Tumor Necrosis Factor-Mediated Hypoalbuminemia inRabbits1 |

BERNHARD HENNIG,*f RON HONCHELf SIMEON E. GOLDBLOMf

AND CRAIG J. McCLAIWf

"Department of nutrition and Food Science, tGraduate Center for Toxicology and ¿Department of

Medicine, university of Kentucky and Lexington Veterans Administration Medical Center, Lexington, KY40536

ABSTRACT The serum albumin concentration is usedclinically as an indicator of nutritional status and as a prognostic indicator. Critically ill patients, who display manyaspects of the acute phase response, frequently have lowserum albumin levels upon hospitalizaron. Cytokines, suchas tumor necrosis factor (TNF), mediate many aspects ofthe acute phase response. One purpose of this study wasto determine if TNF administration to healthy well-nourished rabbits could produce hypoalbuminemia. After intravenous administration of saline or TNF, the TNF-treatedrabbits experienced significant hypoalbuminemia which wasmost prominent at 24 h and was partially corrected by 48h. A second purpose was to evaluate the effects of TNFtreatment on transendothelial movement of albumin usingan in vitro porcine pulmonary artery—endothelial cell system. Exposure to TNF for 24 h resulted in a dose dependent increase in transendothelial passage of albumin.These data suggest that the mechanisms of hypoalbuminemia frequently observed in critically ill patients can beexplained in part by cytokine (TNF)-induced endothelial cellinjury, which results in enhanced endothelial permeabilityto albumin. The hypoalbuminemia observed in many critically ill patients thus may be unrelated to nutritional status,but rather may be related to the patient's underlying diseasestate. J. liutr. 119: 1586-1590, 1988.

INDEXING KEY WORDS:

•albumin •tumor necrosis factor

In hospitalized patients, the serum albumin concentration frequently is used as an indicator of proteinenergy malnutrition, as an indicator of the adequacy ofnutritional support and as a predictor of morbidity andmortality (1-10). The utility of the serum albumin valueas a prognostic indicator may be unrelated to the actualnutritional status of the patient, but instead may bemore closely correlated with the severity of the injury(11-14). Indeed, patients with severe trauma, such asthermal injury or severe head injury, frequently have

hypoalbuminemia at the time of hospitalization andcontinue to have low serum albumin levels in spite ofaggressive nutritional support throughout the hospitalstay (15, 16). Thus, whenever there is evidence of theacute phase response, a fall in plasma albumin may notbe diagnostic of inadequate nutritional support (17).

One possible explanation for the initial hypoalbuminemia observed in septic or trauma patients is thatthere may be an increased leakage of albumin from theintravascular to the extravascular space (12-14). Cytokines, such as interleukin-1 (IL-1) and tumor necrosisfactor (TNF), mediate many aspects of the acute phaseresponse observed in septic or trauma patients (18-23).The purposes of this study were to evaluate whetheror not administration of TNF would cause hypoalbuminemia in healthy well-nourished animals, and to determine if TNF-mediated hypoalbuminemia could bedue to increased leakage of albumin across the vascularendothelium. The second hypothesis was tested bymeasuring the effect of TNF on transendothelial movement of albumin in an in vitro endothelial cell system.

MATERIALS AND METHODS

Rabbit studies. New Zealand White rabbits (ClercoResearch Farms, Cincinnati, OH) having an averageweight of 3.4 kg (range: 2.6-4.2 kg) were housed individually in an environmentally controlled room (temperature: 24 ±1°C)with a 12-h light cycle beginningat 0600. Rabbits were allowed food (Ralston-Purina

'Supported in part by the Veterans Administration, National In

stitutes of Health Grant Nos. NS22712 and HL34423, a grant fromthe American Heart Association, Kentucky Affiliate and the Kentucky Agricultural Experiment Station.

2To whom correspondence and reprint requests should be addressed, at Department of Medicine, Division of Digestive Diseasesand Nutrition, University of Kentucky, Lexington, KY 40536-0084.

0022-3166/88 $3.00 ©1988 American Institute of Nutrition. Received 5 April 1988. Accepted 30 August 1988.

1586

by guest on June 14, 2011jn.nutrition.org

Dow

nloaded from

CYTOKINES AND HYPOALBUMINEMIA 1587

Rabbit Chow, Ralston Purina Co., St. Louis, MO) andtap water ad libitum. At the start of the study, bloodsamples were collected by inserting a 21-gauge butterfly needle into the central artery of the ear. Using a 25-gauge needle, rabbits were then given a bolus treatmentinjection into the marginal vein of the ear. Treatmentswere as follows: controls (n = 4) were injected with 1ml of sterile, nonpyrogenic normal saline (TravenolLaboratories, Deerfield, IL); the low dose group (n = 4)was injected with 250,000 units TNF (donated byGenentech, Inc., South San Francisco, CA) in 1 ml ofsaline; and the high dose group (n = 4) was injectedwith 1,000,000 units TNF in 1 ml of saline. The rTNFapreparation had 0.50 mg TNF/ml at 5.02 x IO7 u/mg

and 200 pg/ml by Limulus Amoebocyte Lysate Assay(Whittaker, M. A., Bioproducts, Walkersville, MD). Foodintake was measured daily throughout the study. At 8and 24 h after injection, blood samples were collectedfrom the central artery of the ear. At 48 h rabbits wereinjected intravenously with 3 ml of Nembutal (AbbottPharmaceuticals, Inc., North Chicago, IL) and bloodsamples were collected via cardiac puncture. For eachserum specimen, albumin, total protein and liver enzymes [serum glutamate pyruvate transaminase (SGPT);serum glutamate oxaloacetate transaminase (SGOT);alkaline phosphatase (AP) were measured, using a Beck-man Astra-8 photometric assay (Beckman Inc., Fuller-

ton, CA). Liver enzymes were unaffected by TNFadministration.

Endothelial cell culture and materials. Endothelialcells were isolated from porcine pulmonary arteries andcultured as previously described (24). The basic culturemedium consisted of M 199 (GffiCO Laboratories, GrantIsland, NY) containing 10% fetal bovine serum (Hy-

Clone Laboratories, Inc., Logan, UT). The experimentalculture medium consisted of M 199 enriched with vitamins, amino acids (GIBCO Laboratories), 5% fetalbovine serum and varying concentrations of TNF (0 to1000 units TNF/ml culture medium). Cell cultures wereverified as endothelial cells by uniform morphology andby quantitative determination of angiotensin-convert-

ing enzyme activity. Cell viability was determined bylactate dehydrogenase (LDH) release, trypan blue exclusion, and morphological assessment using phase-

contrast microscopy. LDH activity was assayed in theculture medium and the cell lysates using the methodof Bergmeyer and Garvehn (25).

For the in vitro albumin transfer experiments, cellsfrom passages 5 through 12 were plated at a density of3 x 10s cells/filter on 13-mm-diameter gelatin-im

pregnated polycarbonate filters (Nucleopore Corporation, Pleasanton, CA) having 0.8 n-m pores and whichwere glued to polystyrene chemotactic chambers(ADAPS Inc., Dedham, MA). After confluent endothelial monolayers were exposed for 24 h to the TNF-enriched media, albumin transfer across the endothe-lium was measured according to a previously publishedprocedure (24). Briefly, chemotactic chambers with at

tached monolayers were placed into 24-well cultureplates thus creating two separate compartments. Eachlower compartment contained 1.5 ml of serum-free M

199, and 0.5 ml of M 199 with 200 (xm of albumin wasadded to each upper compartment. After a 1-h incubation period, media within each compartment weresampled. The respective albumin concentrations weredetermined by measuring the change in absorbance at630 nm after addition of a bromcresol green reagentsolution (Sigma Chemical Company, St. Louis, MO) toan aliquot of the sample.

Statistica! analysis. An analysis of variance table fora repeated measures procedure was conducted for Figures 1 and 2. In the presence of a significant treatmentby time interaction in Figure 1, the mean responsesbetween hours for each treatment were compared usingDuncan's multiple-range test procedure. Statistical sig

nificance was determined at the 0.05 level. For Figure3, a one-way analysis of variance was conducted tocompare the mean response between doses, and then acontrast to detect a linear trend in dose response wasconstructed. Statistical significance was determined atthe 0.05 level.

RESULTS

Animal studies. Figure 1 illustrates the effect of lengthof exposure to saline or to TNF on serum albumin concentration as compared to controls. Hypoalbuminemia

O)4-

3-

15 25 35 45

Hours after InjectionFIGURE 1 Effect of length of exposure to TNF on serum

albumin concentration. Rabbits were administered intravenously either saline or one of 2 doses of TNF (low dose 250,000or high dose 1,000,000 units). Blood was collected at 0, 8, 24and 48 h after injection. Values are mean ±SE,n = 4. -13-, Control; -*-, low dose; -•-, high dose. Injection of thevehicle (control) caused no significant changes in serum albuminconcentration. Injection of low-dose TNF caused a significantreduction in albumin at 24 h compared to all other time periods(P < 0.05). Injection of high-dose TNF caused significant(P < 0.05) hypoalbuminemia at 8, 24 and 48 h compared tobaseline. Furthermore, at 24 h, high dose TNF administrationcaused a significant reduction in albumin compared to low doseTNF administration.

by guest on June 14, 2011jn.nutrition.org

Dow

nloaded from

1588 HENNIG ET AL.

O 24 48

Hours after Injection

FIGURE 2 Effect of exposure to TNF on food intake. Foodintake was measured daily after injection of either saline or oneof 2 doses of TNF. Values are mean ±SE, «= 4. Food intakedid not vary significantly from control values in either treatmentgroup.

in the treatment groups was most prominent at 24 hafter injection,and the albumin concentration approached that of the control group at 48 h. In additionto the time-dependent effect of TNF on serum albuminlevels, the hypoalbuminemia also was dependent onthe TNF dose to which the rabbits were exposed. Amaximum reduction in serum albumin concentrationwas observed in the high dose group 24 h after theintravenqus injection of 1,000,000 units of TNF.

Food intake was measured at all treatment times. As

O

C,

S(OCCO

C

i3

12

10

8

6

4-

2-

250 500 1000

TNF (units/ml)

FIGURE3 Effect of various doses of TNF on subsequent transferof albumin across cultured endothelial cell monolayers. Cellswere washed with M 199 and then incubated for 24 hours in M199, enriched with 5% fetal bovine serum and various doses ofTNF. After washing cells again in M 199, transendothelial movement of albumin was measured over a 1-h period. The concentration of albumin added to the upper chamber during the subsequent 1-h incubation was 200 JIM. Values are mean ±SE,n = 6. Each dose of TNF was significantly different from eachother (P < 0.05) and there was a highly significant linear trendin the dose response (P < 0.0001).

shown in Figure 2, intravenous administration of TNFat 2 doses of 250,000 or 1,000,000 units/ml had nosignificant effect on food intake.

Cell studies. The effect of a 24-h incubation in mediaenriched with various concentrations of TNF on albumin transfer across cultured endothelial cell monolayers is shown in Figure 3. The reduced ability of theendothelium to act as a selectively permeable barrierto macromolecules, as measured by the increased transfer of albumin across cultured endothelial monolayers,was dependent on the TNF concentrations to whichthe cultures were exposed. For example, exposure ofendothelial cells to 1000 units of TNF resulted in afourfold increase of albumin transfer as compared withcontrol cultures.

DISCUSSION

The serum albumin concentration frequently is usedas a biochemical marker of nutritional status for hospitalized patients (1-10). Many hospitalized patientswho are receiving specialized nutritional support arebeing cared for in intensive care units and are amongthe most critically ill. If serum albumin values are tobe used as indicators of a need for specialized nutritional support, then it is important to understand therelative effects that nutritional status and the diseasestate have on serum albumin levels. Several investigators have suggested that the initial hypoalbuminemiaobserved in critically ill patients may not be due tomalnutrition but instead is related to fluid shifts andincreases in vascular permeability (14, 26, 27). Recentstudies have shown that minor elective surgery in well-nourished patients caused decreased serum albuminlevels even though adequate oral nutritional intake wasmaintained (28). Fleck and co-workers demonstrated a

transcapillary leak of radioiodinated albumin in patients undergoing cardiac surgery and patients in septicshock (27). Not only can there be an initial decrease inthe serum albumin level, but also certain other proteinssuch as prealbumin and transferrin may decrease aftertrauma or sepsis (14). Fleck (17) and Fleck, Colley andMyers (14) suggested that stimulation of the synthesisof acute phase reactants such as C-reactive protein (CRP)and the decrease in other proteins such as albumin,may be initiated through a common mediator such asthe cytokine interleukin-1 (IL-1). Critically ill patients,

for example those with thermal injury, head injury,sepsis, multiple trauma etc., demonstrate many aspectsof the acute phase response which is thought to bemediated largely by the cytokines IL-1 and TNF (16,18-22, 29). It is well documented that both IL-1 and

TNF stimulate synthesis of acute phase reactants suchas CRP, and cause a decrease in the mRNA for albumin(30-33). However, the initial rapid decrease in serumalbumin observed in severely ill patients cannot be explained by decreased albumin production alone, be-

by guest on June 14, 2011jn.nutrition.org

Dow

nloaded from

CYTOKINES AND HYPOALBUMINEMIA 1589

cause the half-life of albumin is 18 d. Data from thepresent study support the hypothesis that the cytokineTNF, when given intravenously to otherwise healthyrabbits, produced transient hypoalburninernia. The TNFadministration had no effect on food intake and did notcause metabolic abnormalities such as changes in liverenzyme activities. This is because the high dose of TNFwas quite modest and considerably less than that required to produce hypotension or severe metabolic derangements (23, 34). Because the change in albuminlevels possibly was related to alterations in vascularpermeability, the effects of TNF on transendothelialpassage of albumin in an in vitro cell system were evaluated next. The design of this in vitro system is so welldefined that the effects of individual substances on en-dothelial cell metabolism can be measured with a highdegree of accuracy. It was demonstrated that in thissystem, TNF caused a dose-dependent increase in trans

endothelial movement of albumin.We conclude that the cytokine TNF can cause tran

sient hypoalbuminemia in well-nourished animals, andwe provide in vitro data to support the hypothesis thatthis hypoalbuminemia is caused by a leakage of albumin from the intravascular to the extravascular space.Thus, it is important that clinicians exercise cautionwhen using the serum albumin values as an indicatorof inadequacy of nutritional support in critically ill patients. Clearly, in many instances the serum albuminlevel of critically ill patients will reflect the underlyingdisease state but will not relate necessarily to the patients' nutritional state.

ACKNOWLEDGMENT

We thank Mary A. Stuart for reviewing the manuscript.

LITERATURE CITED

1. HAY, R. W., WHITEHEAD,R. G. & SPICER,C. C. (1975] Serumalbumin as a prognostic indicator in edematous malnutrition.Lancet 2: 427-429.

2. MULLEN,J. L., GERTNER,M. H., BUZBY,G. P., GOODHART,G. L.& RASATO,E. F. (1979| Implications of malnutrition in thesurgical patient. Arch. Surg. 114: 121-125.

3. BLACKBURN,G. L. & HARVEY,K. B. (1982] Nutritional assessment as a routine in clinical medicine. Nutr. Assess. 71: 46-63.

4. BUZBY,G. P., MULLEN,J. L., MATTHEWS,D. C., HOBBS,C. L. &.ROSATO,E. F. (1980] Prognostic nutritional index in gastrointestinal surgery. Am. ¡.Surg. 139: 160-167.

5. BRADLEY,J. A., CUNNINGHAM,K. J., JACKSON,V. }., HAMILTON,D. N. H. & LEDINGHAM,I. M. (1981) Serum protein levels incritically ill surgical patients. Intens. Care Med. 7: 291-295.

6. REINHARDT,G. F., MYSCOFSKI,J. W., WILKENS,D. B., DORBIN,P. B.,MANGAN,J. E. &. STANNARD,R. T. (1980] Incidence and mortality of hypoalbuminemic patients in hospitalized veterans. /.Parenteral &>Enterai Nutr. 4: 357-359.

7. DOMINQUEZDE VlLLOTA, E., MOSQUERA, f. M., RUBIO, J. f., GALDOS,

P., DIEZ BALDA,V., DEAL SERNA,I. L. & TOMAS,M. I. (1980)Association of a low serum albumin with infection and increasedmortality in critically ill patients. Intens. Care Med. 7: 19-22.

8. CHING, N., GROSSI,C. E., ANGERS,J., ZURAWINSKY,H. S., JHAM,G., MILLS,C. B. & NEALON,T. F. (1980) The outcome of surgical treatment as related to the response of the serum albuminlevel to nutritional support. Surg. Gyn. and Obstet. 151: 199-

202.9. APELGREN,K. N., ROMBEAU,I. L., TWOMEY,P. L. & MILLER,R. A.

(1982) Comparison of nutritional indices and outcome in critically ill patients. Grit. Care Med. 10: 305-307.

10. PRABHAKARAN,V. M., PUIARA,S., MILLS,A. J. & WHALEN,V. W.(1986) Can nutritional criteria help predict outcome in hospitalized patients? Clin. Chem. 32: 2077-2079.

11. STARKER,P. M., GUMP,F. E., ASKANAZI,]., ELWYN,D. H. & KINNEY,J. M. (1982] Serum albumin levels as an index of nutritionalsupport. Surgery 91: 194-199.

12. SlLBERMAN, H., ElSENBERG, D., SHOFLER, R., RYAN, J. & WEISS, J.(1982] Nutrition-related factors in acutely injured patients. /.Trauma 22: 907-909.

13. GOLDEN,M. H. N. (1982) Transport proteins as indices of protein status. Am. ¡.Clin. Nutr. 35: 1159-1165.

14. FLECK,A., COLLEY,C. M. & MYERS,M. A. (1985) Liver exportproteins and trauma. Br. Med. Bull. 41: 265-273.

15. YOUNG,B., OTT, L., NORTON,}., TIBBS,P., RAPP,R., MCCLAIN,C.& DEMPSEY,R. (1985] Metabolic and nutritional sequelae inthe non-steroid treated head injury patient. Neurosurgery 17:784-791.

16. BOOSALIS,M. G., SOLEM,L. D., McCALL, J. T., AHRENHOLZ,D. H.& McCLAlN, C. J. (1988] Serum zinc response in thermal injury. /. Am. Coll. Nutr. 7: 69-76.

17. FLECK,A. (1988) Acute phase response: Implications for nutrition and recovery. Nutrition 4: 109-117.

18. DINARELLO,C. A. (1984) Interleukin-1 and the pathogenesis ofthe acute-phase response. N. Engl. ¡.Med. 311: 1413-1418.

19. DINARELLO,C. A. (1986] Interleukin-1 : Amino acid sequences,multiple biological activities and comparison with tumor necrosis factor (cachectin]. In: Year in Immunology, pp. 1-22, Karger, Basel.

20. DINARELLO,C. A. (1988] Biology of interleukin 1. FASEB J. 2:108-115.

21. BEUTLER,B. & CERAMI,A. (1987) Cachectin: More than a tumor necrosis factor. N. Engl. ¡.Med. 316: 379-385.

22. LE, f. & ViLCEK,J. (1987] Tumor necrosis factor and interleukin 1: Cytokines with multiple overlapping biological activities.Lab. Invest. 56: 234-248.

23. REMICK,D. G., KUNKEL,R. G., LARRICK,J. W. & KUNKEL,S. L.(1987) Acute in vivo effects of human recombinant tumor necrosis factor. Lab. Invest. 56: 583-590.

24. HENNIG,B., SHASBY,D. M., FULTON,A. B. & SPECTOR,A. A.(1984) Exposure to free fatty acid increases the transfer of albumin across cultured endothelial monolayers. Arteriosclerosis4: 489-497.

25. BERGMEYER,H. U. & GARVEHN,K. (1983) Láclate dehydroge-nase. In: Methods of Enzymatic Analysis (Donald, W. M., ed.],»pp. 118-126, Academic Press, New York.

26. MYERS,M. A., FLECK,A., SAMPSON,B., COLLEY,C. M., BENT,J. &HALL,G. 11984] Early plasma protein and mineral changes aftersurgery: a two-stage process. /. Clin. Pathol. 37: 862-866.

27. FLECK,A., HAWKER,F., WALLACE,P. L, RAINES,G., TROTTER,J.,LEDINGHAM,I. McA. & CALMAN,K. C. (1985) Increased vascular permeability: A major cause of hypoalbuminemia in disease and injury. Lancet 1: 781-783.

28. BLEIBERG-DANIEL,F., WADE,S., LABARRE,C., BALAGNY,D.,Fl-CHELLE, A., BORY, J. M., DESMONTS, J. M. & LEMONNIER, D.

(1985) Variations in plasma thyroxine-binding prealbumin(TBPA) in relation to other circulating proteins in postoperativepatients during rapid oral refeeding. Clin. Nutr. 39C: 55-62.

29. MCCLAIN,C. J., COHEN,D., OTT,L., DINARELLO,C. A. & YOUNG,

by guest on June 14, 2011jn.nutrition.org

Dow

nloaded from

1590 HENNIG ET AL.

B. (1987) Ventricular fluid interleukin-1 activity in patientswith head injury. /. Lab. Clin. Med. 110: 48-54.

30. GAULDIE,J., SAUDER,D. N., McAoAM, P. W. ]. & DINARELLO,C. A. (1987) Purified interleukin-1 (IL-1) from human mon-ocytes stimulates acute-phase protein synthesis by rodent he-patocytes in vitro. Immunology 60: 203-207.

31. BAUER,f., WEBER,W., TRAN-THI,T.-Õ.,NORTHOFF,G.-H., DECKER,K., GEROK,W. & HEINRICH,P. C. (1985) Murine interleukin 1stimulates a2-macroglobulin synthesis in rat hepatocyte primarycultures. FEES Lett. 190: 271-274.

32. GLIBETIC,M. D. &. BAUMANN,H. (1986) Influence of chronic

inflammation on the level of mRNA for acute-phase reactantsin the mouse liver. /. Immunol. 137: 1616-1622.

33. DARLINGTON,G. J., WILSON,D. R. & LACHMAN,L. B. (1986)Monocyte-conditioned medium, interleukin-1, and tumor necrosis factor stimulate the acute phase response in human hep-atoma cells in vitro. /. Cell. Biol. 103: 787-793.

34. HESSE,D. G., TRACEY,K. ]., FONG,Y., MANOGUE,K. R., PALLADINO,M. A., CERAMI,A., SHIRES,T. & LOWRY,S. F. (1988) Cytokineappearance in human endotoxemia and primate bacteremia. Surg.,Gyn. et) Obstet. 166: 147-153.

by guest on June 14, 2011jn.nutrition.org

Dow

nloaded from