The main goal of this study, id est to demonstrate antibacterial efficacy of induction-heating treatment, was successfully accomplished as shown in our results. This serves not only to corroborate previously published results regarding induction-heating disinfection [8,9,10,11,12], but also to validate the newly developed PDSIH, which presents as a feasible option for operating room conditions and will thus allow for translation to in vivo settings in further studies. To our knowledge, all previous experiments assessing electromagnetic induction heating efficacy have been performed using stationary induction plates or coils [8,9,10,11,12]. This allows for a reproducible and easy-to-monitor technique in the lab environment, but hardly translates to the versatility, handling facility and hygienical conditions needed for further clinical applications, especially when considering the protocolized environment of the operating room.

Also, all the experiments in this study were performed with a validated model of mature biofilm on frequently used metallic alloys, and selected bacterial species are amongst the most common agents of human prosthetic joint infections [5]. Altogether, it represents one step further towards translation of this novel technique into clinical practice.

One main concern about our results is the quantitative difference with previously reported results by Pijls et al. regarding induction heating applied to mature biofilm of S. epidermidis on Ti6Al4V [10]. In their study, the effect was characterized for different temperatures ranging from 50ºC to 90ºC, and bacterial load reduction continuously increased. With the 3.5-minute, 70ºC induction-heating-only protocol 24-hour-biofilm, bacterial load reduction amounted to 5.8 Log10, while our study only demonstrated a 1.22 Log10 CFU/mL reduction. Possible explanations include the biological variability between the S. epidermidis ATCC 35,984 strain from our study, and ATCC 14,990 used by Pijls et al., unaccounted-for differences in the biofilm development technique, or also the different accuracy of the heating protocol administration and thermal image monitoring: whereas Pijls et al. employed a static induction heater placed under the sample, which was automatically controlled by an infra-red temperature sensor monitorization system and updated four times per second, we opted for a manual, less-accurate protocol, closer to that of future in vivo studies within our project. Precise quantitative results, however, should be taken with caution at this stage, for their relative magnitude and significance in the complexity of an in vivo joint infection setting is yet unknown.

In our study, a different quantitative response is insinuated between microorganisms: E coli showed the greatest effect, with a 5.94 and a 3.5 Log10 CFU/mL reduction on Ti6Al4V and CoCrMo respectively, whereas S aureus and S epidermidis presented less than 1.3 Log10 CFU/mL reduction on both materials. This variation could be explained by the fact that Gram-positive bacteria develop thicker, more hydrated biofilms than those of Gram-negative bacteria, which may have an effect on thermal diffusion throughout biofilm matrix. To our best knowledge, all previous studies on induction-heating on mature biofilms were performed using Gram-positive bacteria, and this result has not been reported previously.

Differences between metallic alloys may be suggested by the data, particularly in the case of E coli, which showed an apparently greater effect on Ti6Al4V than on CoCrMo. This may be due to the fact that no viable bacteria could be accounted for in 2 of the Ti6Al4V discs. However, the disc that did present quantifiable CFUs was well inside the range of its CoCrMo counterparts. Differences between biomaterials could not be consistently accounted for in this study.

One of the main limitations of this study is the lack of accuracy in thermal monitoring of the coupon surfaces. As stated above, the selected induction-heating protocol using PDSIH, as well as the thermographic monitoring method, were purposely selected for their feasibility for in vivo settings, and their validation was the main goal of this study. Another limitation is the fact that quantification of viable bacteria in the S epidermidis strain was low compared to other species, even in the control group, in both materials. This could have reduced the observable effect of the induction-heating treatment. However, bacterial load measurements are consistent between the discs, and it did not affect the statistical significance of this reduction. Finally, the simplicity of our study design can be considered as a further limitation: a full comprehension of the relative magnitude and clinical feasibility of induction-heating disinfection will require a multidimensional analysis of the isolated and combined effects of induction-heating and other techniques, such as mechanical cleaning and antibiotic therapy; the first results on this regard have been published in recent years [11, 12]. Furthermore, this study only assessed the effect of induction heating on 3 bacterial strains, while many other species can be less frequently involved in prosthetic joint infection, and induction-heating effect on them cannot be extrapolated yet. Further experiments will be needed to consistently determine the actual effectiveness of induction heating relative to other techniques, as well as possible synergistic effects, in the more complex setting of an in vivo model of joint infection, and to assess the efficacy of this technique on a wider range of microorganisms.