Introduction

The number of cases of colon cancer is increasing in Taiwan and in other developed countries; the number of colectomies performed is also increasing. Malnutrition is a commonly encountered problem when treating colon cancer patients. Serum albumin is closely correlated with the degree of malnutrition and is a regularly used, simple marker of nutrition status [1, 2]. According to some previous studies, patients with hypoalbuminemia who underwent gastrointestinal tract surgery have significantly higher postoperative morbidity and mortality [3, 4]; however, some other studies report no significant differences [5, 6]. Hypoalbuminemia has also been used as a predictor of the survival of colorectal cancer patients [7–11], although the results are controversial.

Hypoalbuminemia in cancer patients does not only result from the cancer itself, but may also result from associated comorbidities [12, 13]. Furthermore, some comorbidities influence postoperative mortality and morbidity [13–16], as well as the survival of cancer patients [16–18]. There is little information concerning the relationships between hypoalbuminemia, comorbidities, and the outcomes of colon cancer patients. Therefore, this study aimed to clarify these relationships and determine if preoperative hypoalbuminemia might be useful for predicting the postoperative mortality, morbidity, and prognosis of colon cancer patients after receiving an elective and curative surgery.

Patients and methods

From January 1995 to December 2008 inclusive, a total of 3,849 consecutive patients with colon cancer underwent elective and potentially curative surgery at Chang Gung Memorial Hospital. Excluding 117 patients without preoperative serum albumin data (3.0% of colon cancer patients in our hospital), the remaining 3,732 patients were included in the analysis of short-term outcomes in this study. A total of 2,529 patients who received surgery more than 5 years ago and survived for more than 30 days after surgery were included in the analysis of long-term outcomes.

Detailed information regarding patient- and tumor-related variables and follow-up statuses was retrieved from the Colorectal Section Tumor Registry at Chang Gung Memorial Hospital. All data in this registry were collected prospectively. Patient demographic data, tumor characteristics, operative details, and short- and long-term postoperative outcomes were included in the data collection.

Patient-related variables consisted of age, gender, serum albumin level, CEA level, and comorbidities. The patients were divided into two age groups: ≤65 years (young) and >65 years (old). Hypoalbuminemia was defined as serum albumin < 35 g/L. CEA levels > 5 ng/ml were defined as abnormal. The following comorbidities were recorded: hypertension, cardiac disease, old cardiovascular accident (CVA), asthma, diabetes, hepatitis, liver cirrhosis, thyroid disease, and other comorbidities, such as peptic ulcer disease, urolithiasis, and gall stones.

Tumor-related factors consisted of location, size, morphology, histology, degree of differentiation, and stage. Tumor stages were determined according to the AJCC TNM staging system (sixth edition)[19]. Tumor location was categorized as right colon (from the cecum to transverse colon) or left colon (from the splenic flexure to sigmoid colon). Tumor morphology was divided into polypoid (include flat and polypoid tumors) and nonpolypoid (include ulcerative and infiltrative tumors).

According to the definition of hypoalbuminemia, patients were divided into two groups: hypoalbuminemic or normal. Each patient underwent standard oncological resection of colonic tumors and received routine postoperative care.

The short-term outcomes of surgery include postoperative morbidity and mortality. Postoperative morbidities were defined as complications within 30 days of the primary surgery and classified as wounds (infection or dehiscence), and pulmonary (atelectasis or pneumonia), cardiovascular (myocardial infarction, stroke, or embolism), urinary (bladder dysfunction or urinary tract infection), gastrointestinal (obstruction, ileus, or bleeding), anastomotic (leakage, stenosis, abscess formation, or peritonitis), and other conditions. Postoperative mortality was defined as death within 30 days of the primary surgery.

The endpoints of the long-term outcome study were overall survival (OS) and relapse-free survival (RFS). OS was calculated by death from any cause, and RFS was calculated by considering any relapses from the index cancer as the only events for survival analyses. Relapses of cancer were confirmed histologically or radiographically.

Statistical methods

Quantitative data were compared using Pearson’s chi-squared and Fisher’s exact tests. For multivariate analysis, logistic regression analysis was used to determine any confounding factors of mortality and morbidity; variables where P < 0.05 in the univariate analyses were used in the multivariate analysis. OS and RFS were calculated by univariate analysis using the Kaplan–Meier method. Survival curves were constructed using the Kaplan–Meier method and were compared using the log-rank test. In order to control for confounding factors, the Cox regression model was used for multivariate analysis in which variables for which P < 0.05 in the log-rank analyses were used. All P values were two-tailed, and were considered statistically significant if <0.05.

Results

The clinicopathologic characteristics of the patients categorized by serum albumin data are presented in Table 1. The mean ages (standard deviation) of patients in short- and long-term outcome analyses were 63.14 years (13.58) and 62.47 years (13.59), respectively. The proportion of hypoalbuminemia in stage I, II, and III colon cancer patients with elective surgery was 18.6%.

Table 1 Characteristics of patients and tumors in short- and long-term outcome analyses
Full size table

In short-term outcome analysis, there were no significant differences in tumor morphology and N stage of the TNM system between the normal and hypoalbuminemic groups. Advanced age, females, abnormal CEA levels, right colon tumors, large tumors (tumor size > 5 cm), mucinous adenocarcinoma, poor differentiation, stage II cancer, and advancing T stage of the TNM system were more likely to be associated with hypoalbuminemia. Similar results were found in the long-term outcome analysis—except there was no significant difference related to gender.

The comorbidities of patients in the two groups are presented in Table 2. There was no significant difference with respect to associated medical diseases between the two groups (52.9% and 56.4% of normal and hypoalbuminemic patients, respectively; P = 0.098). There was no significant difference in the occurrence of hypertension, cardiac disease, asthma, hepatitis, and thyroid disease between the two groups. However, CVA (P < 0.001), diabetes (P < 0.001), liver cirrhosis (P < 0.001), other comorbidities (P < 0.001), and combined comorbidities (P < 0.001) were significantly more likely to be associated with hypoalbuminemia.

Table 2 Associated comorbidities of patients according to preoperative albumin level
Full size table

Short-term outcomes

The analysis of postoperative morbidity and mortality in both groups is presented in Table 3. The postoperative mortality rate was 1.2% (45 of 3,732), and the morbidity rate was 11.7% (436 of 3732). Compared to patients with normal serum albumin, hypoalbuminemic patients have a higher rate of postoperative morbidity and mortality, as well as complications related to wounds, lungs, urinary system, and anastomosis. There was no significant difference between the two groups with regard to complications of the cardiovascular or gastrointestinal systems. In these variables, hypoalbuminemia was the only significant factor related to anastomostic complications. In multivariate analysis, TNM stage, age, sex, each category of comorbidity, CEA level, tumor location, tumor morphology, tumor size, histologic type and grade were taken into analysis if it is significant factor in univariate analysis.

Table 3 Postoperative morbidity and mortality according to preoperative albumin level
Full size table

Hypoalbuminemia was an independent factor for postoperative mortality, morbidity, and complications related to wounds, lungs, and the urinary system (Tables 4 and 5). Advanced age, males, asthma, and hypoalbuminemia were significantly associated with morbidity. Advanced age, diabetes, liver cirrhosis, abnormal CEA level, and hypoalbuminemia were significantly associated with mortality. Old CVA and hypoalbuminemia were significantly associated with complications related to wounds. Advanced age, abnormal CEA level, hypoalbuminemia, nonpolypoid tumor, and poorly differentiated tumors were significantly associated with complications of the lungs. Advanced age and hypoalbuminemia were significantly associated with complications of the urinary system. Males, asthma, mucinous adenocarcinoma, and TNM stage II cancer were significantly associated with complications of the gastrointestinal system.

Table 4 Multivariate analysis of mortality and morbidity
Full size table
Table 5 Multivariate analysis of each type of morbidity
Full size table

Long-term outcomes

For the 2,529 patients who underwent curative surgery between January 1995 and December 2004 and survived more than 30 days after the operation, the 5-year OS and RFS rates were 74.6% and 77.9%, respectively. When stratified by preoperative serum albumin levels, the 5-year OS rates of patients with normal albumin and hypoalbuminemia were 78.0% and 60.0%, respectively (P < 0.0001), and the 5-year RFS rates were 78.9% and 73.5%, respectively (P = 0.0042). Hypoalbuminemic patients had significantly poorer OS (Fig. 1) and RFS rates (Fig. 2) compared to patients with normal serum albumin.

Fig. 1
figure 1

Kaplan–Meier overall survival curves of colon cancer patients with curative resection stratified by albumin level

Full size image
Fig. 2
figure 2

Kaplan–Meier relapse-free survival curves of colon cancer patients with curative resection stratified by albumin level

Full size image

In a univariate analysis of OS and RFS rates, the following factors were examined: TNM stage, age, sex, each category of comorbidity, CEA level, occurrence of morbidity, tumor location, morphology, size, histologic type, and grade. TNM stage (P < 0.0001), age (P < 0.0001), sex (P = 0.0466), hypertension (P = 0.0028), cardiac disease (P = 0.0336), old CVA (P < 0.0001), diabetes (P < 0.0001), liver cirrhosis (P = 0.0097), other medical diseases (P = 0.0012), CEA level (P < 0.0001), occurrence of morbidity (P < 0.0001), tumor morphology (P = 0.0003), histologic type (P = 0.0024), and histologic grade (P = 0.0019) were significant predictors of OS. TNM stage (P < 0.0001), hypertension (P = 0.0236), cardiac disease (P = 0.0366), CEA level (P < 0.0001), tumor morphology (P < 0.0001), histologic type (P = 0.0001), and histologic grade (P < 0.0001) were significant predictors of RFS.

For multivariate analysis, the significant predictors from univariate analysis were introduced into the Cox regression model; the results are shown in Table 6 (OS) and Table 7 (RFS). For the OS rate, the remaining significant predictors were TNM stage (stage III vs. I—HR 2.11, 95% CI 1.57–2.83), age (>65 vs. ≤65—HR 1.92, 95% CI 1.65–2.23), sex (male vs. female—HR 1.16, 95% CI 1.01–1.33), old CVA (yes vs. no—HR 1.76, 95% CI 1.32–2.36), diabetes (yes vs. no—HR 1.23, 95% CI 1.01–1.49), CEA level (≥5 vs <5—HR 1.64, 95% CI 1.42–1.89), albumin level (<35 vs. ≥35—HR 1.75, 95% CI 1.49–2.08), and histologic type (signet-ring cell vs. adenocarcinoma—HR 2.26, 95% CI 1.16–4.38). For the RFS, the remaining significant predictors were TNM stage (stage II vs. I—HR 1.81, 95% CI 1.11–2.96; stage III vs. I—HR 4.21, 95% CI 2.59–6.85), CEA level (≥5 vs. <5—HR 2.04, 95% CI 1.72–2.43), albumin level (<35 vs. ≥35—HR 1.28, 95% CI 1.04–1.56), tumor morphology (nonpolypoid vs. polypoid—HR 1.53, 95% CI 1.20–1.96), and histologic type (signet-ring cell vs. adenocarcinoma—HR 2.53, 95% CI 1.27–5.06).

Table 6 Significant variables from univariate analyses for overall survival, and statistics of univariate and multivariate analyses
Full size table
Table 7 Significant variables from univariate analyses for relapse-free survival, and statistics of univariate and multivariate analyses
Full size table

In both the uni- and multivariate analyses, hypoalbuminemia was a significant predictor of poorer OS and RFS.

Discussion

Malnutrition is a common problem that debilitates cancer patients—including those with colon cancer [12, 20, 21]. Malnutrition in colon cancer patients may result from higher metabolism rates induced by cancer, reduction of food intake, impairment of hepatic protein synthesis, or blood loss, etc. [12, 20, 21]. Although it cannot comprehensively represent the nutritional status of patients [12], serum albumin level is extensively used. Hypoalbuminemia is widely accepted to be a good indicator of malnutrition [1–4, 22]. In the present study, 18.6% of colon cancer patients were hypoalbuminemic. In reality, the proportion of hypoalbuminemic colon cancer patients may be higher than in the present study because stage IV patients were excluded.

The association between hypoalbuminemia and adverse outcomes of elective [23], cardiac [24], and gastrointestinal surgery [3, 4, 25] has been known for many years. However, hypoalbuminemia is not an independent condition of surgical patients; certain types of diseases and characteristics of colon cancer patients also contribute to hypoalbuminemic status [12, 13]. However, previous studies have not explored the association of hypoalbuminemia with conditions such as diseases related to old CVA, diabetes mellitus, or liver cirrhosis, as well as cancer patient characteristics of advanced age, males, right side colon cancer, tumor size, depth of tumor invasion, mucinous type, poor differentiation, and abnormal CEA revealing in this study. These kinds of diseases and characteristics of colon cancer patients were also associated with adverse surgical outcomes; different factors may be associated with different postoperative complications. The confounding influences of hypoalbuminemia, characteristics of colon cancer patients, and diseases on surgical outcomes are seldom discussed in previous studies—the present study aimed to elucidate these issues. The results of this study show that hypoalbuminemia is an independent risk factor for postoperative mortality, morbidity, as well as complications related to wounds, lungs, the urinary system, and anastomosis, but not the gastrointestinal system or other complications. Although some previous studies suggest that hypoalbuminemia is significantly associated with postoperative ileus [26], this can be neither proved nor disproved by this study since postoperative ileus was not isolated from complications of the gastrointestinal system. The results of this study show that other kinds of adverse outcomes are associated with certain organ diseases. Old CVA patients are significantly associated with postoperative complications of wounds. Asthma patients have higher rate of complications of gastrointestinal system. Diabetes mellitus was a risk factor of postoperative mortality.

It is still controversial whether preoperative hypoalbuminemia influences the long-term outcomes of colon cancer patients. Some studies report that by itself, hypoalbuminemia is not an independent factor of survival [9, 10]. However, in conjunction with other factors, such as CRP [9] and CEA [10], they can significantly alter survival rates. Few studies report that hypoalbuminemia is an independent factor for poor survival [7, 11]. However, the analyses used in past survival studies have included overall, disease-free, cancer-specific, and relapse-free survival rates; the results of survival studies vary according to the definition of survival used. It has previously been reported that non-surgical patients with hypoalbuminemia exhibit poor long-term survival [27, 28]. It is reasonable to suggest that hypoalbuminemic colon cancer patients have poor long-term survival rates because they tend to be older and exhibit comorbidities. Whether preoperative hypoalbuminemia influences the further metastasis of colon cancer is not known. In this study, both overall and relapse-free survival rates—representing the possibility of further metastasis—were analyzed. The results revealed that by itself, preoperative hypoalbuminemia in colon cancer patients is an independent factor for poor long-term outcomes, including overall and relapse-free survival—this holds true even according to multivariate analysis, which included the variances in patient and cancer characteristics, and comorbidities. That is to say, colon cancer patients with preoperative hypoalbuminemia have poorer postoperative OS rates and a higher possibility of further cancer metastasis.

Although hypoalbuminemia is a well-known predictor of poor surgical outcomes, pre- or perioperative administration of albumin does not decrease the occurrence of complications [6, 12, 29, 30]. Recent studies revealed that pre- or perioperative nutritional supplements may improve surgical outcomes [31–33]. Whether the administration of albumin or nutritional supplements can improve long-term outcomes or diminish further metastasis remains unknown; further studies are needed.

Conclusion

Hypoalbuminemia is a predictor of poor surgical outcomes of colon cancer patients. It is also a poor prognosis factor for long-term survival of colon cancer patients after curative operation.