Interleukin‐6 concentrations in wound fluids rather than serological markers are useful in assessing bacterial triggers of ulcer inflammation

Abstract

Bacterial pathogenicity, microbial load and diversity are decisive for outcome and therapy of non healing ulcers. However, until now, no routine laboratory parameter is available to assess the inflammatory level caused by chronic wound infections. We thus investigated the usefulness of levels of interleukin (IL)‐6 and tumour necrosis factor alpha (TNFα) in wound fluids for assessing ulcer inflammation in the presence or absence of microbial triggers. In addition, the predictive values of local cytokine analyses were compared with those of C‐reactive protein (CRP) and lipopolysaccharide‐binding protein (LBP) because serological markers are normally used to underline the suspicion of wound infections.

The present data from chronic arterial and venous ulcers (n = 45) clearly show that mixed bacterial infections increased IL‐6 and TNFα concentrations in wound fluids when compared with ulcers with a monomicrobial infection (P < 0·01 and P < 0·05, respectively). IL‐6 was also significantly elevated when a high bacterial load [versus <10⁵ colony‐forming units (cfu)/ml: P = 0·04] or an infection with Pseudomonas was observed (versus isolation of non Pseudomonas strains: P = 0·05). Although distinct proinflammatory triggers may interfere with regard to cytokine levels, sensitivity and specificity were significant in predicting bacterial risk factors, particularly for IL‐6 at a designated cut‐off of 125 pg/ml (sensitivity and specificity for predicting a mixed infection: 70% and 64%, for predicting a bacterial load of >10⁵ cfu/ml: 62% and 57% and for predicting an infection with Pseudomonas: 90% and 57%). In contrast to local cytokine levels, the serological markers CRP and LBP were not associated with the presence of any of the investigated bacterial triggers.

Focusing on the aim of the study, IL‐6 analysed in wound washouts seems to be a useful diagnostic marker for a sensitive and specific assessment of ulcers inflammation with regard to bacterial triggers.

Introduction

Clinical infection of non healing and chronic ulcers is determined by clinical signs. However, diagnostic features like pain, cellulitis and abnormal granulation tissue may be masked if the inflammatory process and the immune system are dysfunctional. Diagnosis of wound infection in such cases relies on a high threshold of clinical suspicion, and interpretation of clinical signs is misleading or varied particularly among wound experts (1). Although reasons for delayed healing of chronic wounds are multifactorial, the impact of microbial aspects like bacterial synergies (2), biofilm formation (3), microbial bioburden 4, 5 and bacterial pathogenicity is indisputable. However, although several soluble mediators assessed as hallmarks of the healing (e.g. platelet‐derived growth factor, epidermal growth factor and transforming growth factor‐β) and the non healing phase [e.g. tumour necrosis factor alpha (TNFα), interleukin (IL)‐6 or IL‐1] of chronic ulcers are in the focus of pathophysiological investigations 6, 7, 8, 9, 10, 11, no objective laboratory parameter is available to assess and monitor the status of inflammation for clinical practice. Therefore, the present pilot study aimed to prove the significance of levels of IL‐6 and TNFα in wound washouts for assessing potent inflammatory triggers like bacterial bioburden, specific pathogens and a polymicrobial infection. In addition, predictive value of local cytokine analyses for inflammatory triggers were compared with that of C‐reactive protein (CRP) and lipopolysaccharide‐binding protein (LBP) because serological parameters are normally used to underline the clinical suspicion of chronic wound infections.

Patients and methods

Patients

A total of 45 stationary patients with non healing chronic venous (n = 25) and arterial/mixed ulcers (n = 20) were consecutively recruited from the Department of Reconstructive and Hand surgery. All patients were admitted for surgical debridement and subsequent skin grafting.

Ulcer size (cm³) was measured using the planimetric (cm²) method and depth (cm), and each ulcer was photographed for wound documentation and follow‐up.

For differentiation between venous and arterial pathogenesis, cut‐off for an adequate circulation of the study foot was defined by an ankle brachial index of 0·7. In addition, venous incompetence was confirmed by duplex ultrasonography. Patients were also screened to exclude the presence of systemic disease including diabetes.

Study protocol was written according to the Ethical Committee of the Bremer Ärztekammer. Written informed consent was obtained from each patient before participating the study.

Serological inflammatory markers

CRP is a classical serological marker for the assessment of an inflammatory process. The present immunoturbidimetric test for qualitative detection of CRP in serum by goat anti‐human antibodies was obtained from Olympus and determined on an auto‐analyser (AU 640; Olympus Diagnostica GmbH, Hamburg, Germany). The lowest detectable level (sensitivity) was estimated at 1·5 mg/l, intra‐assay precision (coeficient of variance (CV)%) was 2·81 (at CRP levels up to 15 mg/dl) and total precision was 5·99. The test was linear within a concentration range of 5–300 mg/l for a six‐point calibration.

LBP is an acute‐phase protein that binds and recruits bacterial endotoxin to the surface of immunoeffector cells. For this reason, it has been suggested that LBP levels may be indicative of exposure to bacteria or endotoxin. In the present study, LBP concentrations were assayed by an automated chemiluminescent test based on styrol beads‐coated goat anti‐human LBP antibodies (DPC, DPC Biermann, Bad Nauhiem, Germany). Within‐run specificity (CV%) on an Immulite 1000 analyser (DPC) was given with 5·8 at a concentration of 14 μg/ml and total precision with 8·1. The lowest detectable level for LBP was given with 0·2 μg/ml. The used LBP assay was highly specific and show no cross‐reactivity with other inflammatory markers such as serum amyloid A, CRP, IL‐6, IL‐8 and TNFα (manufacturer’s instructions).

Sampling of wound fluids

Sampling of diluted wound fluids was performed according to a modified protocol (12). Briefly, a wound area of approximately 2 × 3 cm was washed using a syringe with 5‐ml sterile saline. The washout was collected by needle aspiration and analysed for cytokines.

To verify if cytokine concentrations in wound washouts reflect the inflammatory pattern of deeper tissue, washouts as well as punch biopsies (5 mm) were obtained from a subgroup of patients with chronic ulcers. Punch biopsies were incubated for 1 hour in a phosphate buffer, samples were centrifuged and supernatants were analysed for cytokines. A good correlation was found between concentrations of cytokines released in supernatants of biopsies and in wound washouts (IL‐6, r = 0·81; TNFα, r = 0·83).

Cytokine assays

Concentrations of IL‐6 and TNFα in wound fluids were detected by automated solid‐phase, enzyme‐labelled chemiluminescence sequential immunometric assays (DPC). Sensitivities for IL‐6 and TNFα were given with 2 pg/ml and 1·7 pg/ml, respectively. The within‐run specificities (CV%) were 5·2 for IL‐6 (total 7·4) at a concentration of 112 pg/ml and 3·5 (total 4·0) for TNFα at a concentration of 327 pg/ml.

Semiquantitative analyses of the bacterial load

To assess the bacterial load of chronic ulcers, superficial swab cultures were obtained using a calibrated loop (1 μl) technique. Loops were plated on sterile blood agar (5% sheep blood), McConkey agar and Schaedler agar at bedside and incubated for 24 hours at 37°C. Because bacterial species as well as bacterial load are not equally distributed among the ulcer area, three to four swabs from each ulcer were taken at each visit. After 24 hours, agar plates were inspected visually. Colonies from each germ were counted and the bacterial load [colony‐forming units (cfu)/ml] was estimated as follows – no growth: sterile; 1–3 colonies: <10³ cfu/ml; 3–10 colonies: 10³ cfu/ml; 10–100 colonies: 10⁴ cfu/ml and >100 colonies: >10⁵ cfu/ml. For validation of the technique, a comparison between the Brentano wet culturing technique (13) and the calibrated loop technique was performed. A good correlation was found (data not shown).

Identification of microbes

Microbes, which were cultured from swabs, were biochemically identified by an automated identification system (API Expression; Biomerieux, Nutinger, Germany).

Statistical analyses

Data were analysed using the Statistical Package for the Social Sciences (SPSS, version 10·0 for Microsoft Windows; SPSS, Chicago, IL). Descriptive data are given as mean ± SE. Where appropriate, continuous variables were tested by the Student’s t‐test or the Wilcoxon U‐test. Percentages were compared using chi‐squared test. Spearman’s rho was used to test the correlation between two variables. By using linear regression models involving several independent variables, the best predictive variables for cytokine levels were identified. Finally, calculations for ROC (receiver operations characteristics) analyses were made to evaluate the performance of the tested classification systems (polymicrobial/monomicrobial infection, Pseudomonas/non Pseudomonas infection and bacterial load ≥10⁵ or <10⁵) with regard to the significant diagnostic markers IL‐6 and TNFα. Finally, sensitivity and specificity for cut‐offs were calculated. All given P values are two‐tailed, and a P value of less than 0·05 being considered to be statistically significant.

Results

The mean age of the patients with arterial/mixed and venous ulcers was 72 (4·4) years, 33 were female and 12 were male. The mean history of ulcers was 24·3 (9·2) months without differences between arterial or venous pathogenesis. However, ulcers of arterial genesis had a significantly larger area [40·8 (9·3) versus 9·0 (3·2) cm³, P = 0·001] and higher concentrations of IL‐6 in washouts [626 (208) versus 263 (79) pg/ml, P = 0·049]. No differences were found with regard to TNFα [81 (42) versus 52 (18) pg/ml] or serum CRP [28.2 (7·5) versus 24·2 (3·1) mg/dl] or LBP [16·3 (2·1) versus 17·9 (1·5) mg/dl].

Wound microbiology

After an overnight incubation of the agar plates, a total of 68 isolates obtained from arterial and venous ulcers were identified. They belong particularly to the genus of haemolytic Streptococci (venous ulcer 4% versus 15% in arterial ulcers), Staphylococci [coagulase‐positive (cp) (60% versus 44 %) and coagulase‐negative (cn) (12% versus 5%)], and Pseudomonas (20% versus 32%), the family of Enterobacteriaceae (40% versus 25%), particularly of genus Escherichia and Proteus, and others (16·0% versus 25·0%) including Enterococci, non haemolytic Streptococci, Corynebacterium (non ulcerans and non diphtheriae) and anaerobic isolates. However, no significant difference was found with regard to the bacterial spectrum of arterial and venous ulcers. An average of 1·56 (arterial) and 1·48 (venous) different isolates were found per ulcer. Therefore, 40% of arterial and 48% of venous ulcers were classified to be polymicrobial infected (≥2 pathogens per ulcer). According to the bacterial load of the predominant isolate, 56% of the venous and 60% of the arterial wounds had a bacterial load of >10⁵ cfu/ml.

Serological markers, local cytokines and their relations to each other

A good correlation was found between CRP and LBP within the total group (n = 45; r = 0·47 and P = 0·001). In addition, CRP – but not LBP – significantly correlates with local cytokines (with IL‐6, r = 0·68 and P < 0·001; with TNFα, r = 0·3 and P = 0·04) and with the ulcer area (r = 0·30 and P = 0·03). Local IL‐6 and TNFα levels correlated with each other (IL‐6/TNFα, r = 0·62 and P < 0·001), and ulcer areas were positively correlated with levels of cytokines (TNFα, r = 0·7and P < 0·001; IL‐6, r = 0·52 and P < 0·001).

As mentioned above, a good correlation was found between cytokine concentrations in washouts and releases from biopsies. However, IL‐6 and TNFα release from biopsies were nearly eight and two times higher than intraindividual washout concentrations (data not shown).

Serological markers and cytokine patterns in relation to wound microbiology

With regard to bacterial isolates (cn‐ and cp‐Staphylococci, haemolytic Streptococci, Enterobacteriaceae and Pseudomonas spp.), Pseudomonas species (n = 12) seemed to be associated with the highest local IL‐6 levels (Figure 1A). However, differences were significant [IL‐6: 746 (273) versus 266 (76) pg/mL, P = 0·03] when IL‐6 levels of non‐Pseudomonas‐infected ulcers were grouped (n = 31) and compared with Pseudomonas‐infected wounds (Figure 1A). TNFα, serum CRP, LBP, ulcer area and ulcer history did not differ between both groups.

Figure 1.

IL‐6 concentrations in wound washouts seemed to reflect the level of ulcer inflammation with regard to the presence or absence of microbial triggers (data were given as mean ± SE): (A) IL‐6 was significantly increased in ulcers infected with Pseudomonas spp. (n = 12) compared with non‐Pseudomonas strains (*t‐test). However, after analyses of variance, differences in IL‐6 between the distinct bacterial families (cp‐ and cn‐Staphylococci, haemolytic Streptococci, Enterobacteriaceae and Pseudomonas spp.) did not differ significantly. (B) Furthermore, a high microbial bioburden (≥10⁵ colony‐forming units/ml; n = 28) was also significantly related to elevated IL‐6 concentrations (*P = 0·04). (C) When ulcers were classified according to the number of microbial isolates, IL‐6 and TNFα were increased (**P < 0·01 and *P < 0·05, respectively) when two or more bacterial species were isolated (polymicrobial/mixed infection: n = 22). In contrast to local cytokines, levels of CRP and LBP did not discriminate bacterial triggers of ulcer inflammation (Figure 1B and C) (In sterile ulcers, the lowest levels of cytokines were found. Because the group was small (n = 3), cases were excluded from the analyses). cn, coagulase‐negative; cp, coagulase‐positive; CRP, C‐reactive protein; IL‐6, interleukn‐6; LPS, lipopolysaccharide; TNFα, tumour necrosis factor alpha.

Furthermore, local IL‐6 concentrations in mixed infections (≥2 microbial genus per ulcer; n = 22) were more than three times higher (P = 0·005) in comparison with ulcers with monomicrobial infection (n = 20) (Figure 1B). In addition, TNFα levels (P = 0·05) and ulcer areas (7·8 ± 2·2 versus 34·1 ± 11·8 m³, P = 0·035) were also significantly increased, when a polymicrobial flora was diagnosed. Only three ulcers showed a sterile microbial culture, and these cases were excluded from the analyses.

Bacterial load did significantly determine local parameters of ulcer inflammation. If a bioburden of ≥10⁵ cfu/ml (n = 28) was determined compared with 10⁴ cfu/ml or less (n = 14), local IL‐6 concentrations (P = 0·04, Figure 1B) and ulcer area were significantly higher (7·7 ± 3·2 cm³ versus 26·4 ± 8·6 cm³, P = 0·05). TNFα also tended to be increased from ulcers with the highest bioburden but did not reach a statistically significant level (P = 0·07). Ulcer history and systemic parameters of inflammation (CRP and LBP) were similar in ulcers with high or lower bioburden.

Factors predicting local inflammatory profile

Because distinct bacterial triggers could interfere, multifactorial regression models for IL‐6 and TNFα as dependent variables and distinct potential predictors were performed [ulcer area, an arterial pathogenesis, a polymicrobial infection, a high bioburden (≥10⁵ cfu/ml) and isolation of Pseudomonas spp.]. For IL‐6 concentrations, a polymicrobial infection and the ulcer size predict independently (P = 0·043 and P  = 0·04, respectively) and for TNFα, only the ulcer area was predicted (P = 0·006).

Evaluation of sensitivity and specificity for IL‐6 and TNFα for predicting microbial triggers of ulcer inflammation

With regard to IL‐6, significant area under the curves (AUCs) for discrimination of mixed and monomicrobial infections (area 0·75, P = 0·01), for a high bacterial load versus a low bacterial load (area 0·67, P = 0·05) and for a Pseudomonas infection versus an infection with non Pseudomonas strains (area 0·74, P = 0·023) were obtained. For TNFα, only a significant AUC for a discrimination between a mixed and a monomicrobial infection was calculated (area 0·76, P = 0·008).

The sensitivity and specificity of IL‐6 and TNFα for predicting a polymicrobial infection and for IL‐6, predicting a high bacterial load and an infection with Pseudomonas were analysed at designated cut‐off points (Table 1). These analyses were performed only for cytokines because CRP and LBP did not discriminate for the investigated microbial triggers.

Table 1.

Sensitivity and specificity for IL‐6 and TNFα in predicting a mixed infection and for IL‐6 in predicting a high bacterial load or an infection with Pseudomonas spp

	Parameter	Cut‐off points (pg/ml)	Sensitivity (%)	Specificity (%)
Mixed infection	IL‐6	125	70	64
	TNFα	30	74	62
High bacterial load	IL‐6	125	62	57
Infection with Pseudomonas	IL‐6	125	90	57

IL‐6, interleukin‐6; TNFα, tumour necrosis factor alpha.

When a cut‐off value of IL‐6 was selected equal or greater than 125 pg/ml of wound washout, a sensitivity of 70% with a specificity of 64% was achieved for the evidence of a mixed infection. This sensitivity and specificity were similar to 30 pg/ml using as a cut‐off for TNFα (74% and 62%). However, sensitivity and specificity for IL‐6 at a designated cut‐off >125 pg/ml were 90% and 57%, for predicting a Pseudomonas infection and 62% and 57%, for predicting a bacterial load of >10⁵ cfu/ml.

Discussion

Excessive inflammation plays a crucial role in delayed tissue regeneration, but no diagnostic tool is available for assessing the inflammatory status of non healing ulcers in clinical practice. The present data show evidence that levels of IL‐6 analysed in wound fluids reflect a high bacterial load, a polymicrobial infection or an infection with Pseudomonas spp. However, a mixed infection seemed to be the most significant trigger for local IL‐6 concentrations. Furthermore, TNFα and IL‐6 seemed to be a gauge for the extension of the inflamed area because both cytokines were independently predicted by the ulcer size. Focussing on the aim of the study, IL‐6 analysed in wound fluids is a useful diagnostic marker for a sensitive and specific assessment of potential bacterial risk factors of chronic ulcer inflammation.

In colonised or infected non healing ulcers, bacterial endoproducts and exoproducts are potent inducers of the proteolytic and inflammatory cascade. In addition, persistent microbial triggers in chronic ulcers lead to TNFα expression and release paralleled by IL‐6 rested on high levels when compared with the healing state (14). Therefore, analysing these parameters in wound fluids from chronic ulcers seems to be ideal targets as a diagnostic and monitoring tool for the assessment of the inflammatory level. However, distinct methods exist for sampling of wound fluids 15, 16, but most of them were resources and time consuming. In the present study, concentrations of IL‐6 and TNFα were measured in diluted wound washouts according to a modified protocol (12). For intraindividual validation, cytokine concentrations of wound fluids were compared with concentrations of supernatants from standardised punch biopsies and a good correlation was found (Patients and Methods). Consistently with these results, concentrations of IL‐6 and TNFα in superficial washouts seem to reflect tissue cytokine patterns.

As mentioned above, microbial endoproducts and exoproducts are potent triggers of cytokine release, and chronic ulcers are constantly colonised or infected by various bacteria. According to recent studies 17, 18, 19, 27% and 51% of the investigated ulcers were infected with Pseudomonas spp. and Staphylococcus aureus, respectively. Because Pseudomonas– but not S. aureus or other strains – induces proinflammatory cytokines by the highly potent neutrophils derive of heparin‐binding protein (HBP) (20), it is in line that the present investigation found IL‐6 being significantly increased in Pseudomonas‐infected ulcers (Figure 1A). In this context, it has been shown that endocytosis of HBP by monocytes enhances lipopolysaccharide‐induced cytokine production (21).

Besides effects of specific pathogens, another focus of the present study was set on local cytokine levels in relation to monomicrobial or polymicrobial patterns. In contrast to the serum markers CRP and LBP, IL‐6 and TNFα concentrations of wound fluids significantly discriminated mixed infections from monomicrobial even if Pseudomonas was solely isolated. Furthermore, a polymicrobial infection was the most significant and independent determinant for IL‐6 variability in a multiple regression model. These associations underline the important role of microbial synergy, increasing the net pathogenic effect and hence the severity of infection 22, 23. In this context, Bowler and Davis reported a greater diversity of microorganisms in infected than in non infected ulcers and a lower probability of healing if more bacterial groups were present in any ulcer (12). These observations support an earlier view of Kingston and Seal, who argued that all species associated with a microbial disease should be considered potentially synergistic rather than a single species being causative (24).

Besides the role of microbial synergisms and microbial pathogenicity, several studies exist on the role of the bacterial load in delayed healing of ulcers 25, 26, 27, 28. In the present study, semiquantitative analyses of microbial numbers were performed by a modified superficial swab technique with a calibrated eye as for a urine colony count. The present data showed evidence that a bioburden of 10⁴–10⁵ cfu/ml wound fluid was associated with a sevenfold and threefold increase in local IL‐6 and TNFα, respectively, but no differences in CRP or LBP levels were observed. However, a high bioburden (≥10⁵ cfu/ml) was not independently associated with IL‐6 variability when covariates were included. This fact might be because of the patient selection as most patients had large‐scale ulcers with a long history and a high microbial bioburden. In addition, the assessment of bacterial burden in the present study based on a superficial modified swab technique, which may underestimate the bioburden of infected tissue by factor 10–100. Because Heggers and others 5, 28, 29 postulated a critical level of bacteria of 10⁴–10⁵ cfu/g for a significant bioburden, most of our patients might also have high concentrations of bacteria in tissues.

Conclusions

In this pilot study, significant differences of IL‐6 and – in part – TNFα from wound fluids were identified for ulcers, which are at risk for increased inflammation. Although distinct bacterial triggers may interfere with regard to cytokine levels, sensitivity and specificity were high for IL‐6 and TNFα in predicting a mixed infection and promising for IL‐6 in predicting an infection with Pseudomonas or a high bacterial load (Table 1). Serological markers like CRP and LBP seem to be unsuitable for risk stratification of chronic ulcers with regard to bacterial triggers at least in tissue infections without bone involvement. Because cytokines were also determined by the ulcer size, prospective studies are needed to further validate their clinical usefulness in monitoring the therapeutical and healing process.