuvant chemotherapy (p = 0.0030) ( Table 2).

Survival analysis according to the CONUT and PNI
The five-year relapse-free survival (RFS) rate was 73.0% in the low CONUT group and 53.6% in the high CONUT group (Fig 2), with a significant difference between the groups (p = 0.0018). In addition, the five-year RFS was significantly lower at 51.5% in the low PNI group compared to 70.4% in the high PNI group (Fig 2, p = 0.0162).
The five-year cancer-specific survival (CSS) was 92.7% in the low CONUT group and 81.0% in the high CONUT group (Fig 3), and there was a significant difference between the low and  high CONUT groups (p = 0.0016). Moreover, the five-year CSS was significantly lower at 71.2% in the low PNI group compared to 92.7% in the high PNI group (Fig 3, p = 0.0155).
The correlations between the CSS and the clinicopathological factors are shown in Table 5. A univariate analysis indicated that lymph node metastasis, the preoperative CA19-9 level, the CONUT score and the PNI were significantly associated with the CSS. When a multivariate analysis was performed, T4 tumor, preoperative carcinoembryonic antigen (CEA) level and adjuvant chemotherapy were added as covariates. A multivariate analysis showed that only lymph node metastasis (OR = 3.680, 95%CI; 1.106-14.914, p = 0.0330) and the CONUT score (OR = 4.212, 95%CI; 1.215-13.350, p = 0.0251) were independently associated with the CSS ( Table 5).
The comparison between the CONUT score and factors that comprise the CONUT score The influence of the factors that comprise the CONUT score (albumin level, the total cholesterol level and the total peripheral lymphocyte count) on survival was examined. The cut-off values for each of the factors were determined by their respective ROC curves on cancer-specific survival. The cut-off value for the albumin level was 3.5 g/dL, that for the total cholesterol level was 168 mg/dL and that for the total peripheral lymphocyte counts was 1170/ mm 3 . In the univariate analysis for RFS, the albumin level (p = 0.025) and the total cholesterol level (p = 0.032) were found to be predictive factors (Table 6). In the univariate analysis for CSS, the albumin level and the total peripheral lymphocyte level were found to be predictive factors.
The multivariate analysis for RFS indicated the CONUT score to be a more useful factor than the total cholesterol level and total peripheral lymphocyte counts (Tables 7, 8 and 9). The multivariate analysis showed the CONUT score to be superior to the serum albumin level and the total cholesterol score for predicting CSS. This study suggested that the CONUT score is a more useful factor for predicting survival than the individual factors that comprise the CONUT score (Tables 7, 8 and 9).

Discussion
Various methods to evaluate the immune-nutritional status have been advocated, and recently the PNI has been reported to be associated with the postoperative survival in CRC patients [3-7, 14, 15]. Besides the indicating the nutritional status of a patient [16], the subjective global assessment (SGA) was also reported to be associated with the survival for CRC patients [4]. However, the SGA includes many subjective factors which require expert knowledge to accurately measure them [4], thereby limiting its clinical application. The CONUT score, which was reported to correlate with the SGA, was developed to evaluate the nutritional status more easily and more objectively [12]. However, there have been no previous reports on the relationship between the preoperative immune-nutritional status and the survival after curative surgery for CRC using the CONUT score. This is therefore the first report to evaluate the prognostic significance of the CONUT score in patients with CRC.
The serum albumin concentration is influenced by not only the nutritional status [11], but also by many other factors, such as damage to hepatocytes, infection, inflammation, dehydration or fluid retention status, etc. [10,17]. The lower albumin level in some patients may due to the production of cytokines and CRP, which modulate the production of albumin [11,18]. It was shown that a systemic and chronic inflammatory response to CRC was associated with a reduction in the survival of CRC patients [19,20]. It was also reported that as the inflammation due to cancer increased, the serum albumin concentration of the patients decreased. The poorer prognosis of the CRC patients with lower albumin concentration was associated with the presence of chronic and systemic inflammation [9].
The total peripheral lymphocyte count is one of the indicators of the immunological status [21]. T-lymphocytes play an important role in the immune response to cancer [22,23]. Menges Table 8. The results of the multivariate analysis of the association between the CONUT score and the total cholesterol level with relapse-free survival and cancer-specific survival.  Table 9. The results of the multivariate analysis of the association between the CONUT score and the total peripheral lymphocyte count with relapse-free survival and the cancer-specific survival.  [24]. A decrease in T-lymphocytes was reported to correlate with a poor prognosis because of the inadequate immune response to cancer [22,23]. As above, a decrease in the peripheral lymphocyte count is a poor prognostic factor in CRC patients [25]. Low serum cholesterol levels were reported to be associated with a poorer prognosis in patients with various cancers [11,26,27]. Although it remains unclear why a low serum cholesterol level is associated with a poor prognosis, hypocholesterolemia is not considered to be a cause of cancer, but to be induced by cancer [11]. There is increased expression of LDL receptor mRNA in tumor tissue than in normal tissue. The expression of LDL receptors on tumor cells makes them take up many LDLs [28], which decreases the serum cholesterol level [28]. In addition, the LDL cholesterol taken up into tumor cells increases tumor growth [29][30][31]. This hypothetical mechanism is supported by the reports that the serum cholesterol level increased after tumor resection [28]. A decrease in the serum cholesterol level means not only a lack of caloric intake, but also a loss of cholesterol from the cell membrane [10]. Hypocholesterolemia influences the cell membrane fluidity, which affects the mobility of cell surface receptors and their ability to transmit transmembrane signals [32]. Therefore, even if there are a sufficient number of immunocompetent cells present, they are unable to exert their immunological function against cancer cells due to the changes in their membranes [32,33]. It has been hypothesized that this is why hypocholesterolemia is associated with a poor prognosis.
The PNI, which is the immune-nutritional index calculated using the serum albumin level and the peripheral lymphocyte count, has previously been reported to be associated with the survival in CRC patients [3,9].
In the present report, the CONUT score more accurately predicted the survival in CRC patients than the PNI. Although the CONUT score and the PNI have common factors, they led to different results. Therefore, we examined the reasons why the CONUT score was superior to the PNI in predicting the prognosis. We found that all of the patients with a low PNI were included in the high CONUT group (Table 3). Because the CONUT score could detect the patients who were expected to have a poor prognosis, including some who were not detected by the PNI, the CONUT score was a more accurate prognostic indicator than the PNI. The patients with a high CONUT score who were not included in the low-PNI group had a low peripheral lymphocyte count and/or hypocholesterolemia. This is due to the fact that there is a higher emphasis placed on the peripheral lymphocyte count in the CONUT score. In addition, total cholesterol concentration which is not included in the PNI is an additional factor evaluated in the CONUT score. This is why the CONUT score is considered to be able to detect the patients who will have a poor prognosis more sensitively than the PNI.
Based on the results of the present study, it is thought that the use of the preoperative CONUT score could enable the stratification of risk for poor survival and help to individualize treatments. In clinical practice, patients with a higher risk of death from colorectal cancer can be selected based on the preoperative immune-nutritional status. The administration of more effective adjuvant chemotherapy to reduce the risk of recurrence and shorten the follow-up interval in order to diagnose early recurrence can improve the prognosis.
This study is associated with some limitations. First, this study was a retrospective, singlecenter design with a relatively small number of patients. Further studies, including prospective studies with a larger number of patients, should be performed to confirm our findings. Second, the ROC curve for critical cut-off used a value of 3, which was associated with a low AUC, a non-significant p-value and poor sensitivity; however, the CONUT score was significantly correlated with the outcome in this study. Third, although the CONUT was revealed to be superior to the PNI in this study, this result is based on an analysis of only 27 patients with the high CONUT/high PNI. Therefore, it may be difficult to draw any final conclusions based on such a small sample size. Fourth, there are many countries in which there are large gaps in income among individuals. Income is associated with the nutritional status and medical treatment. Under the Japanese Social Security system, the entire population is guaranteed the right to a minimum standard of living and affordable medical treatment. Therefore, there are fewer problems of malnutrition and poor medical care due to poverty in Japan. Likewise, chemotherapy is widely available and is appropriately used to treat patients. There are no significant gaps in income between the rich and poor in Japan. On the other hand, it might be necessary to consider this situation in other countries.

Conclusion
The results of this study suggest that the CONUT score is a strong independent predictor of survival among CRC patients. Furthermore, the CONUT score might be a more sensitive prognostic factor than the PNI.