Relationship of granulocyte colony stimulating factor with other acute phase reactants in man

Abstract

The non-specific acute phase response in mice is associated with increased resistance to bacterial infection, which is critically mediated by granulocyte colony stimulating factor (G-CSF), but the behaviour of G-CSF in the human acute phase response is not known. Cardiothoracic surgery is a powerful acute phase stimulus and we show here that this procedure caused increased production of G-CSF, in addition to increases in the circulating concentrations of the proinflammatory cytokine interleukin (IL)-6 and the acute phase plasma proteins C-reactive protein (CRP) and serum amyloid A protein (SAA). Values of G-CSF correlated positively with IL-6 concentrations and circulating neutrophil counts, but not with CRP values. These results confirm that G-CSF is a physiological component of the acute phase response in humans that shares some of the same regulatory controls as IL-6, but its downstream effects are on neutrophils, not hepatic acute phase protein synthesis. Our observations are compatible with a protective role against bacterial infection for G-CSF in the human acute phase response, and support investigation of the prophylactic use of G-CSF in at-risk patients.

Introduction

Granulocyte-colony stimulating factor (G-CSF) is a monocyte-derived cytokine that increases circulating neutrophils numbers, cell surface expression of antimicrobial receptors, and enhances phagocytosis, respiratory burst activity and chemotaxis [1]. We have demonstrated in a mouse model that G-CSF production is increased as part of the non-specific acute phase response to sterile inflammation and that this G-CSF is solely responsible for the enhanced survival in otherwise lethal bacterial infection of animals mounting acute phase responses [2]. The remarkable protective effect was critically dependent on the timing of the peak G-CSF concentration with respect to introduction of infection. This has been observed in other animal models [3,4] and may explain why the use of G-CSF for adjunctive treatment of established infection has been unsuccessful in clinical trials [5–7]. However, our observations suggest strongly that if G-CSF has similar effects in humans, pretreatment with G-CSF may provide a novel approach to clinical prophylaxis of bacterial infections. Increased serum concentrations of G-CSF have been detected in acute urinary tract and respiratory infections [8–10], but the relationships to other acute phase proteins and the acute phase response in general have not been examined. Both the classical acute phase response and neutrophilia associated with increased serum G-CSF values have been well documented in patients undergoing elective cardiothoracic surgery [11–19]. Here we report the relationships in such patients between the concentrations of G-CSF, interleukin (IL)-6 and the most sensitive human acute phase plasma proteins, C-reactive protein (CRP) and serum amyloid A protein (SAA).

Patients and methods

All patients over 18 years of age undergoing elective cardiothoracic surgery (coronary artery bypass, heart valve replacement, pulmonary lobectomy or pneumonectomy) at the Hammersmith Hospital (London, UK) between May and July 2001 were invited to participate in the study. Patients unable to give informed consent on their own behalf, those undergoing emergency surgery and patients with burns or traumatic injury, who would have established acute phase responses prior to surgery, were excluded. Among the 57 patients (46 men) studied, median age 67·5 years (range 21–83 years), 44 individuals had coronary artery bypass grafting, six had aortic valve replacement and coronary artery bypass grafting, two had aortic valve replacements alone, four patients had mitral valve replacements and one had aortic coarctation repair. Serum samples were collected from all patients 12–24 h before, and then 3–6 h, 24 h and 48 h after surgery for assay of concentrations of G-CSF and IL-6 by enzyme-linked immunoassay kits from R&D Systems UK (Abingdon, UK), according to the manufacturer's instructions, and the acute phase proteins, CRP and SAA, as described previously [20,21]. The lower limits of detection were 0·2 mg/l for CRP and 0·7 mg/l for SAA. The blood neutrophil count was measured preoperatively and 24 and 48 h postoperatively, as part of the routine clinical care in all patients. Results were analysed using sigmastat version 2·03 (SPSS Inc., Woking, UK). The study was approved by the Hammersmith Hospital Research Ethics Committee.

Results

We observed the typical acute phase plasma protein response of CRP and SAA to the stimulus of cardiothoracic surgery, with little change at 3–6 h but major increases in concentration at 24 h continuing at 48 h (Fig. 1). The concentration of IL-6, which is a key cytokine up-regulator of hepatic acute phase protein synthesis, was increased at 3–6 h and remained high at 24 h and 48 h (Fig. 1). The enzyme-linked immunosorbent assay (ELISA) used for G-CSF was insufficiently sensitive at 100 ng/ml to detect serum G-CSF in most of the patients preoperatively, but there was clearly an acute phase response of this protein, with increased concentrations at 3–6 h and 24 h, falling somewhat by 48 h, in a profile resembling that of IL-6 (Fig. 1).

Fig. 1.

Serum levels of acute phase proteins before and after cardiothoracic surgery. Serum levels of CRP and SAA rose 24 h and 48 h after surgery compared to preceding time-points. Serum levels of G-CSF and IL-6 rose 3–6 h after surgery compared to preoperative levels. *Denotes significant change in serum levels compared to preceding time point (P < 0·05, Kruskal–Wallis one-way analysis of variance on ranks with all pairwise multiple comparisons: Dunn's method). Box-and-whisker plots demonstrate median, standard error and standard deviation of 57 patients.

Correlations between the various acute phase reactants were sought using area under the curve (AUC) analysis as a measure of overall response of each protein (Fig. 2). There were statistically significant positive correlations between production of IL-6 and CRP, IL-6 and G-CSF and SAA and CRP, but not between CRP and G-CSF (Fig. 2).

Fig. 2.

Correlations of serum acute phase proteins and cytokines after elective cardiothoracic surgery. Dot plot correlations of (a) CRP and SAA, (b) CRP and IL-6, (c) CRP and G-CSF and (d) IL-6 and G-CSF from AUC analyses of acute phase protein and cytokine serum levels following surgery in each of 57 patients. Best-fit regression lines are indicated on each plot. Significant positive correlations (Spearman's rank order test) were observed between CRP and SAA (P < 0·001), CRP and IL-6 (P < 0·05), and IL-6 and G-CSF (P < 0·01), but not CRP and G-CSF.

The circulating neutrophil count was significantly higher at both 24 h and 48 h after surgery compared with the preoperative baseline (Fig. 3). AUC analysis for each patient revealed a significant positive correlation between the magnitude of the G-CSF response and the rise in neutrophil count (Fig. 3), compatible with the acute phase response of G-CSF being responsible for the postoperative neutrophilia. No significant positive or negative correlation between circulating neutrophils and IL-6 or CRP was found.

Fig. 3.

Circulating neutrophil counts before and after elective cardiothoracic surgery and correlations with G-CSF responses. (a) Box-and-whisker plot of neutrophil counts before surgery, and at 24 and 48 h after surgery (n = 57). Neutrophil counts at 24 and 48 h after surgery were significantly greater than preoperatively (P < 0·05, Kruskal–Wallis one-way analysis of variance on ranks; all pairwise comparison by Dunn's method). (b) Dot plot correlation and best-fit regression line of G-CSF and neutrophil counts from AUC analyses of serum levels and cell counts following surgery in each of 57 patients, demonstrating a significant positive correlation (P < 0·05, Spearman's rank order test).

Discussion

The similar profile and significant correlation between G-CSF and IL-6 responses supports the hypothesis of shared regulatory mechanisms for these two cytokines, the genes for which are closely co-located on the long arm of human chromosome 17 [13]. There is no evidence that either cytokine stimulates production of the other [22], therefore this correlation is likely to represent common upstream regulatory elements such as nuclear factor IL-6 [23,24]. Despite this observation, their major effector functions are different, IL-6 up-regulating hepatic acute phase proteins, as shown here by CRP and SAA, while G-CSF increases neutrophil numbers and activity. Interestingly, the cytokine, granulocyte macrophage-colony stimulating factor (GM-CSF), is induced by similar stimuli, derived from similar cellular sources and shares some biological effects with G-CSF. It also shows a regulatory relationship to IL-6 [25], but GM-CSF has not been detected in patients with infections [26] or in our murine model of non-specific acute phase response stimulation [2]. We therefore suggest that G-CSF is a specifically important myeloid stimulating factor in the acute phase response.

Here we show, for the first time, that human G-CSF participates in the classical non-specific acute phase response to the powerful stimulus of elective cardiothoracic surgery, similarly to murine G-CSF after induction of acute sterile inflammation. Because we and others have demonstrated in the animals that G-CSF, produced either endogenously as part of the acute phase response or injected in pure form, can mediate powerfully increased host resistance to otherwise fatal bacterial infection, our present findings support the further exploration of prophylactic therapy with G-CSF in patients at risk of bacterial infection.