The Diagnostic Value of Fluorine-18 Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography for the Detection of Surgical Site Infection After Spine Surgery

Tomohide Segawa, MD; Hisashi Koga, MD, PhD; Masahito Oshina, MD; Masayoshi Fukushima, MD; Hirohiko Inanami, MD


Spine. 2021;46(10):E602-E610. 

In This Article


With an increased sample size in this study, the cut-off value for SUVmax calculated by statistical processing was 5.0. The cutoff values obtained in this study were consistent with preliminary cut-off values of our previous report with a smaller sample size. Our results indicated that F-18 FDG PET/CT including both pattern-based diagnosis by two nuclear medical physicians and SUVmax-based diagnosis is an effective diagnostic tool for the prediction of SSI after spine surgery. Both pattern- and SUVmax-based diagnoses demonstrated high accuracy, sensitivity, and specificity with only a limited number of false-negative cases.

Although MRI still remains the first diagnostic modality of choice with favorable sensitivity and specificity, it is often contraindicated due to its disadvantage including metallic implant-related artifacts.[1,15] In addition to F-18 FDG PET/CT, SPECT/CT using technetium-99m diphosphonate and gallium-67 citrate (bone/gallium) are also alternative imaging modalities. However, F-18 FDG-PET/CT offers significant advantages over bone/gallium. Namely, the image quality of F-18 FDG PET/CT is superior to SPECT/CT. Only a single 2-hour visit by the patient to the imaging department is required for F-18 FDG PET/CT, whereas bone/gallium requires multiple visits over several days; furthermore, F-18 FDG PET/CT has lower cost.[15] Thus, F-18 FDG PET/CT can be the procedure of choice for the evaluation of SSI in the spine when other imaging modalities fail to provide a definitive diagnosis (Figure 2).[9,10]

SUVmax-based diagnosis using a cutoff value of 5.0 demonstrated high diagnostic yields comparable to pattern-based diagnosis, irrespective of implantation or interval between surgery and PET/CT analysis (Table 3). Usually, the diagnosis of SSI in the spine based on F-18 FDG PET/CT requires experienced nuclear medical physicians for detailed analysis of F-18 FDG distribution pattern, that is, pattern-based diagnosis.[9,14–17] In this study as well as our previous study, the pattern-based diagnosis was made by two experienced nuclear medical physicians.[20] SUVmax has been reported as a useful quantitative index in diagnosing SSI;[14,16] however, Yu et al[15] reported a significantly lower diagnostic yield in SUVmax-based diagnosis compared to pattern-based diagnosis. Although their study on pyogenic spinal infection was not a report on postoperative infection, their pattern-based diagnosis resulted in a sensitivity of 98% and a specificity of 100%. However, an ROC analysis based on the SUVmax-based diagnosis yielded a maximum accuracy of 85% at an SUVmax cutoff value of 4.2. The sensitivity and specificity were 93 and 68%, respectively, which were significantly lower than the pattern-based approach (P < 0.01). They stated that the poor diagnostic yield of SUVmax-based analysis was associated with the comparable intensity of F-18 FDG uptake between mechanically induced nonseptic inflammation and pyogenic spinal infection. However, some reports in other disciplinary fields have referred to the importance of diagnostic tools for SUVmax and the cut-off value of SUVmax in postoperative infections.[28,29] Mitra et al[29] reported that of the 21 patients who underwent FDG PET/CT scans for suspected prosthetic graft infection, the diagnostic value based on FDG PET/CT analysis of the images had a sensitivity of 92%, specificity of 63%, and PPV and NPV of 80% and 83%, respectively. However, using the SUVmax cut-off at 6.3, the sensitivity of FDG-PET/CT remained at 92%, but the specificity increased to 88% and the PPV and NPV both increased to 92% and 88%, respectively. Tokuda et al[28] investigated the diagnostic value of FDG-PET/CT in infected thoracic grafts and proposed an SUVmax cut-off of 8 to determine the infection status. With an SUVmax cut-off at 8, the sensitivity of their study was 80% with 100% specificity.

In our study, all patients with a cut-off SUVmax of ≥5.0 had infection. To our knowledge, this study is the first report to evaluate the cutoff value of SUVmax in postoperative spinal infections. Therefore, we believe that our study establishes a standard for determining the cut-off value of SUVmax in postoperative spinal infections.

Although several previous studies reported the limitation of F-18 FDG PET/CT in diagnosing infection in the early phase after surgery, our results demonstrated excellent diagnostic yield even in the early phase (Table 3). In rat models with dental implants, increased bone metabolism after implant insertion was associated with increased F-18 FDG uptake without infection.[30,31] The study using a rabbit model of post-surgical osteomyelitis also indicated the limitation of FDG-PET for distinguishing between acute infection and sterile postsurgical inflammation.[18] The clinical study in a limited number of patients also reported the limitation of F-18 FDG-PET/CT due to the formation of granulomatous tissue complicating the diagnosis for the first 3 months.[19] In our study, however, 33 patients (63% of all study subjects) underwent F-18 FDG PET/CT in the early phase (within 12 weeks after spine surgery), and both pattern- and SUVmax-based diagnoses demonstrated a high diagnostic yield. Based on our results together with other reports,[15,32] F-18 FDG PET/CT can be a useful diagnostic tool for SSI in the early phase. The SUVmax values in the false-negative cases based on SUVmax-based diagnosis (4.3 and 4.4) were slightly lower than the cutoff value of 5.0. These cases received antibiotic treatment before the F-18 FDG PET/CT; therefore, antibiotic treatment might influence SUVmax values, resulting in false negative results.[9,10] Both pattern- and SUVmax-based diagnoses demonstrated high diagnostic yields with no statistically significant differences in any parameters (sensitivity: P = 0.16, specificity: P = 1.00, PPV: P = 1.00, and NPV: P = 0.16). However, the number of study subjects may not be high enough to conclude that SUVmax-diagnosis can replace the pattern-based diagnosis by nuclear medical physicians. Further assessment for any confounding factors such as bone microfracture and antibiotic treatment in larger number of populations is warranted.

In conclusion, we retrospectively assessed the diagnostic yield of F-18 FDG PET/CT (pattern-based diagnosis by nuclear medical physicians and SUVmax-based diagnosis) for SSI in 52 subjects who underwent spine surgery. Both pattern- and SUVmax-based diagnoses demonstrated high accuracy, sensitivity, and specificity with only a limited number of false-negative cases.