Tumor Necrosis Factor Soluble Receptors in Patients With Various Degrees of Congestive Heart Failure
Background Tumor necrosis factor alpha (TNF-α) increases in patients with severe congestive heart failure (CHF) and cachexia. Two naturally occurring modulators of TNF-α activity have been identified in human serum. These two soluble proteins are the extracellular domains of the TNF receptors (sTNF-RI and sTNF-RII, respectively). The determination of circulating sTNF-Rs could provide us with some additional information about the activation of this cytokine in CHF.
Methods and Results This study was undertaken to examine the concentration of sTNF-Rs and of bioactive and antigenic TNF-α in 37 consecutive patients with various degrees of CHF compared with that of 26 age-matched healthy subjects. Antigenic TNF-α increased (from 14.3±7.08 to 33.5±13.1 pg/mL, P<.001) in preterminal patients with severe CHF (New York Heart Association [NYHA] class IV). In these patients, sTNF-Rs were also increased (sTNF-RI from 1.17±0.43 to 4.43±2.14 ng/mL and sTNF-RII from 2.2±0.44 to 7.55±2.28 ng/mL, P<.001). When measured by cytolytic bioassay, TNF-α was undetectable (<100 pg/mL). Addition of 625 pg/mL recombinant human TNF-α (rhTNF-α), corresponding in the bioassay to 60% of the lethal dose, to the serum of healthy subjects resulted in a significant increase of the expected cytotoxicity (from 625 to 1290±411 pg/mL, P<.001). Addition of the same dose of rhTNF-α to the serum of patients with mild to moderate CHF (NYHA classes II and III) increased the cytotoxicity from 625 to 877±132 pg/mL, P<.001. In 4 patients with severe CHF (class IV), the expected cytotoxicity was completely inhibited, whereas it was reduced from 625 to 263±198 pg/mL, P<.001, in the remaining 8 patients. Ten patients died within 1 month of entry into the study. They had the highest level of sTNF-RII (8.18±1.92 ng/mL). sTNF-RII was a more powerful independent indicator of mortality than TNF-α, sTNF-RI, NYHA class, norepinephrine, and atrial natriuretic peptide.
Conclusions Measurement of sTNF-Rs, in addition to antigenic and bioactive TNF-α, is essential for evaluation of the activation of this cytokine in CHF. Both sTNF-Rs increase in preterminal patients with severe CHF and might inhibit the in vitro cytotoxicity of TNF-α. Antigenic TNF-α also increases in severe CHF. The increased levels of sTNF-RII independently correlate with poor short-term prognosis.
Recent studies report that tumor necrosis factor alpha (TNF-α) is increased in patients with severe congestive heart failure (CHF).1 2 3 4 5 Earlier studies showed that TNF-α elevation was associated with cachexia1 2 and with a marked activation of the renin-angiotensin system.1 However, a prospective study undertaken to assess serial changes in TNF-α during a period of 1 year showed no correlation between TNF-α, plasma norepinephrine, renin activity, and patient weight.3 In these studies, measurements of TNF-α were carried out either by ELISA or radioimmunoassay methods2 3 to detect antigenic TNF-α or by a cytotoxicity bioassay that detects only biologically active (cytolytic) TNF-α.1 6 7 Irrespective of the methodology used, there was a wide variation in plasma values of TNF-α between and within patients. In several patients, circulating TNF-α was not detected despite the presence of severe CHF.
Two TNF receptors of 55 and 75 kD, TNF-RI and TNF-RII, have been identified on the surface of several cell lines and are thought to mediate and regulate most of the effects of TNF.8 9 The extracellular domain fragments of both receptors shed from cell surfaces can be detected as soluble forms (sTNF-RI and sTNF-RII) in the urine and blood. These soluble proteins are supposed to regulate the TNF-α bioactivity either by inhibiting the binding of TNF trimers to the membrane receptors10 or by preventing TNF-α trimers from dissociation to inactive monomers.11 12 Therefore, measurements of circulating levels of sTNF-Rs provide more complete information on TNF-α activation in CHF.
The objective of this study was to further investigate the activation of this cytokine in CHF. To this end, we determined bioactive TNF-α, antigenic TNF-α, and both the sTNF-Rs in 37 consecutive patients with various degrees of heart failure and then compared these data with those of 26 age-matched healthy subjects.
This study was carried out at the Cardiology Department of the University of Brescia and at the Heart Failure Unit of the Fondazione Clinica del Lavoro, Montescano, Pavia, Italy.
Thirty-seven consecutive patients admitted for investigation or treatment of CHF gave informed consent and were studied in compliance with ethical approval. Table 1⇓ shows the characteristics of these patients. The average age was 50±10 years. Twenty-six were men and 11, women. The cause of heart failure was coronary artery disease in 19 patients, idiopathic dilated cardiomyopathy in 14, and valvular diseases in 4. Fourteen were in New York Heart Association (NYHA) clinical class II; their mean left ventricular ejection fraction, calculated from the two-dimensional echocardiogram, was 26±6%, and their peak oxygen consumption was 16±5 mL · kg−1 · min−1. Eleven patients were in NYHA class III, with an ejection fraction of 22±8% and peak oxygen consumption of 10±4 mL · kg−1 · min−1. Twelve patients were in NYHA class IV. Their mean ejection fraction was 18±5%.
Hemodynamics were measured in the postabsorptive state with a Swan-Ganz catheter. Cardiac output was determined by thermodilution with a Gould model SP 1445 cardiac output computer.
All patients were treated with angiotensin-converting enzyme inhibitors and, when necessary, with digoxin, diuretics, and positive inotropic agents.
No patient had received anti-inflammatory drugs within the preceding 2 weeks. At the beginning of the study, treatment was optimized and kept constant when possible.
Patients with significant concomitant disease such as infection, renal failure, pulmonary disease, thyroid disease, malignancy, or collagen vascular disease were not studied.
The control subjects consisted of 26 healthy volunteers, 18 men and 8 women, 30 to 62 years of age (mean, 48±8 years). None of them were admitted to the hospital, had acute or chronic illness, or reported any symptoms related to the cardiovascular system.
After 30 minutes of supine rest, venous blood was taken and centrifuged within 1 hour for measurement of serum electrolytes, norepinephrine, epinephrine, aldosterone, plasma renin activity, atrial natriuretic peptide (ANP), TNF-α, and sTNF-RI and sTNF-RII. All blood samples were stored at −80°C until the assay.
In 2 critical patients, the TNF system was monitored over time, during the last 4 and 5 days of life, respectively.
The techniques used have been described in detail elsewhere.13 14 15 Plasma norepinephrine and epinephrine levels were measured by high-performance liquid chromatography with electrochemical detection. Levels of plasma renin activity, aldosterone, and ANP were measured by radioimmunoassay. The radioimmunoassay for ANP was preceded by a solid-phase extraction (Sep-Pak C-18 Cartridges, Waters/Millipore). Synthetic hα-ANP and rabbit anti–hα-ANP antibody (Elken Immunochemical) were used as previously described.16
Antigenic TNF-α was determined according to the manufacturer’s instructions (Technogenetics) by a sandwich ELISA with a commercially available kit. The mixture of monoclonal antibodies used does not neutralize TNF-α bioactivity, allowing measurement of the total circulating TNF-α. The sensitivity of the assay is 3 pg/mL. Intra-assay and interassay coefficients of variation (CVs) are <5.2% and <9.9%, respectively.
Cytotoxicity was assessed by means of an L-M fibroblast bioassay as described by Nargi and Yang17 and slightly modified by us. Mouse L-M fibroblasts (ATCC CCL 1.2) were grown in 96-well plates (Falcon, Becton-Dickinson) in minimum essential medium with Earle’s salts (Gibco), supplemented with 5% FCS and 2 mmol/L glutamine at 37°C in a humidified atmosphere of 5% CO2/95% air. One hundred microliters of a suspension of cells (300 000/mL culture medium) was plated in each well and allowed to grow for 24 hours. Twenty-five microliters of serum sample diluted with 25 μL of culture medium was added to each well and incubated overnight in the presence of 50 μL of a solution containing 8 μg/mL actinomycin D (Sigma). Thus, the final dilution of the serum in each well was 1:8. Live cells were then stained with 5 mg/mL thiazolyl blue (MTT, Calbiochem) for 4 hours. After aspiration of the supernatant, 200 μL of dimethyl sulfoxide was added to each well, and the optical density was read by a microplate reader (Biorad) at 595 nm. Bioactive TNF-α in serum samples was calculated by interpolation of the absorbance values on a calibration curve obtained with rhTNF-α standard solutions (Tecnogen). The distribution of the curve is not linear, and semilogarithmic plotting yields a sigmoidal curve. One hundred percent of the lethal dose occurs at 2000 pg/mL. The sensitivity of the assay is 100 pg/mL. Intra-assay and interassay CVs are <15% and <30%.
To confirm whether the cytotoxicity observed was specifically due to TNF-α, the bioassay was repeated with a rabbit polyclonal neutralizing antibody (0.2 μL/well) against rhTNF-α (Genzyme).
The presence of sTNF-Rs in the serum may modulate TNF-α biological activity. To investigate this possibility, in a separate series of experiments, we added rhTNF-α at a final concentration of 625 pg/mL (corresponding in the bioassay to 60% of the lethal dose) to each sample lacking nonspecific cytotoxicity. The cytotoxicity of these spiked samples was then reassessed.
Serum sTNF-RI levels were assessed with a sandwich ELISA. Two monoclonal antibodies against human sTNF-RI, 7H3 and 9B11, that had been generated by our group were used as previously described.18 Microtiter plates were coated with 7H3 by overnight incubation at 4°C with a 7H3 solution, 10 μg/mL in PBS, 100 μL/well. All subsequent steps were carried out at room temperature. After a rinse with PBS, the uncoated plastic surface was blocked by a 2-hour incubation with 200 μL/well of blocking buffer (0.5% BSA, 0.05% Tween 20 in PBS) and then rinsed again. Serum samples were mixed 2:1 with biotin-9B11 (1:2000 in blocking buffer) and preincubated for 1 hour. Then, 150 μL of each mixture was added to different wells and further incubated for 2 hours under gentle mixing by a plate vortex. After thorough washing, the plates were further incubated with 100 μL/well of a streptavidin-peroxidase solution, 1:1000 in blocking buffer (Janssen Biochimica). After a final wash, each well was incubated for 1 hour with 200 μL of a 10 mg/mL ABTS chromogenic solution (Boehringer Mannheim Italia), and the absorbances at 405 nm were measured. The sensitivity of the assay is 0.3 ng/mL. Intra-assay and interassay CVs were <6.4% and <10%, respectively.
Serum sTNF-RII levels were assessed according to the manufacturer’s specifications by an ELISA kit (Quantikine) available from Research and Diagnostics Systems. The minimum detectable dose is 5 pg/mL. Intra-assay and interassay CVs were <5.1% and <10%, respectively.
Values are expressed as mean±SD. The results were considered to be significant if P<.05. The statistical significance was estimated among the various groups by one-way ANOVA. Group-to-group comparison was done with Student’s t test. Correlations between TNF-α, sTNF-Rs, and the other parameters were tested by Pearson analysis. Independent prognostic values of each parameter were assessed by stepwise discriminant analysis (BMDP PC-90 “7M”).
Table 1⇑ shows the clinical and plasma hormone data of patients. The average cardiac index was 0.8, 0.7, and 0.6 times lower than normal in NYHA classes II, III, and IV, respectively. In class IV patients, right atrial and pulmonary pressures were high (13±3 and 46±13 mm Hg) and serum levels of sodium were low (131±7 mmol/L).
In class II patients, the plasma concentration of norepinephrine was within normal limits. It was increased in class III (1.6 times) and IV (2.4 times) patients. All patients received angiotensin-converting enzyme inhibitors, and their plasma renin activity and aldosterone varied greatly. In class IV patients, the mean values for renin activity and aldosterone were 76 and 5.1 times those of healthy subjects, respectively. Plasma concentration of ANP was higher than the normal range for our laboratory (22±20 pg/mL) in class II (3 times), III (5.6 times), and IV (9.1 times) patients. Nine patients of class IV died within 1 month after entry into the study. This indicates that the group of patients with severe CHF we studied was a group of preterminal patients, not representative of all class IV patients.
Individual levels of bioactive and antigenic (ELISA) TNF-α are reported in Fig 1A⇓ and 1B⇓). When measured by bioassay, TNF-α was below the detection limit of our assay (<100 pg/mL) (Fig 1A⇓). In the sera of 12 patients and of 5 healthy subjects, a cytolytic bioactivity was detectable. However, this cytotoxic effect was unrelated to TNF-α, since a rabbit polyclonal antibody against rhTNF-α added to these samples failed to inhibit their cytolytic effect on L-M cells.
When the more sensitive ELISA method was used in class IV, preterminal, patients, the mean values of TNF-α were higher than in healthy subjects, with a degree of overlapping (from 14.33±7.08 to 33.52±13.12 pg/mL, P<.001). In class II and III patients, mean values of TNF-α were practically the same as normal values.
In class IV patients, both sTNF-Rs were higher (P<.001) than in healthy subjects (3.8 times for the sTNF-RI and 3.4 times for sTNF-RII) (Fig 1C⇑ and 1D⇑). In class II and III patients, the mean values of sTNF-Rs were not different from those of control subjects but were significantly lower (P<.001) than those of class IV patients.
Ten patients died within 1 month (mean time to death, 13.7±8.2 days) after TNF determination. Nine were in class IV and one in class III. Fig 2B⇓ and 2C⇓ shows that these patients had significantly higher levels of antigenic TNF-α (P<.05) and of sTNF-Rs (P<.001). The same was not true for norepinephrine, ANP (Fig 2D⇓ and 2E⇓), renin activity, and aldosterone (data not shown). There was a correlation between sTNF-RII values and duration of survival (r=.67, P<.05). The discriminant stepwise analysis of all parameters considered in this study showed that sTNF-RII was the most important single independent variable predicting death (F[1, 29]=86.24, 96% prediction of outcome). The other parameters, included NYHA clinical classification, had a lower value of predictivity (F[1, 29]<40).
Two of the 10 patients in class IV deteriorated substantially immediately after entry into the study. In these patients, we were able to monitor the TNF system every day for 4 and 5 days, respectively, up until the time they died. The data are reported in Fig 3⇓. In both patients, TNF-α bioactivity remained undetectable. The antigenic TNF-α varied widely. sTNF-Rs progressively increased (sTNF-RI from 2.2 to 4.6 ng/mL and sTNF-RII from 2.1 to 15.8 ng/mL in patient 1; sTNF-RI from 3.1 to 7.9 ng/mL and sTNF-RII from 4.0 to 10.9 ng/mL in patient 2).
Addition of rhTNF-α to the serum of either healthy volunteers or patients with mild to moderate CHF (classes II and III) resulted in a 106±66% (from 625 to 1290±411 pg/mL, P<.001) and 40±21% (from 625 to 877±132 pg/mL, P<.001) increase, respectively, of the expected cytotoxic activity (Table 2⇓). Conversely, the same addition of rhTNF-α to the serum of preterminal patients with severe CHF (class IV) resulted either in a complete inhibition of cytotoxic activity (4 patients) or in a 58±32% (from 625 to 263±198 pg/mL, P<.001) reduction of the expected cytotoxicity (Table 2⇓).
Our data suggest that the TNF system is activated in preterminal patients with heart failure in the absence of “cardiac cachexia,” defined as a generalized wasting syndrome with a weight loss of at least 15% to 20% of the ideal body weight.19 20 The activation of the TNF system is complex, however, and measurement of sTNF-Rs is necessary to assess it fully. Antigenic TNF-α showed a great variability and, as a mean value, was significantly increased in patients with severe CHF, confirming previously reported data.2 3 5 Bioactive TNF-α, measured by a cytolytic assay based on L-M cells, was undetectable in most of the patients. Although the serum of some class II and III patients (12 of 37) was toxic to these cells, we were unable to neutralize these effects with a polyclonal rabbit antiserum specific for TNF-α, suggesting that the cytotoxic activity was not related to TNF-α. It follows that we were unable to detect bioactive TNF-α in any of the patients studied. Sindhwani et al5 reported an increase of bioactive TNF-α in patients with severe CHF (14±6 U/mL) in the presence of antigenic levels of TNF-α (48±8 pg/mL), very close to those we found. Similarly, Levine et al1 reported an increase (>39 U/mL) of TNF-α bioactivity on L-929 cells in patients with CHF and cachexia. Taking into consideration the modest increase of antigenic TNF-α in our patients and the detection limit of our assay (<100 pg/mL), we cannot exclude that an increase of TNF bioactivity occurred in our patients as well. A further analysis using a bioassay with higher sensitivity is necessary to clarify this point.
However, it has been suggested that the bioactivity of TNF-α in the serum could be modulated by sTNF-Rs.11 12 Interestingly, circulating sTNF-Rs were unchanged in our patients with mild and moderate CHF but clearly elevated in class IV, preterminal, patients, supporting the hypothesis that bioactivity of TNF-α was modulated by the increased levels of the soluble receptors, which directly bind to the TNF-α molecule or prevent its binding to cell receptors.
This latter hypothesis is supported by our data, which show that the addition of rhTNF-α to the serum of these patients with high levels of sTNF-Rs reduced and, in some cases, completely inhibited the expected TNF-α bioactivity. Conversely, the addition of the same dose of rhTNF-α to the sera of healthy subjects or patients with less severe cardiac failure (classes II and III) resulted in an increase of the expected TNF-α bioactivity. There was a correlation between absolute levels of sTNF-Rs and the percent potentiation-inhibition changes after spiking. This finding suggests that sTNF-Rs can inversely modulate TNF-α activity, depending on the circulating concentration and time of exposure, providing an idea about the complexity of the TNF system.
TNF-α is produced by inflammatory cells and is a mediator of systemic responses to sepsis and injury. It induces a cascade of endogenous mediators involved in several immunological functions.21 Thus, on the one hand, TNF-α may be considered an essential element in host defense against injury. On the other, its excessive production mediates detrimental systemic and cardiac effects and precipitates a syndrome similar to that of septic shock and cachexia.22
sTNF-Rs, the naturally occurring inhibitors of TNF-α activity, can exert a counteraction that could be either advantageous or injurious for the organism. When present in the serum at physiological levels, they can protect trimeric TNF-α from monomerization and subsequent inactivation or can prolong the half-life of circulating TNF.11 Hypothetically, sTNF-Rs could have exerted this protective action in healthy subjects or in patients with moderate CHF. This would explain the increase in the expected activity that we found after the in vitro spiking with rhTNF-α of the serum of this population.
It has been shown that at physiological concentrations, sTNF-Rs may act as a “slow-release reservoir” of bioactive TNF, thus increasing its half-life.11 12 The mechanism for the spontaneous denaturation of TNF and the way it is inhibited by the sTNF-Rs are not understood. The stabilization of TNF by its soluble receptors is reminiscent of the stabilization of enzymes by their substrates. In both interactions, stabilization of the protein may be due to an induced conformational change that, in the case of TNF-Rs, is likely to occur in the extracellular, ligand-binding domains of the receptors.
When present at higher concentrations, as in our group of preterminal patients in class IV, sTNF-Rs could inhibit the pathological increase of TNF-α activity. We observed inhibition of the expected activity in class IV patients after addition of rhTNF-α. Under these conditions, sTNF-Rs could act as anti-TNF molecules by forming complexes with high affinity to the cytokine.11 12 The shedding of these receptors and the resultant decrease in their concentration on the cell surface could also prevent cell damage. Administration of sTNF-Rs to experimental animals protects against shock and mortality induced by the TNF-α challenge.23 24 Alternatively, since TNF-α induces the shedding of its soluble receptors, it is also possible that increased sTNF-Rs simply reflect activation of the cytokine at a local level. In this latter case, sTNF-Rs could be sensitive “serum markers” of local TNF-α activation. Interestingly, there was a positive correlation between antigenic TNF-α and sTNF-Rs.
It is known that the levels of sTNF-Rs correlate with the severity of various diseases.25 26 27 28 29 This was also true in our small group of patients with CHF and particularly so for levels of sTNF-RII associated with a short-term prognosis. This is not surprising, since the levels of sTNF-RII were strongly correlated with natremia, ejection fraction, cardiac index, right atrial pressure, norepinephrine, and ANP, which from large-scale trials appear to be prognostic factors for CHF. The stepwise discriminant analysis showed that sTNF-RII was the most important single independent variable in predicting death, better than the clinical classification (NYHA). There are, however, several limitations for considering sTNF-RII a prognostic indicator for CHF. First, the number of patients we studied was very limited. Second, we analyzed plasma levels of TNF-α and sTNF-Rs at a given moment in time. We did not investigate the biological turnover or the dynamic behavior of the system. This limits the pathophysiological significance of our findings, since, for example, the rate of shedding of sTNF-Rs could be normal but their metabolic turnover or renal clearance slowed in the terminal stages of heart failure. Third, it is not clear at the moment whether the increase of sTNF-RII concentration is a casual phenomenon or whether the complex of TNF-α and sTNF-RII has a direct pathological effect. Fourth, our patients in class IV were preterminal and not representative of all class IV patients.
We conclude that measurement of sTNF-Rs in addition to that of antigenic and bioactive TNF-α is essential for evaluation of the TNF system in CHF. Both sTNF-RI and sTNF-RII are increased in preterminal patients with severe heart failure and might modulate the in vitro cytotoxicity of TNF-α. Antigenic TNF-α also increases in severe CHF. The increase of sTNF-Rs, and particularly that of sTNF-RII, correlates with poor prognosis. At present, it is not clear whether the elevation of sTNF-Rs in terminal failure is due to an actual increase or to a reduced breakdown or elimination of these receptors. Further explorations are needed to more precisely define the meaning, molecular basis, and interaction of sTNF-Rs and TNF-α in CHF.
This study was supported by National Research Council (CNR) target project “Prevention and Control Disease Factors” 93.00656 PF 41/115.19070 and by CNR target project “Biotechnology and Bioinstrumentation.” Monoclonal antibodies 7H3 and 9B11 were produced under a National Research Program on Advanced Biotechnology contract stipulated by Tecnogen with the Italian Ministry of Research and Technology. The authors thank Tamara Bettini for technical assistance and Roberta Bonetti for secretarial assistance in preparing the manuscript. We thank Dr Marco Pagani for statistical analysis of the data and Dr Bill Dotson Smith for his editing of the manuscript.
- Received November 1, 1994.
- Revision received February 16, 1995.
- Accepted February 25, 1995.
- Copyright © 1995 by American Heart Association
McMurray J, Abdullah I, Dargie HJ, Shapiro D. Increased concentrations of tumor necrosis factor in ’cachectic’ patients with severe chronic heart failure. Br Heart J. 1991;66:356-358.
Dutka DP, Elborn JS, Delamere F, Shale DJ, Morris GK. Tumor necrosis factor α in severe congestive cardiac failure. Br Heart J. 1993;70:141-143.
Katz SD, Rao R, Berman JW, Schwarz M, Demopoulos L, Bijou R, LeJemtel TH. Pathophysiological correlates of increased serum tumor necrosis factor in patients with congestive heart failure: relation to nitric oxide-dependent vasodilation in the forearm circulation. Circulation. 1994;90:12-16.
Sindhwani R, Yuen J, Hirsch H, Tedguy A, Galvao M, Levato P, LeJemtel TH. Reversal of low flow state attenuates immune activation in severely decompensated congestive heart failure. Circulation. 1993;88(pt 2):I-255. Abstract.
Ruff MR, Gifford GE. Purification and physico-chemical characterization of rabbit tumor necrosis factor. J Immunol. 1980;125:1671-1677.
Engelmann H, Aderka D, Rubinstein M, Rotman D, Wallach D. A tumor necrosis factor–binding protein purified to homogeneity from human urine protects cells from tumor necrosis factor toxicity. J Biol Chem. 1989;264:11974-11980.
Aderka D, Engelmann H, Maor Y, Brakebusch C, Wallach D. Stabilization of the bioactivity of tumor necrosis factor by its soluble receptors. J Exp Med. 1992;175:323-329.
Anand IS, Ferrari R, Kalra GS, Wahi PL, Poole-Wilson PA, Harris P. Edema of cardiac origin: studies of body water and sodium, renal function, hemodynamic indexes, and plasma hormones in untreated congestive cardiac failure. Circulation. 1989;80:299-305.
Ferrari R, Anand IS, Kalra GS, Wahi PL, Poole-Wilson PA, Harris P. Body fluid compartments, renal function, hormone levels and hemodynamics in untreated congestive heart failure. Circulation. 1988;28:106. Abstract.
Poiesi C, Rodella A, Mantero G, Cannella G, Ferrari R, Albertini A. Improved radioimmunoassay of atrial natriuretic peptide in plasma. Clin Chem. 1989;35:1431-1434.
Morrison WL, Edwards RHT. Cardiac cachexia. BMJ. 1991;302:301-302.
Mohler KM, Torrance DS, Smith CA, Goodwin RG, Stremler KE, Fung VP, Madani H, Widmer MB. Soluble tumor necrosis factor (TNF) receptors are effective therapeutic agents in lethal endotoxemia and function simultaneously as both TNF carriers and TNF antagonists. J Immunol. 1993;151:1548-1561.
Barrera P, Boerbooms AM, Janssen EM, Sauerwein RW, Gallati H, Mulder J, de Boo T, Demacker PN, van de Putte LB, van der Meer JW. Circulating soluble tumor necrosis factor receptors, interleukin-2 receptors, tumor necrosis factor alpha, and interleukin-6 levels in rheumatoid arthritis: longitudinal evaluation during methotrexate and azathioprine therapy. Arthritis Rheum. 1993;36:1070-1079.
Aukrust P, Liabakk NB, Muller F, Lien E, Espevik T, Froland SS. Serum levels of tumor necrosis factor-alpha (TNF-alpha) and soluble TNF receptors in human immunodeficiency virus type 1 infection: correlations to clinical, immunologic, and virologic parameters. J Infect Dis. 1994;169:420-424.
Zangerle R, Gallati H, Sarcletti M, Weiss G, Denz H, Wachter H, Fuchs D. Increased serum concentrations of soluble tumor necrosis factor receptors in HIV-infected individuals are associated with immune activation. J Acquir Immune Defic Syndr. 1994;7:79-85.