Biofilms and delayed healing – an in vitro evaluation of silver‐ and iodine‐containing dressings and their effect on bacterial and human cells

Abstract

This study investigated whether there are differences in the ability of wound dressings to modulate certain factors known to affect wound healing. A selection of antimicrobial dressings (AQUACEL ^® Ag Extra^™, AQUACEL ^® Ag+ Extra^™, IODOFLEX ^™, ACTICOAT ^™ 7 and PROMOGRAN PRISMA^™ matrix) were tested for their effect on both bacterial bioburden and human dermal fibroblasts. Some dressings underwent further evaluation for activity against Pseudomonas aeruginosa biofilms using a colony‐drip flow reactor model. The ability of in vitro biofilms to produce proteases, and the effect of PROMOGRAN PRISMA matrix on such proteases, was also investigated. All antimicrobial dressings tested reduced vegetative bacterial load; however, only PROMOGRAN PRISMA matrix was able to significantly reduce biofilm populations (P = 0·01). Additionally, PROMOGRAN PRISMA matrix was the only dressing that did not inhibit dermal fibroblast growth. All other dressings were detrimental to cell viability. In vitro biofilms of Pseudomonas aeruginosa were demonstrated as being capable of releasing bacterial proteases into their surroundings, and incubation with PROMOGRAN PRISMA matrix led to a 77% reduction in activity of such proteases (P = 0·002). The unique ability of PROMOGRAN PRISMA matrix to reduce in vitro vegetative bacteria, biofilm bacteria and bacterial proteases while still allowing dermal fibroblast proliferation may help rebalance the wound environment and reduce the occurrence of infection.

Introduction

Wound healing is a complex and dynamic process consisting of distinct phases. Following injury, haemostasis, inflammation, granulation, epithelisation, contraction and remodelling follow a timely and orderly continuum to achieve wound closure 1. Wounds can progress both forwards and back through the phases depending upon intrinsic and extrinsic forces at work within the patient, and disruption, at any point, may cause the process to be delayed or stop 2. The result is a non‐healing or chronic wound, which may persist, in some instances, for months or years 3. Management of such wounds can therefore become costly, both in terms of patient suffering and health care resource 4.

Prolonged inflammation appears to be a major contributor to wound chronicity, resulting in an enzymatically hostile wound environment 5. Over‐production of proteases, particularly matrix metalloproteinases (MMPs) and serine proteinases, is thought to impair wound healing 5, 6. In vitro studies with fibroblasts isolated from human diabetic skin have shown that these cells secrete abnormally high levels of these enzymes 7, while clinical studies have shown elevated levels of these destructive enzymes in chronic wound fluids compared with acute wound fluids 8.

The presence of microorganisms can also delay the healing of a wound 9. Infection may prolong the inflammatory phase, leading to biochemical imbalance. Thus, microorganisms cause harm and delay healing in both a direct (production of toxins, metabolic wastes, alteration in pH and oxygen levels) and indirect (activation of host neutrophils) manner 10. It is increasingly acknowledged that bacteria in a wound may form biofilms; indeed, chronic wounds provide ideal conditions for attachment and proliferation 11. The wound bed often contains necrotic tissue and debris, aiding bacterial adherence, whilst exudate provides nutrients for bacteria to grow 12. Additionally, chronic wounds are often associated with an impaired host immune response, increasing susceptibility to infection 12. A study by James et al. using microscopy techniques reported that 60% of chronic wound specimens contained biofilm, compared to only 6% of acute wound samples examined 11. Researchers using similar techniques reported evidence of biofilm in 47–59% of chronic wounds tested, correlating well with James' data 13, 14.

Wound biofilm is often discussed in the context of wound infection. However, if biofilm is present in up to 60% of chronic wounds as suggested 11, 13, 14, it is clear that, in many cases, the presence of biofilm impacts wound healing without leading to active infection 15. It is also likely that biofilm may be present in some wounds that go on to heal without intervention. This suggests that a balance exists between host response and bacterial action, with the presence of biofilm in a wound shifting the equilibrium in favour of the microbe. If biofilm bioburden is controlled (although not eradicated) by the host environment, then infection may not ensue, and in some instances, healing may proceed, albeit more slowly 15. In other cases, the balance of the wound may need to be restored before healing can begin. For these wounds, with biofilm‐delayed healing, there may be several factors contributing to the wound imbalance. It is likely that bacteria present in the biofilm are impacting the wound environment in a similar way they do in an infected wound but to a lesser extent, with the presence of biofilm leading to an elevated host response. Additionally, bacteria may produce proteases that promote chronic local inflammation and degrade tissue 16.

When discussed in these terms, a wound with biofilm‐delayed healing appears to share many characteristics with a critically colonised wound. The term ‘critically colonised’ has been used to describe wounds that fail to heal even with low numbers of planktonic bacteria (≤10⁵ colony‐forming units, CFUs) and in the absence of obvious indicators of inflammation or infection 17. It has been hypothesised that the term critical colonisation may actually be describing wounds that have biofilms 18, a sensible assumption considering the ubiquitous nature of biofilms.

Currently, strategies for reducing biofilm in wounds have focussed on management of infection, using antimicrobial wound dressings with or without systemic antimicrobial treatment 19. Consequently, such dressings are designed to contain high levels of antimicrobial agents, which are bactericidal but may also be inherently cytotoxic to host dermal cells. However, the use of such dressings on wounds that have biofilm‐delayed healing rather than overt infection may actually negatively impact the wound environment; the bactericidal agents may cause further damage to host tissue, impacting wound healing. We propose that dressings containing antimicrobial agents at levels not detrimental to host cells, which effectively reduce biofilm bioburden and also have the ability to rebalance the local wound environment, may provide the maximum benefit to such stalled chronic wounds.

PROMOGRAN PRISMA™ matrix may have potential as such a therapy. It is comprised of collagen and oxidised regenerated cellulose dressing (ORC/Collagen), which has been shown in in vitro studies to significantly reduce levels of key proteases associated with hard to heal wounds, including diabetic foot ulcers 20, 21, 22. In addition, the inclusion of silver gives the material broad‐spectrum antimicrobial properties 23.

Materials and methods

Dressings

Dressings used in this study were nanocrystalline silver dressing (ACTICOAT 7; Smith & Nephew, Hull, UK); silver‐containing hydrofibre dressing (AQUACEL Ag Extra; ConvaTec, Deeside, UK); Silver‐containing hydrofibre dressing with enhanced antimicrobial efficacy (AQUACEL Ag+Extra; ConvaTec); Cadexomer iodine dressing (IODOFLEX; Smith & Nephew), collagen/ORC dressing with silver (PROMOGRAN PRISMA™; Systagenix, Gargrave, UK), Gauze (TOPPER 8; Systagenix, Gargrave, UK).

Experimental design

In this study, we have compared collagen/ORC with silver [PROMOGRAN PRISMA™, subsequently referred to as PRISMA matrix] to several commercially available antimicrobial wound dressings with previously published data against in vitro biofilms 24, 25, 26, 27. For each dressing, antimicrobial activity against vegetative bacteria and effects on human dermal fibroblasts was determined. Three dressings, PRISMA matrix, AQUACEL Ag+ Extra and IODOFLEX, underwent further evaluation to determine their efficacy against in vitro biofilms of P. aeruginosa using a colony‐drip flow reactor wound model (C‐DFR). Additionally, we demonstrated bacterial protease production by in vitro P. aeruginosa biofilms and investigated the ability of PRISMA matrix to modulate Bacterial Protease Activity (BPA)

Microorganisms and reagents

Pseudomonas aeruginosa ATCC 27312 and ATCC 15442 and Staphylococcus aureus ATCC 6538 were obtained from the ATCC national culture collection (LGC Standards, Middlesex, UK). Strains were grown on Tryptone Soy Agar (TSA) (Cherwell Laboratories, Bicester, UK) at 37°C. Simulated Wound Fluid (SWF) consisted of maximum recovery diluent (Oxoid, Altrincham, UK) supplemented with 10 g/l Bovine Serum Albumin (Sigma, Dorset, UK) and 25% HI‐Foetal Bovine Serum (HI‐FBS) (Life Technologies, Warrington, UK) (aseptically added post autoclaving). Microorganisms were cultured on Tryptone Soya Agar (TSA) (Cherwell Laboratories, Biscester, UK) and in Tryptone Soya broth (TSB) (Oxoid, Hampshire, UK) unless stated otherwise.

Antimicrobial efficacy – vegetative cells

Shake flask assay

The ability of antimicrobial wound dressings to reduce populations of P. aeruginosa ATCC 27312 and S. aureus ATCC 6538 was evaluated by shake flask log reduction assay in separate experiments. Briefly, bacteria were grown overnight and used to prepare an approximately 10⁶ CFU/ml bacterial suspension in 0·1% Bacto^™peptone (BD Biosciences, Oxford, UK). This initial suspension was serially diluted and enumerated to give bacterial counts prior to dressing exposure. Ten millilitre volumes of this suspension were added to flasks containing aseptically prepared 2·5‐cm² samples of each test dressing. All dressings were evaluated in triplicate. Flasks were immediately placed in a shaking incubator (37°C, 150 rpm). At relevant time points (0, 1, 3 and 6 hours), a 1‐ml sample was removed from each flask, serially diluted and inoculated in triplicate onto TSA to enumerate surviving bacteria. For samples containing antimicrobial, Dey–Engley neutralising media (Southern Group Laboratories, Northamptonshire, UK) was utilised. Inoculated agar plates were incubated at 37 °C for 24 hours, and colonies were counted at the appropriate dilution. Log₁₀ values were calculated from mean colony counts and log₁₀ reductions determined.

Cell proliferation

Culture of human dermal fibroblasts

Human adult dermal fibroblasts, isolated from a male donor (CRL‐2522, ATCC), were grown and maintained in Dulbecco's Modified Eagle's Medium (DMEM) (Life technologies, Warrington, UK) containing 10% (v/v) foetal bovine serum [Life technologies, Warrington, UK], 100 units/ml penicillin, 100 µg/ml streptomycin and 0·25 µg/ml amphotericin B (antibiotic/antimycotic solution, Life technologies, Warrington, UK). These cells were routinely sub‐cultured and used for experimental testing between the passages 3 and 10.

Cell proliferation assay

Human adult dermal fibroblasts were harvested at 95 % confluence and re‐seeded in DMEM containing 10% FBS at a cell density of 2·5 × 10³ cells/100 µl/well in a 96 well microtitre plate. The cells were allowed to adhere and spread to the plate surface for 24 hours in a humidified incubator at 37°C, 5% CO₂. The media was then removed by aspiration and the cell monolayer washed with serum‐free DMEM. Dressing extracts, prepared by incubating 6‐mm punch biopsies of each dressing with 0·5 ml of serum‐free DMEM overnight at 37°C, 5% CO₂, were then added to the cell monolayer (100 µl per well). Six extracts of each dressing were prepared and tested in duplicate. DMEM containing 10% FBS, which provides ideal conditions for cell proliferation, was included as a positive control, while serum‐free diluent alone was tested to determine basal cell growth. All samples were incubated with the cells for 72 hours at 37°C, 5% CO₂. After this incubation period, the test media were removed through aspiration and replaced with serum‐free DMEM. The labelling solution from a commercially available proliferation kit (XTT, cell proliferation kit II; Sigma) was added to this. Once the labelling solution was added, an initial absorbance reading was obtained at 450 nm, after which the microtitre plate was incubated at 37°C, 5% CO₂ and the absorbance monitored over 5 hours. Data from three independent experiments was combined, and results were analysed for statistical significance using a paired student's t‐test.

Digital images

Before the addition of a labelling reagent from the commercially available proliferation kit, digital images of the cell monolayer were obtained. A Nikon eclipse TE2000‐U microscope (Nikon Ltd, Kingston upon Thames, UK) was used to capture the pictures, which were then analysed using Lucia G on DMX1200 version 4·81 software (Nikon Ltd).

Antimicrobial efficacy – biofilm populations

Colony‐drip flow reactor

The ability of three dressings, AQUACEL Ag+Extra, IODOFLEX and PRISMA matrix, to reduce biofilm populations was investigated using a C‐DFR based on that described previously by Lipp et al. 28 To prepare the reactor apparatus, 25‐mm² absorbent pads (Millipore, Consett, UK) were glued with silicon‐based aquarium sealant to clean glass microscope slides and placed in the channels of the C‐DFR (Biosurface Technology, Bozeman, MT). The entire set‐up was autoclaved and maintained sterile until use. A non‐antimicrobial dressing (gauze) was included as a control in each experiment.

Experiments began by hydrating the absorbent pads with 0·5 ml of SWF, and then, 0·22‐µm porous polycarbonate membranes (Sigma) were placed on these absorbent pads. Next, the membranes were inoculated with 10 µl of a TSB‐diluted overnight culture (0·5 McFarland standard suspension). The system was left undisturbed for 30 minutes, while the inoculum was allowed to dry. The reactor was then attached to a medium reservoir, and SWF was pumped through the system at 5 ml/hour/channel. This reactor and set‐up allowed the medium to drip down the microscope slide and absorb into the pad, which then supplied nutrients to the bacteria growing on the top side of membrane. The bacteria were then allowed to grow for 24 or 72 hours at room temperature (average temperature of 22·5°C, range 20·1–24·7°C). Iodoflex was tested against 72‐hour‐old biofilms only.

After the growth period, one biofilm/membrane per model was subjected to plate counting (see below) to enumerate the biofilm population's pre‐antimicrobial exposure. For each of the other channels, a sterile dressing sample was placed directly on top of the biofilm/membrane. Dressings were moistened with SWF to simulate clinical usage. The assay continued for a further 24 hours (flow rate 5 ml/hour/channel) before the dressings were removed and the biofilm/membranes examined with plate counts to enumerate remaining biofilm after antimicrobial exposure. Samples of biofilm/membrane pre‐antimicrobial exposure were also subjected to scanning electron microscopy (SEM).

Plate counting – enumeration of biofilm

After removal from the C‐DFR, biofilm/membranes were rinsed thrice with sterile PBS to remove any adherent vegetative cells. Samples were added to Dey–Engley neutralising broth to negate any residual antimicrobial effect resulting from dressing contact. The samples were then subjected to 3 minutes of high‐speed vortexing. Serial 10‐fold dilutions were made using sterile DPBS, and the dilutions were plated on TSA. After 24 hours of incubation at 37 °C, the plates were counted, and the number of colony‐forming units (CFU) per membrane was calculated. For each antimicrobial dressing, a minimum of five replicates from three independent experiments were evaluated. Mean averages and standard deviations (SD) were then generated. For two‐group comparisons, statistical analysis for significance was determined using a two‐tailed t‐test assuming equal variance.

Scanning electron microscopy

High‐resolution SEM analysis was used to confirm the presence of mature biofilm on untreated membrane filters after the growth period. The membrane was placed (biofilm side‐up) on an absorbent pad saturated with 2% glutaraldehyde (Sigma). The samples were fixed overnight at room temperature and then washed in phosphate buffer. Membrane filters were removed and placed on a stainless steel mesh, post‐fixed in 1% osmium tetroxide for 1 hour, then washed in phosphate buffer. The samples were then dehydrated in an ethanol series consisting of 25%, 50%, 70% and 90% ethanol for 30 minutes each, followed by three changes of 100% ethanol for an hour each and then HMDS (Sigma) for 30 minutes. Samples were dried overnight in a desiccator. The membrane filters were mounted on stubs with carbon sticky tabs and coated with around 7 nm of gold‐palladium before viewing in the JEOL JSM‐6490LV microscope (JEOL USA, Peabody, MA).

Production of bacterial proteases by biofilms in vitro

Biofilms of P. aeruginosa ATCC 27312 were grown on 6‐mm cellulose discs (BD, Oxford, UK). Briefly, stationary phase cultures of P. aeruginosa were adjusted to 0·5 McFarland (c. 1·5 × 10⁸ CFU/mL) and used to inoculate sterile discs housed in a 24‐well microtitre plate. A total of 500 microlitres of SWF were added to each well, and samples were incubated for 48 hours at 37°C with 80 rpm shaking to allow biofilm formation. SEM has previously confirmed the presence of mature biofilms following preparation in this manner 29.

After incubation, biofilm discs were rinsed thrice with PBS to remove adherent vegetative cells. Biofilm discs were then transferred to new wells containing 1 ml of TSB supplemented with 300 µg/ml gentamycin (Sigma). This gentamycin concentration was 20 times the Minimum Inhibitory Concentration (MIC) for this strain of P. aeruginosa (MIC previously determined to be 15·6 µg/ml by broth microdilution in TSB, data not shown). Biofilm discs were re‐incubated for 24 hours at 37°C with 80 rpm to allow protease production to occur whilst restricting the growth of vegetative cells. After incubation, biofilm supernatants were removed from the wells, and biofilm discs were transferred to fresh gentamycin‐containing media, before being incubated again as described above. After this final incubation, supernatants were again collected. All supernatants were filter sterilised through a 0·22‐µm filter to remove any cell debris.

Proteolytic activity of biofilm supernatants was determined by fluorometric casein analysis. Briefly, 10 µl of a test sample was added to 190 µl casein assay buffer (25 mM Tris, 150 mM Sodium Chloride, pH 7·2) containing casein substrate (23267; Thermo, Warrington, UK) at 25 µg/ml. A standard curve of endoproteinase Glu‐C from Staph V8 (Merck, Hertfordshire, UK) was also included to allow comparison between assays. Plates were read immediately (Fluorometer settings; Excitation: 485 nm, bandwidth 12. Emission: 525 nm, bandwidth 12. Kinetic scan: 30 minutes, interval 1 minute. Shaking: 5 seconds before each reading). Enumeration of biofilm populations on the discs was also performed as described previously.

Collagen ORC with silver‐additional assessment

Evaluation of ability of PRISMA matrix to reduce BPA in vitro

Bacterial supernatants with BPA were prepared from overnight cultures of P. aeruginosa ATCC 15442. Cultures were centrifuged (4500 g, 15 minutes) and the supernatant filter sterilised. To evaluate the ability of PRISMA matrix to reduce BPA, triplicate samples of each dressing were pre‐wet by submerging in DPBS for 10 seconds, before being incubated in P. aeruginosa supernatant (100 µl, 37°C, 18 hours). Supernatant alone with no dressing was also included as a positive control. After incubation, the remaining proteolytic activity of supernatants was determined by fluorometric casein analysis as described previously. Data from three independent experiments was combined, and percentage reduction was compared to the positive control calculated. Statistical analysis for significance was determined using a two‐tailed t‐test.

Results

Effect on cell proliferation

Human adult dermal fibroblasts exposed to media and exposed to PRISMA matrix demonstrated increased proliferation, evidenced by significantly greater absorbance levels than the negative control of serum‐free DMEM (P = 0·0003) (see Figure 1). This observation was supported by microscopic observations of cellular morphology (see Figure 2A and E). Fibroblasts were observed to be confluent, have good adherence and to be bipolar.

Figure 1.

Effect of antimicrobial dressing exposure on cellular proliferation of human dermal fibroblasts in vitro. DMEM containing 10% FBS, which provides ideal conditions for cell proliferation, was included as a positive control, while serum‐free media alone was included as a negative control to determine basal cell growth. The dotted line indicates the threshold between cell proliferation (above the line) and decrease in cell viability (below the line). PRISMA matrix demonstrated a proliferative effect on the cells compared to the negative control. AQUACEL Ag+Extra, AQUACEL Ag Extra, ACTICOAT 7 and IODOFLEX were all shown to have a cytotoxic effect on cells, resulting in a reduction in viability compared to the negative control.

Figure 2.

Human dermal fibroblasts exposed to antimicrobial dressings were visually examined for cell viability and morphological changes. Microscopic observations demonstrated cellular damage or mortality after exposure to AQUACEL Ag Extra, AQUACEL Ag+Extra and ACTICOAT 7 (B–D) (IODOFLEX was not examined in this assay). Cells were condensed with poorly defined edges, and many were free floating within the media. Exposure to PRISMA matrix (E) did not appear to affect the normal cell morphology. Under these conditions, fibroblast morphology was very similar to that of the serum‐free DMEM control (A). Cells were adherent and bipolar.

Dressing extracts of ACTICOAT 7, AQUACEL Ag Extra, AQUACEL Ag+Extra and IODOFLEX all had a negative effect on human dermal fibroblasts, with no proliferative effects demonstrated. Each of these dressings produced a significantly lower absorbance than the serum‐free DMEM control (P < 0·001 for all of the individual dressings). These observations were supported by microscopic observations of cell morphology; cells appeared spherical and condensed, with poor adherence (see Figures 2 B–D).

Antimicrobial efficacy – vegetative bacteria

PRISMA matrix, ACTICOAT 7, AQUACEL Ag Extra, AQUACEL Ag+Extra and IODOFLEX all demonstrated bactericidal activity (>4·00 log₁₀ unit reduction) against both Staphylococcus aureus and Pseudomonas aeruginosa vegetative cultures within 3 hours (Figures 3A and B) when compared to an unexposed control.

Figure 3.

(A) Effect of antimicrobial dressing exposure on in vitro bacterial populations in the shake flask assay. Log₁₀ values were calculated from mean colony counts, and log₁₀ reductions were calculated compared to a control unexposed to antimicrobial. PRISMA matrix, ACTICOAT 7, AQUACEL Ag Extra, AQUACEL Ag+Extra and IODOFLEX all demonstrated bactericidal activity (>4·00 log₁₀ unit reduction) against Pseudomonas aeruginosa vegetative cultures within 3 hours.

Antimicrobial efficacy – biofilm bacteria

C‐DFR biofilm maturity

There was no difference between control biofilm TVC at the beginning and end of the exposure period (Figure 4A), with 72‐hour‐old biofilm populations appearing stable. In contrast, average populations of 24‐hour‐old biofilms prior to dressing application were 2·49 log₁₀ units lower than those left to mature for 72 hours and continued to expand after the control dressing (gauze) was applied (Figure 4B). SEM analysis of 72‐hour‐old modelled biofilms revealed complex bacterial biofilm microstructures along with the presence of extracellular polymeric substance (EPS) indicative of stable, mature biofilms (Figure 5). It was noted that data generated using 24‐hour‐old biofilms showed increased variability between independent experiments compared to data from older biofilms.

Figure 4.

(A) Effect of 24 hours of AQUACEL Ag+Extra dressing, IODOFLEX dressing or PRISMA matrix exposure on 72‐hour‐old pre‐formed Pseudomonas aeruginosa biofilms in the C‐DFR. PRISMA matrix reduced biofilm populations by 1·49 log₁₀ units compared to pre‐exposure levels (P = 0·01). AQUACEL Ag+Extra dressing and IODOFLEX dressing exposure resulted in a reduction in biofilms population of 0·50 and 0·82 log₁₀ units, respectively, during the same period. (B) Effect of 24 hours of AQUACEL Ag+Extra dressing or PRISMA matrix exposure on 24‐hour‐old pre‐formed Pseudomonas aeruginosa biofilms in the C‐DFR. Both AQUACEL Ag+Extra dressing and PRISMA matrix inhibited the growth of the biofilm during the exposure period, compared to a gauze control dressing.

Figure 5.

SEM images of Pseudomonas aeruginosa biofilm on membrane substrate after 72 hours in the C‐DFR (pre‐exposure populations).

Effect of dressing application on Pseudomonas biofilms

The 24‐hour continuous exposure of AQUACEL Ag+Extra or PRISMA matrix to 24‐hour‐old biofilms in the C‐DFR resulted in biofilm populations of 6·62 and 6·47 log₁₀ units, respectively, at the end of the exposure period. These values were 2·62 and 2·77 log₁₀ units lower than those of the biofilm that had received a gauze control (P = 0·03, Table 1). However, when values are compared to the control biofilm prior to dressing exposure (6·73 log₁₀ units), it is clear that these figures represent an inhibition of growth as opposed to a true reduction in biofilm population, with an overall reduction from pre‐exposure biofilm populations of ≤0·26 log₁₀ units after exposure to AQUACEL Ag+Extra or PRISMA matrix (Figure 4B, Table 1).

Table 1.

Summary of antimicrobial efficacy after 24 hours of continuous exposure to dressings on Pseudomonas biofilms

	Mean log CFU/ml (±SD)	95% CI log cfu/ml	Change in mean log CFU/ml compared to pre‐exposure	Change in mean log CFU/ml compared to gauze control	Significance (P value)
24‐hour‐old biofilms
Biofilm pre‐exposure	6·73	5·73–7·73	—	—	—
Gauze	9·24	9·09–9·39	+2·51	—	0·02*
Aquacel Ag+Extra	6·62	4·90–8·34	−0·11	−2·62	NS* 0·03†
PRISMA matrix	9·47	4·64–8·30	−0·26	−2·77	NS* 0·03†
72‐hour‐old biofilms
Biofilm pre‐exposure	9·30	8·62–9·98	—	—	—
Gauze	9·20	8·69–9·70	−0·10	—	NS*
Aquacel Ag+Extra	8·80	8·25–9·35	−0·50	−0·40	NS*, †
PRISMA matrix	7·81	7·29–8·34	−1·49	−1·39	0·01* 0·002†
Iodoflex	8·48	7·71–9·22	−0·82	−0·72	NS* 0·02†

NS, not significant.

^*Compared to pre‐exposure.

^†Compared to gauze.

When the assay was performed using 72‐hour‐old biofilms, the application of a gauze control dressing for 24 hours had minimal impact on biofilm population (9·30 versus 9·20 log₁₀ units) (Figure 4A, Table 1). In contrast, 24‐hour continuous exposure to PRISMA matrix resulted in a 1·49 log₁₀ unit reduction in biofilm population, a significant reduction compared to both pre‐exposure levels (P = 0·01) and AQUACEL Ag+extra (P = 0·004), exposure to which reduced biofilm TVC by only 0·50 log₁₀ units during the same time period (not significant). IODOFLEX application reduced P. aeruginosa populations by 0·83 log₁₀ units, which was not a significant reduction compared to pre‐exposure counts.

BPA production by in vitro Pseudomonas biofilms and the ability of PRISMA matrix to reduce BPA

P. aeruginosa biofilms contained approximately 10⁹ CFU/disc prior to exposure to gentamycin. After the first incubation with gentamycin, approximately 10⁶ viable bacteria remained; this decreased to 10⁴ bacteria after the second incubation. In both instances, the supernatant surrounding the biofilm discs remained clear, indicating that growth of vegetative bacteria was inhibited. Caseinolytic analysis of this supernatant confirmed bacterial protease activity (BPA) in both instances (Figure 6). When P. aeruginosa supernatant, characterised as demonstrating high‐level BPA, was exposed to PRISMA matrix, a significant reduction in activity (77%, P = 0·002) compared to an unexposed control sample was observed (Figure 7).

Figure 6.

BPA of supernatants surrounding Pseudomonas biofilms after first and second incubations in gentamicin‐containing media, detected by fluorometric casein assay. Protease activity was detected after both incubation steps.

Figure 7.

Effect of incubation with PRISMA matrix on BPA of Pseudomonas aeruginosa supernatant. Incubation with PRISMA matrix resulted in a 77% decrease in BPA compared to supernatant, which had no dressing added (P = 0·002).

Discussion

Antimicrobial wound dressings commonly contain either silver or iodine, both of which are well characterised as broad‐spectrum antiseptic agents 30. Wound dressings containing antiseptics are considered particularly useful because of their effectiveness against antibiotic‐resistant microorganisms, such as methicillin‐resistant S. aureus (MRSA) 31. Antibiotic resistance has a genetic basis, occurring through mutation and transfer/acquisition of resistance genes 32. However, bacteria may also display phenotypic resistance or ‘tolerance’ to antimicrobial therapy when present as a biofilm. With biofilm‐related antimicrobial tolerance, efficacy of both antibiotics and antiseptics such as silver and iodine may be impacted 33.

Currently, strategies for reducing biofilm in wounds have focused on the management of infection, using antimicrobial wound dressings with or without systemic antimicrobial treatment and debridement 34. Consequently, such dressings are designed to contain high levels of antimicrobial agents, which are bactericidal but may also be inherently cytotoxic to host dermal cells. The prolonged use of such dressings on wounds that have suspected biofilm rather than overt infection, however, may negatively impact the wound environment, causing further damage to host tissue and further delaying wound healing, particularly as the relationship between the presence of biofilm and delayed healing in wounds is not yet clearly understood.

Here, we have characterised the in vitro cytotoxic and antimicrobial properties of several dressings containing either silver or iodine. All the dressings selected for this study have previously demonstrated activity against biofilms in vitro. 24, 25, 26, 27, 35, 36 In our study, all test dressings were bactericidal against both the Gram‐positive and Gram‐negative vegetative test microorganisms in vitro. It has been suggested that silver levels that are cytotoxic to bacteria are also cytotoxic to human cells 37, and this was the case for three of the four silver dressings evaluated here as well as the iodine‐containing dressing (ACTICOAT 7, AQUACEL Ag Extra, AQUACEL Ag+Extra, IODOFLEX), all of which were found to have a detrimental impact on human dermal fibroblasts. In contrast, PRISMA Matrix demonstrated high‐level antimicrobial efficacy with no negative impact on host cells and, in fact, allowed host cell proliferation. These results suggest that silver can be present at levels sufficient to manage bioburden without damaging the fibroblast monolayer; however, excess silver can have detrimental effects.

Three dressings, PRISMA matrix, IODOFLEX and AQUACEL Ag+Extra, underwent further evaluation for activity against Pseudomonas aeruginosa biofilms in the C‐DFR biofilm model. Both silver‐containing dressings effectively inhibited the expansion of 24‐hour‐old biofilms compared to a gauze control, but this did not equate to an actual reduction in populations compared to pre‐exposure levels. It was noted that there was variability in dressing efficacy between experiments; we suggest that this may be because the biofilm has not yet reached a mature steady state, as evidenced by continued growth of the control during the exposure period. It is possible that the physiologically heterogeneous immature biofilms are influenced by minor variations in local conditions, exhibiting variable responses to antimicrobial exposure until steady state is reached. When biofilms were allowed to mature for 72 hours before dressing exposure, biofilms had reached steady state, with pre‐ and post‐exposure control populations stable and the presence of mature biofilm microstructures confirmed by SEM analysis. Of the three dressings evaluated against 72‐hour biofilms, only the PRISMA matrix significantly reduced biofilm populations compared to pre‐exposure populations (1·49 log₁₀ unit reduction, P = 0·01). AQUACEL Ag+Extra and IODOFLEX achieved only 0·50 and 0·82 log₁₀ unit reductions, respectively, which were not significant by statistical analysis.

The C‐DFR model utilised in this study is designed to reflect conditions in a highly exudative wound environment. It is very challenging due to the constant flow of proteinaceous media across the surface of the biofilm, which allows robust biofilm formation while also potentially washing away or inactivating antimicrobials 38. The model was originally described by Lipp et al. 28 and was used to evaluate the biofilm prevention properties of wound dressings; the antimicrobial exposure period began immediately after the substrate was inoculated with vegetative cells. The authors reported that the antimicrobial dressings tested were minimally effective, with increased biofilm TVC occurring during the exposure period, albeit with less growth than under the control dressing 28. Here, we report, for the first time to our knowledge, the use of this model to measure the disruption of mature, pre‐formed P. aeruginosa biofilms.

Whilst chronic wounds are likely to be colonised with multiple bacterial species, the presence of P. aeruginosa has particularly been associated with delayed healing and increased wound area 39, 40. Additionally, P. aeruginosa can be considered a model organism for the study of bacterial mechanisms contributing to bacterial persistence 41. The P. aeruginosa strain utilised in this study was an ATCC isolate originally recovered from an infected wound. When interpreting data from this model, it is important to consider both the differences in behaviour of culture isolates versus those obtained clinically and that interactions between different bacterial populations that may occur in a mixed biofilm culture are not represented in a single species reactor.

The traditionally accepted benchmark for bactericidal activity of antimicrobial dressings when evaluated in vitro is a reduction in bacterial population by >3 log₁₀ units. However, the assays used to measure such activity are almost universally ‘static culture’ models, whereby exposure to test material occurs in a small, fixed volume of media, meaning that released active antimicrobials accumulate in the system. This contrasts markedly with the dynamic set‐up of the C‐DFR model described above. With this as context, the reduction in the Pseudomonas biofilm population of 1·49 log₁₀ units achieved by application of PRISMA matrix to 72‐hour pre‐formed biofilms in this study may represent substantial in vitro anti‐biofilm efficacy. The significantly better performance of PRISMA matrix compared to AQUACEL Ag+Extra was particularly interesting as PRISMA matrix contains a much lower concentration of silver, 0·25% w/w (2500 ppm) versus 1·2% w/w (12 000 ppm).

In addition to recalcitrance to antimicrobial therapy, biofilms in wounds may contribute to wound imbalance and chronicity through the production of proteases. Protease production has been associated with a variety of bacteria known to colonise wounds 16; however, only a few studies have focused on protease production by biofilms 42, 43. In our study, proteases were detected from supernatant surrounding in vitro Pseudomonas biofilms grown in the presence of gentamicin to restrict the growth of vegetative bacteria and kill any biofilm bacteria not exhibiting an antibiotic‐tolerant biofilm phenotype. The detection of BPA after the second incubation in antibiotic‐containing media confirmed that the detected activity was likely to be a result of production by bacteria present in the biofilm rather than the release of intracellular proteases through cell lysis upon initial gentamicin exposure. The production of such proteases by biofilm bacteria in a wound may further contribute to wound chronicity, causing tissue damage and inflammation. In our in vitro assay, incubation with PRISMA matrix significantly reduced BPA (77% reduction, P = 0·002). Collagen/ORC is already known to reduce host proteases, including MMP‐9 and human neutrophil elastase (HNE) in vitro; 20, 21, 22 however, this is the first demonstration of such an effect on bacterial proteases in supernatant.

In summary, in this study, we demonstrated that PRISMA Matrix was unique in its ability to reduce in vitro bacterial bioburden while still allowing dermal fibroblast proliferation. ACTICOAT 7, IODOFLEX, AQUACEL Ag Extra and AQUACEL Ag+Extra were detrimental to the in vitro fibroblast monolayer. It is important to note that in wounds presenting with overt signs of infection, any one of the dressings evaluated in this study could be appropriately utilised; when active infection is present, healing cannot occur, so any potential cytotoxic effect is negated by the need to reduce wound bioburden. Additionally, despite in vitro evidence that higher concentrations of antiseptics are cytotoxic to human fibroblasts 37, there is little evidence of a corresponding effect when such antiseptics are used clinically 44. The potential for cytotoxicity during use of such dressings is, however, still acknowledged by practitioners; a recently published expert working group consensus document recommended that silver dressings be used in the context of a 2‐week challenge, with regular review of treatment strategies to ensure that they remain conducive to wound healing 45. Clinical diagnosis of infection in chronic wounds is not always clear and is complicated by the increasing evidence that the presence of biofilm may delay healing in the absence of many of the normal clinical signs of infection. In these cases, prolonged use of wound dressing containing high levels of antimicrobial may be more harmful than beneficial, eliminating not only bacteria but also fibroblasts and keratinocytes 46.

When evaluated in vitro, PRISMA matrix contains enough silver to effectively reduce populations of important wound pathogens without any negative effect on host cells, such as dermal fibroblasts. Additionally, PRISMA Matrix significantly reduced populations of in vitro Pseudomonas biofilms and reduced the activity of bacterial proteases produced by bacteria in such biofilms. This dressing could, therefore, help rebalance both the bacterial load and any proteolytic imbalance, reducing the occurrence of infection. This in vitro work suggests that low silver concentrations may also allow this dressing to be used on many indicated wound types, minimising the risk of causing indiscriminate damage to the cells of the healing wound.