Limitations of serum albumin level as a marker of nutritional status in hemodialysis patients

Introduction: Albumin is widely used to evaluate nutritional status. However, its usage on patients with chronic kidney disease (CKD) remains controversial. This study was conducted to determine whether serum albumin levels can be used indicator of nutritional status in endstage CKD patients undergoing routine hemodialysis. Methods: This cross-sectional study included 535 subjects who underwent hemodialysis twice per week in 2017 from three hospitals in Jakarta, Indonesia. Data were collected from the medical records of the subjects and analysed using descriptive statistics and mean comparisons of independent samples. Results: Based on body mass index as a marker of nutritional status, the majority (85.9%) of subjects were not underweight. Almost an equal proportion of underweight subjects had normal serum albumin (15.1%) compared with hypoalbuminemia (14.4%). Hypoalbuminemia was found in 29.7% and 28.6% of non-underweight and underweight subjects, respectively. Moreover, there was no significant difference in serum albumin levels between underweight and non-underweight patients (p = 0.886, Mann-Whitney test). Conclusions: Our data do not support a role for serum albumin level as a marker for nutritional status in CKD patients.


INTRODUCTION
Malnutrition is a potential health problem for both adult and pediatric patients. Malnutrition contributes to increased morbidity and mortality rates, particularly among those with chronic pathological conditions. Malnutrition is also associated with increased healthcare cost. A study involving 709 adult patients, conducted in Brazil, reported a significantly higher incidence of medical complications in regions with higher rates of malnutrition (27%) compared with those with a lower prevalence of malnutrition (17%). Moreover, the mortality rate (12.4%) was also higher in patients with malnutrition than in those with good nutritional status (4.7%). 1 Low nutritional status or malnutrition are often observed in patients with chronic kidney disease, usually in the form of protein-energy wasting (PEW), a medical condition in which there is a low level of protein, or backup energy, within the body. The leading cause of PEW is decreased functional capacity due to metabolic stress or inadequate nutritional intake. During chronic kidney disease, various mechanisms can lead to PEW, including non-specific inflammatory processes, nutritional loss, metabolic acidosis, and due to the use of dialyser membranes. These conditions resulted in increased protein catabolic process via suppression of the insulin/insulin-like growth factor 1 (IGF-1) molecular pathway; and activation of the ubiquitin-proteasome system. 2,3 Although the measurement of albumin is established as a method to evaluate the nutritional status of patients with malnutrition, the predictive and diagnostic values of serum albumin in patients with chronic kidney disease remain controversial. The reason was due to the limited number of studies and because of the contributions of multiple confounding factors towards serum protein concentration. The relatively long half-life of albumin in plasma and its characteristic as an acute-phase protein make its levels particularly susceptible to physiological and pathological processes, such as infection, medication, and processes that alter the hepatic metabolism, among others. This study aimed to evaluate the association of serum albumin level and nutritional status and to determine whether serum albumin level is a suitable nutritional marker in end-stage chronic kidney disease patients undergoing routine hemodialysis twice per week.

METHODS
The study design was cross-sectional. The inclusion criteria applied for this study were those patients on routine hemodialysis twice per week in the relevant medical facilities at the time of the investigation. Subjects were excluded if they died or stopped hemodialysis treatment during the period of data collection. The minimum sample size was calculated based on a previous study by Takahashi et al. (2012), which investigated the combination of

ORIGINAL ARTICLE
serum albumin, C-reactive protein, and body mass index (BMI) as predictors of long-term mortality in patients undergoing hemodialysis. 4, 5 The study includes 535 subjects attending three hospitals with hemodialysis units in Jakarta-Indonesia (Cipto Mangunkusumo Hospital, Koja District Hospital, and Cengkareng District Hospital). Demographic and dialysis-related data were collected from the medical records. BMI calculation was conducted according to Asian BMI classification. The albumin cut-off in this study is 3.5 g/dL. The data gathering was conducted after the study process had been explained, and the subjects had signed informed consent. Patients were also provided with assurance about data confidentiality and their right to withdraw their participation from the study.
Data analysis include descriptive statistics (frequency, mean, median, mode, standard deviation, chi-square test) and mean comparisons by independent T-test or Mann-Whitney test (non-parametric data). Kolmogorov-Smirnov method used to define the normality of the data and Levene test used to asses homogeneity of variance. Results were considered statistically significant if p-values were < 0.05. Statistical analysis was performed with SPSS version 21.
All ethical issues had been reviewed according to the latest version of the Declaration of Helsinki. Ethical clearance was obtained from the Health Research Ethics Committee, Faculty of Medicine, Universitas Indonesia, Jakarta, Indonesia (Number of Letter: 0612/UN2.F1/ETIK/2018).

RESULTS
In total, 535 samples were obtained from the medical facilities. However, based on the removal of specific samples due to incompleteness of the medical records, and data availability, there were some differences between the numbers of samples analysed for each variable in this study. Due to insufficient completeness of medical records, only 254 of 535 samples had serum albumin data.
As detailed in table 1, the subjects in this study were recruited evenly from the three medical facilities. The overall subject who underwent hemodialysis in this study were non-underweight (86%) and were mostly within the normal range of BMI (49.2%).
Demographic and dialysis-related data were presented in The characteristics of the study population with serum albumin data, in terms of the type of dialyser membrane used and the presence of infection, were also analysed ( Table 3). In total, there were 216 samples from patients who underwent dialysis with single-use membranes and 53 from those using reusable membranes. Most subjects using single-use dialyser membrane had normal serum albumin levels (78.2%), with low levels of serum albumin in only 21.8% of these subjects. In contrast, 67.9% of samples from patients using reusable dialyser membranes had low serum albumin levels. Based on infection status, 77.8% and 67.5% of subjects with or without infections, respectively, had normal albumin levels.

ORIGINAL ARTICLE
Next, we tested whether there was an association between BMI classification and serum albumin levels in study subjects, and the result was non-significant (p = 0.847; chi-square test) ( Table 4).
Mann-Whitney Test was performed since the data was not normally distributed (Kolmogorov-Smirnov, p>0,05). We also found no significant difference (p-value > 0.05; Mann-Whitney test) between the mean rank of underweight patients (126.24) and that of non-underweight patients (127.71). Therefore, we conclude that there was no significant difference in serum albumin levels between underweight and non-underweight patients who underwent hemodialysis twice per week in this study (Table 5).

DISCUSSION
The use of serum albumin levels as a marker to assess nutritional status, especially in patients undergoing hemodialysis, remains controversial. The advantages of serum albumin levels as a diagnostic tool for malnutrition are that they are simple and affordable to measure and that the results are highly reproducible. Serum albumin levels also respond consistently to follow-up intervention after their measurement, and levels remain stable following surgery. In contrast, the disadvantages of serum albumin level measurement include the existence of numerous factors that can influence albumin levels in the serum, specifically inflammatory responses, infection status, hepatic abnormalities, medication, sepsis, and history of previous trauma (e.g., burn injuries). 6 The published evidence indicates a lack of data supporting the use of serum albumin levels for diagnostic and evaluation of malnutrition in subjects either with or without specific pathological conditions. For example, a meta-analysis of 63 studies of the effects of starvation on healthy subjects, involving a total of 2125 patients, showed that serum albumin levels remain normal in most subjects with an extreme history of starvation (mean BMI <12 kg/m 2 or duration of starvation > 6 weeks) and clinical manifestations of malnutrition. 7 Another case-control study comparing 14 subjects with anorexia nervosa with 15 healthy subjects showed that there was no significant variation between the serum albumin levels in the two groups after one year of follow-up, indicating a lack of diagnostic utility of serum albumin level as an indicator for malnutrition. 8 There have been few reports of assessment of the value of serum albumin levels as a nutritional marker in patients with end-stage chronic kidney disease undergoing routine hemodialysis. Plasma protein concentration can be affected by several factors, including the rate of protein synthesis, fractional catabolic rate (FCR), the volume of distribution, or loss of proteins due to external factors. Under conditions of protein restriction, decreased FCR is a homeostatic response to decreased serum albumin concentrations. In contrast, in patients undergoing routine hemodialysis, this homeostatic response does not occur because of the multiple inflammatory responses commonly observed in patients with end-stage renal failure (particularly those with known source of infection). 9 In another observational study of patients undergoing routine hemodialysis, conducted by Kaysen et al., low serum albumin levels were significantly associated with systemic inflammation but did not exhibit any significant relationship with inadequate nutritional status. 10 The results of this study also demonstrate no significant correlation between serum albumin  In the past, low serum albumin levels have been considered as a potential indicator of low nutritional status. However, our results do not support this hypothesis. Instead, in this study, hypoalbuminemia was more common among non-underweight subjects (29.7%) than underweight subjects (28.6%). This study has several limitations, including the use of different types of dialyser membranes, which may be a confounding factor. The use of semi-permeable membranes to clear solutes through diffusion and ultrafiltration processes are affected by the type of membrane used: low-flux (membranes with lower water permeability) and high-flux (membranes with higher water permeability). The use of high-flux membranes is more effective for the removal of molecules of moderate to large sizes (10000-15000 Daltons), such as inflammatory proteins and lipoproteins. 11 Hence, it could be argued that the results may be confounded by the type of membranes used in the hemodialysis process. We found that hypoalbuminemia was more common in subjects undergoing dialysis with single-use membranes compared to those using reusable membranes (Table 3). However, similar results were also found in subjects with normal albumin levels, thus suggesting that these findings may be insignificant. These findings may be attributable to the higher proportion of single-use membrane users in the study population or the higher number of total serum albumin level data points for subjects undergoing dialysis with single-use membranes. Another potential confounding factor was the infection status. However, the conclusion about the association between serum albumin levels cannot be drawn out since the number of subjects who did not have any infection and with normal serum albumin level is also higher compared to the number of subjects with the presence of infection and hypoalbuminemia (Table 3).
The primary limitation of this study is the incompleteness of the data, particularly the serum albumin levels. Although data were collected in three medical centres (one of which is a national teaching and top referral hospital in Indonesia), the amount of data was insufficient, due to lack of routine measurement of serum albumin levels in patients undergoing regular hemodialysis.

CONCLUSION
Despite the presence of multiple confounding factors, which could theoretically affect levels of serum albumin, this study support showed that serum albumin levels are not significantly associated with nutritional status (based on BMI classification) in patients with end-stage chronic kidney disease undergoing routine hemodialysis. Further studies to confirm these results may include cohort studies, with subjects specifically chosen to undergo routine examinations with 1 year of follow-up during their routine hemodialysis processes, to ensure the collection of sufficient data. We also recommend comparisons of the use of albumin as a nutritional marker in end-stage chronic kidney disease patients undergoing routine hemodialysis with use of other nutritional laboratory markers, such as pre-albumin, transferrin, or retinol-binding protein, to identify other more effective nutritional markers.