Sonication is superior to scraping for retrieval of bacteria in biofilm on titanium and steel surfaces in vitro

Background and purpose Low-virulence implant infections are characterized by bacterial colonization of the implant with subsequent biofilm formation. In these cases, soft tissue biopsies often prove to be culture negative. Consequently, detachment of the causative adherent bacteria is crucial for correct microbiological diagnosis. Using an in vitro model, we compared 4 methods of biofilm sampling from metal surfaces. Methods Discs of titanium and steel were incubated in the presence of Staphylococcus aureus, Staphylococcus epidermidis, Enterococcus faecalis, and Propionibacterium acnes in Mueller Hinton broth. Non-adherent bacteria were removed by repeated rinsing of the discs. 10 parallels of each disc were subjected to 1 of 4 methods for bacterial recovery: (A) sonication of the discs, (B) scraping of the discs using surgical blades followed by streaking of the blades onto agar plates, (C) scraping of the discs followed by vortex mixing of the surgical blades, and (D) scraping of the discs followed by sonication of the surgical blades. Quantitative bacterial cultures were performed for each sampling method. Results With the exception of S. epidermidis on steel, sonication efficiently and reliably dislodged biofilm bacteria. The scraping methods employed did not detach bacteria embedded in biofilm. Interpretation Scraping of metal surfaces is not an adequate method for sampling of biofilm bacteria in vitro.

 Prosthetic joit ifectio (PJI) is a devastatig complicatio occurrig i about 1% ad 2% of patiets receivig a hip or kee prosthesis, respectively (Wilso et al. 1990, Espehaug et al. 1997, Phillips et al. 2006. Whereas diagosis of a early postoperative PJI is usually straightforward, the diagosis of a late or chroic PJI is otoriously difficult. Both a late, chroic PJI ad aseptic prosthetic looseig preset cliically with implat looseig ad joit pai, ad there are usually few sigs of iflamatio (Zimmerli et al. 2004). I late PJI, the sesitivity of culture of periprosthetic biopsies is i the rage of 65-89% (Atkis et al. 1998, Spagehl et al. 1999, Padey et al. 2000.The iadequacy of culture i this settig is probably best explaied by the biofilm mode of growth of bacteria o a biomaterial surface (Gristia ad Costerto 1985). Correct idetificatio ad susceptibility testig of bacteria causig a PJI is essetial for the successful treatmet of PJI (Hasse ad Spagehl 2004).
I order to circumvet the obstacle of the biofilm i retriev-ig bacteria from the implat surface, alterative strategies have bee developed (Trampuz et al. 2003). I the past decade, soicatio-i.e. ultrasoic eergy applied directly to the biomaterial surface to disrupt adheret biofilm-has bee reported to be a more reliable tool for the diagosis of PJI (Tuey et al. 1999, Nguye et al. 2002, Trampuz et al. 2007, Esteba et al. 2008). Soicatio of a large explated prosthesis is, however, techically demadig ad carries a substatial risk of cotamiatio durig hadlig (Trampuz et al. 2006). Cosequetly, soicatio has ot bee implemeted as a stadard procedure for diagosis of chroic prosthetic joit ifectio.
Mechaical scrapig of surfaces is ofte used for specime collectio ad is the method of choice for certai ifectios, e.g. ifectious keratitis (Hall ad York 2004). I theory, scrap-ig the surface of a removed implat represets a techically easy alterative for mechaical removal of adheret biofilm bacteria. It has bee hypothesized that the results of scrapig could be improved by either vortex mixig or soicatio of the scrapig product (Costerto et al. 1986). Scrapig, eve followed by vortex mixig or soicatio of the surgical blade, is cosiderably less complicated tha soicatio of a large prosthesis. However, we are ot aware of ay experimetal i vitro or i vivo study o scrapig as a method for detach-met of bacteria from metal implats. To our kowledge, oly oe cliical study has dealt with the use of scrapig of joit prostheses i order to improve bacterial detectio. Neut et al. (2003) foud that as a diagostic tool, scrapig was better tha covetioal methods, i.e. culture of periprosthetic soft tissue biopsies. From that study, oe was led to ifer that bacterial detectio by scrapig was superior to the results preseted i most soicatio studies.
If equally effective, scrapig would be preferable to soica-tio due to its techical simplicity. We here report a i vitro compariso of the recovery of biofilm bacteria from metal surfaces by soicatio ad by 3 differet scrapig techiques.
Cofocal scaig laser microscopy (CSLM) was employed to cofirm the 24-hour biofilm formatio ability of each strai. 8 study groups were examied (Table 1). Bacteria were sus-peded i 25 mL of Mueller Hito broth (BD, Frakli Lakes, NJ) ad icubated at 35ºC util a spectrophotometric desity of approximately 1 × 10 8 coloy formig uits/mL (CFU/mL) had bee reached i the expoetial growth phase. A batch of 40 discs (oe study group) was immersed i this bacterial suspesio bath ad icubated at 35ºC for 24 h o a getly stirrig agitator (20 rpm).
To remove o-adheret bacteria, the discs were rised 6 times i sterile salie. First, the discs for each study group were placed i a sterile plastic tube (Sarstedt, Norway) cotaiig 25 mL salie ad getly vortex mixed (MS2 Miishaker; IKA Works Ic., Wilmigto, NC) at 100 rpm for 10 secods. The discs were the trasferred to aother tube, ad the procedure was repeated twice. Each sigle disc was the trasferred to a sterile glass test tube cotaiig 5 mL salie ad subjected to vortex mixig at 100 rpm. The sigle disc risig was also repeated 3 times.
Aliquots of 50 µL salie were icubated o agar (Merck, Darmstadt, Germay) with 5% ox blood at 35ºC for 3 days. For culture of P. acnes, FAA agar (Merck) was icubated i a aaerobic cabiet for 7 days. The bacteria cultured were eumerated by coloy coutig. The umber of CFU after fial risig was recorded as a quatitative baselie, facilitatig evaluatio of the differet detachmet methods.
Each experimetal group (10 discs) was subjected to 1 of 4 methods for biofilm detachmet ad bacterial recovery. The experimetal desig is summarized i Table 1.

Method A (sonication of discs)
Each sigle disc was trasferred ito a sterile plastic cotaier (Miigrip, Segui, TX) cotaiig 5 mL salie. The cotaier was sealed ad immersed i a ultrasoic bath (TPC-120; Tel-soic AG, Broschhofe, Switzerlad). Soicatio at 30 kHz with a power output of 300W, as specified by the maufacturer, was performed at 37°C for 5 mi. After soicatio, aliquots of 50 µL were icubated as described above.

Method B (scraping of discs with direct culture)
Thorough scrapig of the complete surface of the disc was performed usig a sterile surgical blade. The disc was fixed Figure 1. Titanium discs (left) and steel alloy discs (both from Scandinavian Customized Prosthesis AS, Trondheim, Norway) served as surfaces for biofilm formation, with roughness of Ra = 2.5 µm and Ra < 0.5 µm for titanium and steel discs, respectively. All discs were manufactured with a diameter of 17 mm and a thickness of 2 mm. betwee the thumb ad the idex figer. 1 surgical blade was used for each disc. Seedig was doe by streakig both sides of the surgical blade oto a agar plate, followed by icuba-tio as described above.
Method C (scraping of discs followed by vortex mixing of the surgical blade) Scrapig was performed as described for method B. After scrapig, each sigle surgical blade was trasferred ito a glass test tube cotaiig 5 mL salie before vortex mixig at 2,000 rpm for 30 secods. After vortex mixig, a aliquot of 50 µL from each tube was icubated as described above.
Method D (scraping of discs followed by sonication of the surgical blade) Scrapig was performed as described for method B. After scrapig, each surgical blade was soicated i 5 mL salie ad aliquots of 50 µL were icubated as described above.
To prevet cotamiatio, sterile forceps ad sterile surgical gloves were used durig hadlig of the discs ad the surgical blades. All procedures were performed i a lamiar airflow cabiet.
Compariso of methods A, B, C, ad D was doe for S. aureus, S. epidermidis, ad E. faecalis. Culture of P. acnes requires aaerobic coditios. Due to the restraied storage capacity i a aaerobic cabiet, we were compelled to reduce the umber of agar plates icubated. Hece, oly methods A ad C were compared for P. acnes.

Statistics
The 4 methods were compared usig a Kruskal-Wallis test (SPSS software). 2 group comparisos were computed with Ma-Whitey test. Statistical sigificace was cosidered at p ≤ 0.05.

Results
The risig procedure efficietly removed o-adheret bacteria ( Figure 2). The results were uiform i 7 of the 8 study groups: S. aureus o titaium ad steel discs, S. epidermidis o titaium discs, E. faecalis o titaium ad steel discs, ad P. acnes o titaium ad steel discs. Firstly, soicatio of discs (method A) allowed retrieval of more bacteria tha ay of the scrapig techiques (p < 0.001, Figure 3). Secodly, the umber of CFUs detected after soicatio of discs was higher compared to culture of the salie used for the fial ris-ig step (p < 0.001). Thirdly, all scrapig techiques allowed recovery of fewer bacteria compared to the fial risig step. Fially, oly soicatio recovered bacteria i 10 of 10 parallels, i cotrast to scrapig which yielded highly variable results (Table 2).
For S. epidermidis o steel discs, bacterial recovery was geerally low (Figure 3). Soicatio of steel discs did ot yield more S. epidermidis tha fial risig (p = 0.3), ad bacterial growth was observed i oly 7 of 10 parallels (Table 2).
There was o statistically sigificat differece i the umber of S. epidermidis coloies recovered by soicatio of steel discs (method A) ad scrapig followed by soicatio of the surgical blade (method D) (p = 0.4; Figure 3). However, both method A ad method D allowed recovery of more bacteria tha scrapig ad direct culture (method B) ad scrapig followed by vortex mixig of the surgical blade (method C) (p < 0.05).

Discussion
Our study was udertake to retrieve bacteria from biofilms o titaium ad steel surfaces usig a i vitro model, ad to compare differet methods of biofilm disruptio for subse-quet culture. With the exceptio of S. epidermidis o steel, our fidigs clearly show that soicatio is capable of detachig bacteria i this model. The various scrapig methods proved to be isufficiet by demostratig a lower yield ad highly icosistet results. This coclusio is based o the cocept that i order to disrupt bacteria from a biofilm, the umber of bacteria recovered must be icreased compared to what ca be observed after the fial step of repeated washig.
These results also preset circumstatial evidece of the successful establishmet of biofilms o the metal surface. The oe study group that was differet quatitatively was S. epidermidis o steel. Here, we caot rule out the possibility that there was a poorly developed biofilm cotaiig a low umber of bacteria-eve though i a separate study biofilm matrix was evidet by CSLM after calcofluor white staiig. A alterative iterpretatio is that ot eve soica-tio effectively dislodges a S. epidermidis biofilm from steel surfaces. Figure 2. Arithmetic mean of the number of CFUs cultured after rinsing steps 4, 5, and 6 (final step) in all study groups. For 2 study groups (P. acnes on titanium and steel discs) culture was performed only after the final step due to limited anaerobic culture capacity. There are several methodological cosideratios relevat to our study. A umber of i vitro models for biofilm formatio exist (Mombelli 1999). We used a dyamic versio of a closed system biofilm method with stadard coditios of fluid motio ad temperature to culture biofilms simultaeously o 40 discs (Christese et al. 2000, Sissos et al. 2000. The time allowed for biofilm formatio also varies cosiderably. Overight (18-hour), 24-or 48-hour icubatio periods have bee recommeded for biofilm formatio i vitro (A ad Friedma 1997). Most ivestigators use icubatio periods from just a few hours up to 24 hours (Christese et al. 1987, Tollefso et al. 1987, Olso et al. 1988, Khardori ad Yassie 1995, Vadecasteele et al. 2003, va der Borde et al. 2004). The bacterial species we studied, S. aureus, S. epidermidis, E. faecalis, ad P. acnes, are all commo causative microbes i chroic prosthetic ifectios (Tsukayama et al. 1996, Berbari et al. 1998, Letio 2003, Zeller et al. 2007. A disadvatage of usig cliical isolates as opposed to referece strais is the icomplete kowledge of their detailed biofilm characteristics ad limited applicatio i laboratories (Oliveira et al. 2007). The cliical strais we employed had a priori demostrated their ability to establish cliically relevat biofilm ifectios o joit prostheses.
I routie orthopedic surgery, several differet foreig materials are regularly implated, e.g. boe cemet, polyethylee compouds, ad differet metal alloys. Biomaterials have dif-feret affiities for bacteria (Oga et al. 1993). I geeral, a icrease i surface roughess ehaces bacterial coloizatio ad early biofilm formatio (Arold ad Bailey 2000). Both the surface roughess of the discs ad the metal alloy we used to make them were similar to those foud i commoly used hip prostheses (Fures et al. 2007).
The purpose of soicatio to improve the diagosis of PJI is to detach biofilm bacteria o the implat surface usig ultra-soic eergy (above 20 kHz). For subsequet culture, the bacteria must still be viable after soicatio. Before the study, we had soicated each of the 4 bacterial strais for 20 mi without

S. aureus
Titanium 10/10 10/10 7/10 5/10 Steel 10/10 9/10 2/10 0/10 S. epidermidis Titanium 10/10 5/10 1/10 8/10 Steel 7/10 1/10 0/10 6/10 E. faecalis Titanium 10/10 10/10 10/10 4/10 Steel 10/10 9/10 4/10 6/10 P. acnes Titanium 10/10 1/10 Steel 10/10 1/10 observig ay bactericidal effect (data ot show). Kobayashi et al. (2007) recommeded a soicatio time of betwee 1 ad 5 mi as beig ideal for dislodgig biofilm bacteria without affectig bacterial viability. Ivestigators usig soica-tio should report the maufacturer ad model umber of the equipmet, the output power, oscillatio frequecy, reactio volume, fluid temperature, ad soicatio time (Christese et al. 1995). Soicatio at about 50 Hz is sometimes reported (Tuey et al. 1999, Nguye et al. 2002, Baysto et al. 2007. Reportig this pretrasduced electric frequecy is i our opi-io misleadig ad cofusig, because o dislodgemet of biofilm will occur at 50 Hz. I a cliical cotext, soicatio of a large explated prosthesis is techically demadig ad carries a sigificat risk of cotamiatio durig hadlig. Thus, the ecouragig results obtaied from scrapig the prosthesis ad directly ioculatig the agar plate idicated that scrapig is a beeficial method due to its techical simplicity (Neut et al. 2003). However, these results have ot bee reproduced by other research groups. The results from our i vitro study idicate that scrap-ig, eve followed by post-scrapig procedures for disruptig the biofilm, is ot a efficiet techique for dislodgemet of biofilm bacteria.
To coclude, with the exceptio of S. epidermidis o steel, we have demostrated that soicatio efficietly dislodges bacteria from biofilms geerated i vitro o titaium ad steel surfaces. Scrapig of metal surfaces caot be recommeded as a method for diagosis of biofilm-related ifectio. Further studies o soicatio are warrated, as are studies comparig methods for biofilm samplig i vivo.