• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of iaiPermissionsJournals.ASM.orgJournalIAI ArticleJournal InfoAuthorsReviewers
Infect Immun. Apr 2000; 68(4): 1942–1945.

Predictive Value of Nuclear Factor κB Activity and Plasma Cytokine Levels in Patients with Sepsis

Editor: R. N. Moore


The relationship between fluctuating cytokine concentrations in plasma and the outcome of sepsis is complex. We postulated that early measurement of the activation of nuclear factor κB (NF-κB), a transcriptional regulatory protein involved in proinflammatory cytokine expression, may help to predict the outcome of sepsis. We determined NF-κB activation in peripheral blood mononuclear cells of 34 patients with severe sepsis (23 survivors and 11 nonsurvivors) and serial concentrations of inflammatory cytokines (interleukin-6, interleukin-1, and tumor necrosis factor) and various endogenous antagonists in plasma. NF-κB activity was significantly higher in nonsurvivors and correlated strongly with the severity of illness (APACHE II score), although neither was related to the cytokine levels. Apart from NF-κB activity, the interleukin-1 receptor antagonist was the only cytokine tested whose level in plasma was of value in predicting mortality by logistic regression analysis. These results underscore the prognostic value of early measurement of NF-κB activity in patients with severe sepsis.

Many reports have focused on aspects of the proinflammatory cytokine network, which is believed to be central to the pathophysiology of the sepsis syndrome (5, 8). However, the cytokine responses in patients with sepsis appears to vary so much between individuals (10) that the prognostic usefulness of circulating cytokine concentrations is often less than that of clinical variables, such as the acute physiology and chronic health evaluation (APACHE) II or III (9). Other studies indicate that the problem in overwhelming sepsis is not that inflammatory cytokines are expressed but, rather, that their expression is not properly modulated by anti-inflammatory mediators (16, 17). Recent investigations by others (3) and ourselves (1) searching for new clinically reliable markers in patients with sepsis have shown that circulating leptin levels, whose secretion is closely linked to the activation of the cytokine cascade (1), may help to predict mortality in sepsis and septic shock.

Among several transcriptional regulatory factors involved in immunoregulatory genes expression, nuclear factor kappa B (NF-κB) acts at a critical step for directing the transcription of many proinflammatory genes in animal models of inflammatory diseases (6, 7). Investigations regarding the role of NF-κB in human inflammatory diseases are scarce (2, 15). So far, no study has aimed to examine in patients with sepsis the relationship between the concentrations of some components of the proinflammatory and anti-inflammatory cytokine response in plasma, NF-κB expression in peripheral blood mononuclear cells, and clinical outcome. We hypothesized that severe, fatal sepsis could be distinguished from less severe sepsis by demonstrating greater NF-κB activation and decreased anti-inflammatory response. Thus, this study compared the prognostic value of combining measurements of NF-κB activity in circulating blood cells and the cytokine profile in plasma in patients with severe sepsis.



The study population was recruited from a series of 100 patients with a clinical diagnosis of sepsis, consecutively admitted to the Department of Internal Medicine or the Intensive Care Unit over a 4-month period. This study was approved by the local Ethics Committee. The patients were screened daily for severe sepsis criteria according to the recommendation of the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference Committee (13). A total of 34 patients with bacteriologically documented infections (24 with positive blood cultures and 10 with positive bronchial fluid or urine cultures) were included in the study when they met the severe-sepsis criteria for the first time. The time between admission and inclusion in the study varied from 1 to 12 days (median, 5 days). No patient died during the first 3 days of entry into the study. Written informed consent was obtained from all subjects. The following exclusion criteria were imposed: malignancy and chronic inflammatory diseases, treatment with steroids or immunosuppresive drugs during the last month, hepatic failure (serum aspartate aminotransferase and/or alanine aminotransferase level, >100 IU/liter; prothrombin time, <60%; total bilirubin level, >60 μmol/liter), renal insufficiency (creatinine level in plasma, 200 μmol/liter), AIDS, and gestation. The control group consisted of 20 healthy individuals before elective surgery, who had normal concentrations of orosomucoid in serum (reference value, <1.1 g/liter) at the time of blood sampling, to exclude an ongoing acute-phase reaction.

Laboratory methods.

Body mass index (weight in kilograms divided by the square of height in meters) and APACHE II score (12) were calculated on entry into the study. Plasma samples were simultaneously obtained at the time of systolic hypotension and every 6 h thereafter for 24 h. Tumor necrosis factor alpha, interleukin-1β (IL-1β), IL-6, and IL-10 levels were determined by an enzyme-linked immunoassay (Medgenix Diagnostics, Fleurus, Belgium), and concentrations of soluble TNF receptor type I (sTNF-RI) and IL-1 receptor antagonist (IL-1ra) were measured using a quantitative sandwich enzyme immunoassay (Quantikine; R&D systems, Minneapolis, Minn.). The detection limits of the assays were 10 (TNF-α), 21 (IL-1β), 8 (IL-6), 12 (IL-10), 90 (sTNF-RI), and 250 (IL-1ra) pg/ml. All cytokine assays were performed in duplicate and had intra- and interassay variations lower than 8 and 10%, respectively.

Nuclear protein extraction and electrophoretic mobility shift assay (EMSA).

Human peripheral blood mononuclear cells (PBMC) were isolated from freshly drawn heparinized blood by centrifugation on Lymphoprep (Nycomed Pharma AS, Oslo, Norway) as specified by the manufacturer. Nuclear protein extracts were individually obtained from PBMC of the 34 patients and 20 healthy control subjects by the method of Montaner et al. (14). Briefly, cells were lysed in cold buffer A (20 mM HEPES [pH 8], 1.25% Nonidet P-40, 10 mM KCl, 0.15 mM EGTA, 0.15 mM EDTA, 1 mM dithiothreitol [DTT], 0.2 mM phenylmethylsulfonyl fluoride). Nuclei were pelled by centrifugation at 400 × g and 4°C for 5 min and washed in cold buffer B (20 mM HEPES [pH 8], 50 mM NaCl, 25% glycerol, 0.15 mM EGTA, 0.25 mM EDTA, 1.5 mM MgCl2, 1 mM DTT, 0.2 mM phenylmethylsulfonyl fluoride). After centrifugation, nuclear proteins were extracted by incubation for 30 min in cold buffer C (buffer B but with 400 mM NaCl). Then, nuclear extracts from all control subjects were pooled and used subsequently as the control sample. Double-stranded oligonucleotide probes of the immunoglobulin gene containing the NF-κB binding site (18) with the following sequences were synthesized: 5′-TCGACGAGCTCGGGACTTTCCGAGC-3′        3′-GCTCGAGCCCTGAAAGGCTCGAGCT-5′

The DNA binding reaction was carried out by incubation of 15 μg of nuclear proteins with 32P-labeled double-stranded oligonucleotides (0.3 ng) in a final volume of 20 μl of reaction mixture [5 mM HEPES (pH 7.8), 50 mM KCl, 0.5 mM DTT, 5 mM MgCl2, 10% glycerol, 0.15 μg of poly(dI-dC)-poly(dI-dC) per μl]. The specificity of binding was confirmed by competition with a 400-fold excess of unlabeled κB oligonucleotide in the reaction mixture. After incubation, the samples were loaded onto a 5% polyacrylamide (in 0.5× Tris-borate-EDTA buffer) gel, which was run at 10 V/cm. One control sample was always loaded onto each gel along with several (two to four) patient samples. After fixation with 10% acetic acid, the gels were dried and radioactivity was quantified using an Instant Imager (Packard). The NF-κB activity value for each patient sample was expressed relative to the NF-κB binding activity obtained in the control sample run in the same gel (taken to be 100%). All samples from septic patients were assayed in two or three different gels; the final value considered for each sample was the mean of the individual values obtained, which were quite reproducible. Additionally, dried gels were subjected to autoradiography to visualize the NF-κB activity and the specificity of the binding.

Statistical analysis.

Mean cytokine levels were tabulated by adding serial values and dividing by 4, the total number of determinations. Data are presented as geometric means and extremes. Comparisons were performed by one-way analysis of variance. Scheffé's simultaneous confidence intervals were determined to account for multiple comparisons. Effectiveness in predicting outcome was compared by calculating the sensitivity and specificity of the best cutoff value for each variable, and the accuracies of the various variables were determined by constructing their receiver operating characteristic (ROC) curves with software from MedClac. Stepwise logistic regression analysis was used to compare survivors and nonsurvivors and to determine which variables best predicted mortality.


Eleven patients died within the first 10 days, and the other 23 patients survived and were discharged from hospital. NF-κB binding activity on study admission and average serial cytokine concentrations in plasma over the first 24 h are shown in Tables Tables11 and and2.2. Figure Figure11 shows a representative experiment of the EMSA for NF-κB activity in control subjects and in two patients with sepsis. Compared with survivors, patients who died had significantly higher NF-κB activity in peripheral blood cells and APACHE II scores (P < 0.01); moreover, there was a strong correlation between these two variables (r = 0.616; P < 0.001). Differences in NF-κB activity between patients with positive blood cultures and those without proven bacteremia (364% activity [range, 135 to 820%] versus 231% activity [range, 123 to 740%]) did not reach statistical significance. All patients had increased levels of IL-6, which were manyfold greater than normal levels, and there was evidence of significantly higher levels in those who died (Table (Table1).1). In contrast, the IL-10 concentrations in plasma were significantly higher in survivors than in nonsurvivors. TNF-α and IL-1β were detected in only 6 of the nonsurvivors (55%) compared with 14 of the survivors (61%) (χ2 = 2.7; P = 0.1). sTNF-RI and IL-1ra were detected in plasma from all patients in concentrations greatly in excess of the TNF-α and IL-1β concentrations, respectively. Survivors had significantly higher concentrations of sTNF-RI and IL-1ra and a higher ratio of IL-1ra to IL-1β than did nonsurvivors (Table (Table2).2).

Demographic data, NF-κB binding activity, and IL-6 and IL-10 levels in patients with severe sepsisa
Concentrations of TNF-α, IL-1β, and their endogenous inhibitors in plasma in patients with severe sepsisa
FIG. 1
Representative EMSA analysis for NF-κB binding activity in PBMC from healthy controls and patients with sepsis. Nuclear extract (15 μg) from the control sample (lanes 1 and 2), from a survivor with sepsis (lanes 3 and 4), or from a nonsurvivor ...

Comparisons of ROC curves for the APACHE II score, NF-κB activity, and IL-1ra assay are shown in Fig. Fig.2.2. Evaluation of these ROC curves indicated that 21 was the best cutoff point of the APACHE II score for predicting mortality (sensitivity, 96%; specificity, 92%), while 372%, expressed as the relative percentage of the value obtained with nuclear extracts pooled from 20 healthy control subjects, was the best cutoff value for NF-κB activity (sensitivity, 97%; specificity, 70%). The area under the curve for NF-κB activity was significantly higher than that for the APACHE II score and IL-1ra assay.

FIG. 2
ROC curves for NF-κB activity in PBMC, APACHE II score, and IL-1ra levels in plasma. The areas under the curve (AUC) for these variables were compared. Overall, NF-κB activity was the best predictor of mortality.

Significant linear correlations were found between NF-κB activity and the IL-1ra concentration in plasma (r = 0.556; P < 0.001) or the IL-1β/IL-1ra ratio (r = 0.37; P < 0.05). A stepwise logistic regression analysis revealed that only NF-κB activity (β = 1.90; standard error = 0.69; odds ratio = 6.7; 95% confidence interval = 1.7 to 26.3) and IL-1ra (β = 1.35; odds ratio = 3.2; 95% confidence interval = 0.6 to 9.7) were found to predict mortality (P < 0.001). No other variable was of sufficient predictive value to be included in the model.


Mortality in patients with sepsis reflects a multifactorial pathology, and neither cytokine concentrations in plasma nor even the APACHE II score can be expected to accurately predict patient outcomes. Therefore, we chose to focus on NF-κB as an important transcriptional regulatory factor that regulates the expression of multiple cytokine genes in animal models of sepsis (6, 7). Previous observations showed a significantly higher NF-κB activity in the blood monocytes of 5 patients with sepsis who died than in 10 patients who survived, but the predictive value of NF-κB was not compared with that of other sepsis mediators (2). In this study, by comparing the predictive value of measuring NF-κB activity in PBMC and the concentrations of some pro- and anti-inflammatory mediators in plasma in patients with severe sepsis, we found that the NF-κB activity measured in a single blood sample on admission into the study was a better overall predictor of mortality than the balance and time course of pro- and anti-inflammatory cytokines released in plasma. The APACHE II score correlated strongly with NF-κB activity on study admission, but neither was related to cytokine levels in plasma. The analysis of the ROC curves showed that NF-κB activity had a better predictive value for mortality than did the APACHE II score and IL-1ra assay. Apart from NF-κB activity in mononuclear cells, the IL-1ra level in plasma was the only cytokine concentration determined that had an independent correlation with prognosis in the logistic regression analysis.

The predictive value of the levels of circulating proinflammatory cytokines and their inhibitors in human sepsis is controversial, since patients with sepsis do not represent a homogeneous population (4, 5). In addition, to be most useful for prognosis, daily determinations should be made because the peak value is the most informative. In this study, we measured various mediator concentrations at particular time points, i.e., every 6 h over the first day after the onset of systolic hypotension, when the concentrations in plasma probably better reflect their paracrine activity. Although significant differences in the concentrations of the cytokines tested were observed between survivors and nonsurvivors in the group as a whole, the degree of overlap limited the predictive value of these concentrations. Interestingly, we noted a profound imbalance between the levels of cytokines and their antagonists. Surviving patients had levels of IL-1ra in plasma in great excess, approximately a mean value of 75-fold higher than those of IL-1β, whereas nonsurvivors had significantly less overproduction of IL-1ra. A high IL-1ra level or IL-1ra/IL-1β ratio thus appears to modulate inflammatory responses by inhibiting IL-1 activity and is associated with recovery. This finding is consistent with data obtained for experimental endotoxemia, suggesting that high circulating IL-1ra levels could have a modulating effect as a defense mechanism (11).

In summary, unlike measurements of pro- and anti-inflammatory cytokine levels in plasma at particular time points, which only indirectly reflect tissue secretion, our data showed that NF-κB activity in PBMC assessed once on study admission better reflects the complex pattern of immunologic events in patients with sepsis and may accurately predict the outcome of infection. An overproduction of IL-1ra appears to play a protective role. Thus, both parameters used together provided excellent prognostic information.


This work was supported by a grant (expediente 082/0010/1997) from the Direccion General de Investigación, Comunidad Autónoma de Madrid, Madrid, Spain.

We are grateful to David Cabestrano and Manuel Cidoncha (Intensive Care Unit) for helping with data collection. We thank the patient care staff and our patients for their willing involvement in the study.


1. Arnalich F, López J, Codoceo R, Jiménez M, Madero R, Montiel C. Relationship of plasma leptin to plasma cytokines and survival in sepsis and septic shock. J Infect Dis. 1999;180:908–911. [PubMed]
2. Böhrer H, Qiu F, Zimmermann T, Zhang Y, Jilmer T, Mannel D, Böttiger B W, Stern D M, Waldherr R, Saeger H D, Ziegler R, Bierhaus A, Martin E. Role of NF-κ B in the mortality of sepsis. J Clin Investig. 1997;100:972–985. [PMC free article] [PubMed]
3. Bornstein S R, Licinio J, Tauchnitz L R, Engelman L, Negrao A B, Gold P, Chrousos G P. Plasma leptin levels are increased in survivors of acute sepsis: associated loss of diurnal rhythm in cortisol and leptin secretion. J Clin Endocrinol Metab. 1998;83:280–283. [PubMed]
4. Calandra T, Baumgartner J D, Grau G E, Wu M M, Lambert P H, Schellekens J, Verhoef J, Glauser M P. the Swiss-Dutch J5 Immunoglobulin Study Group. Prognostic values of tumor necrosis factor/cachectin, interleukin-1, interferon-α and interferon-γ in the serum of patients with septic shock. J Infect Dis. 1990;161:982–987. [PubMed]
5. Casey L, Blak R A, Bone R C. Plasma cytokine and endotoxin levels correlate with survival in patients with sepsis. Ann Intern Med. 1993;119:771–778. [PubMed]
6. Chen F, Castranova V, Shi X, Demers L M. New insights into the role of nuclear factor-κB, a ubiquitous transcription factor in the initiation of diseases. Clin Chem. 1999;45:7–17. [PubMed]
7. Christman J W, Lancaster L H, Blackwell T S. Nuclear factor-κB: a pivotal role in the systemic inflammatory response syndrome and new target for therapy. Intensive Care Med. 1998;24:1131–1138. [PubMed]
8. Damas P, Canivet J L, De Groote D, Wrindts Y, Albert A, Franchimont P, Lamy M. Sepsis and serum cytokine concentrations. Crit Care Med. 1997;25:405–412. [PubMed]
9. Friedland J S, Porter J C, Daryanani S, Bland J M, Screaton N J, Vessely M J J, Griffin G E, Bennett E D, Remick D G. Plasma proinflammatory concentrations, acute physiologic and chronic health evaluation (APACHE) III scores and survival in patients in an intensive care unit. Crit Care Med. 1996;24:1775–1781. [PubMed]
10. Gardlund B, Sjolin J, Nilsson A, Roll M, Wickerts C J, Wretlind B. Plasma levels of cytokines in primary septic shock in humans: correlation with disease severity. J Infect Dis. 1995;172:296–301. [PubMed]
11. Granowitz E V, Santos A A, Poutsiaka D D, Cannon J G, Wilmore D W, Wolff S M, Dinarello C A. Production of interleukin-1-receptor antagonist during experimental endotoxaemia. Lancet. 1991;338:1423–1424. [PubMed]
12. Knaus W A, Draper E A, Wagner D P, Zimmermann J E. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13:389–393. [PubMed]
13. Members of the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference Committee. Definitions for sepsis and organ failure and guidelines for the use of innovative therapy in sepsis. Crit Care Med. 1992;20:864–874. [PubMed]
14. Montaner S, Ramos A, Perona R, Esteve P, Carnero A, Lacal J C. Overexpression of PKC in NIH 3T3 cells does not induce cell transformation nor tumorigenicity and does not alter NF κB activity. Oncogene. 1995;10:2213–2220. [PubMed]
15. Schwartz M D, Moore E E, Moore F A, Shenkar R, Moine P, Haenel J B, Abraham E. Nuclear factor-κB is activated in alveolar macrophages from patients with acute respiratory distress syndrome. Crit Care Med. 1996;24:1285–1292. [PubMed]
16. Van der Poll T, de Waal Malefyt R, Coyle S M, Lowry S F. Antiinflammatory cytokine responses during sepsis and experimental endotoxemia: sequential measurements of plasma soluble interleukin-1 receptor type II, IL-10, and IL-13. J Infect Dis. 1997;175:118–122. [PubMed]
17. Walley K R, Lukacs N W, Standiford T J, Strieter R M, Kunkel S L. Balance of inflammatory cytokines related to severity and mortality of murine sepsis. Infect Immun. 1996;64:4733–4738. [PMC free article] [PubMed]
18. Zabel U, Schreck R, Baeuerle P A. DNA binding of purified transcription factor NF-κB: affinity, specificity, Zn2+ dependence, and differential half-site recognition. J Biol Chem. 1991;266:252–260. [PubMed]

Articles from Infection and Immunity are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...