Chronic osteomyelitis frequently occurs as a result of direct contamination from open fracture, open surgery or as a result of spread (haematogenous or direct) from other sources of infection. Chronic osteomyelitis is classified according to its transverse extent and the ‘host status’, which is dependent on the immune status of the patient. Systemic and local factors affecting immune competence, metabolism and local vascularity are taken into account with the Cierney and Mader classification (Table 1).

The treatment of chronic osteomyelitis consists of excision of devitalised material, skeletal stabilisation, obliteration of dead space, obtaining good soft tissue cover and reconstruction of the bone, all in conjunction with antibiotics. The antibiotics are frequently given for 6 weeks intravenously, followed by a further 6 weeks orally. There has been increasing interest in systems to deliver antibiotics locally. Collatamp G, a gentamicin-impregnated collagen matrix, can be inserted intra-operatively following surgical debridement. After insertion, gentamicin quickly reaches local concentrations exceeding the minimum inhibitory concentrations (MIC) of most causative organisms; it also assists in reducing dead space. Importantly serum concentrations of gentamicin never exceed dangerous levels [1].

The purposes of this cohort study were to examine the effectiveness of surgical marginal resection in combination with local application of antibiotics (Collatamp G- gentamicin in a collagen fleece). With the information gathered, we aimed to compare our cohort with a historical cohort, in order to determine whether local application of antibiotics leads to a lower rate of disease recurrence.

This is a single surgeon, single centre, cohort study at the Royal Infirmary of Edinburgh, Scotland, United Kingdom. Patients treated for clinically confirmed osteomyelitis with Collatamp G, between January 2003 and January 2010, were identified from the hospital database and cross-checked with theatre records. All patients in this cohort were treated with marginal resection, followed by insertion of Collatamp G into area of resection. Post-operatively patients received intravenous antibiotics whilst they were in-patients and were converted to oral therapy when they were ready for discharge.

Medical documentation from in-patient records and out-patient consultations were reviewed to determine host status, post-operative outcomes, complications and recurrence. Laboratory databases were accessed for peri-operative C-reactive protein (CRP), erythrocyte sedimentation rate (ESR) and haemoglobin levels. Microbiology reports were studied to determine the causative organism.

The historical cohort selected for comparison is from the paper by Simpson and colleagues [2]. This cohort had marginal resection of infected tissue, followed by 6 weeks of intravenous administration of antibiotics and 6 weeks of oral antibiotics.

Statistical analysis including basic descriptive analysis, Kruskall-Wallis tests for post-operative CRP and ESR values, Kaplan-Meier plots for survival probability were performed with SPSS v18 (SPSS Inc, Chicago, USA).

50 patients were identified. Of these patients, 24 ‘A’ hosts and 26 ‘B’ hosts were identified. Demographics are summarised in Table 2 and are recorded in Tables 3 and 4 according to host status. 70% (n=35) of patients were male and the age range of the group was 15 to 77 years (mean 40.9, SD 15.9). The most commonly affected bones were the femur (38%) and tibia (36%). Length of admission ranged from 1 to 77 days (mean 9.8, SD 11.4). ‘B’ hosts spent an average of 12.3 days in hospital whereas ‘A’ hosts spent 7.1 days. The most commonly isolated organisms were Staphylococcus aureus (32%, n=16), Coagulase-negative staphylococci (22%, n=11) and Pseudomonas aeruginosa (8%, n=4). In 16% (n=8) patients, no organism was cultured. Details of post-operative complications are recorded in Table 5.

Those with a recurrence of infection required further conservative or surgical treatment. Further treatment was considered indicated if there was a recurrence of symptoms and signs, or if there was a rise in CRP or ESR accompanied by localised warmth or pain. In total, 6 patients (12%) suffered a recurrence of infection. 5 underwent further surgery for the infected focus and 1 was managed conservatively on an out-patient basis. Half of the patients were ‘A’ hosts and the other half were ‘B’ hosts. No patient required amputation and there was no mortality as a result of recurrence.

All but 2 patients had less than 2 weeks of intravenous antibiotics post-operatively. One patient continued on daily intravenous Ertapenum for a total of 5 weeks, including after discharge from hospital, as this was the only antibiotic the causative organism was sensitive to, and he was the only patient to have out-patient systemic antibiotic treatment. The other patient who had more than 2 weeks intravenous therapy had a two-stage operation during the same admission and had prolonged administration of systemic antibiotics in order to cover both operations. 4 patients (8%) received oral antibiotics only. All patients, except those with active fracture healing, had a total of 6 weeks of antibiotic therapy.

Table 6 shows the proportion of patients with abnormal pre-operative CRP and ESR levels. 45% (n=17) of these patients demonstrated elevated CRP and ESR levels. However, 34% (n=13) patients had normal CRP and ESR levels. For 12 patients, pre-operative inflammatory markers were not available, while for 10 patients the post-operative inflammatory markers were not available. Results sampled during the post-operative period were examined. No relationship was found between the mean CRP or ESR values and disease recurrence, either within 2 weeks post-operatively or after 2 weeks post-operatively.

As in the tumour field, there is a variable length of follow up and disease-free interval. Therefore, survival analysis, a technique developed in the tumour field has been utilised to analyse the current cohort of patients. For comparison, we have constructed a similar disease-free plot with data from our study and the historical cohort from Simpson and colleagues’ study [2]and this is shown in Fig. (1). This shows a lower probability of recurrence in the current cohort, compared to the historical cohort.

Fig. (1). Disease-free plot of current cohort and cohort (Group 2) from Simpson and colleagues (2001).

Chronic deep bone infection is difficult to treat, with a poor response rate to systemic antibiotics alone. Disadvantages include potential toxicity, difficulty in achieving high concentrations at the site of infection, and compliance problems [3]. Despite advances in therapy, long-term recurrence rates have been reported to be 20-30% [4].

A Cochrane review of antibiotics for treating chronic osteomyelitis reported that the remission rates at 12 months were not found to be different between oral and parenteral antibiotics [4]. The disadvantages and poor success rate of systemic antibiotics has prompted development of systems to deliver antibiotics directly to the infected focus. The need for a local antibiotic delivery system has been recognised for many years.

Polymethyl methcrylate (PMMA) bone cement beads loaded with gentamicin have been widely used in clinical practice for over 30 years [5]. Combining antibiotics such as gentamicin with acrylic cement allows gradual release of the drug, typically over weeks to months [6, 7]. This provides sustained postoperative antibacterial cover. Pharmacokinetic studies show that on average, 5.78% of the gentamicin implanted is eluted from the cement [6]. Serum concentrations remain low, but local to the implant, concentrations of gentamicin exceed the minimum inhibitory concentrations (MIC) of the most likely pathogens. However there are several drawbacks with cement as various factors restrict antibiotic choice. The drug must be water soluble to allow diffusion out of the cement. It also has to be heat-stable at temperatures of 100°C which occur during the exothermic curing process of the cement. The antibiotic also has to be bactericidal at low doses, as when higher doses of antibiotic are mixed with cement, the mechanical strength of the mixture is compromised [8, 9]. Also, PMMA is not biodegradable and so a second surgical procedure is required in order to remove the implants after the infection has been treated. The removal surgery is often more difficult due to local tissue scarring and adhesions may lead to further infection. In addition, a second operation adds further expense and poses the risk of further pain and anaesthetic complications.

Another local delivery system is the Lautenbach method. This involves the insertion of a double-lumen, suction-irrigation system after reaming and debridement of the intramedullary canal. This establishes both a local antibiotic delivery system and allows the volume of the cavity to be measured and the drainage fluid to be cultured. The patient is declared as free of infection when the irrigate produces three consecutive clear cultures with improved inflammatory markers and obliteration of the cavity volume [10]. A recurrence rate of 12% has been reported which is no lower than rates of patients treated with debridement and systemic antibiotics. Due to the duration the drainage tube has to remain in situ, there have been concerns of infection being introduced and further organisms have been grown from the drainage fluid that were not present in the operative samples. Hashmi and colleagues report a mean hospital stay of 27 days (range = 14-48 days) when using this technique [11]. This is a considerably longer hospital stay than our cohort of patients (mean = 9.8 days).

Collagen sponge matrices as carriers for gentamicin were developed in the early 1980s and have several advantages over other systems. Collagen is considered to be fully biodegradable and locally re-modellable. However in one case ‘fibro-gelatinous’ looking material was removed from the medullary canal of a tibial case that had suffered a recurrence and histology of this material was consistent with it being derived from the implanted collagen. Currently, therefore we use gentamicin collagen in areas where it will be in contact with tissue macrophages, such as the cortical window. This system achieves high local gentamicin levels with low serum concentrations. In rats with osteomyelitis, treatment with gentamicin-collagen has been proven to reduce bacterial colony count and to give significant therapeutic effect [12]. The effect was less marked with gentamicin-PMMA beads.

In humans, gentamicin-impregnated collagen has been shown to minimise wound infection following cardiac surgery. When compared to intravenous antibiotics only, insertion of collagen-gentamicin sponges between the two sternal halves reduced the incidence of wound infection [13]. Similarly, perineal wound healing is improved by the insertion of gentamicin-collagen fleeces after excision of rectal cancer [14].

Cierny and Mader attempted to predict which patients had a greater risk of recurrence. The ‘host status’ is determined by the presence or absence of systemic and local factors which compromise the immune system’s ability to elicit an effective response to infection [15]. This classification system allows treatment to be scaled to biological grade of disease as in the surgical management of malignancies. ‘A’ hosts have healthy immune systems and responded very well to marginal resection (resection margin < 5mm) in a previously reported series. ‘B’ hosts however, have compromised defences and have poorer success rates with similar clearance margins.

We examined the perioperative serological markers. Nearly half of patients (45%, n=17) with available blood results had raised CRP and ESR levels, whereas a third of patients with chronic osteomyelitis had normal CRP and ESR (34%, n=13) levels. Inflammatory markers such as CRP and ESR have been used to quantify the degree of tissue damage and invasiveness of a procedure. They have also been used to detect post-operative complications such as infection or loosening of an implant. Peak CRP levels are reached two days post-operatively [16, 17] and usually normalise within 3 weeks [18]. Peak ESR levels are detectable five days following surgery, after which they decrease in a slow and irregular manner. ESR can remain abnormally high for up to 42 days after uncomplicated, elective, orthopaedic surgery and so CRP is considered to be a more reliable aid in detecting post-operative complications than ESR [19]. We also attempted to distinguish the relationship between CRP and ESR levels with disease recurrence. However, no significant correlation was found.

With regards to the survival analysis, in the cohort reported in this paper, all of the patients had marginal resection and had received much shorter total antibiotic treatment and particularly shorter systemic administration. It is therefore interesting to note that the disease-free probability is higher in the cohort from this study.

We have reviewed the results of a cohort of patients with chronic osteomyelitis treated with a local delivery system of gentamicin. The disease-free probability in our cohort, treated with marginal resection and a shorter post-operative course of antibiotics, compared favourably with a similar cohort treated with prolonged systemic and oral antibiotics. This local delivery system of gentamicin is a valuable tool for treating patients with chronic osteomyelitis.