Single‐stage versus two‐stage bone flap reconstruction in chronic osteomyelitis: Multicenter outcomes comparison

Chronic osteomyelitis is an invalidating disease, and its severity grows according to the infection's particular features. The Cierny‐Maiden criteria classify it according to the anatomical aspects (I to IV) and also by physiological class (A host being in good immune condition and B hosts being locally (L) or systemically (S) compromised). The surgical approach to chronic osteomyelitis involves radical debridement and dead space reconstruction. Two‐stage management with delayed reconstruction is the most common surgical management, while one‐stage treatment with concomitant reconstruction is a more aggressive approach with less available literature. Which method gives the best results is unclear. The purpose of this study is to compare single and two‐stage techniques.

(p = .0086)and in Pseudomonas aeruginosa sustained infections (p = .0208),but in the latter case, the treatment strategy did not influence the outcome (p = .4000).
Conclusions: Bone union and infection healing rates are comparable between one and two-stage approaches.Pseudomonas aeruginosa infections have a higher risk of infection relapse, with similar effectiveness of one-and two-stage strategies.B-hosts have a higher infection recurrence rate without comparable data between the two approaches.Further studies with a larger sample size are required to confirm our results and define B-hosts' best strategy.
Level of Evidence: Level III of evidence, retrospective cohort study investigating the results of treatments.

| INTRODUCTION
Chronic osteomyelitis is a long-standing, recurrent bone infection characterized by low-grade inflammation involving bone and surrounding soft tissues (Lew & Waldvogel, 1997, 2004).The treatment combines surgical procedures and antibiotic therapy (Doi et al., 1995), aiming for infection healing, skeletal stability with bone reconstruction if needed, and skin closure (Cierny III, 2011).The surgical management consists of radical debridement and management of resulting tissue defect called "dead space."(Cierny III, 2011) The entity of debridement depends on the Cierny-Mader clinical stage (Cierny III et al., 2003;Mader et al., 1997).Therefore, it leads to more significant tissue in more advanced settings that require bone and sometimes soft tissue reconstruction (Marais et al., 2014).
Single-stage and multi-stage surgery strategies differ in timing between radical debridement and reconstruction.The two-stage approach includes a first surgical step of radical debridement, with temporary skeletal stabilization, filling of dead space with spacers if needed, and usually immediate wound closure.Specific antibiotic therapy follows this first stage.When infection healing occurs, a second surgical procedure definitively reconstructs the bone defect and, if further debridement is required, the resultant soft-tissue defect (Dinh et al., 2009;Tulner et al., 2004).Two-stage treatment is often preferred, especially in advanced stages (Dinh et al., 2009;Pande, 2015): it permits obtaining infection healing before definitive bone reconstruction, to proceed with further and more radical debridement if needed, and to employ temporary antibiotic spacers to bring local antibiotics and create a receiving space for a further bone flap (Dinh et al., 2009;Tulner et al., 2004).This surgical strategy has disadvantages of several surgical procedures and patients' prolonged morbidity (Rhomberg et al., 2003).
The one-stage procedure consists of radical debridement and concomitant defect reconstruction.Carrying well-vascularized tissue to the damage site can prevent further damage caused by exsiccation, infection, and fibrosis and promote bone union (Godina, 1986;Philandrianos et al., 2018).It is a more aggressive approach, with the risk of losing the grafted tissue and sometimes the entire limb.However, it reduces patients' morbidity, the need for multiple surgical procedures and prolonged inhospital stay, and improves vascularity thanks to immediate reconstruction, local antibiotics, and oxygen delivery (Doi et al., 1995;Wood & Cooney, 1984).Moreover, it has the advantage of permitting a generous debridement without regard to the size of the resulting defect because concomitant reconstruction solves the trouble of temporary management of dead space (Doi et al., 1995;Wood & Cooney, 1984).One-stage approach suits selected cases: the need for short hospitalization and bone defects with concomitant significant skin defects that can be reconstructed in single-stage using composite bone-soft tissue flap (Yajima et al., 2004).Limited literature is available about one-stage procedures with vascularized bone flaps in chronic osteomyelitis in small case series, with good reported outcomes in terms of bone union and infection healing.(Ciclamini et al., 2020;Doi et al., 1995;Rhomberg et al., 2003;Sun et al., 2010) This study compares the outcomes of one-stage and two-stage approaches with bone reconstruction using vascularized bone flaps in chronic limb osteomyelitis.The inclusion criteria were: a diagnosis of chronic osteomyelitis of one limb; the surgical treatment with radical debridement and resultant bone tissue defect requiring reconstruction with bone flap (free or pedicled, bone alone or composite); a minimum follow-up of 24 months.

| PATIENTS AND METHODS
The authors collected patients' names and their information about medical history (infection onset timing and etiology, skin lesions, previous surgical procedures) and clinical course in hospital datasets.The authors classified the osteomyelitis stages according to Cierny-Mader (Cierny III et al., 2003).They grouped the patients into A-hosts and B-hosts according to the presence of predisposing immunodeficiency factors proposed by Cierny-Mader (Cierny III et al., 2003).When soft-tissue loss was present, a muscular or a fasciocutaneous flap covered the spacer.After a minimum of 3 months, a second surgery (two stages) removed the cement spacer and reconstructed the bone defect.Also, in the two-stage procedure, the authors performed bone reconstruction stabilization using ORIF or external fixation, depending on the need for stability of bone.

| RESULTS
We retrieved 27 chronic limb osteomyelitis cases in 26 patients.Four patients were lost to follow-up before 24 months and excluded from the study.Therefore, 23 cases in 22 patients were finally evaluated (Figure 1).Patients were retrospectively assessed, with an average follow-up of 56.2 months (24-109 months).Ten cases (43.5%) sustained a two-stage approach: radical debridement, concomitant primary soft-tissue closure, and antibiotic PMMA spacers implanted in 7 patients.Temporary skeletal stabilization was necessary, using external fixation in 8 cases and splint in the other two patients.
Antibiotic therapy followed the first surgical procedure.A second surgery completed the treatment after a mean period of 116 days (14-323 days), and it consisted of bone reconstruction (with bone or composite flaps), bone coverage, and skeletal stabilization.Six patients took further antibiotic therapy after the second surgery.(range 16-62 years).Sex and age distributions between one and two-stage groups did not significantly differ (p = .4347and p = 7798, respectively).The upper limb was affected in 12 cases, and the lower limb was involved in 11 patients.In the one-stage group, the inferior limb was more affected than the superior one, while the two-stage group was the opposite (p = .0360).
Twenty-two cases were Cierny-Mader stage 4; only one case (heel osteomyelitis in one-stage group) was a Stage 3. Four cases were compromised hosts (17.4%) because of drug addiction, alcoholism, smoking, and viral hepatitis, and they entered the study as B-hosts (Calhoun & Manring, 2005).All B-hosts joined the one-stage approach group, and this choice was because of the critical soft-tissue defect that needed simultaneous reconstruction with bone defect using composite vascularized bone flaps.
Intraoperative samples for bacterial cultures taken during debridement resulted in a negative in 8.7%.The other 91.3% was positive for 1 to 4 microorganisms, 5 of which were positive for Pseudomonas aeruginosa (4 in the one-stage group and 1 in the two-stage group).Unequal distributions of B-hosts and Pseudomonas aeruginosa infections between the two groups were not statistically confirmed ( p = .1044and p = .3393,respectively).
All 23 cases were contiguous or direct penetration osteomyelitis: 56.5% caused by exposed fracture, 30.4% originated by previous osteosynthesis surgery, and 13.0% (forearm osteomyelitis) caused by drug extravasations in drug-addicted patients.Soft tissues were intact in only 8.7% (two cases) and compromised in 91.3% of patients (ulcer, fistula, and exposition of underlying bone).59.3% (16 patients) sustained one to three previous surgical procedures before enrolling in the study.
The average onset time was 23.4 months (range 1-93 months), with a mean onset time of 32.0 months (range 93-1 month) in the one-stage group and 13.1 months (range 31-2 months) in the two-stage group ( p = .7159).
After bone reconstruction, 22 cases required skeletal stabilization with plates, external fixation, Kirschner wires, or a combination (Table 2).In only 1 case, the bone flap was press-fitted.Immediate wound closure was achievable in 52.2% of all cases.In the remnant patients, soft tissue defects were more significant, and definitive soft tissue reconstruction was possible only using skin grafts, soft tissue flaps, or bone composite flaps (Table 3).Direct closure was possible in 38.5% of the one-stage group, while the two-stage group achieved it in 70.0% of the cases.Overall, bone union occurred in 73.9% (17 cases) in a mean period of 9.0 months (range 2-28 months) (Table 4).
In the two-stage group, bone union was higher (80.0%) and faster (mean of 6.8 months) than in the one-stage group (69.2% union rate in a mean time of 11.0 months).Still, a statistical difference was not demonstrated ( p = .6600for bone union rate and p = .1731for bone union timing).Two of the six cases (33.3%) of bone nonunion occurred in compromised hosts (representing only 17.4% of our sample).The B-hosts bone union rate was 50.0%, while it reached 78.9% in A-hosts, but the difference was not statistically significant (p = .5392)(Table 5).The bone union seems to be not related to Infection healing was obtained in 82.6% (19 cases), while infection recurrence occurred in 17.4% (4 cases) in a mean time of 8.0 months (range 2-17 months) (Table 4).All relapses occurred in the one-stage group, where the infection recurrence rate was 30.8%.
There was no recurrence in the two-stage group (0.0% recurrence rate), but the difference was not statistically confirmed ( p = .1044).
The infection recurrence rate in B-hosts was higher than observed in A-hosts (75.0% and 5.3%, respectively), with a significant statistical difference ( p = .0086)(Table 5).No comparable data are available between one and two-stage approaches in B-hosts because they all had one-stage management.
Furthermore, infection relapse was higher in Pseudomonas aeruginosa infection (60.0%) if compared with osteomyelitis caused by other microorganisms (5.6%), with a significant statistical difference ( p = .0208)(Table 5).In Pseudomonas aeruginosa high-risk forms, one or two-stage approaches did not influence the outcome ( p = .4000).

| DISCUSSION
In our overall series, one and two-stage approaches' outcomes are comparable and do not differ significantly in bone union and infection healing (Figure 2).The bone union rate in the two-stage group seems higher, and the infection healing is higher than in the one-stage case group.Still, despite this, the analysis does not give a statistically significant result.As such, the surgical approach does not affect the bone union rate significantly.All four infection relapses occurred in the one-stage group, with a recurrence rate of 30.8% in the one-stage approach and a 0.0% relapse rate in the two-stage treatment group, but without a statistically confirmed difference.
Patients' comorbidities and Pseudomonas aeruginosa correlate to statistically higher infection recurrence rates.A significantly higher recurrence rate occurred in B-hosts (75.0%, while it reached only 5.3% in A-hosts).Because of the absence of B-hosts treated with two-stage management, comparing the infection healing rate in these high-risk patients between the two different timing approaches is impossible, representing a limit of the study.As previously explained, surgeons treated all B-hosts with the one-stage method to reconstruct bone and soft tissue defects with composite bone flap in a single surgery, reducing the need for multiple surgical procedures as much as possible.Besides the compromised immune system, drug or alcohol-addicted patients often demonstrate poor compliance to therapeutic processes and follow-up checkups, especially when a stable caregiver is absent.A more conservative two-stage management could be indicated in B-hosts high-risk patients, providing that adequate social care is present in drug-addicted patients to avoid patients' loss between the first and second surgical steps.
Disease sustained by Pseudomonas aeruginosa showed 60.0% infection relapse, while the recurrence rate was 5.6% in other microorganisms' infections.Furthermore, in Pseudomonas aeruginosa infections, the choice between one and two-stage approaches did not influence the infection healing.In our series, all 4 B-hosts and 4 out of 5 Pseudomonas aeruginosa infection cases enrolled in the one-stage group (Figure 1).This irregular distribution between two groups of high-risk patients may act as a confounding factor and explain the higher, but not statistically relevant, difference between the infection healing rate in one and two-stage groups.This study compares the more aggressive and less employed one-stage reconstruction in limb osteomyelitis to the more widespread two-stage strategy; meanwhile, the few other studies available in the literature analyze one-stage reconstruction outcomes without a statistical comparison with other treatments.This new comparison represents the novelty of this study (Ciclamini et al., 2020;Doi et al., 1995;Rhomberg et al., 2003;Sun et al., 2010).
Our study also presents some limits.It is an observational retrospective study with a small sample.As an observational retrospective multicentric study, there was intrinsic bias, and confounding factors may influence results: the surgeon's choice between the two approaches was not random but was based on bone and soft tissue defect features with consequent non-randomized patients' distribution.The therapeutic strategies and decision-making may differ in the different centers.Our limited sample size is due to the infrequent advanced osteomyelitis stages requiring bone flap reconstruction, but it is comparable with other similar studies in the literature (Ciclamini et al., 2020;Doi et al., 1995;Rhomberg et al., 2003;Sun et al., 2010).
Future research will need a larger cohort, making it possible to include the etiology of osteomyelitis.Also, the factors affecting osteomyelitis should be studied separately (systemic and local factors in Cierny-Mader classification, such as vascular status of the limbs).
Further studies with larger samples must confirm our results and define the best timing approach in high-risk B-hosts.

| CONCLUSIONS
Bone union and infection healing rates are comparable between onestage and two-stage approaches.No specific risk factors influenced bone nonunion.We identified two infection recurrence risk factors: Pseudomonas aeruginosa sustained infections and B-compromised hosts.A more conservative multi-stage surgical management should probably be preferable in compromised patients.
An observational, retrospective, multicentric, non-randomized cohort comparative study was designed, including patients with limb chronic osteomyelitis diagnosis treated at the Trauma Center in Turin (CTO Hospital) and the Center for Bone Infections in Albenga between February 2010 and February 2019.The study was designed according to the STROBE panel.The study conformed to the World Medical Association Declaration of Helsinki (June 1964), and our ethical local committee approved the study protocol.
Patients in the one-stage group sustained radical surgical debridement, resectioning all the infected bone and soft tissue, and concomitant dead space reconstruction with pedicled or free vascularized bone flaps when needed.The bone stabilization was realized with ORIF or external fixation (depending on the need and characteristic of the osteosynthesis).In the case of soft-tissue loss, concomitant softtissue reconstruction with bone coverage (osteocutaneous bone flaps) fulfilled the requirements of soft-tissue repair.In the case of free flaps, the authors performed end-to-side anastomosis in the upper and lower limbs to preserve the paramount vascular axises integrity.Patients in the two-stage group sustained a first surgical procedure with radical debridement of septic bone and soft tissue and temporary dead space management using antibiotic cement spacers.
Evaluated primary outcomes were bone union (monitored by periodical x-ray) and infection healing (defined as the absence of local fistula and signs and symptoms of infection, assessed with clinical monitoring).The analysis looked for correlations between outcomes and type of treatment.Other potentially confounding factors' distribution and influence on the outcomes were also evaluated, such as patients' comorbidities and type of infection.The authors performed a statistical analysis using nonparametric tests.Fisher's exact test measured the associations between outcomes and qualitative factors.Mann-Whitney U-test was employed to assess correlations between outcomes and continuous variables.p-values less than .05were considered statistically significant.
Thirteen cases (56.5%) sustained a single-stage treatment consisting of a single surgery of radical debridement, concomitant soft-tissue coverage, and bone reconstruction.Antibiotic therapy always follows single-stage surgery.The sample included 21 males (1 with bilateral forearm osteomyelitis) and one female, with an average age of 40.7 years F I G U R E 1 Flow diagram of eligible, excluded, and included participants.
other potential confounding factors like patient age ( p = .5988),internal or external devices ( p = .9999),polymicrobial infections (p = .6213),Pseudomonas aeruginosa infection p = .5756),disease duration ( p = .2086),and disease location ( p = .9999).Two of the six T A B L E 1 Flaps used in bone reconstruction.nonunion had revision surgery.Another patient presented aseptic delayed consolidation in previous femoral osteomyelitis, and because of functional impairment, he sustained an over-knee amputation.