An investigation into the effects of prosthesis relevant surfaces on the formation of Staphylococcus aureus biofilms

The aim of this study was to investigate the effects prosthesis relevant surfaces and finishes have on the formation of Staphylococcus aureus (S. aureus) biofilms, the leading causative pathogenic bacteria of periprosthetic joint infection (PJI). Microbiological biofilm analysis was conducted to quantify S. aureus biofilm growth on prosthesis relevant surface coatings and finishes. Through the use of a CDC Biofilm Reactor (CDC-BR), biofilms were grown under constant shear conditions on three different titanium surface finishes, including Plasma Spray (PS), Grit Blasted (GB) and an As-fabricated (AF) surface used for comparative and control purposes. Advanced metrological techniques were applied to characterise each surface. This advanced approach to surface characterisation, including functional volume parameters has been carried out to provide a detailed quantifiable description of the surface and one that better relates to the nature of growth upon a surface. Results show that the PS surface exhibits a significant increase in biofilm growth in comparison to the GB and AF surfaces, with the AF surface showing the lowest amount of biofilm growth. Additionally, the morphology of the features of the PS coating allows for the biofilm accumulation to flourish in the re-entrant features across its topography. These findings highlight the difficulties of biofilm eradication and further complicate the design process of prosthetics, where features implemented to promote osseointegration simultaneously offer favourable locations for bacterial cell attachment and subsequent biofilm development, leading potentially, to PJI. There is a general consensus throughout literature regarding an apparent trend between an increase in Sa and an increase in biofilm formation (Zheng et al 2021 Frontiers in Bioengineering and Biotechnology 9 643722; Bridgens et al 2015 Surface and Coatings Technology 284; James et al 2019 Aesthetic Plast. Surg. 43 490–497). This study has shown that whilst this may generally be the case, the location of this surface area increase within the topography may be a more important observation.


Introduction
The requirement for total arthroplasty procedures, in particular total hip and knee arthroplasties (THA and TKA respectively) is predicted to reach four million by the year 2030.According to data published within the National Joint Registry 17th annual report (2020), within the UK, over 208,319 arthroplasty procedures were conducted in 2019.This figure reduced to 110,621 in 2020 due to the COVID 19 pandemic, this data has been omitted since it does not portray a true representation of the annual procedures undertaken.A breakdown of the arthroplasty procedures carried out in the UK within 2019 can be found in table 1 below, along with the volume of revision procedures and those due to infection.Data was extracted from patient characteristic information available from the NJR [1], it should be noted that the percentages representing infection related revision (table 1) may not be the sole indicator for the revision procedure.During 2018 over 1.3 million THA and TKA surgeries were performed within the US and are associated with some of the highest costs of any surgeries carried out [2].
The incidence of infection is relatively low for THA and TKA procedures, being 1% and 2% respectively, with an ever-increasing populous requiring total arthroplasty procedures, it poses an ever-increasing threat to the patient and health care industry alike [3].

Periprosthetic joint infection (PJI)
The complications associated with PJI are numerous and grievous, having devastating effects on the patient's morbidity and quality of life.The burdens connected with PJI placed on the over-encompassing health service both fiscally and in terms of treatment and rehabilitation is significant [4].Within the UK the average cost of THA and TKA is £6000 and £6500 respectively [5].The costs associated with these surgeries performed in the US is significantly higher with reported average costs of $30,000 and $31,000 for THA and TKA procedures, respectively [6,7] If PJI related revision surgeries are required, this cost is significantly increased, Levack et al reported a five-fold increase on the initial surgery, resulting in a cost of revision surgery equating to approximately £30,000 within the UK [8].In 2017 the approximate total burden to the UK NHS, for primary and hospital care concerning THR and TKR was £2B [9].This further emphasises the impact of arthroplasty procedures and the requirement to reduce any possible revision events occurring.Infection is the second highest indication for a TKR revision procedure behind aseptic loosening.For THR procedures, infection remains the third highest indication for a revision procedure behind aseptic loosening and periprosthetic fracture [1].Infection related revision is associated with one of the highest costs for revision procedures with an approximate cost three times that of aseptic revision [10].
The implantation of a foreign body within the human environment results in a dramatic increase in the susceptibility of infection of the juxtaposed tissue.Research performed by Kapadia et al shows that in the presence of a foreign body, the bacterial load required to initiate an infection is reduced by an approximate 100,000 times [3].This significant reduction is corroborated by Southwood et al who showed that the concentration of S. aureus required to establish an infection reduced from 10 4 to 10 2 Colony Forming Units (CFUs) when a foreign body is present [11,12].Many factors influence the pathogenesis of an infection, with early cases postulated to occur through the implantation phase of the arthroplasty, from both exogenous and endogenous sources.The most common causative factors for early onset infection are accredited to contaminants, both within the surgical environment and those present within the skin microbiota [13].Early onset infection is classified as less than 3 months post-surgery, after this point but before 12 or 24 months, the infection is deemed a delayed-onset infection.Any infection occurring past this point is classified as a late-onset infection [12].These late onset cases of PJI are linked to infection spread through hematogenous mechanisms from other infection sites.An alternative classification system is simply, acute and chronic.Acute infection is postulated to occur within 4 weeks and chronic at any time past this [14,15].This lack of coordination regarding classification further relays the challenges faced in the treatment of PJI and complicates the decision process of subsequent treatment approaches.
The pathogenesis of PJI and biofilms share an intrinsic link.Biofilms can exist in an innocuous state, having a gestation period of up to several years whilst exhibiting few clinical signs [16].Biofilms consist of community-based microorganisms encased in a selfproduced structural matrix.The formation of biofilms has developed as an evolutionary adaptation to a hostile environment.This allows pathogens in their planktonic state to evade eradication from the host immune response and exist within a multi-cellular community [17].The primary constituents of the biofilm consist of polysaccharides, proteins and nucleic acids which constitute a matrix of extracellular polymeric substances (EPS) [4].The EPS encasement serves to provide the encased bacteria with mechanical stability and also contributes to the evasion of the host immune response whilst providing a source of nutrients to allow for the proliferation of the bacteria [17].The quiescent state within the biofilm renders the encapsulated colony recalcitrant, with protection from the biofilm casing preventing the absorption of antibiotics [18].The general consensus throughout literature is that the development of a biofilm can be delineated into three distinct stages: initial cell attachment, maturation and proliferation and finally dispersal [19,20].
Interest has increased within the past decade regarding novel and innovative methods for the reduction of medical device-healthcare associated infections (MD-HAIs) by medical device companies and researchers alike.

Methodology and materials
Titanium (Ti-6Al-4v) alloy coupons (diameter 12.7 mm, 3 mm average thickness) were used for all microbiological analysis, dimensions chosen as to match the specifications of the CDC biofilm reactor.Two prosthesis relevant surface finishes were provided, a Plasma Sprayed Ti6Al4V coating (PS) and a Grit Blasted finish (GB).These are routinely used in hip replacement systems [21].A third As-Fabricated (AF) finish was provided and used for comparative and control purposes.All surfaces were engraved with a unique identifier on the reverse of the sample.Nonphysical markings would be removed through the sterilisation process.All surfaces were subjected to an initial cleaning and sterilisation cycle and autoclaved at 121 °C for a minimum of 30 min.

Areal surface characterisation
The interaction between bacteria and the topography of a surface has been extensively studied with the surface features dictating the behaviour of the bacteria.The scale of the surface features plays a pivotal role in the attachment behaviour of said bacteria.Features at the macroscale, matching those of the bacterial cell have been generalised to implicit hydrodynamic effects on the bacteria.Features within the nanoscale have effects on the attachment behaviour through physiochemical forces, cell membrane damage and localised chemical affects [22].Quantitative analysis of the surface topography through advanced surface characterisation can provide a deeper understanding of the surface and its impact on the initial cell attachment and subsequent biofilm behaviour.
In order to characterise the surface, a series of areal surface parameters were calculated.Regarding height parameters, the following have been included; Arithmetic mean height (Sa), Root mean square height (Sq), Kurtosis (Sku), Skewness (Ssk), Max peak height and Max valley depth (Sp & Sv, respectively).The roughness of the surface (Sa & Sq) is commonly reported and often used solely to characterise a surface, to gain a better understanding of the feature morphology and overall topology of the surface profile, parameters such as Sku and Ssk have been selected.The kurtosis of the surface describes the geometry of the features upon the surface with higher values indicating a tip geometry of a peak having a reduced angle, therefore a sharper peak.The skewness of the surface concerns the deviation around a mean line, a positive value relates to a surface that is peak dominated and a negative value, one that is valley dominant.These additional parameters are often used for lubrication analysis but are useful indicators to help understand the behaviour of a surface at the implant-bone interface [23,24].As Sa and Sq utilise average height values, they are not and should not be solely used as indicative parameters to characterise a surface, due to the nature of average data, surfaces that exhibit substantially different topographies, including opposing skewness values can result in the same Sa/ Sq values.
Whilst typically used within the automotive field and to obtain useful tribological information of a surface, the volume parameters defined from the material ratio curve and found within the functional parameter field, have been used within this study with hope to quantify the available space for bacterial attachment and subsequent biofilm formation.Although typically used for wear and lubrication analysis such as the volume for available fluid retention and highlighting the volume/ area of material that will experience the most wear [25], volume parameters have been applied to applications within numerous fields and have seen novel use for the characterisation of open pore surfaces for additively manufactured components [26].Parameters that define the valley and core volumes (Vvv, Vvc respectively) represent the relationship of the material interface and air at a given material ratio threshold.These thresholds (mr1 & mr2), set at 10% and 80% define the peak and valley zones, respectively.These defaults assume the surface exhibits three distinct zones and that the material falls within these   regions, peak material represented by peak material volume (Vmp (0%-10%)), core material (Vmc, Vvc (10%-80%)) and void valley range (Vvv (80%-100%)) [26][27][28].Although these material ratio thresholds are flexible and can be set at functional specific values, the values of 10% and 80% have been utilised for this study and applied to all surface types.A representation of the material ratio curve can be seen in figure 1.
Functional volumetric parameter values are extracted from the surface measurement prior to the application of filters, thus are obtained at the earliest opportunity to reduce the influence of filtration.For the sake of the PS coating, this was extracted from the primary surface.The remaining, AF and GB surfaces, a 0.5%-99.5% threshold was applied prior to the volume parameter extraction.This was necessary as to reduce the negative effects of optical measurement techniques, producing spikes within the data from features such as high slope angles and increased localised reflection [31].
To further characterise the surface and provide information of the available space for colonisation, the topographical surface area of the test surfaces has been determined.This value encompasses the complete topographical profile of the surface, including the area contained within the valleys etc.The cross-sectional representation of the PS surface within figure 3(a) below shows the true nature of the topographical details of the surface.The topographical surface area was calculated through MeshLab software on the extracted surfaces (STL.) and utilised for all further calculations.Using the topographical surface area in conjunction with volume parameters and the traditional amplitude parameters, it provides a more realistic representation of the true surface of the component.

Plasma spray (PS)
The porous coating allows for cementless fixation of the implant and permits the implant to have a primary focus on osseointegration whilst focusing on a scratchfit fixation with the cortical bone.The surface consists of irregular pores with a distribution range of 100-1000 microns, typical pore dimensions are shown in figure 2.
The topography of the surface is highly irregular with re-entrant features and steep slope angles, these are demonstrated in figure 3 above.This image was captured through a reconstructed model after microcomputerised tomography (μCT) analysis.Figures 3(b)-(d) shows a typical section of the coating captured through the Zeiss EVO MA10 Scanning Electron Microscope (SEM).The nature of the surface and the presence of re-entrant features dictated the use of μCT thusly.The characteristic features of the PS coating render optical and contact measurement methods unsuitable for their use in surface characterisation of this surface type.The μCT was conducted on the Nikon XT H 225 industrial CT scanner.The surface was determined, extracted, and reconstructed within CT Pro and VG studio 3.2 and exported as a Stereolithography (STL) file.The measurement settings can be found in table 2 below.

Grit blasted (GB)
The nature of the grit blasted coated samples allows for more traditional measurement & characterisation approaches to be conducted, due to the surface topography being within line of sight, with no reentrant features.In order to conduct quantitative analysis, focus variation methods were used.Datasets were acquired using the Alicona G5 InfiniteFocus instrument (Bruker Alicona Graz, Austria).A 20x objective lens was used with a vertical resolution of 500 nm and lateral resolution of 1 μm.A sampling distance of 0.4 × 0.4 μm was used throughout the analysis.Downsampling of the data was not required when using the Alicona G5.Although this results in significantly larger data files, it allows for a considerable increase in measured points (180 M).The coaxial light source was used with a scan area of 5 × 5 mm with a Z-range of 200 μm, centred within the diameter of the surface.The surface height map data was transferred into MountainsLab 8.0 software for subsequent surface analysis.

As-fabricated (AF)
The Alicona G5 was used to acquire surface data of the AF surface.The same scan area of 5 × 5 mm was used with a Z-range of 200 μm.The coaxial light was used with a vertical resolution of 500 nm and a lateral resolution of 1 um.The unfinished surface had visible tooling marks, circumferential in nature, a resultant of the turning process.These can be seen in table 4.
Upon completion of all surface measurements, in order to quantify the measurements, filtering was applied in accordance with ISO 25178-3.The parameters generated were in accordance with ISO 25178-2 [29,30].Tabulated generated data can be found in tables 3-5 below.

Bacteria culture and growth
The laboratory procedure concerning this study is based on a modified version of the standard test method E2562-17 and E3161-18 [32,33].These outline a standard practice for growing a S. aureus biofilm on a test substrate within a CDC Biofilm reactor [32,33] with subsequent quantification.A total of 24 samples (n = 24 total) were analysed, this consisted of 8 per group (n = 8 per group).
Bacterial cultures of S. aureus were used for all microbiological analysis.Cultures were taken from a freeze-dried ampoule from the National Collection of Type Cultures (NCTC, 10788).The Ampoule was used in conjunction with freshly prepared Trypticase Soy Agar (TSA, Neogen Corp.) plates to provide a master plate, from which all samples were taken.Plates were incubated at 36 ± 2 °C for 24 ± 2 h.Isolated colonies were taken and aseptically transferred into 300 ml of sterile Trypticase Soy Broth (TSB, Neogen Corp.) before being placed in a shaking incubator for a further 24 ± 2 h at 36 ± 2 °C.
Upon completion of the incubation period, spectroscopic analysis was conducted with a baseline spectrogram acquired from an uninoculated sample of TSB.The media was then diluted with TSB until a reading of OD 620 = 0.3 was achieved, equating to an approximate concentration of 10 8 CFU/ mL.

Biofilm growth
Prior to the growth phase, the samples were placed within the reactor, 24 in total, 8 for each surface type.The assembled reactor was autoclaved at 121 °C for a minimum of 30 min 400 ml of the diluted media was transferred into the reactor and incubated at 36 ± 2 °C with the baffled stir bar rotational speed set at 60 ± 5 r min −1 for 48 ± 2 h.The surfaces identified for plating were aseptically removed, scraped with a disposable loop, rotated 60°and scraped.This was repeated until the entire surface was covered, the loop was rinsed in a universal vial containing 9 ml of Phosphate Buffered Saline (PBS).A 10-fold serial dilution was conducted with subsequent plating carried out via the pour plate method on TSA plates.A standard countable range of 30 to 300 was followed and the arithmetic mean of duplicated plates were calculated.The viable bacterial colonies are represented as CFU/mL of the original sample, prior to the dilution.Also, the log alternative has been included in addition to the log CFU/cm 2 .The calculated total topographical surface area has been used to determine the colony count per unit area.

Results and conclusion
Areal surface characterisation results PS Following surface reconstruction within VGStudio Max 3.2, the STL file was transferred to Mountains-Map 8.0 surface analysis software.An 8.2 × 8.2 mm area was extracted from the central region of the surface.The surfaces were leveled via the least squares plane (LSPL) method and an S-filter (λs) index of 25.00 μm.A Gaussian Regression L-filter (λc) of 8.00 mm was applied.An example of the extracted surface can be found in table 3 below.

AF and GB
Surface analysis was conducted on MountainsMap 8.0.A 4.1 × 4.1 mm area was extracted to negate edge effects and levelled via the LSPL method.A λs filter index of 2.5 μm was applied followed by a λc Gaussian filter of 0.8 mm.This was repeated a total of 3 times per individual coupon, ensuring a wider range of surface features were captured.All 8 coupons that were subject to microbial analysis were characterised for their surface topography.Tabulated surface texture measurements can be found for the AF and GB surfaces in tables 4-5 respectively below, this represents the averaged data concerning all surfaces measured for each surface type.

Biofilm quantification results
The following results were obtained from three replicates with 24 coupons (n = 24 total), 8 (n = 8 per group) of each surface type and plating conducted in triplicate, with duplicates taken for each.The biofilms were removed and quantified as the per the method outlined above.A countable range practice of 30-300 was conducted, subsequent to a 10-fold serial dilution.The colony forming units for each surface type can be found in table 6 below.These values have been calculated using equations (1)-(3).Significant differences were indicated amongst the surface types regarding the colony forming units noted, one-way ANOVA statistical analysis was conducted to identify the statistical significance, resulting in a p-value of 0.0001.The PS surfaces exhibited a significant increase in the colony counts in comparison to the remaining two surfaces.The lowest counts were found on the AF surfaces (table 6).

Discussion
The role of PJI and in particular the development of antibiotic resistant biofilms upon a prosthetic surface encompasses ever increasing complications to the patient and health service.Bacteria in the biofilm state in contrast to their planktonic counterparts pose a risk to the success of the orthopaedic implant.To better understand the relationship between the bacteria behaviour, be it in biofilm development or their initial attachment and the surface of the prosthetic, a deeper characterisation of this surface is required.Traditionally, the surface roughness is often solely quoted as the defining parameter of the surface and often obtained through profile traces represented by Ra.This parameter alone, does not give a true representation of the surface and should be used with caution for this purpose.Two surfaces, with opposing topographies with one displaying periodic peaks will give the same roughness value as one displaying periodic valleys, due to the averaged nature of this parameter.Therefore, this study has tried to demonstrate the importance of additional surface parameters such as those found in the functional volume family.
This study has investigated the effect prosthesis relevant surfaces have on the formation of S. aureus biofilms.Three surfaces have been examined, a plasma spray aimed primarily at osseointegration, a grit blasted finish used to achieve a scratch-fit fixation to the juxtaposed bone and a third, as-fabricated surface used for control and comparison purposes.These surfaces all exhibited different surface parameter values and incorporate different topographies and feature morphologies.The heightened interest porous coatings have seen to exploit their osseointegrative properties, carries with it further challenges in the characterisation of these surfaces.In order to overcome the re-entrant features present on these surfaces, traditional techniques such as those reliant upon line-of-sight or stylus-based approaches are unsuitable for these surface types.Therefore, computerised tomography has been used within this study and is often the only solution for this application.
Results from this study shows the porous coating of the PS exhibited the largest CFU/mL count of all the surfaces examined, by a considerable margin.This surface type incorporates re-entrant and porous sections within its topography, highlighted in figures 2-3.This surface type also incorporates the greatest height parameter and functional volume values, shown in figures 4 and 5(a).These findings are in contrast to the AF surface that showed the lowest CFU/ mL count and the lowest surface parameter values.Interestingly, the height parameters of the AF and GB surface do not differ significantly (figure 4), with only the parameters defining the available surface below the mean plane showing a significant increase concerning the GB surface.
Although the similar topographical surface areas of the AF and GB samples does not best represent the increase in the CFU count, using it in conjunction with the volume parameters highlights the importance of the available volume within the lowest voids of the surface topography, where an increase in the Vvv values correlates with the increase in the CFU/ mL per surface type.Within literature, an observation has been noted regarding the increase in surface area from highly textured surfaces could potentially harbour amplified bacterial loads [22,34].The findings from this study show that this is not necessarily the case, shown in figure 5(b) below, the topographical surface area does not necessarily dictate an increase in the biofilm accumulation but more where the increased available surface area is situated within the surface.This is shown in the insignificant increase in surface area of the GB surface to the AF surface, not correlating to large increase in the CFU/ mL of the same surfaces.
An area of note regarding this study is that it represents in vitro conditions, therefore will not be fully representative of a clinical situation.The locations of these prosthesis relevant surfaces within the human body differ and are exposed to different environments.The porous coatings are bone-contacting surfaces and may fair better in relation to biofilm contamination due to the vascular nature of bone, offering more ready access to any antibiotic therapies that may be administered.In addition, only S. aureus has been used within this study, this study could be expanded upon to include additional, relevant bacterial species to ascertain the surface effects over different species.

Conclusion
The growth of S. aureus biofilms has been shown to be affected by the surface topography of three distinct surface types.Advanced surface characterisation has been conducted on two prosthesis relevant surface coatings and a third control surface.Quantification of the biofilm on these surface types suggests the importance of the functional volume parameter family, highlighting the available space available for colonisation away from the topmost surface.The difference in the CFUs/ mL counted are significant across the surfaces with the PS coating exhibiting over 200 times that of the AF surface and over 60 times greater than that of the GB surface.Relating this to the CFU per unit area, the change in bacterial count for the surface types is further highlighted.The CFU/ cm 2 for the PS coating was 2.8 ×10 6 in contrast to the 41 × 10 3 CFU/ cm 2 for the AF surface.
This study shows that the magnitude of the available space within the lowest points of the surface topography plays a crucial role in the volume of bacteria that could potentially colonise that surface.Utilising surface parameters alone, such as Sa and Ra do not give a true representation of that surface type.The increase in surface area offered by the increase in surface roughness is not as important as where that surface area increase lies within the surface topography.

Figure 2 .
Figure 2. Example of manual measurements of a typical pore feature on the PS sample.SEM image processed in MountainsMap 8.1.

Figure 3 .
Figure 3. (a) Cross sectional view of PS surface type, captured through XCT, (b)-(d) SEM images at increasing magnification showing feature morphology.SEM images obtained on Zeiss EVO MA10.

Figure 4 .
Figure 4.A comparison of the surface height parameters against the CFU/ mL for all surface types.

Figure 5 .
Figure 5. (a) and (b) -(a) chart showing comparison of surface volume parameters against CFU/mL all surface types.(b) comparison of total topographical surface area against the CFU/ mL for all surface types.

Table 2 .
CT parameters for all measurements of PS surfaces.

Table 3 .
Isometric view of surface and parameters for the PS surfaces.

Table 4 .
Isometric view of surface and parameters for the AF surfaces.

Table 5 .
Isometric view of surface and parameters for the GB surfaces.

Table 6 .
Comparison of Colony forming unit information for all surface types.