Assessment of Protein Nitrogen Appearance in Chinese Peritoneal Dialysis Patients—Which Method to Use?

Abstract

Objective. We compared the Bergstrom's and Randerson's formula for PNA determination, and compared the normalization of PNA by ideal body weight (IBW) and standard body weight (SBW) as estimated by the Watson's formula. Methods. We studied 208 Chinese PD patients. Two 24-h dialysate and urine collections were performed six months apart. Protein nitrogen appearance was determined by the Randerson's formula (PNA-Rand) and Bergstrom's formula (PNA-Berg), the latter used as the gold standard. PNA-Berg was normalized with IBW and SBW, denoted as NPNA-IBW and NPNA-Watson respectively. The change of PNA over six months, denoted as ΔPNA-Rand and ΔPNA-Berg, were calculated. The results were compared by the Bland and Altman's method. Results. At zero month, the average PNA-Berg was 61.8 ± 14.8 g/day, and the average PNA-Rand was 58.1 ± 14.5 g/day. The value of PNA-Rand was consistently lower than the corresponding PNA-Berg. The bias of PNA-Rand was −3.7 g/day. The limits of agreement were −9.2 to +1.8 g/day. When NPNA-Watson was compared to NPNA-IBW, the bias of NPNA-Watson, using NPNA-IBW as gold standard, was 0.01 g/kg/day; the limits of agreement were −0.22 to +0.23 g/kg/day. The difference between NPNA-Watson and NPNA-IBW correlated with the body mass index (r = −0.820, p<0.001) and body weight (r = −0.834, r<0.001). After six month, there was a significant reduction in urine protein loss. However, total protein loss was only slightly reduced (7.3 ± 3.0 to 6.9 ± 2.8 g/day, p = 0.029). The correlation between ΔPNA-Berg and ΔPNA-Rand remained excellent (r = 0.983, p<0.001). The bias of ▵PNA-Rand was +0.3 g/day; the limits of agreement were −4.7 to +5.2 g/day. Conclusion. Ideal body weight that is validated for specific ethnic group, rather than the Watson's formula, should be used for normalization of PNA. Although the Randerson's formula under-estimates PNA when compared to the Bergstrom's formula, it is a reliable method for serial PNA monitoring because dialysate protein loss is stable in most patients.

• Protein nitrogen appearance
• Nutrition
• Peritoneal dialysis

Introduction

Protein nitrogen appearance (PNA) is an important nutritional index in patients receiving peritoneal dialysis (PD).[[1]] Nevertheless, the method for determination and normalization of PNA remains a choice of individual nephrologist.[[2]], [[3]] Both the Randersons's and the Bergstrom's formula have been recommended for the determination of PNA,[[3]], [[4]], [[5]], [[6]], [[7]] and either the Watson's formula or the ideal body weight could be used to calculate normalized PNA.[[8]], [[9]] In fact, the ideal body weight is often very different from the body weight as estimated by the Watson's formula in many ethnic groups. In the present study, we examined a large group of Chinese PD patients to compare the methods of PNA determination and normalization.

Patients and Method

Study Design

We studied 208 prevalent PD patients in our dialysis unit. Baseline data including age, sex, underlying renal disease, presence of diabetes, hepatitis B status, PD regimen, and duration on dialysis were recorded. For each patient, standard clearance study was performed at 0 month and then repeated at six months. The method of clearance study has been reported previously.[[10]] Briefly, 24-h dialysate and urine collections were performed. If there was an acute medical problem such as peritonitis, this study was obtained one month after resolution of the problem. The daily drained volume of dialysate was mixed, measured volumetrically and sent for measurement of urea, creatinine, and glucose concentrations. Daily protein loss in dialysate (DPr) was measured. Similarly, 24-h urine volume and urinary concentrations of urea and creatinine were determined, and daily urinary protein loss (UPr) was measured. Daily total protein loss (TPr) was defined as the sum of DPr and UPr. At the end of the 24-h collection, body weight was measured, and a blood sample was obtained for measurement of plasma urea, creatinine, and albumin concentrations. Serum albumin level was measured by bromocresol purple method. Body surface area (BSA) was determined from body weight and height by Dubois' formula.[[11]]

Protein Nitrogen Appearance

Urea nitrogen appearance (UNA) was determined from the total amount of urea in 24-h dialysate and urine collection. Protein nitrogen appearance was calculated by the Bergstrom's formula (PNA-B)[[3]], [[4]] and the Randerson's equation (PNA-R)[[6]] as follows:while UNA and TPr were in gram per day.

PNA-Berg were further normalized by the ideal body weight (IBW) and standard body weight (SBW) as follows:Ideal body weight was determined by the body height and sex according to a standard formula validated in Southern Chinese[[12]]:Standard body weight was defined as total body water (V) divided by 0.58. Total body water was determined by the Watson's formula[[13]]:

Total weekly Kt/V and weekly creatinine clearance normalized to BSA (CCr) were also determined by standard method.[[14]] Residual GFR was calculated as average of 24-h urinary urea and creatinine clearance as described.[[15]] Creatinine concentration in dialysate was corrected for interference by glucose according to reference formula determined in our laboratory.[[16]]

Serial Change in PNA

The change in PNA over the six-month period was calculated as follows:

The change in total protein loss (ΔTPr), dialysate protein loss (ΔDPr), urinary protein loss (ΔUPr), and glomerular filtration rate (ΔGFR) were calculated in similarly.

Statistics

Statistical analysis was performed by SYSTAT 7.0 for Windows software (SPSS Inc., Chicago, IL). All data were expressed in mean ± SD unless otherwise specified. PNA-Berg and PNA-Rand at 0 month were compared by the Bland and Altman's method,[[17]] using PNA-Berg as the gold standard. In this analysis, the difference between the two methods of PNA determination was plotted against the corresponding PNA-Berg. Similarly, NPNA-IBW and NPNA-Watson at 0 month were compared, using NPNA-IBW as gold standard; ΔPNA-Berg and ΔPNA-Rand were compared, using ΔPNA-Berg as gold standard. Continuous variables were compared by Pearson's correlation coefficient. A p-value of less than 0.05 was taken as statistical significant. All probabilities were two-tailed.

Result

We studied 208 patients. The demographic and baseline clinical characteristics are summarized in Table 1. The median duration of dialysis was 14 months (range 1–154 months). The mean daily exchange volume was 6.5 ± 1.1 L; 164 patients (78.8%) had 6 L/day, 40 patients (19.2%) had 8 L/day, and 4 patients (1.9%) had 10 L/day dialysis exchange. At 0 month, the average PNA-Berg was 61.8 ± 14.8 g/day, and the average PNA-Rand was 58.1 ± 14.5 g/day. Dialysis adequacy and nutritional parameters are summarized in Table 2.

Table 1. Demographic and baseline clinical data

No. of patients	208
Sex (M:F)	104:104
Age (years)	57.0 ± 12.1
Body height (m)	1.60 ± 0.08
Body weight (kg)	58.0 ± 11.1
Renal diagnosis
Glomerulonephritis	71 (34.1%)
Diabetic nephropathy	63 (30.3%)
Polycystic kidney	9 (4.3%)
Hypertensive nephrosclerosis	8 (3.8%)
Obstructive uropathy	5 (2.4%)
Others/unknown	52 (25.0%)
Major comorbidity
Diabetes	71 (34.1%)
Cardiovascular disease	47 (22.6%)
HBsAg positive	16 (7.7%)

Table 2. Dialysis adequacy and nutritional status at 0 month and 6 months

	0 Month	6 Months	p Value
Protein nitrogen appearance
PNA-Berg (g/day)	61.8 ± 14.8	59.8 ± 14.3	p = 0.034
PNA-Rand (g/day)	58.1 ± 14.5	56.3 ± 14.0	p = 0.069
NPNA-IBW (g/kg/day)	1.13 ± 0.26	1.10 ± 0.25	p = 0.039
NPNA-Watson (g/kg/day)	1.14 ± 0.26	1.11 ± 0.25	p = 0.055
IBW (kg)	54.6 ± 5.7	54.6 ± 5.7	p = NS
SBW (kg)	54.8 ± 9.3	54.7 ± 9.3	p = 0.114
Protein loss
TPr (g/day)	7.3 ± 3.0	6.9 ± 2.8	p = 0.029
DPr (g/day)	6.5 ± 2.6	6.4 ± 2.5	p = 0.086
UPr (g/day)	0.8 ± 1.2	0.5 ± 0.9	p<0.001
Kt/V (total, weekly)	1.97 ± 0.52	1.91 ± 0.51	p = 0.09
Kt/V (peritoneal, weekly)	1.52 ± 0.38	1.54 ± 0.38	p = NS
CCr (L/week, total)	69.4 ± 31.0	62.2 ± 28.2	p<0.0001
Residual GFR (mL/min)
Mean ± SD	2.13 ± 2.22	1.67 ± 2.09	p<0.001
Median	1.67	0.88

Data were compared by paired Student's t test.

Baseline PNA-Berg and PNA-Rand

At 0 month, the correlation between PNA-Berg and PNA-Rand was excellent (r = 0.982, p<0.001) (Fig. 1). PNA-Rand was then compared to PNA-Berg by the Bland and Altman plot (Fig. 2). The bias of PNA-Rand, using PNA-Berg as gold standard, was −3.7 g/day. The limits of agreement were −9.2 to +1.8 g/day.

Figure 1. Scatter plot of PNA-Rand vs. PNA-Berg at 0 month. The line of equality is shown for reference.

Figure 2. Bland and Altman's plot of the difference between PNA-Rand and PNA-Berg vs. PNA-Berg at 0 month. Closed circles indicate cases with total protein loss (TPr) below 10 g/day; open circles, TPr 10–15 g/day; closed triangles, TPr above 15 g/day. Dotted lines indicate the overall limits of agreement.

When only cases with TPr less than 10 g/day were analyzed (n = 178), the bias of PNA-Rand was −2.9 g/day; limits were agreement were −6.3 to +0.6 g/day. When cases with TPr above 10 g/day were analyzed (n = 30), the bias of PNA-Rand was −8.7 g/day; limits of agreement were −13.7 to −3.7 g/day. When only cases with TPr above 15 g/day were analyzed (n = 7), the bias was −12.4 g/day; limits of agreement were −14.5 to −10.2 g/day.

NPNA-IBW vs. NPNA-Watson

There was a fair correlation between IBW and SBW (r = 0.819, p<0.001). At 0 month, the correlation between NPNA-IBW and NPNA-Watson was excellent (r = 0.904, p<0.001) (Fig. 3). When NPNA-Watson was compared to NPNA-IBW by the Bland and Altman method (Fig. 4), the bias of NPNA-Watson, using NPNA-IBW as gold standard, was 0.01 g/kg/day. However, the limits of agreement were −0.22 to +0.23 g/kg/day. The difference between NPNA-Watson and NPNA-IBW was significantly correlated with the body mass index (r = −0.820, p<0.001) and body weight (r = −0.834, p<0.001).

Figure 3. Scatter plot of NPNA-Watson and NPNA-IBW at 0 month. The line of equality is shown for reference.

Figure 4. Bland and Altman's plot of the difference between NPNA-Watson and NPNA-IBW vs. NPNA-IBW at 0 month. Dotted lines indicate the overall limits of agreement.

Serial Change of PNA

After six months, there was a small but significant reduction in TPr and UPr as a result of decline in residual renal function (see Table 2). Nevertheless, the correlation between ΔPNA-Berg and ΔPNA-Rand was excellent (r = 0.983, p<0.001) (Fig. 5). ΔPNA-Rand was then compared to ΔPNA-Berg by the Bland and Altman plot (Fig. 6). The bias of ΔPNA-Rand, using ΔPNA-Berg as gold standard, was +0.3 g/day. The limits of agreement were −4.7 to +5.2 g/day.

Figure 5. Scatter plot of ΔPNA-Rand vs. ΔPNA-Berg after 6 months. The line of equality is shown for reference.

Figure 6. Bland and Altman's plot of the difference between ΔPNA-Rand and ΔPNA-Berg vs. ΔPNA-Berg after 6 months. Closed circles indicate cases with total protein loss (TPr) below 10 g/day; open circles, TPr 10–15 g/day; closed triangles, TPr above 15 g/day. Dotted lines indicate the overall limits of agreement.

When only cases with TPr at 0 month less than 10 g/day were analyzed (n = 178), the bias of PNA-Rand was −0.2 g/day; limits were agreement were −4.5 to +4.2 g/day. When cases with TPr above 10 g/day were analyzed (n = 30), the bias of PNA-Rand was +2.8 g/day; limits of agreement were −2.6 to +8.3 g/day. When only cases with TPr above 15 g/day were analyzed (n = 7), the bias was +3.9 g/day; limits of agreement were −1.9 to +9.7 g/day.

Discussion

In patients with stable body weight, protein nitrogen appearance is an indirect measurement of dietary protein intake.[[1]], [[18]] It is commonly used as an index of nutritional status in PD patients in clinical trials[[10]], [[19]] and in routine clinical practice.[[5]], [[20]]

We found that there was a substantial discrepancy between ideal body weight (IBW) and standardized body weight (SBW). It should be noted that SBW was derived from the Watson's formula, which has not been validated in Chinese. In fact, the discrepancy between IBW and SBW was proportional to body mass index and measured body weight of the patient. In other words, SBW tends to over-estimate body size, especially in obese patients. Our observation is different from that reported in western population.[[21]] Nevertheless, we believe that IBW should be used for normalizing PNA because IBW has been validated in Chinese population, while the Watson's formula is not. Our findings highlight the importance of using ideal body weight that has been validated for specific ethnic groups for the normalization of PNA.

Unfortunately, IBW does not take into account of the patients' body built. It has been suggested that lean body mass should be the ideal parameter for normalizing PNA, at least in western country.[[9]], [[22]] However, lean body mass derived from creatinine kinetics has substantial mathematic coupling with PNA, because both parameters are derived from the same 24-h dialysate and urinary sample.[[23]] In addition, the formula described by Forbes and Bruining[[24]] has not been validated in Chinese. Our previous study found that lean body mass determined by anthropometric method had poor correlation with that derived from creatinine kinetics and may not be reliable in Chinese.[[25]]

The Bergstrom's formula is now regarded as the ideal method for calculation of PNA.[[3]], [[4]] Although the DOQI guidelines also recommend the Randerson's formula,[[4]], [[5]] there are methodological concerns. Dialysate and urinary protein loss are estimated rather than calculated in the Randerson's formula.[[6]] Therefore, the estimation would not be accurate in patients with significant residual renal function and possibly substantial proteinuria. As expected, we find that the Randerson's formula consistently gives a lower estimate of PNA than the Bergstrom's formula in the present study. Furthermore, the higher the total protein loss, the higher was the error of estimation (see Fig. 2). Both observations suggest that the Randerson's formula consistently under-estimates dialysate and urinary protein loss.

One of the major concerns of the Randerson's formula is that urinary protein loss is expected to decline with time as residual renal function goes off. As a result, the error of estimation by the Randerson's formula, at least theoretically, would not be constant. It could be argued, therefore, that Randerson's formula is not suitable for serial monitor of PNA, especially in patients with significant residual renal function. Contrary to the theoretical concern, we find that Randerson's formula and Bergstrom's formula give highly consistent measurements in the change of PNA with time, and the error was only moderately dependent on total protein loss. In fact, the bias and errors of estimations (see Fig. 6) were quite negligible when the values are normalized to NPNA. The reason is obvious: urinary protein loss only has a small contribution to the total daily protein loss. In fact, dialysate protein loss, which was remarkably constant over time (see Table 2), accounts for almost 90% of the total protein loss. However, it should be noted that the period of observation was only six months in the present study. It would require another study to examine the reliability of the Randerson's formula for serial PNA monitoring over an extended period of time. As a matter of fact, there are limited data on the longitudinal change of dialysate protein loss in PD patients.

In summary, ideal body weight that is validated for specific ethnic group, rather than the Watson's formula, should be used for normalization of PNA. Although the Randerson's formula gives a somewhat biased estimate in PNA, it remains a reasonable method for serial PNA monitoring of individual patient, especially in patients with little residual renal function.

Acknowledgment

This study was supported in part by the CUHK research accounts 6901031 and 6901112.