Kidney International, Vol. 62 (2002), pp. 276–281

Serum prealbumin is higher in peritoneal dialysis than inhemodialysis: A meta-analysis

PHILIP GOLDWASSER, JOSEPH G. FELDMAN, and ROBERT H. BARTH

Department of Medicine, VA New York Harbor Heathcare Center, and Department of Preventive Medicine andCommunity Health, State University of New York, Downstate Medical Center, Brooklyn, New York, USA

Serum prealbumin is higher in peritoneal dialysis than in hemo- strongly with the serum level of albumin, and correlatesdialysis: A meta-analysis. at least as well as albumin does with other nutrition

Background. Although not widely appreciated, the reported markers in both hemodialysis (HD) and peritoneal dial-concentration of serum preabumin, like that of serum choles-ysis (PD) [abstract; Koomen et al, Perit Dial Int 14(Suppl 1):terol, tends to be higher in patients on peritoneal dialysis (PD)S32, 1994] [1–9]. Serum prealbumin also predicts survivalthan on hemodialysis (HD), despite the substantial loss of pro-

tein during PD. at least as well as does serum albumin in HD [1, 2, 5,Methods. The mean difference in serum prealbumin was 9], although not quite as well as serum albumin in PD

quantified by meta-analysis of the mean differences found in [10, 11].six cohorts with both PD and HD patients (set 1; N � 639)

K/DOQI clinical outcome goals have been set for theusing a fixed-effects model, and meta-analysis of the meanserum levels of both prealbumin (�30 mg/dL) and albu-prealbumin values reported in 23 cohorts of unselected dialysis

patients on a single modality (set 2; 9 PD cohorts, 14 HD min (�4.0 g/dL) [12]. It is well known that serum albumincohorts; N � 12,256) using a mixed model. For comparison, concentration tends to be lower in PD than in HDthe mean difference in serum albumin concentration between [13, 14], probably due to the substantial peritoneal lossPD and HD also was estimated in sets 1 and 2 using the same of albumin [15–19]. It is not surprising, therefore, that amethods.

serum albumin level of 4.0 g/dL or greater was achievedResults. In set 1, the mean prealbumin difference (PD-HD)by 32% of HD patients but only 17% of PD patients inin the individual cohorts ranged from 3.6 to 14.7 mg/dL (P �

0.05 in five cohorts), and the weighted mean difference was a recent large national survey [20]. Although not widely5.4 mg/dL (95% CI, 3.8 to 7.0 mg/dL). In set 2, weighted mean appreciated, PD also alters the serum level of prealbu-prealbumin was 8.1 mg/dL (95% CI, 5.2 to 10.9 mg/dL) higher min. However, serum prealbumin, like serum choles-in PD than in HD in the entire data set, and 6.9 mg/dL (95%

terol, tends to be higher in PD than in HD [1, 9, 13,CI, 5.2 to 8.6 mg/dL) higher in a sensitivity analysis that ex-14, 21], despite the loss of prealbumin and lipoproteinscluded two outlying HD studies. By contrast, weighted mean

serum albumin concentration was significantly lower in PD during PD [19, 22–24]. If PD induces a higher serumthan in HD in both sets 1 and 2; the mean difference was prealbumin level and lower serum albumin level com-0.25 g/dL (95% CI, 0.14 to 0.36 g/dL) in set 1 and 0.28 g/dL pared to HD, clinical targets specific to PD for these(95% CI, 0.14 to 0.42 g/dL) in set 2.

proteins may be required.Conclusions. Serum prealbumin level is approximatelyThe purpose of this study was to quantify the mean6 mg/dL higher in PD than HD, perhaps due to the stimulation

of hepatic synthesis by PD albumin loss, while serum albumin difference in serum prealbumin between PD and HD byis approximately 0.3 g/dL lower in PD. Different reference performing a pooled analysis of mean serum prealbuminranges and clinical targets (such as, K/DOQI guidelines) are values in two sets of published reports. Set 1 comprisedneeded for PD and for HD.

articles or abstracts that reported serum prealbumin incohorts containing both PD and HD patients, and set 2,articles that measured prealbumin in unselected dialysis

The concentration of serum prealbumin is a useful patients on a single modality. For comparison, the meanindicator of prognosis in patients on dialysis. It correlates difference in serum albumin between PD and HD also

was estimated in each of these data sets.Key words: prealbumin, albumin, peritoneal dialysis, hemodialysis,prognostic indicator, end-stage renal disease, kidney.

METHODSReceived for publication November 27, 2001 Literatureand in revised form February 12, 2002Accepted for publication February 14, 2002 Two literature sets were defined for meta-analysis.

Articles or abstracts reporting mean serum prealbumin in 2002 by the International Society of Nephrology

276

Goldwasser et al: Serum prealbumin in PD and HD 277

cohorts with both PD and HD patients were includedin set 1, and articles reporting mean serum prealbumin incohorts of unselected dialysis patients on a single mod-ality were included in set 2. Published articles that re-ported serum prealbumin in dialysis patients were lo-cated by searching PUBMED using “prealbumin” as thesearch word. Additional set 1 studies published as ab-stracts were identified by a manual search of the abstractpublications of the American Society of Nephrology(1982 to 2001), and those of the International Society forPeritoneal Dialysis and the Annual CAPD Conferencepublished in PD International between 1989 and 2001.The analysis was limited to publications located throughthe end of August 2001, reporting prealbumin concen-tration in units of mg/dL; when multiple follow-up re-ports were available on the same cohort, the earliestpublication was chosen. Assay methods for prealbumin Fig. 1. Serum prealbumin (�SE) in patients treated with hemodialysisand albumin were recorded when available. (HD; �) or peritoneal dialysis (PD; �) in six set 1 cohorts.

Six cohorts were selected for set 1 comprising 198 pa-tients on PD and 441 patients on HD. In the first cohort,the PD and HD experiences were reported in separate

weighing each diffi by the inverse of its variance (weighti),publications [1, 21]; the second [14], third [9], and fourthusing a method based on a fixed-effects model [40]. The[25] cohorts were each reported in single articles, andhomogeneity of the distribution of the differences aboutthe fifth and sixth were reported as abstracts (abstracts;the grand mean was tested with a chi-square test [Q �Koomen et al, op. cit.; Fassinger et al, Perit Dial Int� weighti � (diffi � grand mean)2].12(Suppl 1): 116, 1992) Three set 1 cohorts reported the

The set 2 study means were examined using a mixedprealbumin assay used (nephelometry [1, 9, 21]; immuno-model, with dialysis modality as a fixed factor, study asdiffusion [24]); serum albumin data were available ina random factor, and the inverse of the variance of thefour cohorts (abstract; Koomen et al, op. cit.) [1, 14,mean (N � SD2) as study weight. For any set 2 study that21, 25], and the method used in one (bromcresol greenreported the mean values of prealbumin (or albumin) for[1, 21]). Twenty-three cohorts were selected for set 2,subgroups only, we computed a single value for meanincluding nine PD cohorts comprising 490 patientsand a single pooled SD value by combining the subgroup(range 8 to 147) [11, 19, 24, 26–31], and fourteen HDdata. The results of the meta-analysis presented are thecohorts comprising 11,766 patients (range 17 to 7123)estimates of the mean (�SE) in the PD and HD studies[2–7, 32–39]. The prealbumin assays used were report-and of the mean difference. Sensitivity analyses wereed in eighteen studies and did not significantly differperformed in order to test the robustness of the resultsbetween PD and HD (PD, immunodiffusion 3, immuno-relative to exclusion of potential outliers defined by aprecipitation 1, immunonephelometry 3; and HD, immu-t statistic 3.5. Computations were performed using SASnodiffusion 3, immunoprecipitation 2; immunonephe-version 8 (SAS Institute, Cary, NC, USA).lometry 4; antigen-antibody complex 2; P 0.64 by

chi-square). Twenty-two set 2 studies (all except refer-ence 24) reported albumin data, and the method was RESULTSavailable in eleven (PD, bromcresol green 2, immuno-

Prealbuminchemical 1; HD, bromcresol green 6, immunodif-In set 1, the mean concentration of serum prealbuminfusion 2).

was higher in PD than in HD in each of the six cohorts,Statistics and data analysis the difference ranging from 3.6 to 14.7 mg/dL (P � 0.05

in 5 of the cohorts; Table 1 and Fig. 1). There was aFor each cohorti in set 1 (where i � cohort identifica-direct correlation between the HD and PD study meanstion number, ranging from 1 to 6), we tabulated the mean(Spearman correlation coefficient � 0.81 P � 0.05), whichvalues of serum prealbumin (or albumin), the standardsuggests that the calibration of the assays varied betweendeviation (SD), and the number of subjects (N) sepa-the studies. The weighted mean difference between PDrately for PD and HD, and computed the absolute differ-and HD was 5.4 mg/dL (95% CI, 3.8 to 7.0 mg/dL).ence of the means (diffi) and tested its significance by

The set 2 study means also showed a clear trend tothe t test, and, lastly, computed the weighted mean of thesix diffi values and the 95% confidence interval (CI), increased values in PD (Fig. 2), which was highly signifi-

Goldwasser et al: Serum prealbumin in PD and HD278

Table 1. Mean serum prealbumin in set 1 cohorts

PDa HDa

Cohort Mean � SD (N) Mean � SD (N) Differencea P valueb

1 30.4�8.0 (33) 26.5�6.4 (125) 3.9 �0.0052 32.2�9.5 (71) 28.6�9.3 (53) 3.6 �0.0383 37 �9 (55) 31�10 (192) 6 �0.00014 46.0�7.5 (3) 31.3�8.9 (21) 14.7 �0.0135 44 �12 (28) 35�10 (41) 9 �0.0026 40.8�9.9 (8) 35.0�2.6 (9) 5.8 �0.12

a mg/dLb t test

Fig. 3. Mean serum albumin plotted against mean serum prealbuminin patients treated with HD (�) or PD (�) in four set 1 cohorts. Thenumbers used to identify the cohorts are the same as in Table 1.

Fig. 2. Prealbumin (�SE) in 14 HD studies (open symbols) and 9 PDstudies (filled symbols) in set 2. Triangles, squares, and circles representsamples of 100, 101-1000, and 1000 patients, respectively. Data aremean � SE.

Table 2. Weighted estimates of mean serum prealbumin in PD andHD and of the difference in set 2 studies

PD (N) HD (N) PD-HD P value

40.7�1.2 (9) 32.7�0.8 (14) 8.1�1.4 mg/dL �10�4 Fig. 4. Mean serum albumin plotted against mean serum prealbumin40.7�1.0 (9) 33.1�0.7 (13)a 7.6�1.2 mg/dL �10�4 in patients treated with HD (open symbols) or PD (filled symbols) in40.8�0.9 (9) 33.9�0.1 (12)b 6.9�0.9 mg/dL �10�4 twenty-two set 2 studies. Triangles, squares, and circles represent sam-

ples of 100, 101 to 1000, and 1000 patients, respectively.N denotes the number of studies. Data are mean � SE.a Sensitivity analysis, excluding the largest outlier (see text)b Sensitivity analysis, excluding the two largest outliers (see text)

Albumin

Mean serum albumin was significantly higher in HDcant on meta-analysis (Table 2). The estimate of the pre-than in PD in each data set on meta-analysis. Thealbumin difference was 8.1 mg/dL (95% CI, 5.2 to 10.9weighted mean difference was 0.25 g/dL (95% CI, 0.14mg/dL) in the analysis of all twenty-three studies. Sensi-to 0.36 g/dL) in set 1, and 0.28 � 0.07 g/dL (P 0.001)tivity analyses also were performed excluding one orin set 2. Prealbumin and albumin serum levels are knownboth of the most extreme outlying values, derived fromto correlate directly within a group of dialysis patientstwo HD studies that were large and, hence, heavilytreated with a single modality. However, when the albu-weighted, that is, studies 1 (N � 1618 [38]) and 3 (N �min and prealbumin study means were plotted together,1053 [34]) in Figure 2. Eliminating both these studiesan inverse trend was evident between the dialysis modal-yielded a conservative, lower estimate of the mean preal-

bumin difference (6.9 mg/dL; 95% CI, 5.2 to 8.6 mg/dL). ities in each data set (Figs. 3 and 4).

Goldwasser et al: Serum prealbumin in PD and HD 279

DISCUSSION suboptimal (each PD cohort mean satisfied the prealbu-min target but none satisfied the albumin target), and,Meta-analysis of two sets of publications yielded simi-similarly, that PD therapy is both better (higher prealbu-lar estimates of the effect of dialysis modality on serummin) and worse (lower albumin) than HD. On the otherprealbumin level. The mean concentration difference be-hand, if these differences are regarded as intrinsic to PD,tween PD and HD was 5.4 mg/dL in set 1 and 6.9–8.1as we suggest below, then the laboratory reference rangesmg/dL in set 2. The mean effect of dialysis on serumand clinical target values should be set accordingly. Toalbumin concentration was also similar in both studybe consistent in both HD and PD, the respective targetssets: 0.25 to 0.28 g/dL higher in HD.should be set approximately 6 mg/dL higher for prealbu-The meta-analytic models accounted for the variationmin and approximately 0.3 g/dL lower for albumin inin study means caused by sampling error and dialysisPD, with the exact target values varying by laboratory.modality, but it is important to consider how the results

The cause of the depression in serum albumin in PDmight be affected by other factors, such as publicationis presumably the peritoneal loss of albumin (4 to 6 g/day)bias, confounding, and study artifacts [41]. It is unlikely[17, 19, 22, 23, 45], a high proportion of the amount ofthat publication bias affects the present analysis becausealbumin normally synthesized (10 to 15 g/day) [46, 47].the included studies were neither conducted nor pub-Consistent with this supposition, the rate of loss of albu-lished for reasons related to whether dialysis affects pre-min (or total protein) in PD correlates inversely withalbumin concentration. Some of the variation betweenserum albumin [15–18]. However, the cause of the in-studies may be the result of confounding by differencescrease in prealbumin concentration in PD is unknown.in patient characteristics. For example, sex, diabetes, andIn fact, PD might be expected to lower serum prealbuminproteinuria may influence serum prealbumin (abstract;because the rate of peritoneal loss of prealbumin (50 toMaru et al, J Am Soc Nephrol 8:72A, 1997) [1, 42, 43].110 mg/day) [19, 24] represents a sizable fraction of theTo minimize this problem, set 2 included no cohorts thatnormal synthesis rate (500 to 650 mg/day) [47]. There-had been assembled based on any of these characteristicsfore, the increased prealbumin level in PD must be due

(for example, all non-diabetics). Given the specificity ofto some combination of reduced volume of distribution,

immunoassays, it is unlikely that an interferent exists in increased rate of synthesis, and/or decreased rate of cata-the serum having differing activities in PD and HD. bolic clearance in PD compared to HD. Prealbumin andHowever, assay calibration differences will inevitably albumin normally have similar volumes of distributionexist between studies [44]. They result in a spurious in- [47]. Even if there were systematic differences betweencrease in the variation among the study means, and pre- PD and HD in the factors that influence distributionsumably explain the direct correlation observed between volume, that is, capillary permeability and state of hydra-HD means and PD means in set 1. Since the assay meth- tion, they would be unlikely to selectively raise the con-ods for PD and HD were identical within each set 1 centration of prealbumin and not that of albumin be-study, the impact of calibration on the estimate of the cause of the similarity of the two molecules in size andconcentration difference between PD and HD should charge [48]. One plausible, though speculative mecha-be minimized. In set 2, the similarity of the assays used nism for the increase in serum prealbumin in PD is ain the PD and HD studies also should reduce the impact change in the metabolism of prealbumin (increased he-of calibration on the estimate of the difference, unless, patic synthesis or reduced systemic catabolism) due toby chance, a large weight were assigned to an outlying peritoneal loss of a regulatory molecule (such as, albu-value, which in fact occurs in this data set. This problem min), similar to the known alteration in cholesterol me-was addressed in the sensitivity analysis that excluded tabolism. Observations in PD patients that support athe two most extreme such outliers, producing a result causal role for albumin in regulating hepatic metabolism(6.9 mg/dL) close to that of the set 1 analysis. Finally, any of other proteins or lipoproteins include: (a) the perito-inflation of the variance, due to interstudy differences in neal rate of loss of albumin (or total protein) correlateseither patient or assay characteristics, widens the confi- directly with the elevation in serum cholesterol [15, 45];dence intervals of the estimated difference. and (b) intravenous albumin infusion causes the serum

One implication of these concentration differences is levels of fibrinogen, factor 7, and lipoprotein(a) to fallthat the use a single value for either serum prealbumin [18]. Moreover, in nephrotic patients with normal renal(�30 mg/dL) or albumin (�4.0 g/dL) as a clinical target function, it has been analogously found that the serum[12, 20] is confounded by dialysis modality. A single tar- levels of prealbumin and cholesterol are both elevated,get value would not be expected to perform consistently despite overt losses of prealbumin and lipoproteins, andin both PD and HD. For example, using these targets to it has been hypothesized that, like serum cholesterol,classify the prognosis and nutritional state of the cohorts serum prealbumin is elevated as a result of the stimula-contained in sets 1 and 2 combined (Table 3) might lead tion of hepatic synthesis by the large renal loss of albumin

and/or the resultant hypoalbuminemia [43, 48, 49]. Fu-one to conclude that PD therapy is both optimal and

Goldwasser et al: Serum prealbumin in PD and HD280

Table 3. K/DOQI outcome goals for serum levels of prealbuminand albumin applied to mean values from sets 1 and 2

Mean PD HD P valuea

Prealbumin �30 mg/dL 100% (15/15) 75% (15/20)b �0.048Albumin �4 g/dL 0% (0/12) 39% (7/18) �0.016

a One-tailed Fisher’s exact testb For the 18 HD cohorts that had serum albumin data, this rate is 72% (13/18)

targets should be set for PD and for HD, with the exactvalues varying by laboratory.

Fig. 5. Proposed bimodal distribution of prealbumin and albumin se-rum levels in dialysis patients contrasting the direct correlation within- ACKNOWLEDGMENTSgroups (solid lines) with the inverse correlation between-groups (brokenline). This material is based upon work supported by the Department of

Veterans Affairs (VA) and was presented at the World Congress ofNephrology in San Francisco, CA, USA, on October 15, 2001. Theauthors thank Ms. Francine Tidona, Chief Librarian, and Mr. RobertHall, Health Sciences Librarian, at the VA NY Harbor Healthcare

ture studies are needed to directly examine whether al- Center (Brooklyn, NY) for their invaluable assistance with this study.terations in metabolism or in volume of distribution

Reprint requests to Dr. Philip Goldwasser, VA NY Harbor Health-explain the higher serum prealbumin in PD comparedcare Center-Brooklyn, 800 Poly Place (111-F), Brooklyn, New York

to HD. 11209, USA.E-mail: Phillip.Goldwasser@med.va.govThe concentration of prealbumin correlates directly

with that of serum albumin within a group of patients ona single form of dialysis (abstract; Koomen et al, op. cit.) REFERENCES[1–3, 6–8]. This probably reflects the similar effects on 1. Goldwasser P, Michel M-A, Collier JC, et al: Prealbumin and

lipoprotein(a) in hemodialysis: Relationships with patient and vas-prealbumin and albumin of variations in the states ofcular access survival. Am J Kidney Dis 22:215–225, 1993nutrition, inflammation, and hydration. However, if al-

2. Cano N, Fernandez JP, Lacombe P, et al: Statistical selectionbumin and prealbumin serum levels reflect only these of nutritional parameters in hemodialyzed patients. Kidney Int

32(Suppl 22):S178–S180, 1987three factors, the inverse correlation observed between3. Jacob V, Le Carpentier JE, Salzano S, et al: IGF-I, a markerdialysis modalities in Figures 3 and 4 raises a question.

of undernutrition in hemodialysis patients. Am J Clin Nutr 52:39–How can PD patients simultaneously have lower albu- 44, 1990

4. Oksa H, Ahonem K, Pastrnack A, Marnela K-M: Malnutritionmin, implying poorer nutrition, more inflammation, and/orin hemodialysis patients. Scand J Urol Nephrol 25:157–161, 1991more overhydration than in HD patients, and have

5. Aparicio M, Cano N, Chauveau P, et al: Nutritional status ofhigher prealbumin, implying the opposite? This apparent haemodialysis patients: A French national cooperative study.

Nephrol Dial Transplant 14:1679–1686, 1999paradox can be resolved by recognizing two additional6. Duggan A, Huffman FG: Validation of serum transthyretinPD-specific factors that influence the serum levels of

(prealbumin) as a nutritional parameter in hemodialysis patients.prealbumin and albumin: the direct effect of PD losses, J Renal Nutr 8:142–150, 1998

7. Neyra NR, Hakim RM, Shyr Y, Ikizler TA: Serum transferrinand a secondary alteration in prealbumin metabolismand serum prealbumin are early predictors of serum albumin in(for example, the hypothesized stimulation of the liver). chronic hemodialysis patients. J Renal Nutr 10:184–190, 2000

This hypothesis could explain the “paradoxical” inverse 8. Visser R, Dekker FW, Boeschoten EW, et al: Reliability of the7-point subjective global assessment scale in assessing nutritionalcorrelation between groups as well as the known directstatus of dialysis patients. Adv Perit Dial 15:222–225, 1999correlation within a modality, illustrated in Figure 5. 9. Holland DC, Meers C, Lawlor ME, Lam M: Serial prealbumin

Peritoneal albumin losses and increased hepatic prealbu- levels as predictors of outcomes in a retrospective cohort of perito-neal and hemodialysis patients. J Renal Nutr 11:129–138, 2001min synthesis shift the within-group albumin vs. prealbu-

10. Burdick CO: Prealbumin and prediction of survival in dialysismin regression line downward and rightward in PD, cre- patients. (letter) Am J Kidney Dis 31:195, 1998ating the inverse correlation between groups. Of note, 11. Herzig KA, Purdie DM, Chang W, et al: Is C-reactive protein a

useful predictor of outcome in peritoneal dialysis patients? J Amthe between-group and within-group correlations of se-Soc Nephrol 12:814–821, 2001rum albumin with serum cholesterol in dialysis patients 12. K/DOQI clinical practice guidelines for nutrition in chronic renal

follow the same pattern shown in Figure 5 [13]. failure. Am J Kidney Dis 35(Suppl 2):S20–S22, 200013. Avram MM, Goldwasser P, Burrell DE, et al: The uremic dyslip-We conclude that serum prealbumin level is approxi-

idemia: A cross-sectional and longitudinal study. Am J Kidney Dismately 6 mg/dL higher in PD than in HD, perhaps due 20:324–335, 1992to a stimulatory effect of peritoneal albumin loss on 14. Rodrıguez-Carmona A, Fontan MP, Cordido F, et al: Hyperlepti-

nemia is not correlated with markers of protein malnutrition inhepatic prealbumin synthesis, while serum albumin ischronic renal failure: A cross-sectional study in predialysis, perito-

approximately 0.3 g/dL lower in PD. Different prealbu- neal dialysis, and hemodialysis patients. Nephron 86:274–280, 200015. Broyer M, Niaudet P, Champion G, et al: Nutritional and meta-min and albumin serum reference ranges and clinical

Goldwasser et al: Serum prealbumin in PD and HD 281

bolic studies in children on continuous ambulatory peritoneal dial- 32. Young GA, Swanepoel CR, Croft MR, et al: Anthropometryand plasma valine, amino acids, and proteins in the nutritionalysis. Kidney Int 24(Suppl 15):S106–S110, 1983

16. Kagan A, Bar-Khayim Y, Schafer Z, Fainaru M: Kinetics assessment of hemodialysis patients. Kidney Int 21:492–499, 198233. Ikizler TA, Wingard RL, Harvell J, et al: Association of morbid-of peritoneal protein loss during CAPD: I. Different characteris-

tic for low and high molecular weight proteins. Kidney Int 37:971– ity with markers of nutrition and inflammation in chronic hemodial-ysis patient: A prospective study. Kidney Int 55:1945–1951, 1999979, 1990

17. Kagan A, Bar Khayim Y: Role of peritoneal loss of albumin in 34. Owen WF, Lowrie EG: C-reactive protein as an outcome predictorfor maintenance hemodialysis patients. Kidney Int 54:627–636, 1998the hypoalbuminemia of continuous ambulatory peritoneal dialysis

patients: Relationship to peritoneal transport of solutes. Nephron 35. Chazot C, Laurent G, Charra B, et al: Malnutrition in long-termhaemodialysis survivors. Nephrol Dial Transplant 16:61–69, 200171:314–320, 1995

18. Kim SB, Yang WS, Sang KL, et al: Effect of increasing serum 36. Docci D, Bilancioni R, Pistocchi E, et al: Evolution of serumprealbumin following hemodialysis: Effect of different dialysisalbumin on haemostatic factors synthesized in the liver in CAPD

patients. Nephrol Dial Transpl 13:2053–2058, 1998 membranes. Nephron 62:145–149, 199237. Fleishmann E, Teal N, Dudley J, et al: Influence of excess weight19. Young GA, Brownjohn AM, Parsons FM: Protein losses in pa-

tients receiving continuous ambulatory peritoneal dialysis. Neph- on mortality and hospital stay in 134 hemodialysis patients. KidneyInt 55:1560–1567, 1999ron 45:196–201, 1987

20. ESRD clinical performance measures project. Am J Kidney Dis 38. Chertow GM, Ackert K, Lew NL, et al: Prealbumin is as importantas albumin in the nutritional assessment of hemodialysis patients.37(Suppl 3):S20–S45, 2001

21. Avram MM, Goldwasser P, Erroa M, Fein PA: Predictors of Kidney Int 58:2512–2517, 200039. Kauffman P, Smolle KH, Horina JH, et al: Impact of long-termsurvival in continuous ambulatory peritoneal dialysis patients. Am

J Kidney Dis 23:91–98, 1994 hemodialysis on nutritional status in patients with end-stage renalfailure. Clin Invest 72:754–761, 199422. Kagan A, Bar-Khayim Y, Schafer Z, Fainaru M: Kinetics of

peritoneal protein loss during CAPD: II. Lipoprotein leakage and 40. Petitti DR: Meta-Analysis, Decision Analysis, and Cost-Effective-ness: Methods for Quantitative Synthesis in Medicine (2 ed). Newits impact on plasma lipid levels. Kidney Int 37:980–990, 1990

23. Dulaney JT, Hatch FE: Peritoneal dialysis and loss of proteins: York, Oxford University Press, 2000, pp 119–123.41. Hunter JE, Schmidt FL: Methods of Meta-Analysis: CorrectingA review. Kidney Int 26:253–262, 1984

24. Mydlık M, Derziova K, Valek A, et al: Vitamins and continuous for Error and Bias in Research Findings. Newbury, Sage Publica-tions, 1990ambulatory peritoneal dialysis. Int Urol Nephrol 17:281–286, 1985

25. Miller DG, Levine S, Bistrian B, D’Elia JA: Diagnosis of protein 42. Ingenbleek Y, Young V: Transthyretin (prealbumin) in health anddisease: Nutritional implications. Ann Rev Nutr 14:495–533, 1994calorie malnutrition in diabetic patients on hemodialysis and peri-

toneal dialysis. Nephron 33:127–132, 1983 43. Mydlık M, Dershiova K, Bratova M, Havris S: Serum vitaminA, retinyl esters, and vitamin E in nephrotic syndrome. Int Urol26. Harty JC, Boulton H, Venning MC, Gokal R: Is peritoneal

permeability an adverse risk factor for malnutrition in CAPD pa- Neph 23:399–405, 199144. Goldsmith BM, Munson S: Rate nephelometry and radial immu-tients? Miner Electrolyte Metab 22:97–101, 1996

27. Ikizler TA, Wingard RL, Breyer JA, et al: Short-term effects of nodiffusion compared for measuring serum prealbumin. Clin Chem33:161–163, 1987recombinant human growth hormone in CAPD patients. Kidney

Int 46:1178–1183, 1994 45. Heimburger O, Stenvinkel P, Berglund L, et al: Increased plasmalipoprotein(a) in continuous ambulatory peritoneal dialysis is re-28. Palop L, Martinez JA: Cross-sectional assessment of nutritional

and immune status in renal patients undergoing continuing ambula- lated to peritoneal transport of proteins and glucose. Nephron 72:135–144, 1996tory peritoneal dialysis. Am J Clin Nutr 66:498S–503S, 1997

29. Jacob V, Marchant PR, Wild G, et al: Nutritional profile of 46. Rothschild MA, Oratz M, Schreiber SS: Albumin synthesis (Sec-ond of two parts). N Engl J Med 286:816–821, 1972continuous ambulatory peritoneal dialysis patients. Nephron 71:

16–22, 1995 47. Gerschengorn MC, Glinoer D, Robbins J: Transport and metab-olism of thyroid proteins, in The Thyroid Gland, edited by De30. Bonnefont-Rousselot D, Jaudon MC, Issad B, et al: Antioxi-

dant status of elderly chronic renal patients treated by continuous Visscher M, New York, Raven Press, 1980, pp 81–12348. Beetham R, Dawnay A, Ghany C, et al: A radioimmunoassay forambulatory peritoneal dialysis. Nephrol Dial Transplant 12:1399–

1405, 1997 urinary prealbumin. Ann Clin Biochem 30:377–382, 199349. Tormey WP, O’Brien PA: Clinical associations of an increased31. Kang D-H, Lee R, Lee H-Y, et al: Metabolic acidosis and com-

posite nutritional index in CAPD patients. Clin Nephrol 53:124– transthyretin band in routine serum and urine electrophoresis. AnnClin Biochem 30:550–554, 1993131, 2000