Are we overfeeding hemodialysis patients with protein? Exploring an alternative method to estimate protein needs

Are we overfeeding hemodialysis patients with protein? Exploring an alternative method to estimate protein needs Manouk Dam a, , Eva Anne Hartman , Hinke Kruizenga , Brigit C. van Jaarsveld , Peter J.M. Weijs b, d a Department of Nutrition and Dietetics, Amsterdam University Medical Centers, Location AMC, Meibergdreef 9, 1105, AZ, Amsterdam, the Netherlands b Department of Nutrition and Dietetics, Amsterdam University Medical Centers, Location VUmc, Boelelaan 1117, 1081, HV, Amsterdam, the Netherlands c Department of Nephrology, Amsterdam University Medical Centers, Location AMC, Meibergdreef 9, 1105, AZ, Amsterdam, the Netherlands d Amsterdam University of Applied Sciences, Faculty of Sports and Nutrition, Department of Nutrition and Dietetics, Dr. Meurerlaan 8, 1067, SM, Amsterdam, the Netherlands


Introduction
Protein-energy wasting (PEW) is a syndrome that occurs in hemodialysis (HD) patients and has a negative impact on clinical course [1e3]. Multiple mechanism can cause a state of PEW, such as uremic toxicity, inflammation and insufficient protein intake. Ensuring an optimal protein intake is considered an important part of treatment in PEW [1,2,4].
In daily practice, dietitians take these factors of PEW into account and subsequently calculate patients' protein needs based on (ideal) bodyweight (BW) which is in accordance with several guidelines [4e7]. In previous studies, a mean protein intake of at least 1.0e1.2 g/kg/day showed a neutral or positive nitrogen balance [8e10]. In case of severe PEW, amounts up to 1.5 g/kg/day are also considered (11). In the Netherlands often actual BW is used instead of ideal BW and in previous studies is suggested to adjust BW in under-and overweight patients, using body mass index (BMI) [11,12]. This is based on a study investigating protein needs in ICU patients, in which protein mass, thus fat free mass (FFM), was measured precisely with in vivo neutron activation analysis [13]. In this population, 1.5 g of protein/kg FFM corresponded with 1.2 g of protein/kg BW and 1.9 g of protein/kg FFM with 1.5 g of protein/kg BW. In subjects with a healthy weight, this calculation was accurate, though in patients with under-and overweight this appeared incorrect. Protein requirements in underweight patients are often underestimated, because the body contains relatively more protein per kg BW. In overweight patients the opposite occurs, and consequently protein needs are overestimated when calculated with their current BW. Therefore, arbitrarily in practice, in patients with a BMI below 20, BW is adjusted to BMI ¼ 20 and in patients with a BMI above 27.5, BW is adjusted to BMI ¼ 27.5.
Considering body protein mass is most represented in FFM, why is BW still used when calculating protein needs? This is a topic of discussion in the Netherlands. There is a good rationale to apply FFM instead of BW, because BW doesn't provide information on body composition and individual differences can be expected. A Dutch study investigated these methods in hospital in-and outpatients, and great under-and overestimation in protein needs was seen when using BW, especially in under-and overweight patients [14]. In our clinical practice we noticed these same differences between HD patients. This is of great interest, since insufficient protein intake could contribute to developing PEW [15,16] and excessive protein intake can induce hyperphosphatemia, contributes to metabolic acidosis and in the long-term vascular calcifications [5,17].
It is of great interest to investigate which differences are seen when calculating protein needs based on FFM versus BW. Therefore, the aim of this study is to compare protein needs based on either FFM or BW in HD patients.

Study design and population
In this prospective study, measurements of body composition were performed between June 2009 and May 2019 as part of regular care. Data of these measurements were included in this study. Patients on maintenance HD, with a conventional or nocturnal schedule, were included. Three dieticians were trained to perform nutritional assessment and regularly performed nutritional assessment in all patients. All patients filled out a written informed consent. This study is in accordance with the Declaration of Helsinki and has been approved by the Medical Ethics Committee (METc).

Body composition and anthropometric measurements
Body composition was assessed with multi-frequency bioelectrical impedance spectroscopy (BIS), using a Body Composition Monitor (Fresenius Medical Care, Bad Homburg, Germany). Measurements were mostly performed about 30 min after HD treatment, with the intention to measure the most optimal stability in a patient's body fluid compartments. More details regarding body composition and anthropometric measurements were previously described [18].

Estimation of protein needs
Protein needs were estimated with data on FFM and BW and in this study the estimations of protein needs based on FFM were chosen as reference method. The following equations were used: FFM in kg x 1.5 g/kg BW per day compared with BW x 1.2 g/kg/ day, as well as BMI adjusted BW (adjustments were made when BMI was below 20 kg/m 2 or above 27.5 kg/m 2 ). FFM in kg x 1.9 g/kg BW per day compared with BW x 1.5 g/kg/ day, as well as BMI adjusted BW as previous described.

Statistical analysis
Patient characteristics were reported as means with standard deviations, medians with interquartile ranges or proportions when appropriate. A consensus was reached by the research group, that an estimation of protein needs within ±10% of the reference method was considered accurate. Below 90% was considered underpredicted, above 110% as overpredicted. Independent samples t-tests were used to determine differences between males and females in subject characteristics and protein needs. Repeated measures ANOVA was used to assess differences in protein needs between males and females and between different BMI groups. Pairwise comparisons of the means were analysed with post hoc Bonferroni test. BlandeAltman plots were created with scatterplots of the differences between methods against the mean between methods. The limits of agreement were established as two standard deviations above and below the mean difference. Statistical analysis were performed using SPSS statistics, version 26.0 (SPSS, Chicago, IL, USA). Statistical significance was considered at the level of P 0.05. Table 1 shows baseline characteristics of the study population. A total of 115 HD patients were included in this study, with 64% males (n ¼ 73). The mean age of the total group was 54.5 ± 15.2 years. Mean BW was of 80.5 ± 17.5 kg and a BMI of 25.8 ± 5.2 kg/ m 2 in males and mean BW was 70.8 ± 17.9 kg and a BMI of 26.6 ± 6.1 kg/m 2 in females. Mean FFM was 47.6 ± 9.3 kg in males and 31.8 ± 8.5 kg in females (P < 0.001). Table 2 shows the estimates of protein needs of the total group by sex. Mean protein needs estimated with BW and BMI adjusted BW all showed higher protein needs in g/day, compared with FFM. In males, comparing FFM x 1.5 g/kg with BW and BMI adjusted BW, mean protein needs were 26 ± 22 g/day and 21 ± 18 g/day higher. In females, higher protein needs were found of 37 ± 26 g/ day with BW and 31 ± 16 g/day with BMI adjusted BW. Using FFM x 1.9 g/kg, similar results were found, as shown in Table 2. FFM x 1.5 g/kg and FFM x 1.9 g/kg, compared with both equations in either sex, showed overpredictions of 80% up to 100%. Figure 1A,B shows the differences in protein needs (g) of FFM x 1.5 g/kg and FFM x 1.9 g/kg compared with BW and BMI adjusted BW, by sex and across BMI groups. In both FFM equations, comparisons with BW and BMI adjusted BW, the difference in protein needs rises with BMI. In males and females, protein needs calculated by BW and BMI adjusted BW were significantly higher compared to FFM, across all BMI categories (P < 0.01). In overweight (BMI 25 to 30) and obese (BMI >30) males and females, a large significant difference was found (P < 0.01): in overweight males mean protein needs were found to be 30 ± 15.6 g/day higher calculated with BW and 29 ± 15.4 g/day higher when using BMI adjusted BW, in obese males higher protein needs were found of 51 ± 20.4 g/day and 23 ± 20.8 g/day, respectively. In overweight and obese females, mean protein needs were 37 ± 15.5 g/day higher with BW and 35 ± 13.5 g/day higher in BMI adjusted BW, and mean protein needs of 69 ± 17.4 g/day and 45 ± 9.3 g/day in obese females, respectively. Comparisons with FFM x 1.9 g/kg showed corresponding P-values and differences in mean protein needs. Figure 2AeD shows BlandeAltman plots of mean protein needs and differences in protein needs between BW, BMI adjusted BW and FFM, by sex. In males, FFM x 1.5 g/kg compared to BW showed a mean difference of 25.2 with limits of agreements (LOAs) of À17.3 to 67.7. In the comparison with BMI adjusted BW the mean difference was 20.5 with LOAs of À14.3 to 55.2. In females, FFM x 1.5 g/kg with BW a mean difference was found of 37.2 with LOAs of À12.8 to 87.2. Compared with BMI adjusted BW, a mean difference was found of 31.0 with LOAs of 0.8e62.8.

Discussion
This study is discussing the topic on protein estimations by comparing estimations of protein needs based on FFM with estimations based on BW in HD patients. Our results show large differences between these two methods. Exploring different BMI groups, we found enormous overestimations of protein needs, especially in obese patients and more so in females, when considering FFM as reference method.
Our data show large differences in FFM between males and females (males 47.6 kg and females 31.8 kg, P < 0.001) and therefore differences in protein needs based on FFM among all BMI groups (P < 0.01). These differences in FFM are not unexpected; large differences in FFM between individuals exist, due to biologic factors, such as sex and age, and environmental influences such as daily physical activity [19,20]. However, in most practices dietitians do not measure body composition with e.g. BIS, but often only use BW, BMI or maybe a screening tool to diagnose PEW, such Subjective Global Assessment (SGA). In previous studies a high BMI is associated with better survival in HD patients [21,22] and SGA is assessed as a valid tool to measure a patients' nutritional status [23,24], but BMI or SGA do not distinguish FFM or provide any insight into differences in body composition. Therefore, certainly combined with higher values of BW and BMI, a patient could incorrectly be diagnosed with a good nutritional status whilst having a low FFM. FFM seems to be a more important factor for survival [21,25] and is considered as one of the most meaningful criteria for diagnosing PEW [26]. It seems of great importance that dietitians should perform more nutritional assessments in order to avoid this issue and to gain a complete representation of all important components that are part of protein-energy wasting in HD patients.
It is however, challenging to measure FFM in a reliable and reproducible way. Ideally, FFM is established with an indirect  method such as, Dual-Energy X-ray Absorptiometry (DXA) or a computed tomography (CT) scan [27e29] instead of a double indirect method, such as BIS. Unfortunately, these methods are, besides expensive and time-consuming, not available for nutritional assessment. An easier, more convenient method to estimate FFM is by BIS and previous literature shows a good agreement among DXA and bio-electrical impedance techniques [30e33]. In a chronic kidney disease (CKD) population, BIS is considered appropriate, because it can discriminate intra-and extracellular body fluids adequately and is found to give a complementary insight in nutritional status [30,34e39]. BIS is relatively cheap, easy to obtain by a dialysis center and even though BIS is an indirect method for measuring FFM, it still seems more appropriate than using BW and thereby not taking FFM into account at all.
In conclusion, the main goal of this paper is to discuss and create awareness on discrepancies between estimations of protein needs based on FFM and BW. This is a current topic of discussion in the Netherlands, since concerns arise around the use of BW to calculate protein needs. Evidence on this topic is scarce and there are several arguments in favor to use FFM instead of BW. Therefore, it is suggested to use measured FFM as basis for protein needs in order to avoid protein overfeeding in this vulnerable patient group, as well as avoid protein underfeeding as muscle wasting is an ongoing threat. This study shows large differences when using FFM, and when considering FFM as preferable choice, we might be overfeeding patients with protein. Although, there are challenges regarding measuring FFM accurately, we recommend to address this issue more and investigate this topic in further research.

Funding sources
No funding was received for this manuscript.

Declaration of competing interest
The authors declare that they have no conflict of interest.