Background

There are more than 225,000 cases of ovarian cancer diagnosed worldwide, accounting for almost 4% of all cancer diagnosed in women [1]. It is estimated that there will be almost 21,880 new cases and 13,850 deaths due to ovarian cancer in the United states in 2010 [2].Ovarian cancer accounts for more deaths that all other gynaecological cancer put together, this is partly due to presentation at late stage and lack of specific symptoms [3]. Advanced stage disease is often associated with omental cake, ascites resulting in poor nutritional status [4]. Malnutrition is associated with increased post-operative complications [5], poor clinical outcome [6] and death [7].

Various prognostic markers such as prognostic nutritional indices, serum albumin, total protein, transferrin, haemoglobin and anthropometric measurements have been used to assess nutritional status in gynaecological cancer patients. Out of these, serum albumin as an objective parameter often used to measure long-standing malnutrition [4].

Albumin is produced by the liver and almost 60% is present in the extravascular space. It helps to maintain intravascular oncotic pressure, facilitate transport of substances and acts as a free radical scavenger [8]. Malignant disease has been shown to be associated with low albumin due to inhibitory effect on its synthesis from liver [9] and sequestration in ascites or pleural effusion.

Albumin has been shown to be a prognostic marker in colorectal cancer [10], glioblastoma multiforme [11], gastric cancer [12] and breast cancer [13]. In this study, we have investigated the role of serum albumin as an independent prognostic predictor of survival in patients with ovarian cancer.

Methods

Retrospective data were collected for patients undergoing surgery for ovarian cancer at Derby City General Hospital from 1988 to 1998. The patients were treated initially with total abdominal hysterectomy, bilateral salpingo-oophorectomy and omentectomy. All the patients with stage Ic to IV disease received chemotherapy following surgery. The patients were followed up 3 months for first 2 years and 6 months thereafter for a period of 5 years. At each visit, the patients were assessed by both clinical and ultrasound examination with CA 125 evaluation to detect recurrence.

Clinico-pathological variables recorded included preoperative serum albumin, age, FIGO stage, tumour grade, extent of cytoreduction, histological subtype, details of adjuvant treatment and disease specific survival (DSS). DSS was calculated from operation date to 31 December 2005, when remaining survivors were censored. The database was audited to ensure its validity. Patients who had received prior chemotherapy were excluded from the study.

Statistical analysis

All data were analysed using SPSS version 7 (SPSS Inc, Chicago, IL). Continuous data were analysed using median, interquartile range and 95% confidence interval (CI). Categorical data were compared using chi square or Fisher exact test. Survival variation between groups was analysed using Kaplan–Meier curve using log rank test to estimate sizes of differences in survival. Multivariate analysis was done using Cox’s proportional hazards model to determine relative risk and independent significance of individual factors. A P value ≤ 0.05 was considered to be statistically significant.

Results

A total of 235 patients were identified during the study period. The median age of patient was 62 years with a median survival of 24.5 months. Almost 60% of the patients had stage 3 and 4 whilst 58% had optimal debulking surgery (Table 1). Patients with serum albumin level of <25 g/l had a median survival of 4.8 months, 25–35 g/l had a median survival of 15.2 months and in those with >35 g/l, the median survival was 43.2 months (Table 2; Fig. 1).

Table 1 Patients characteristics
Full size table
Table 2 Median survival based on different levels of preoperative serum albumin
Full size table
Fig. 1
figure 1

Kaplan–Meier survival curves for overall survival in patients with ovarian cancer showing progressively poor survival with reducing level of serum albumin

Full size image

On univariate analysis, variables including age at diagnosis, FIGO stage, grade, residual disease and serum albumin were associated with significantly poor survival whilst CA125, chemotherapy and histological were not (Table 3).

Table 3 Analysis of survival based on prognostic markers as single variable
Full size table

Multivariate Cox regression analysis was then performed on the variables that were associated with poor survival on univariate analysis. Age at diagnosis, FIGO stage and serum albumin retained their significance whilst residual disease and grade did not (Table 4).

Table 4 Cox proportional hazards model of various markers for survival in ovarian cancer patients
Full size table

Further analysis was done by comparing the three groups based on serum albumin levels (<25, 25–35 and >35 g/l) to other prognostic indicators of ovarian cancer. This demonstrated no difference in the proportion of patients receiving chemotherapy (P = 0.06) and CA 125 level (P = 0.23). Furthermore, serum albumin did not show any significant association with residual disease but was associated with FIGO stage, grade and histological type of the tumour (Table 5).

Table 5 Prognostic markers for ovarian cancer stratified according to serum albumin levels (g/dl) (Fisher exact test)
Full size table

Discussion

Ovarian cancer patients are more likely to present with poor nutritional status and cachexia due to the metabolic effects of tumour mass, ascites and small bowel obstruction [14]. Malnutrition leads to reduced muscle mass and subsequently affecting the functional status of the individual. Moreover, malnourished patients have defective scarring mechanisms leading to increased post-operative wound dehiscence and infections. In these patients, the tumour is also resistant to the effect of chemotherapy resulting in poor overall survival [15].

There have been various parameters used to measure the nutritional status of the patients with gynaecological cancer with varied results. They include anthropometric measurements (weight loss, body mass index (BMI), triple skin fold thickness and arm circumference), biochemical (serum prealbumin, albumin, transferrin, haemoglobin, and vitamins) and immunological (skin sensitivity tests) measurements [16].

Various prognostic scoring systems such as prognostic nutritional index (PNI) [17], prognostic and nutritional inflammatory index [18] and patient-generated subjective global assessment (PG-SGA) [19], consisting of combination of different prognostic nutritional markers have also been used to the state of nutrition in critically ill and cancer patients. Of the above markers, PG-SGA, serum albumin and skin fold thickness have been shown to be accurate in predicting malnutrition and subsequent survival in gynaecological cancer patients [16]. We therefore used preoperative serum albumin as predictor of survival in patients with ovarian cancer.

In our cohort, serum albumin levels of <25 g/l associated with median survival of 4.8 months compared to 43.2 months in patients with serum albumin of >35 g/l with clear separation of survival curves. This correlation of poor survival with decreasing levels of serum albumin has also been demonstrated by Parker et al. [20]. They have also shown that serum CA125 in addition to albumin is also an independent prognostic predictor of survival in patients with ovarian cancer. In our study, there was no correlation of CA125, chemotherapy and histology with survival.

On Cox regression model of multivariate analysis, serum albumin retained its capability as an independent prognostic marker for poor survival in patients with ovarian cancer along with age and FIGO stage. The presence residual disease after surgery is a well-known independent prognostic marker for survival in patients with ovarian cancer [21], but in our study the presence of residual disease lost its significance as poor prognostic marker on multivariate analysis.

Further stratification of various variables with respect to serum albumin, showed that there was no difference in patients receiving chemotherapy, presence of residual tumour and CA125 level although there was a significant association of serum albumin with stage, grade and type of tumour. This indicates the low albumin is associated with aggressive disease and is associated with poor survival.

The levels of serum albumin can also be affected by other associated conditions such as liver disease, uraemia, hypothyroidism, alcohol abuse, corticosteroids and trauma [22]. Moreover, albumin has a large body pool and a half life of 20 days and once the pool is depleted, it takes typically 14 days to return to normal on adequate nutrient supplementation [23].

Early recognition and preoperative nutritional support in malnourished patients has been shown to reduce post-operative complications and improve outcomes in patients with gastrointestinal cancers [24] and in patients undergoing pancreaticoduodenectomy [25].

As this study is retrospective analysis, it forms a good basis for further investigation on the use of serum albumin as prognostic marker in prospective trials. There is also a need for prospective studies evaluating the effect of optimisation of serum albumin in the peri-operative period, by adequate nutritional supplementation on overall survival in ovarian cancer patients.