Bioelectrical impedance analysis as a nutritional assessment tool in Autosomal Dominant Polycystic Kidney Disease

Objective Autosomal dominant polycystic kidney disease (ADPKD) patients with massive organomegaly suffer from pressure-related complications including malnutrition. In this study, we analyzed the efficacy of segmental bioelectrical impedance analysis (BIA) for objective and quantitative nutritional assessment in ADPKD patients Design and methods We conducted a cross-sectional study, to evaluate the clinical utility of segmental BIA for assessing the nutritional status of ADPKD patients. BIA measurements was assessed according to modified subjective global assessment (SGA) scores and were compared with data from a healthy population. The association between BIA measurements and the height adjusted kidney and liver volumes (htTKLV), were analyzed. Subjects A total of 288 ADPKD patients, aged ≥ 18 years old, were analyzed Main outcome measures Nutritional status was evaluated with SGA and segmental BIA. The htTKLV were measured in each patients using computed tomonography images. Results Higher ratios of extracellular water to total body water (ECW/TBW) in the whole-body (ECW/TBWWB), trunk (ECW/TBWTR), and lower extremities (ECW/TBWLE) and lower phase angle of lower extremities (PhALE) correlated with lower SGA scores in the ADPKD population and in both gender. The four parameters, ECW/TBWWB, ECW/TBWTR, and ECW/TBWLE of >0.38 and PhALE of <5.8 θ were associated with malnutrition in ADPKD patients. These correlations were preserved in the subgroup analysis for chronic kidney disease stages 1-3A. Compared to healthy populations’ data, body fluid parameters and segmental ECW/TBW values, except for the upper extremities (ECW/TBWUE), were greater in ADPKD patients. Increased htTKLV was an independent risk factor for malnutrition in ADPKD. The highest correlation with htTKLV was observed for the ECW/TBWTR (r = 0.466), followed by ECW/TBWWB (r = 0.407), ECW/TBWLE (r = 0.385), PhALE (r = -0.279), and PhATR (r = 0.215). Conclusions These results demonstrated that segmental BIA parameters of ECW/TBWWB, ECW/TBWTR, ECW/TBWLE and PhALE provide useful information on nutritional status including the impact of organomegaly in ADPKD.


Introduction
Malnutrition in chronic kidney disease (CKD), also known as protein-energy wasting is one of the strongest predictors of mortality and morbidity [1,2]. Not only anorexia or inadequate intake of nutrients due to uremic symptoms in CKD patients but also inflammatory conditions and oxidative stress increase malnutrition risk in CKD patients [3]. In previous studies, nutritional markers such as serum albumin, creatinine, body mass index (BMI), and subjective global assessment (SGA) score were independent predictors of death and treatment failure in CKD [4,5]. Besides, the pre-transplant nutritional status in CKD patients are known to affect the outcomes of kidney transplantation [6,7]. Therefore, efforts have been made to establish guidelines for proper nutritional assessment and intervention to improve the outcome of CKD patients [8].
In autosomal dominant polycystic kidney disease (ADPKD), renal function decreases and the kidney and/or liver volume increases as the disease progresses. In addition to decreased renal function, organomegaly due to cyst growth is another risk factor for malnutrition in ADPKD, causing pressure-related symptoms and complications [9,10]. Therefore, regular assessment of nutritional status and timely intervention are important in ADPKD patients [11,12].
Subjective global assessment (SGA) is a standard nutritional assessment tool that has been well-validated in CKD patients [13,14]. However, the practical applications of SGA are limited, as it is a non-continuous and subjective scoring system that cannot detect subtle individual changes during regular follow-ups. Thus, other methods for monitoring nutritional status in ADPKD patients are required in addition to SGA.
Bioelectrical impedance analysis (BIA), which uses electrical currents and impedance to assess fluid status and body composition, could be an option for assessing nutritional status in ADPKD patients. Its use has been validated in various conditions, including CKD [15][16][17][18].
Increased ratio of extracellular water to total body water (ECW/TBW), change in body composition, and reduced phase angle (PhA) are known to be related to malnutrition [16,19]. The main advantages of BIA over other methods (Dual-energy X-ray absorptiometry, biomarkers, etc.) are its ease of use, short measuring time, and non-invasiveness. Additionally, because it is small and mobile and quantifies continuous parameters, BIA is suitable for detecting changes over time at the outpatient clinic.
However, the one-cylinder BIA model could underestimate total body water (TBW) in ADPKD patients because they have cystic fluids in the abdomen [20]. To overcome this, we employed segmental BIA with a 5-cylinder model that considers the upper extremities (UE), lower extremities (LE), and trunk as 5 separate cylinders for independent impedance measurements. Using segmental BIA, we were able to obtain segmental data for ECW/TBW, lean mass (LM), and PhA from each of the 4-extremities and the trunk separately.
In our previous study, ADPKD patients, especially those with large abdominal cystic organ volume, are found to be at risk for malnutrition [10]. If the height-adjusted total kidney and liver volume (htTKLV), the sum of height-adjusted TKV (htTKV) and height-adjusted TLV (htTLV), is �2,340 (mL/m), the risk of malnutrition increases 8.7 times after adjusting for renal function [10]. This is a follow-up study to evaluate segmental BIA as a nutritional assessment tool for ADPKD patients. First, we assessed the relationships of various BIA parameters with SGA to determine the most suitable BIA parameters to reflect nutritional status in ADPKD patients. Second, we compared the BIA data of ADPKD patients with those of healthy subjects from the general population. Thirdly, to analyze the effect of abdominal cystic mass on malnutrition in ADPKD patients, we obtained the htTKV and htTLV of all patients using computed tomography (CT) and analyzed the effect on segmental BIA data. We performed this study among pre-dialysis patients with stage 1 to 4 CKD to exclude the effect of severe renal dysfunction on nutritional status.

Materials and methods
This is a cross-sectional study of ambulatory patients who visited the outpatient ADPKD clinic following HOPE-PKD (coHOrt for genotype-PhenotypE correlation in ADPKD) protocol of Seoul National University Hospital. In the HOPE-PKD, a total of Korean 288 ADPKD patients were registered during December 2013 to March. 2014. Inclusion criteria of HOPE-PKD was subjects of 18 years or older and satisfied the Unified Criteria. The nutritional status of ADPKD patients were evaluated using SGA and BIA in our outpatient PKD clinic per standardized protocols. Patients with active cancer, infection, or renal replacement therapy or with a history of liver resection or transplantation were excluded. Detailed clinical information and reasons for liver resection or transplantation are discussed in a previous publication [21].
A database of BIA profiles from healthy subjects in the general population (Inbody Co., Ltd.) was used to compare with ADPKD patients after 1:1 matching. Data for 281 healthy subjects who were matched with the case patients for sex, age, and height ±2 cm were extracted from the database of the healthy population. PhA data was not included for the healthy population.
We used modified SGA with a seven-point scale, which has been validated in various studies in CKD patients [22]. Detailed nutritional assessment using modified SGA was described in our previous report [10]. The seven-point scale reflecting the dietician's subjective judgment of the patient's overall nutritional status was interpreted as follows: 7, well-nourished; 6, at risk; 5, mildly malnourished; 3-4, moderately malnourished; and 1-2, extremely malnourished.
Inbody S10 (Inbody Co., Ltd, Seoul, Korea), a multi-frequency segmental BIA analysis instrument, was used in this study [23]. In the outpatient's clinic, the measurements were performed in the standing position with 4-electrodes connected to both hands and feet of patient's. The BIA data of body fluid status, body composition parameters, and PhA of 50kHz were collected. In this study, we adjusted several quantitative segmental BIA parameters of the TBW, intracellular water (ICW), extracellular water (ECW), fat free mass (FFM), LM and fat mass (FM) by height, respectively (TBW/Ht, ICW/Ht, ECW/Ht, FFM/Ht, LM/Ht and FM/ Ht). UE segmental parameters were defined as the average of the left and right arm data, and LE segmental parameters were defined in the same manner. Whole-body PhA (PhA WB ) was defined as the PhA on the right-side of the body and was calculated from right-side body impedance, as described previously [24].
In our ADPKD clinic, abdominal CT scans were performed every other year [25], and the most recent abdominal CT scan was used to measure htTLV and htTKV. Total liver volume was calculated as the sum of the products of slice thickness and area measured on a set of contiguous image using Rapidia 2.8 CT software (INFINITT Healthcare Co. Ltd, Seoul, Korea) and the total kidney volume was estimated with the ellipsoid method [26]. The htTKLV, was defined as the sum of htTLV and htTKV. [10].
Anthropometric measurements (height, weight, and body mass index (BMI)) and laboratory measurements (serum hemoglobin, creatinine, total protein, albumin, and total cholesterol) were performed on the same day as the SGA and BIA evaluation. The estimated glomerular filtration rate (eGFR) was calculated with the CKD Epidemiology Collaboration (CKD-EPI) equation [27]. Continuous variables including age, height, weight, body mass index, serum hemoglobin, eGFR, and albumin showed normal distributions and were presented as mean value and standard deviation. However, htTKV, htTLV and htTKLV showed a non-normal distribution, and were therefore expressed as median value and interquartile range (IQR).
Since no subject had an SGA score less than 4 in the outpatient environment, all patients were classified into three groups: mildly to moderately malnourished (SGA score of �5), at risk (SGA score of 6), and well-nourished (SGA score of 7). For statistical analysis, we used the linear association test or the Jonckheere-Terpstra test to evaluate the P for trend among the three SGA groups. The value of P for trend <0.05 were interpreted as statistically significant. In the comparison study with healthy population data, paired t-test was used for normally distributed variables and Wilcoxon signed rank test for non-normally distributed variables. Receiver operating characteristic (ROC) curve analysis was used to evaluate BIA parameters for their ability to discriminate malnourished patients from well-nourished patients and to discriminate patients with significant htTKLV, which is � 2,400mL/m. The Youden index was used to determine the optimal cut-off value in MedCalc for Windows version 14 (MedCalc Software, Ostend, Belgium). Correlation analysis was performed to determine the relationships of BIA parameters with htTKLV. Due to its skewed distribution, htTKLV was natural logtransformed to ln htTKLV. Log odds graphs of definite malnutrition over normal nutrition with respect to various BIA parameters were plotted with STATA statistical software package version 13 (Stata Corp., College Station, TX, USA). All other statistical analyses were conducted with SPSS version 22 (IBM Corporation, Armonk, NY, USA).
To exclude the renal function effect on BIA result, subgroup analysis of early stage CKD patients (CKD stage 1-3a) was conducted. In this subgroup analysis, the BIA data was compared among SGA groups and the association between BIA parameters and abdominal cystic organ volume was also analyzed. Additionally, the adjusted log odds graph of definite malnutrition over normal nutrition with respect to various BIA parameters were plotted in whole study population after adjusted with age, hemoglobin level and renal function.
This study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No; 1205-112-411 and 1407-083-594). All participants provided written informed consent.

Association of BIA parameters with SGA in ADPKD patients
When we analyzed segmental BIA parameters across SGA score groups (4 or 5 vs. 6 vs. 7), the groups with lower SGA scores exhibited higher ECW/TBW WB , ECW/TBW TR , and ECW/ TBW LE (all for P for trend <0.001). The P for trend was statistically significant for UE ECW/ TBW (ECW/TBW UE ), but the trend was less robust than in the other segments of the body (P for trend = 0.035) ( Table 2, Fig 1).
Segmental PhA values also correlated with SGA score. The UE PhA (PhA UE , P for trend <0.001) and PhA LE (P for trend <0.001) differed significantly among the three SGA groups, with the mean value decreasing as the SGA score decreased. The P for trend was also significant for the trunk PhA (PhA TR ) across the SGA score groups, but the PhA TR trend was less robust than in the other segments of the body (P for trend = 0.015). The PhA WB , which has been reported as a nutritional marker in previous studies in other disease conditions [24], did not differ significantly among the three SGA groups (P for trend = 0.478) ( Table 2). For each body fluid parameter, such as ICW/Ht, ECW/Ht, TBW/Ht, the P for trend was statistically significant, but the mean values failed to exhibit definite trends across the SGA score groups. Regarding body composition parameters, the FFM/Ht and the LM/Ht for the WB, UE, and LE exhibited statistically significant P for trend values (all P for trends <0.01) without definite trends in the mean values across the SGA groups. FM/Ht did not differ significantly among the three groups. Only the trunk LM/Ht (LM/Ht TR ) exhibited a definite trend of decreasing mean value with decreasing SGA score (P for trend = 0.002) (S1 Table and S1  Fig and S2 Fig).

Differences between the genders in the association of BIA parameters with SGA
In our previous study [10], several parameters that were associated with SGA score in males (such as age, body weight, BMI, and hemoglobin level) showed no association in females. Therefore, in this study, subgroup analyses were undertaken to determine whether there were gender differences in the associations of BIA parameters with SGA score.
Four parameters of higher ECW/TBW WB , ECW/TBW TR , and ECW/TBW LE and lower PhA LE values correlated with lower SGA score in both male and female ADPKD. However, the other body fluid parameters and body composition parameters correlated with SGA score only in the male subgroup. ECW/TBW UE and PhA TR values, for which the P for trend values were statistically significant in the total population, did not exhibit such trends in either gender subgroup ( Table 2, S1 Table).

ROC curve analysis of BIA parameters with SGA in ADPKD patients
ROC curve analysis was conducted to identify the most suitable BIA parameter for predicting definite malnutrition (SGA �5). The parameters that differed significantly (P for trend <0.05) in both male and female patients and in the total population were analyzed, and ROC curves were generated using the data from subjects with SGA score of 7 (well-nourished) or �5 (malnutrition).
The areas under the curve (AUCs) and the ROC curves of the four relevant parameters are shown in Fig 2(A). The AUC for ECW/TBW WB was the largest (0.762), and the AUCs for all the other parameters were also greater than 0.7, in the following order: ECW/TBW TR (0.758), ECW/TBW LE (0.747) and PhA LE (0.741).

Log odds analysis of BIA parameters to predict malnutrition in ADPKD patients
When the log odds of malnutrition (SGA score of �5) were plotted across the values of each of the four BIA parameters that were significant in each sex and in the total population, the mean (95% confidence interval) log odds increased significantly as the ECW/TBW WB , ECW/ TBW TR , and ECW/TBW LE increased. On the contrary, PhA LE correlated negatively with log odds (Fig 3). The cut-off values for positive likelihood of malnutrition were ECW/TBW WB >0.380, ECW/TBW TR >0.379, ECW/TBW LE >0.383 and PhA LE <5.8. The same trends were observed in the subgroup analysis of both males and females ( Table 3). Similar log odds graphs were obtained after adjustment for age, hemoglobin level, and renal function (S3 Fig).

Differences in BIA parameters between the healthy population and ADPKD patients
To characterize the BIA parameters in ADPKD, we compared the BIA data of ADPKD patient with healthy population's by extracting 1:1 matched, 281 healthy subjects from the database of Inbody Co., Ltd, [15]. After the subjects were matched for sex, age, and height ±2 cm, there was no significant difference in weight or BMI between the ADPKD and healthy groups (p = 0.086 and 0.088, respectively).

Fig 2. ROC curve of BIA parameters for the diagnosis of malnutrition and significant htTKLV. Predicting A)
malnutrition defined as SGA score of 4 and 5 and B) significant htTKLV of �2,400 mL/m using BIA parameters. In the ROC curve analysis of BIA parameters to predict malnutrition, all AUCs were larger than 0.7 and the AUC for ECW/ TBW WB was the largest (0.762). In the ROC curve analysis of BIA parameters to predict significant htTKLV values of �2,400 mL/m, all AUCs of parameters shown in the figure were larger than 0.6 and the AUC for ECW/TBW TR was the largest (0.726). AUC; area under curve, BIA; bioelectrical impedance analysis, ECW/TBW WB ; ratio of extracellular water to total body water of whole-body, ECW/TBW TR ; ratio of extracellular water to total body water of trunk, ECW/TBW LE ; ratio of extracellular water to total body water of lower extremities, PhA TR ; Phase angle of trunk, PhA LE ; Phase angle of lower extremities, ROC; receiver-operating characteristics, SGA; subjective global assessment.
https://doi.org/10.1371/journal.pone.0214912.g002 . Log odds of having malnutrition increased as the ECW/TBW WB , TR and LE increased. On the contrary, log odds decreased with PhA LE increased. BIA; bioelectrical impedance analysis, ECW/TBW WB ; ratio of extracellular water to total body water of whole-body, ECW/TBW TR ; ratio of extracellular water to total body water of trunk, ECW/TBW LE ; ratio of extracellular water to total body water of lower extremities, PhA LE ; Phase angle of lower extremities, SGA; subjective global assessment.
The mean values of body fluid parameters including ICW/Ht, ECW/Ht, and TBW/Ht were higher in ADPKD patients than in the general population (P<0.001 for all). The mean values of ECW/TBW WB , ECW/TBW TR , and ECW/TBW LE were higher in ADPKD patients than in healthy subjects (P<0.001 for all).
Among the body composition parameters, FFM/Ht was higher in ADPKD patients than in healthy subjects and FM/Ht was significantly lower in ADPKD patients (P<0.001 for both). Whole-body LM/Ht (LM/Ht WB ) and lower extremity LM/Ht (LM/Ht LE ) were also higher in ADPKD patients than in the healthy population (P<0.001 for both). Increased LM/Ht WB and LM/Ht LE might reflect the edematous condition in ADPKD resulted from the decreased renal function and/or inferior vena cava compression by cystic organs. On the contrary, trunk LM/ Ht (LM/Ht TR ) was lower in ADPKD patients (P<0.001). (Table 4).

Subgroup analysis of early-stage CKD patients
We conducted subgroup analysis (n = 221 patients) of subjects with CKD stages 1 to 3a to exclude the effects of uremia on the nutritional and fluid status of ADPKD patients. The results of the subgroup analysis were similar to those of the overall analysis. (S2 Table and S3 Table) Associations between BIA parameters and abdominal cystic organ volume (htTKLV) We used segmental BIA data to elucidate the correlations of various BIA parameters with htTKLV. Fig 4 displays the scatter plots and Pearson's correlation coefficients for the relationships between ln htTKLV and BIA parameters such as the ECW/TBW TR , ECW/TBW WB and ECW/TBW LE , PhA TR and PhA LE . The highest correlation was observed for the ECW/TBW TR (r = 0.466), followed by ECW/TBW WB (r = 0.407), ECW/TBW LE (r = 0.385), and PhA TR (r = 0.215). The PhA LE correlated negatively with ln htTKLV (r = -0.279). In the subgroup analysis of patients with eGFR �45 mL/min/1.73 m 2 , the correlations of ln htTKLV with various BIA parameters in this subgroup were similar to those in the total population. The highest correlation was still noticed for ECW/TBW TR (r = 0.416), followed by ECW/TBW WB (r = 0.357), ECW/TBW LE (r = 0.337), PhALE (r = -0.250) and PhA TR (r = 0.232) in the subgroup analysis of patients with eGFR �45 mL/min/1.73 m 2 (Fig 4B).
ROC curve analysis was conducted to identify the most suitable BIA parameter for predicting significant htTKLV values of �2,400 mL/m. From previous study, htTKLV values �2,340 mL/m increased malnutrition risk 8.7 time [10]. In this study, to simplify the number, we used the cut-off value for significant htTKLV as 2,400 mL/m. ECW/TBW TR showed the largest AUC (0.726) with the following order; ECW/TBW WB (0.697), ECW/TBW LE (0.683), PhA LE (0.637) and PhA TR (0.601). The cut off value of ECW/TBW TR to predict significant htTKLV values � 2,400 mL/m was 0.387 with sensitivity of 57.7% and specificity of 76%. (Fig 2(B))

Discussion
In a previous study, we found that even in the early stages of CKD, about 30% of patients had SGA score below 6, and that an increased htTKLV was an independent risk factor for malnutrition in ADPKD patients [10]. Therefore, nutritional assessment is important in ADPKD patients, especially for those with large abdominal volume [12]. In addition to SGA, BIA can be used as a nutritional assessment tool, with the advantage of having objective, continuous variables. In ADPKD, with cystic organs, conventional one-cylinder-model BIA might underestimate the TBW, due to the truncal geometry and tissue interfacing [20]. To overcome the limitation of one-cylinder-model BIA in the use of ADPKD patients, we evaluated segmental BIA, which is based on 5-cylinder model, as a quantitative tool for nutritional assessment in ADPKD patients in this study. The ECW/TBW WB , ECW/TBW TR , and ECW/TBW LE and PhA LE correlated significantly with SGA score of ADPKD patients. Even though the difference was minimal, ECW/TBW WB showed the best correlation with the SGA score. These findings correspond to the results of other studies in various clinical settings. ECW/TBW, which reflects an edematous status and/ or malnutrition, is an important clinical factor related with the outcomes of CKD, liver cirrhosis, heart failure, and intensive care unit patients [28][29][30][31]. PhA, defined as the vector angle All parameters are presented as mean ± standard deviation. BIA; bioelectrical impedance analysis, ECW/TBW WB ; ratio of extracellular water to total body water of whole-body, ECW/TBW TR ; ratio of extracellular water to total body water of trunk, ECW/TBW UE ; ratio of extracellular water to total body water of upper extremities, ECW/TBW LE ; ratio of extracellular water to total body water of lower extremities, difference between the resistance (R) and reactance (Xc), reflects cell viability and membrane stability and is also known as a useful nutritional marker. Reduced PhA values are associated with malnutrition and poor outcomes in patients with CKD, cancer, and liver cirrhosis [32][33][34][35]. Our results indicated that high ECW/TBW and low PhA are clinically useful BIA parameters for the detection of malnutrition in ADPKD patients. (Fig 3). Decreased PhA has been regarded as a one of the best nutritional marker. However, except PhA LE , none of other segmental PhA parameters showed any correlation with the nutritional status in both gender of ADPKD patients. These results could represent the impact of organomegaly, since PhA TR was positively correlated with htTKLV. Therefore, in ADPKD patients, among segmental PhA parameters, only PhA LE, where the effect of organomegaly are minimal are suitable for nutritional parameters.
Since segmental BIA can measure the segmental fluid volume, higher ECW/TBW TR was positively associated with htTKLV. Other parameters such as ECW/TBW WB, ECW/TBW LE, and PhA TR also showed positive correlations and PhA LE demonstrated negative correlations with htTKLV (Fig 4). Our previous study showed that htTKLV was an independent risk factor for malnutrition in ADPKD [10]. ROC analysis showed that ECW/TBW TR of > 0.387 was related to significant htTKLV �2,400 mL/m. These data suggested that ECW/TBW TR could reflect nutritional status in ADPKD patients mainly caused by the mass effect of organomegaly.
Deterioration of renal function is another significant factor of malnutrition in ADPKD. To exclude effect of reduced renal function on nutritional status and BIA parameters, we created an adjusted likelihood graph. After adjustment for age, hemoglobin level, and CKD stage, higher ECW/TBW WB , ECW/TBW TR , and ECW/TBW LE and lower PhA LE values were associated with a greater likelihood of malnutrition (S3 Fig). Also in the subgroup analysis with CKD in stages 1-3a population, the correlations of BIA parameters with SGA scores were similar to those in the total population (S2 Table and S3 Table). This implies that, aside from reduced renal function, an increased htTKLV influences the nutritional status of ADPKD patients independently, and that segmental BIA can detect this malnutrition even in earlystage CKD patients. BIA parameters were also directly associated with ln htTKLV in ADPKD patients even in early CKD stages (Fig 2(B)). An increased htTKLV might affect BIA parameters by causing pressure-related gastrointestinal symptoms that lead to malnutrition [9]. Other pressure-related complications such as caval compression and LE edema could also influence BIA results.
The body composition parameters did not exhibit trends of association with SGA score in ADPKD population as in other CKD patients [36]. In the male subgroup, all the body composition parameters had significant positive correlations with the SGA score, while none were meaningful in females. This discrepancy between the genders could be related to the lower muscle mass in females than in males. Since our study was undertaken in Koreans, and in an outpatient setting where only mild to moderate malnourishment was observed, the degree of sarcopenia might have been too mild to be detected with statistical significance. Also ECW increase in CKD patients due to sodium retention and volume overload can result in overestimation of LM measured by BIA adopting 2-compartment body composition model (FFM and FM) [37,38]. However in the analysis of the study patients to see the association between ECW/TBW and skeletal muscle index, defined as the sum of appendicular lean mass divided by height square, the majority of patients were in the range of normal volume status (ECW/ TBW between 0.36-0.40) and non-sarcopenic status suggesting the minimal effect of volume overload on LM value from BIA in our study patients [39,40]. (S4 Fig) It was previously unknown how polycystic masses would be presented in BIA, therefore we compared the BIA results between ADPKD patients and the healthy population. Greater values for body fluid parameters and ECW/TBW WB , ECW/TBW TR , and ECW/TBW LE were observed in the ADPKD group than in the healthy population. Increased ECW/TBW WB and ECW/ TBW TR in ADPKD may demonstrated the influence of fluid-filled cysts in the abdominal organs. In case of ECW/TBW LE , where the effect of abdominal cystic mass were minimal, may represent nutritional status per se. However, compression effect of organomegaly on intra-abdominal vessels which cause LE edema in ADPKD also participate in the increment of ECW/TBW LE .
This study has a few limitations. Since this study was conducted in an outpatient clinic, patients with severe malnutrition have been excluded. Also there was no control subjects with CKD other than ADPKD. The Kidney and liver volumes were measured using CT instead of MRI. Additionally, information on the menstrual cycle in females was not gathered. Even though, this is the first study to evaluate BIA as a nutritional assessment tool in ADPKD patients and to identify the correlation of BIA parameters with abdominal cystic organ volume. In addition, the strength of the present study is that the kidney and liver volumes were measured in every patient to characterize their effect on the nutritional status and their relationship with segmental BIA parameters in ADPKD patients. Follow up studies to determine the association of malnutrition with clinical outcomes are needed. Using other malnutrition criteria such as malnutrition inflammation score [13] and newly developed GLIM criteria [41] would provide more information on the nutritional status in ADPKD patients. Studies of other nutritional biomarkers and body composition measures using a prospective cohort would enhance our understanding and management of malnutrition in ADPKD patients.
In conclusion, the ECW/TBW WB and other BIA parameters of ECW/TBW TR , ECW/ TBW LE and PhA LE were qualified malnutrition parameters. The ECW/TBW TR was not only associated with malnutrition, but also with htTKLV. Other parameters such as ECW/TBW WB , ECW/TBW LE and PhA LE also showed correlation with htTKLV but with less significance.

Conclusions
These results showed that segmental BIA can be a suitable tool for assessing nutritional status as well as the impact of abdominal cystic organs in ADPKD patients, which provides continuous and segmental parameters. All body composition parameter did not show significant association with SGA scores in female population. � P; P for trends SGA 7, well-nourished; SGA 6, at risk; SGA 5, mildly malnourished; SGA 3-4, moderately FM/Ht; height-adjusted fat mass, FFM/Ht; height-adjusted fat free mass, LM/Ht WB ; heightadjusted lean mass of whole-body, LM/Ht TR ; height-adjusted lean mass of trunk, LM/Ht UE ; height-adjusted lean mass of upper extremities, LM/Ht LE ; height-adjusted lean mass of lower extremities, SGA; subjective global assessment. In the subject group with both ECW/TBW WB > 0.4 and skeletal muscle index higher than the cutoff for sarcopenia, the lean mass could be overestimated because over-hydration are included in the lean mass. However, in our study population, majority of subjects were not included in this group suggesting the minimal effect of volume overload on lean mass measured by BIA. BIA; bioelectrical impedance analysis, ECW/TBW WB ; ratio of extracellular water to total body water of whole-body. (TIF)