This retrospective study showed that aseptic delayed union is characterized by a lower metabolic activity than septic delayed union. Integrating this metabolic activity, expressed as SUV_max in the interpretation of the images helps in the differentiation of both conditions. In this study of 30 delayed union fractures, 13 aseptic delayed unions had a mean SUV_max of 3.23 and 17 septic delayed unions had a mean SUV_max of 4.77, which is a significant difference (P value 0.016). Normal contralateral bones had a mean SUV_max of 0.76.

For the use of the FDG-PET/CT scan as a diagnostic tool, it is essential to weight the importance of discriminating and excluding septic delayed union from aseptic delayed union in trauma patients. Higher sensitivity may lead to unnecessary surgery, while higher specificity could cause undertreatment of septic delayed union. A cut-off SUV_max set at 4.0 indicates an approximately equal and acceptable sensitivity and specificity of 65 and 77%, respectively. SUV_max set at 3.0 indicates a good sensitivity of 76%, but simultaneously gains at the expanse of the specificity reducing to 38%. Higher SUV_max of 5.0 has an excellent specificity of 92%, but has a very poor sensitivity of 29% in contrast. Therefore, in our opinion a SUV_max set at 4.0 is the best cut-off point for the use of this diagnostic tool.

To understand the process of FDG-uptake in case of abnormal bone healing, one should take into account the dynamics of normal bones and uncomplicated fracture healing expressed by FDG-uptake. In our first study, we therefore determined the FDG uptake in normal bones. We concluded that normal FDG uptake of the long bones of the lower extremity have a SUV_mean < 0.5 and a SUV_max < 0.8 [14]. Two studies have described the patterns of physiological FDG-uptake in bones during uncomplicated fracture healing. Zhuang et al. demonstrated in 37 patients that FDG-uptake may be normal within 12 weeks following fracture [15]. Shon et al. reported a study of four patients with uncomplicated fracture healing that FDG-uptake normalized within 8 weeks [16]. These two studies confirm that FDG-uptake of physiological bone healing will normalize in time based on visual assessment of the PET image. Unfortunately, no studies have been published defining the FDG-uptake pattern, let alone SUV measures, in the process of complicated fracture healing. In the setting of complicated fracture healing such as aseptic delayed fracture healing and septic delayed union, reasons for increased metabolism results into poor differentiation between the origin of increased FDG-uptake. In our study we included patients with delayed fracture healing to differentiate septic delayed union from aseptic delayed union.

Some discrepancies were found in the rates of diagnostic accuracy parameters between this study and previous studies. Schiesser et al. described a sensitivity, specificity and diagnostic accuracy of 100, 87.5 and 95% respectively for detection of osteomyelitis. This study included soft tissue infections as well, which could explain the higher diagnostic accuracy. Noteworthy, one of the false-positive cases described in this study turned out to be a delayed union [17]. Winter et al. found a sensitivity, specificity and diagnostic accuracy of 100, 86 and 93% respectively for detection of osteomyelitis in the peripheral skeleton. Two of four false positive findings were related to recent surgery within the previous 6 months [13].

Hartmann et al. used visual assessment and found a sensitivity, specificity and diagnostic accuracy of 100, 85 and 91% respectively for detection of osteomyelitis in the peripheral skeleton. FDG-uptake was expressed into intensity of greyscale and was graded into a five-point scale [18]. Wenter et al. also used visual assessment in addition to SUV measurements and found a sensitivity, specificity and diagnostic accuracy of 85, 86 and 86% respectively [19]. In our opinion, the visual scoring of FDG-uptake is less objective than the use of SUVs and may be more sensitive to inter- and even intraobserver differences. Therefore we call for an unbiased predictive SUV cut-off point to differentiate between normal bone uptake, uncomplicated fracture healing, aseptic delayed union and septic delayed union.

Guhlmann et al. did use SUV measurements for FDG-uptake for diagnosing osteomyelitis. This study found a very high sensitivity, specificity and diagnostic accuracy; 100, 92 and 97%. The high diagnostic accuracy found in this study can be explained by the long follow-up of patients in whom ongoing bone healing can be excluded and the risk for more fulminant osteomyelitis increases. Mean SUV found in peripheral fractures with osteomyelitis was 3.6 (SD 2.0), which approaches the SUV found in our study; SUV_max 4.77 [20]. However, it is not consistent to compare these results with our results, because it is unclear if they used SUV_mean of SUV_max to calculate the FDG-uptake.

Overall it can be stated that in previous studies a higher diagnostic accuracy of FDG-PET/CT for detecting osteomyelitis has been found compared to our diagnostic accuracy for detecting septic delayed union. This can be explained by differences in patient population and the different methods to quantify FDG-uptake.

Our data demonstrate that FDG-PET/CT is able to differentiate between aseptic and septic delayed union in the lower extremities in trauma patients by using the cut-off SUV_max set at 4.0. These findings broaden the domain in which FDG-PET/CT can provide additional information for guiding surgical therapy into an earlier and clinical relevant detection period. Early diagnosis has an obvious advantage for guiding surgical treatment which will lead to reduced morbidity compared to an unrecognized septic delayed union.

A larger prospective cohort study is needed to establish the distribution pattern of FDG-uptake during fracture healing in order to develop a diagnostic window for FDG-PET/CT in septic delayed union after trauma and osteosynthesis.