Surgical Treatment of Diabetic Foot Ulcers Complicated by Osteomyelitis with Gentamicin-Loaded Calcium Sulphate-Hydroxyapatite Biocomposite

Diabetic foot ulcers, complicated by osteomyelitis, can be treated by surgical resection, dead space filling with gentamicin-loaded calcium sulphate-hydroxyapatite (CaS-HA) biocomposite, and closure of soft tissues and skin. To assess the feasibility of this treatment regimen, we conducted a multicenter retrospective cohort study of patients after failed conventional treatments. From 13 hospitals we included 64 patients with forefoot (n = 41 (64%)), midfoot (n = 14 (22%)), or hindfoot (n = 9 (14%)) ulcers complicated by osteomyelitis. Median follow-up was 43 (interquartile range, 20–61) weeks. We observed wound healing in 54 patients (84%) and treatment success (wound healing without ulcer recurrence) in 42 patients (66%). Treatment failures (no wound healing or ulcer recurrence) led to minor amputations in four patients (6%) and major amputations in seven patients (11%). Factors associated with treatment failures in univariable Cox regression analysis were gentamicin-resistant osteomyelitis (hazard ratio (HR), 3.847; 95%-confidence interval (CI), 1.065–13.899), hindfoot ulcers (HR, 3.624; 95%-CI, 1.187–11.060) and surgical procedures with gentamicin-loaded CaS-HA biocomposite that involved minor amputations (HR, 3.965; 95%-CI, 1.608–9.777). In this study of patients with diabetic foot ulcers, complicated by osteomyelitis, surgical treatment with gentamicin-loaded CaS-HA biocomposite was feasible and successful in 66% of patients. A prospective trial of this treatment regimen, based on a uniform treatment protocol, is required.


Introduction
In people with diabetes mellitus, foot ulcers are a major complication, with a lifetime prevalence of 19-34% [1]. An estimated 18.6 million people are currently affected by diabetic foot ulcers [2]. Approximately 60% of diabetic foot ulcers are infected at presentation, and 40% of the noninfected diabetic foot ulcers become infected before healing [3,4]. Infected

Experimental Section
We conducted a retrospective multicenter cohort study of patients treated between February 2017 and June 2019. Retrospective data collection was completed in December 2019. Dutch hospitals in which surgical treatments of DFO with gentamicin-loaded CaS-HA biocomposite were performed were contacted. The physicians (vascular surgeons and/or trauma surgeons) who performed the surgical treatments were assigned as local investigators. Patients provided written informed consent for data collection, analysis and storage. We performed all study procedures according to the Declaration of Helsinki and the Dutch Personal Data Protection Act. The Medical Ethics Committee Twente ruled this study exempt from formal ethical approval because of the retrospective observational design (METC Twente project K18-33).
We included patients with diabetes mellitus with one or more foot ulcers complicated by suspected or confirmed osteomyelitis who underwent a surgical procedure with gentamicin-loaded CaS-HA biocomposite (Cerament G™; BoneSupport, Lund, Sweden). We defined a foot ulcer as a discontinuation of the skin that minimally includes the epidermis and part of the dermis [8]. In accordance with the International Working Group on the Diabetic Foot (IWGDF) guidelines and the Dutch national guidelines, suspected osteomyelitis was defined as the presence of suggestive clinical signs (e.g., positive probe-to-bone test, exposed bone and/or intraosseous pus at intervention), elevated serum inflammatory markers and suggestive findings of osteomyelitis on imaging assessment (X-ray imaging, computed tomography, magnetic resonance imaging or radionuclide imaging) [14,30,32]. Confirmed osteomyelitis was defined as bone samples with cultures positive for microbiological pathogens [14,30,32].
Prerequisites for performing surgical procedures with gentamicin-loaded CaS-HA biocomposite were that the soft tissues and skin around the ulcer were adequate for closure of surgical wounds by primary closure or by reconstructive techniques. All patients were treated with gentamicin-loaded CaS-HA biocomposite as a last resort after conventional treatments of DFO failed. Failed conventional treatment was defined as a persistent foot ulcer with unresolved osteomyelitis after prolonged antibiotic therapy (>6 weeks [14]) or a persistent wound with unresolved osteomyelitis after surgical resection of DFO or minor amputation with adjuvant antibiotic therapy (approximately 1 week [14]). Minor amputation was defined as any resection through or distal to the ankle, in accordance with the IWGDF definition [8]. Osteomyelitis was in these cases diagnosed as described above. Only the first procedure was included if patients underwent multiple treatments with gentamicin-loaded CaS-HA biocomposite.
We excluded patients with severe chronic limb ischemia, irrespective of preoperative revascularization, because of the low probability of postoperative wound healing [33]. Severe chronic limb ischemia was defined in accordance with the wound, ischemia and foot infection classification as an ankle-brachial index ≤0. 39, an ankle systolic pressure <50 mmHg, a systolic toe pressure <30 mmHg or a transcutaneous oxygen pressure <30 mmHg [33].
When this study was conducted, no uniform treatment protocol existed for the surgical treatment of DFO with gentamicin-loaded CaS-HA biocomposite. The surgical procedures with gentamicin-loaded CaS-HA biocomposite generally consisted of the following steps: After excision of one or more foot ulcers, macroscopic necrotic bone and necrotic and infected soft tissues were resected ( Figure 1A). In most cases, surgeons obtained one or more bone samples for examination of microbiological pathogens. The remaining dead space was irrigated with saline solution and filled with CaS-HA biocomposite loaded with gentamicin (17.5 mg/mL) via injection or as pellets ( Figure 1B,C). Guidance using x-ray imaging was used at the surgeons' discretion. After the dead space was filled with gentamicin-loaded CaS-HA biocomposite, surgical wounds were closed by primary closure or reconstructive techniques (e.g., local transposition flaps). In certain patients with forefoot or midfoot DFO, complete resection of necrotic bone as part of the surgical procedure with gentamicin-loaded CaS-HA bio- After the dead space was filled with gentamicin-loaded CaS-HA biocomposite, surgical wounds were closed by primary closure or reconstructive techniques (e.g., local transposition flaps). In certain patients with forefoot or midfoot DFO, complete resection of necrotic bone as part of the surgical procedure with gentamicin-loaded CaS-HA biocomposite resulted in a transmetatarsal amputation. These amputations were performed on a level as distally as possible, with the aim of preventing subsequent extensive proximal amputations.
After resection, the intramedullary canals of the residual metatarsals were filled with gentamicin-loaded CaS-HA biocomposite, followed by surgical closure of the amputation wounds. Temporary or definitive fixation (e.g., external fixation or Kirschner wires) methods were also required in certain patients because of biomechanical instability after bone resection. After resection of a joint, the residual dead space was filled with gentamicinloaded CaS-HA biocomposite to create a semi-rigid or rigid arthrodesis, irrespective of fixation methods used.
Decisions regarding postoperative antibiotic therapy were made independently by the treating physicians in the absence of a uniform treatment protocol. In general, postoperative antibiotic therapy was only administered to patients with extensive DFO in whom adequate surgical resection of all necrotic and infected tissues was difficult. Postoperative offloading was advised in all patients until postoperative wound healing was observed. Again, lack of a uniform treatment protocol resulted in individual decisions of treating physicians regarding the use of offloading devices and postoperative wound care (e.g., bandages).
In the participating centers, local investigators selected patients according to the inclusion criteria and retrospectively collected data from electronic health record systems which were registered in a secured database (OpenClinica LLC, Version 3.13, Waltham, MA, USA). Data regarding demographics, comorbidities, index ulcers and affected feet were collected. We defined the index ulcer as the clinically most important foot ulcer with the clearest association to the underlying osteomyelitis as judged clinically and based on imaging findings. Index ulcers located around the metatarsals, the phalanges and associated soft tissues were classified as "forefoot index ulcers", index ulcers located around the cuboid, navicular, cuneiform bones and associated soft tissues as "midfoot index ulcers," and index ulcers around the talus, calcaneum and associated soft tissues as "hindfoot index ulcers" [8]. Additionally, we classified index ulcers according to the "Site, Ischemia, Neuropathy, Bacterial infection, Area, Depth" (SINBAD) classification [34].
We collected data of loss of protective sensation and deformities of the affected foot. Loss of protective sensation was defined as absence of pressure sensation of a 10-g monofilament [13]. We classified deformities as mild (pes cavus, hallux valgus, hallux limitus or hammer toes), moderate (hallux rigidus, claw toes or prominent metatarsal heads) or severe (Charcot neuroarthropathy-related deformity, previous ankle arthrodesis or previous partial calcanectomy) [35]. The deformity graded most severe determined the classification [35]. We also registered previous contralateral major amputations (defined as any resection proximal to the ankle in correspondence with the IWGDF definition) and previous ipsilateral minor amputations [8,35]. Moreover, we collected data regarding the surgical procedures and postoperative treatments. Furthermore, we collected data of the microbiological culture results of intraoperatively obtained bone samples, including gentamicin-resistance of pathogens which was investigated using the minimal inhibitory concentration breakpoints of the European Committee on Antimicrobial Sensitivity Testing (EUCAST) (Växjö, Sweden) [36].
Follow-up was completed until minor or major amputation after the initial surgery, repeated surgical resection of DFO, death or the last-mentioned consultation in the electronic health record system. We registered data regarding postoperative wound healing, ulcer recurrence, minor and major amputations, readmissions, reoperations, functional results and adverse events, including postoperative fractures, Charcot neuroarthropathy exacerbations and deaths. Wound healing was defined as macroscopic complete epithelialization after removal of abundant callus without drainage or requirement of wound dressings, maintained for a minimum of 2 weeks [8,15]. We defined ulcer recurrence as development of an ulcer on the same location as the index ulcer after initial postoperative wound healing irrespective of the presence of (ongoing) osteomyelitis. Functional results were classified as unable to mobilize weight-bearing, able to mobilize weight-bearing with a walking aid or able to mobilize weight-bearing without a walking aid. Information was also collected regarding footwear used at the final follow-up.
As primary outcomes, we assessed postoperative wound healing and ulcer recurrence. Treatment success was defined as uncomplicated wound healing without ulcer recurrence, and treatment failure was defined as presence of a persistent wound at final follow-up or ulcer recurrence after initial postoperative wound healing. A persistent wound was defined as absence of postoperative wound healing after primary closure or closure by reconstructive techniques. As secondary outcomes, we assessed minor and major amputations and functional results at final follow-up.
Categorical data are reported as numbers of patients with corresponding percentages and continuous data as means with standard deviations (SD) or medians with interquartile ranges (IQR) when nonparametric. We compared characteristics regarding demographics, comorbidities, index ulcers, affected feet, surgical procedures and postoperative treatments between patients with treatment success, patients with treatment failure due to persistent wounds, and patients with treatment failure due to ulcer recurrence after initial postoperative wound healing. These comparisons were made using one-way analyses of variances (ANOVA) for parametric continuous data, Kruskal-Wallis tests for nonparametric continuous data and Fisher exact tests for categorical data. All tests were performed two-sided (α = 0.05). Post hoc analyses were performed using a Bonferroni correction.
A univariable Cox regression analysis was performed to investigate associations between treatment failures and all above-mentioned characteristics. We used a Kaplan-Meier curve to demonstrate treatment failures, in which patients with persistent wounds were indicated as having an event at day 1. SPSS 23.0 software (IBM, Armonk, NY, USA) was used for all statistical analyses.

Patients and Procedures
From 13 hospitals, we included 64 patients, of whom 49 (77%) had confirmed osteomyelitis and 15 (23%) had suspected osteomyelitis ( Figure 2, Table 1). The surgical procedure with gentamicin-loaded CaS-HA biocomposite involved minor amputations in five patients (8%) with forefoot DFO and three patients (5%) with midfoot DFO ( Table 1). Details of the surgical procedures are listed in Appendix A: Table A1.

Patients and Procedures
From 13 hospitals, we included 64 patients, of whom 49 (77%) had confirmed osteomyelitis and 15 (23%) had suspected osteomyelitis ( Figure 2, Table 1). The surgical procedure with gentamicin-loaded CaS-HA biocomposite involved minor amputations in five patients (8%) with forefoot DFO and three patients (5%) with midfoot DFO (Table 1). Details of the surgical procedures are listed in Appendix A: Table A1.   Table A1. Gram-negative. OA: Obligate anaerobic. * Postoperative follow-up, which was completed until amputation, repeated surgical resection of DFO, death, or the last-mentioned consultation in the electronic health record system, was significantly (p = 0.002) shorter in patients with persistent wounds. ** Distribution of Charcot osteoarthropathy was not significantly different between patient groups. *** Distribution of limb ischemia was not significantly different between patient groups. † Index ulcer duration was significantly longer in patients with ulcer recurrence (p = 0.014). † † Of patients with persistent wounds, a significantly larger proportion underwent surgical procedures with gentamicin-loaded CaS-HA biocomposite that involved a minor amputation (p = 0.004). ‡ Of patients with persistent wounds, a significantly larger proportion had gentamicin-resistant DFO (p = 0.007).

Primary Outcomes
Median postoperative follow-up was 43 (IQR, 20-61) weeks (Table 1). We observed uncomplicated wound healing in 54 patients (84%) and a median time to wound healing of 9 (IQR, 5-16) weeks ( Figure 2). Of 10 patients (16%) with persistent wounds, four underwent minor amputations during follow-up, four underwent major amputations, one underwent repeated surgical resection of DFO without antibiotic-loaded CaS-HA biocomposite, and one still had an ongoing wound at a final follow-up of 14 weeks (Figure 2). Recurrent foot ulcers were observed in 12 of 54 patients (22%) after initial postoperative wound healing, and the median time to ulcer recurrence was 24 (IQR, 16-46) weeks ( Figure 2). Thus, the rate of treatment success was 66% (42 patients) overall (Figure 2).

Secondary Outcomes
After a median follow-up of 8 (IQR, 5-23) weeks, minor amputations were performed because of treatment failures in four patients (6%), of whom two initially underwent surgical procedures with gentamicin-loaded CaS-HA biocomposite that involved minor amputations (Figure 2). Major amputations were performed after a median follow-up of 17 (IQR, 7-41) weeks in seven patients (11%), of whom five initially underwent surgical procedures with gentamicin-loaded CaS-HA biocomposite that involved minor amputations (Figure 2). At the final follow-up, 50 patients (78%) could mobilize weight-bearing, including 11 patients (17%) who had treatment failures ( Figure 2). Of these 50 patients, 48 (96%) used custom-made or prefabricated therapeutic footwear, and two (4%) used prefabricated footwear at final follow-up. Three of 14 patients (22%) who were unable to mobilize weight-bearing had pre-existent incomplete paraplegia ( Figure 2).

Treatment Failures
Patients with persistent wounds (defined as treatment failure at day one for timebased analyses) and patients with ulcer recurrence after initial postoperative wound healing are demonstrated in the Kaplan-Meier curve in Figure 3. In univariable Cox regression analysis, factors that were independently associated with treatment failure were gentamicinresistant DFO, index ulcer location and surgical procedures with gentamicin-loaded CaS-HA biocomposite that involved a minor amputation (Table 2).

Treatment Failures
Patients with persistent wounds (defined as treatment failure at day one for timebased analyses) and patients with ulcer recurrence after initial postoperative wound healing are demonstrated in the Kaplan-Meier curve in Figure 3. In univariable Cox regression analysis, factors that were independently associated with treatment failure were gentamicin-resistant DFO, index ulcer location and surgical procedures with gentamicin-loaded CaS-HA biocomposite that involved a minor amputation ( Table 2).

Adverse Events
Readmissions, reoperations and adverse events are listed in Figure 2. Charcot osteoarthropathy exacerbations were observed in two of 18 patients (11%) with Charcot osteoarthropathy during follow-up ( Figure 2). After 23 weeks of follow-up, one patient with postoperative wound healing without ulcer recurrence died of cardiac disease. No other patients were lost to follow-up.

Discussion
In this multicenter, retrospective cohort study, we investigated the treatment of DFO with gentamicin-loaded CaS-HA biocomposite in patients where conventional treatment had failed. Treatment success was observed in 66% of patients during median 43 weeks of follow-up. Treatment failure due to a persistent wound or ulcer recurrence after initial postoperative wound healing was observed in 15% and 19% of patients, respectively. After treatment failure, minor and major amputations were required in 6% and 11% of patients, respectively. Furthermore, 78% of patients could mobilize weight-bearing at final followup, including 17% patients with treatment failures. These findings confirm results from previous studies and show that surgical treatment with gentamicin-loaded CaS-HA biocomposite is feasible for patients with DFO after failed conventional treatments [23,24,28,29].
Our findings correspond with previous publications of this treatment regimen regarding wound healing rates (details are listed in Table 3) [23,24,28,29]. However, there are important differences between our study and previous publications regarding locations of DFO, surgical procedures and postoperative treatments. First, contrary to previous studies in which patients with midfoot or hindfoot DFO were predominantly included, we mainly included patients with forefoot DFO [24,28,29].      Second, the surgical procedures with gentamicin-loaded CaS-HA biocomposite involved minor amputations in 13% of patients in our study, whereas these were only reported in one previous study in two of 70 patients [23]. This is probably the result of the high proportion of patients with forefoot DFO in our study, in whom complete surgical resection sometimes can only be performed by minor amputation.
Third, only 41% of patients in our study received postoperative antibiotic therapy, whereas all patients were administered postoperative antibiotic therapy for several weeks in previous studies [23,24,28,29]. This difference is probably caused by the lack of uniform treatment protocols. The observed success rate, obtained in a study population of which more than half was not treated by postoperative antibiotic therapy, could suggest that systemic antibiotic therapy is not indicated in all patients after treatment with gentamicinloaded CaS-HA biocomposite. However, further investigation of the role of systemic antibiotic therapy in this treatment regimen is required before recommendations can be made.
Finally, vacuum-assisted closure of surgical wounds was not performed in our study, whereas this was performed in up to 50% of patients in previous studies [23,24]. These differences limit comparisons of our study with previous publications, and indicate that uniform protocols are needed regarding patient selection, surgical procedures and postoperative treatment for this treatment regimen.
The ultimate treatment goal in people with DFO is to become ulcer-free. Therefore, our definition of treatment failure includes ulcer recurrence. Because only postoperative wound healing was considered in previous studies of this treatment regimen, reported success rates might be overestimated [23,24,29]. We postulate that ulcer recurrence should be included in the definitions of treatment failure in future studies. Furthermore, we recommend reporting details regarding offloading and other ulcer prevention strategies in future treatment protocols, because inadequate offloading is one of the possible causes for ulcer recurrence [35].
In this study, we explored potential risk factors for treatment failures. In a univariable analysis, gentamicin-resistant DFO, hindfoot DFO and surgical procedures with gentamicin-loaded CaS-HA biocomposite that involved minor amputations were associated with treatment failures. These results should be interpreted with caution given the high confidence intervals and the small sample size with a limited number of events. However, since these potential associations might be clinically relevant, they should be investigated further. These investigations should include postoperative offloading, since the potential association between hindfoot DFO and treatment failures could be the result of offloading difficulties [38][39][40].
Amputations are frequently performed in patients with DFO after unsuccessful conventional treatments [10,11]. Minor and major amputations were performed in 6% and 11% of patients, respectively, after failed treatments with gentamicin-loaded CaS-HA biocomposite in our study. These rates are higher than the rates of minor (0-3%) and major (7-9%) amputations reported in previous studies of this treatment regimen [23,29]. This is probably caused by the inclusion of patients after failed conventional treatments, who had no options left after failure of treatment with gentamicin-loaded CaS-HA biocomposite. However, considering the selection of patients after failed conventional treatments, the observed rates of minor and major amputations were lower than expected. These findings warrant further research into this new treatment regimen, in which amputations performed after treatment failures should also be investigated.
Our study has several limitations. First, the retrospective design imposes a risk of bias. Second, the absence of current protocols for the surgical procedures and postoperative treatments resulted in a heterogeneous study population. Even though we included almost all patients treated with gentamicin-loaded CaS-HA biocomposite, and therefore obtained a representative study population for current clinical practice in the Netherlands, the resulting heterogeneity introduces various confounding factors which limit the conclusions that can be drawn. Third, the included study population was relatively small. Even though it is larger than in most previous studies, the small study population increases statistical errors. Therefore, the analyses performed should be considered exploratory and be interpreted cautiously. Fourth, our study is limited by the lack of a control group. In future prospective studies, a uniformed treatment regimen should be compared to a control cohort of patients treated by conventional treatments. Fifth, adequate investigation of ulcer recurrence was limited by the median follow-up of 43 weeks, which should ideally be 12 or 18 months after initial wound healing for this purpose [1]. Sixth, specific information regarding diabetes mellitus (i.e., glycemic control) and other risk factors for vascular disease (e.g., dyslipidemia or smoking status) was lacking, as well as details regarding the preoperative discontinuation of antibiotic therapy and the number and exact sites where bone samples were obtained. Seventh, we did not include a minimal postoperative duration in the definition of persistent wounds, which should be considered in further prospective studies. Eight, pre-operative functional status was not investigated in the assessment of functional results. Finally, assessment of persistent or recurrent osteomyelitis in patients with treatment failures was not possible, since additional investigations (e.g., imaging, bone samples) were not performed consistently. Future prospective studies will overcome these limitations.
Uniform treatment protocols are required for the treatment of DFO with gentamicinloaded CaS-HA biocomposite. For a uniform treatment protocol, suggestions for patient selection should include patients with DFO after unsuccessful treatment by antibiotic therapy for at least 6 weeks [14]. Patients with severe limb ischemia should be excluded [33]. Regarding the surgical procedures, we advocate thorough surgical resection of DFO and obtaining multiple bone samples. Regarding postoperative treatments, we suggest offloading in accordance with the IWGDF guidelines until postoperative wound healing is observed, at least several days of postoperative antibiotic therapy based on the results of microbiological analysis of bone samples, and structural follow-up in a multidisciplinary setting [14,41]. Prospective investigation of treatment protocols, based on the insights reported in our study and previous studies, is required [23,24,28,29]. This prospective investigation should consider postoperative wound healing, ulcer recurrence, amputations and functional results in comparison to the pre-operative functional status.
In conclusion, surgical treatment with gentamicin-loaded CaS-HA biocomposite was feasible in this study of patients with DFO and successful in 66% of patients. A prospective trial of this treatment regimen, based on uniform treatment protocols, is required.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to confidentiality agreements. Supporting data can only be made available to bona fide researchers subject to a non-disclosure agreement. Details of the data and how to request access are available from Kor H. Hutting

Surgical Procedures No. (%)
Amputation of metatarsal head(s), debridement of medullary canal(s) and filling with gentamicin-loaded CaS-HA biocomposite. 8 (13) Arthrotomy of tarsometatarsal joint(s), resection of joint with distal part of one or multiple tarsals and base of one or multiple metatarsals, dead space filling with gentamicin-loaded CaS-HA biocomposite.