2.1 Clinical

Patients may exhibit clinical manifestations pathognomonic for FRI, such as fistula, sinus tract, wound breakdown, or visible pus from wound at the site of the fracture, which are part of the confirmatory diagnostic criteria established by AIGEC.³ A retrospective cohort study by Sigmund et al. found sinus tract to have a sensitivity of 45.7 % and specificity 100.0 %, and visible pus to have a sensitivity of 50.0 % and specificity of 96.8 %.⁴ Other clinical signs and symptoms of pain, erythema, swelling, warmth, fever, persistent post-operative drainage from wound, and new joint effusion, are clinical criteria that may suggest the presence of FRI. However, they are often difficult to distinguish from the expected inflammatory responses to the traumatic soft tissue damage from the primary injury or surgical procedure, delayed fracture healing due to other patient factors, and pain from post-traumatic arthrosis. Hence, the presence of suggestive criteria should prompt the clinician to further evaluate for FRI through laboratory and imaging investigations.

2.2 Serum markers

Elevated serological biomarkers of C-reactive protein (CRP), white blood cell (WBC) count and erythrocyte sedimentation rate (ESR) may suggest FRI. Sigmund et al. evaluated the diagnostic value of CRP, which demonstrated a 67 % sensitivity and 60 % specificity.⁴ Other inflammatory conditions such as hospital acquired infections, post-fracture and post-surgery inflammatory responses also increase CRP. Hence, relying solely on CRP may lead to overtreatment, which may include unindicated invasive revision surgery and prolonged antibiotic therapy. Furthermore, infection by a low virulence organism or biofilm formation on implant surfaces which shelter pathogens from the host immune response can result in false negative results and hence missed diagnoses.⁵ Instead, these serum biomarkers may demonstrate more utility in patients who develop late FRI, whereby a meta-analysis found CRP to have a 77.0 % sensitivity and 67.9 % specificity in the detection of late FRI, WBC count with a 51.7 % sensitivity and 67.1 % specificity, and ESR with a 45.1 % sensitivity and 79.3 % specificity.⁶ CRP may also have a role in prognostication, whereby the mean CRP level of less virulent versus high virulence micro-organisms was significantly lower at 10.04 mg/L in former, as compared to 26.31 mg/L in the latter.⁴ Arguably, the sensitivities and specificities of these serological markers are still low in early and late FRI. Hence, they should not serve as confirmatory tests, but rather prompt further evaluation when elevated.

2.3 Histopathology and microbiology

Use of Gram stain for Gram positive or negative bacteria, Ziehl-Neelsen stain for tuberculosis, and Grocott methenamine silver stain for invasive fungal infections, can identify the presence of microorganisms in deep tissues obtained during an operative intervention.

The positive culture of distinct pathogens obtained from at least two separate deep tissue or implant (sonication) specimens is considered a confirmatory criterion of FRI. The identification of the pathogen can aid with prognostication (eg. whether the organism is low or high virulence) and guide appropriate culture-directed antibiotic therapy. To note is that pathogens causing late onset FRI are often skin commensals, and hence interpretation of the results may be confounded by the possibility of contamination if inappropriate sampling protocols are utilized. Furthermore, organisms may be slow growing in a biofilm, and thus result in false negative results.

Sonication of implant parts is the utilisation of sound waves to dislodge biofilm on implant surfaces for culture. A semiquantitative cut off point for the number of colony forming units helps differentiate infection from contamination. A systematic review by Onsea et al. that included five studies on sonication fluid culture concluded that sonication may be a useful adjunct, but that there is no evidence that it is superior to tissue culture.⁷ Even existing studies of sonication in the context of PJI have found that sonication has a lower sensitivity and specificity than tissue cultures,⁸ and such results could be extrapolated to the context of FRI.

The original consensus definition of FRI did not include the presence of polymorphonuclear neutrophils (PMNs) visualized under high-power field as part of the confirmatory diagnostic criteria. However, a recent study has demonstrated that for patients with late onset or chronic FRI, the complete absence of PMNs has a high correlation with aseptic nonunion, while >5 PMNs per HPF (at 400x magnification) is always associated with infection,⁹ and this criterion has now been included as a confirmatory sign of FRI.

2.4 Imaging

Plain radiographs are often the first line imaging to monitor for fracture healing. Radiological findings of bone lysis, implant loosening, sequestration, non-union and periosteal reaction at non-fracture sites are suggestive of FRI. A study by Li et al. found that radiolucent lines and periosteal reaction were associated with infection, with a low sensitivity (16.4 % and 12.1 % respectively) and high specificity (90.5 % and 94.8 % respectively).¹⁰ However, studies have demonstrated that within the first two weeks after fracture fixation, bone does not show signs of osteomyelitis or osteolysis despite having an infected implant.¹¹ Hence, diagnosis of FRI on plain radiographs may lead to delayed diagnosis.

Three-phase bone scan has a limited role in FRI. With a high sensitivity but low specificity,¹² it can be utilized as a screening tool in patients with a low pre-test probability of infection, whereby a negative bone scan result likely excludes the presence of FRI.

Computed tomography (CT) scans can provide more bony detail, better demonstrating the presence of poor fracture healing, implant loosening, osteolysis, sequestration, and periosteal reaction. However, the sensitivity and specificity for FRI are poor (47 % and 60 % respectively).¹³

Magnetic resonance imaging (MRI) is more useful at delineating soft tissue pathology. However, despite utilizing metal artefact reduction sequencing techniques, metal implants would still result in significant scattering, hence obscuring details and making interpretation of scan images challenging. In addition, features of infection, inflammation and post-surgical changes appear similar on MRI, hence resulting in a poor specificity and limited utility for the detection of FRI.

3.1 Serum markers

Besides routinely used inflammatory markers of CRP, WBC count and ESR, other inflammatory cytokines can be studied to evaluate their utility and cut off values that would support the diagnosis of FRI. A literature search on Pubmed on the role of inflammatory cytokines yielded four studies. Interleukin-6 (IL-6) is a pleiotropic cytokine that participates in tissue repair and regeneration. During tissue injury and infection, IL6 is rapidly generated, and high IL-6 levels may suggest underlying FRI. Zhao et al. has found IL-6 to have a similar diagnostic efficacy as ESR and CRP.¹⁴ Baertl et al. found a more significant increase in IL-1β, IL-6, IL-10 and interferon-γ for high virulence Staphylococcus aureus than low virulence strains in mice models.¹⁵ Sabate-Bresco et al. found upregulation of IL-17A in all mice infected with Staphylococcus epidermidis.¹⁶ Hence, inflammatory cytokines may be useful for diagnosis of FRI, may correlate with the virulence of the organism, and elevations in specific inflammatory cytokines could potentially suggest the type of causative pathogen.

Developing a panel of inflammatory biomarkers for diagnosis of FRI can be considered. A prior study had evaluated plasma levels of 49 protein inflammatory biomarkers using enzyme-linked immunosorbent assays and mid-infrared spectral patterns to compare between 13 FRI patients and 13 matched controls,¹⁷ showed promising results. Future studies with a larger sample size can help identify key inflammatory biomarkers to be included in a diagnostic panel for FRI.

Hypocalcemia and thrombocytosis have also been observed to be associated with FRI. Liu et al. found that the incidence of asymptomatic hypocalcemia was significantly higher in FRI patients than patients with aseptic nonunion or controls (22.94 % vs 6.92 % vs 8.82 % respectively, p < 0.001).¹⁸ Platelets are acute phase reactants, whereby platelet count (PC) increases and mean platelet volume (MPV) decreases as a result of an infectious or inflammatory process. Hence, PC/MPV ratio has been found to be a promising tool for ruling out FRI with a specificity of 55.6 % and sensitivity of 100 %.¹⁹

3.2 Histopathology, microbiology and molecular diagnostics

In the context of PJI, studies have validated surgical sampling protocols and demonstrated that they improve culture yield.^20,21 Strategies include cessation of antibiotics for at least two weeks prior to obtaining deep tissue culture, obtaining adequate tissue samples, and methods to avoid cross contamination such as the no touch technique when sampling. In the context of fractures whereby external forces have resulted in the primary injury, there may be superficial wounds that may or may not extend into the fracture site. This differs from PJI whereby there are often no superficial traumatic wounds. Hence, in the context of superficial traumatic injuries, one should avoid going through sites of superficial injury when obtaining cultures so as to avoid contamination and misdiagnosis of superficial infection as FRI.

Obtaining intramedullary cultures may also be considered. Onsea et al. evaluated the use of the Reamer-Irrigator-Apirator (RIA) system (Depuy Synthes; Johnson & Johnson Co, Inc,NJ), whereby RIA debris of patients with FRI demonstrated similar sensitivity to tissue cultures (71 % vs 67 %).²² In 4 out of 24 FRI patients, RIA cultures yielded additional relevant pathogens not found on standard tissue cultures. In the control group of aseptic nonunion patients, RIA cultures had no false positive.

Laboratory culture techniques can be optimized. Altering incubation conditions and processing techniques have been found to improve culture yield in PJI.²³ The pathogens can be slow growing and low in numbers, and hence enriched cultures with prolonged incubation are necessary to improve yield.²⁴ Vortexing with sterile glass beads may also have a role in facilitating biofilm disruption in FRI to improve detection.^25,26 As pathogens in FRI and PJI have been demonstrated to have a similar pathogen distribution,²⁷ applying the aforementioned concepts studied in PJI to FRI may improve culture yield as well. In summary, employing strict tissue sampling and laboratory processing protocols have been demonstrated to be effective in improving diagnosis in PJI, and thus should be implemented and evaluated in the context of FRI as well.

Previous studies have demonstrated that up to 15 % of PJI cases are culture negative,^28,29 yet over 90 % of samples in culture negative patients demonstrate bacterial nucleic acid when molecular diagnostics are utilized.³⁰ This is likely the case for FRI as well, whereby there can be a large proportion of false negative cultures. Real-time polymerase chain reaction (rt-PCR) panels can be performed to test for common pathogens in FRI. A search on Pubmed yielded eight studies on PCR techniques in FRI. Renz et al. compared sonication fluid PCR vs cultures, and found PCR to be comparable to cultures but with the added advantage of a shorter processing time and being a fully automated procedure.³¹ However, the performance of swab PCR is inferior to conventional tissue cultures.³²

Metagenomics, which is the study of genetic material from a mixed community of organisms, may also aid with early detection of pathogens.³³ It can be used to identify a wide range of pathogens inclusive of bacteria, viruses, fungi and parasites, and can also detect the presence of drug resistant genes in pathogens so as to guide pharmacological therapy. A 2021 expert consensus by the Laboratory Medicine Branch of the Chinese Medical Association had in fact recommended the use of metagenomic next-generation sequencing³⁴ to identify pathogens in suspected infection sites whereby no causative organism has been identified through traditional laboratory techniques. However, molecular diagnostic techniques are costly, and hence should be utilized judiciously, perhaps only in patients whereby conventional culture techniques have yielded inconclusive results. Limitations of metagenomics include the fact that pathogens identified may be dead or dormant microorganisms that may not be of clinical significance, contamination by environmental pathogens can confound results, and that for intracellular pathogens, appropriate cell wall rupture methods must be utilized so that the intracellular organism can be identified. Hence, further large-scale trials are required to assess the clinical efficacy and cost effectiveness of molecular methods such as metagenomics.

Measuring wound alpha-defensin (AD) level has also showed promising results. AD is an antimicrobial peptide that is secreted by neutrophils in response to pathogenicity, so as to aid the host immune system with rapid killing of the pathogen. A search on Pubmed of utility of AD in FRI yields only one study. The study is a small prospective cohort study by Kumar et al.,³⁵ wound AD levels were significantly elevated at a 2.6-fold increase in FRI patients as compared to controls. Future larger scale studies are required to further validated these results before wound AD can be used as a biomarker for diagnosis of FRI. Other wound biomarkers such as wound neutrophil esterase 2, bactericidal/permeability-increasing protein, neutrophil gelatinase-associated lipocalin and lactoferrin had previously shown promising results as biomarkers for PJI,³⁶ and may be valuable investigating in the context of FRI as well.

MicroRNAs (miRNA) regulate proliferation and function of osteoblasts and osteoclasts, which are required for bone healing in fracture. In FRI, the inflammatory reaction as a result of infection impairs bone healing, and hence may be reflected by miRNA levels. It has been demonstrated that miRNA-331-3p is downregulated in rabbit models with infected nonunion fractures.³⁷ Human studies are required to evaluate whether similar downregulation is observed in human subjects, and to explore whether miRNA-331-3p may be a potential therapeutic target to prevent nonunion caused by FRI.

An interesting study by Zhao et al. found dysbiosis of the gut microbiota to be associated with FRI.³⁸ In the study, 50 participants with FRI, healed fracture with no FRI and healthy controls, underwent analysis of their gut microbiota present in fecal samples. Patients with FRI were found to have a higher abundance of bacteria of the genus Escherichia and Streptococcus, and a relative decrease in Roseburia as compared to subjects with healed fractures. Given the small study size and the observational case control nature of the study by Zhao et al., future prospective studies with a larger study population are required to further evaluate the causational relationship between FRI and gut dysbiosis, as opposed to the presence of gut dysbiosis being a risk factor for patients to develop FRI. Some hypotheses posit that stressors such as infection may cause intestinal ischemia-reperfusion, which affects the intestinal epithelial and mucosal layers, thereby driving gut dysbiosis. Other studies have also demonstrated how diseases such as ischaemic stroke, inflammatory bowel disease and ankylosing spondylitis can affect gut microbiota,^39–41 and hence future studies to evaluate other confounding factors when using gut microbiota as a diagnostic tool for FRIs are needed.

3.3 Imaging

To achieve early detection of FRI, medical imaging techniques and modalities need to identify the presence of infection, the location of infection, and the extent of the infection. The diagnostic imaging principles utilized to identify PJI cannot be extrapolated to FRI as bone metabolism is increased in fracture healing, and hence may affect interpretation of infection on imaging.

Nuclear imaging is an up-and-coming field of imaging in modern day medicine. A Pubmed search yielded 15 studies on nuclear imaging in FRI. WBC scintigraphy can detect and localize metabolically active cells. Govaert et al.’s large retrospective single-center study found WBC scintigraphy to have a high diagnostic accuracy whereby the duration of the time interval between surgery and the scan, and concomitant use of antibiotics did not affect the diagnostic accuracy.⁴² By identifying the pattern of radiolabelled WBC accumulation over time, active infection can be discriminated from other causes of inflammation such as post-surgical changes and foreign body reactions, hence demonstrating a high specificity of 97 % and low sensitivity of only 79 %.⁴³ PET/CT utilises fluorodeoxyglucose which is taken up by immune cells that use glucose as an energy source, thereby identifying highly metabolic areas. However, this does not allow for discrimination between infectious and inflammatory cells, and thus results in a poorer specificity of 91 % when evaluated in patients with osteomyelitis.⁴⁴ It is likely that in FRI patients, fracture and recent surgical fixation would further confound the findings and lead to a poorer specificity, for which future studies can evaluate.

Future developments in nuclear imaging techniques should aim at identifying the causative pathogen of FRI. In addition, identifying specific resistance mechanisms of the pathogen, such as thickness and maturation of the biofilm, can provide information on challenges that may be faced in eradication of the pathogen, such that treatment methods can be titrated accordingly. Britton et al. utilized a radiolabelled antibiotic, ciprofloxacin, to identify specific bacteria, but only demonstrated a low specificity of 81.7 %.⁴⁵ Future studies can evaluate the use of other radiolabelled antibiotics. Other infection specific imaging tracers can also be considered, such as compounds that would bind to bacterial cells (prokaryotic) but not to mammalian cells (eukaryotic).

Recently, fluorescent tracers have been gaining traction in tumour resection surgery.⁴⁶ Fluorescent dye administered to patient pre-operatively results in illumination of the tumour of interest during cancer surgery, thereby detecting the site and extent of tumour to guide resection margins. When an external light source of a specific wavelength is used, the fluorescent tracer which is bound to the pathogen of interest would generate an optical signal on excitation, thereby illuminating the target tissue.⁴⁷ A Pubmed search revealed four studies on use of fluorescence in FRI. Lopez-Alvarez et al. evaluated the use of a near-infrared fluorescent tracer composed of antitbiotic vancomycin and fluorophore IRDye800CW applied to extracted orthopaedic implants, which allowed prompt diagnosis of FRI to within <30min, and is a potential means for on table diagnosis of FRI.⁴⁸ Future research can evaluate if such methods may similarly be applicable to FRI through developing targeted fluorescent tracers with a targeting moiety that has binding affinity for specific pathogen-associated molecular targets or biomarkers. Fluorescence image-guided surgery can also enable the orthopaedic surgeon to identify the presence of infection intra-operatively by looking for tissues that illuminate, and carry out sufficient surgical debridement and decontamination such that the visible pathogen load viewed via in vivo optical imaging is reduced to a minimum prior to wound closure, so as to achieve optimal source control. Rupp et al. evaluated the use of VELscope, an intraoperative tool to differentiate between viable and necrotic bone using fluorescent technology, and found the technology to have good correlation with pathohistological findings of bone necrosis, thereby aiding adequate intra-operative bone resection in patients with chronic FRI.⁴⁹