Abstract
Background/Aim: Positron emission tomography (PET) using 18fluorine-labelled fluorodeoxyglucose (FDG), is the most widely applied molecular imaging technique in oncology. The present study assessed the efficacy and limitations of FDG-PET by comparing FDG accumulation in bone and soft tissue lesions, as well as histopathological features. Patients and Methods: The study included 122 patients with 165 lesions, as assessed by histopathological examinations. The maximum standardized uptake values (SUVmax) of benign lesions were compared to those of primary, recurrent, or metastatic sarcomas, as well as those of other malignancies. Results: The sensitivity, specificity, and accuracy of SUVmax for differentiation between benign lesions and primary sarcomas were 67.9%, 92.9%, and 80.4%, respectively. There were no significant differences between benign lesions and recurrent or metastatic sarcomas. Conclusion: Although FDG-PET is a useful imaging modality to differentiate primary sarcomas from benign lesions, it is difficult to differentiate residual or metastatic sarcomas from benign lesions.
Diagnostic imaging techniques including radiography, computed tomography (CT), magnetic resonance imaging (MRI), 201thallium scintigraphy, 99mTc-MIBI scintigraphy, and positron emission tomography (PET) are widely used to assess bone and soft tissue lesions (1-5). Anatomic imaging techniques such as radiography, CT, and MRI are useful for identifying the location, extent, and association between musculoskeletal tumors and normal tissues. Although histopathological examination is the gold standard for diagnosis, biopsy and excision of the lesion are invasive, especially for patients with benign tumors. Several studies have assessed the efficacy of radiological examinations to detect malignant lesions (6-10), however, false-positive and false-negative lesions occur when differentiating bone and soft tissue lesions using radiological examinations. 201Thallium scintigraphy is also used to differentiate benign and malignant bone and soft tissue lesions (11). However, giant cell tumor (GCT), Langerhans cell histiocytosis (LCH), pigmented villonodular synovitis, and osteomyelitis have high accumulation profiles using 201Thallium scintigraphy, which often results in misdiagnoses of malignant tumor (12-18). In contrast, chondrosarcoma, well-differentiated liposarcoma, chordoma, and malignant peripheral nerve sheath tumor have low accumulation profiles of the tracers.
PET with 18fluorine-labelled fluorodeoxyglucose (18F-FDG), which is the molecular imaging technique most widely applied in oncology, provides quantitative, qualitative, and functional information regarding tumor cells, based on their increased rate of glucose metabolism. FDG-PET effectively and sensitively detects, stages, and restages malignancies in patients with diverse types of cancer (19-24). Furthermore, the combination of PET and whole-body CT provides functional and anatomical information, resulting in accurate detection and staging of malignancies.
Several studies have reported on the efficacy of PET/CT for detection of malignant tumors and differentiation between malignant and benign lesions (25-29). Although 18F-FDG accumulates preferentially in malignant tumors, it is a non-specific tracer and accumulation also occurs in other physiological and non-malignant processes, including infective and inflammatory conditions such as osteomyelitis and arthritis (30). The aim of the present retrospective study was to determine whether integrated PET/CT reliably differentiates between malignant lesions and benign lesions, by comparing 18F-FDG accumulation in bone and soft tissue lesions.
Patients and Methods
Patients. The present study included 122 patients (63 men, 59 women patients; mean age: 57.0 years; age range=13-88 years) with 165 bone and soft tissue lesions, which were assessed by PET/CT and histopathological examinations from July 2010 to November 2015. The 165 lesions included 137 malignant lesions and 28 benign lesions. The 137 malignant lesions comprised 28 primary tumors, 34 recurrent or residual tumors, 43 metastatic tumors, and 32 other malignancies including carcinomas and lymphomas. The study was approved by the Kanazawa University Ethical Committee. All participants provided written informed consent.
Integrated PET/CT acquisition. Integrated PET/CT was performed after fasting for at least 6 h, if the serum glucose level was ≤200 mg/dl. Each patient was injected with 4.0 MBq/kg (0.11 mCi/kg) of 18F-FDG, and whole-body PET/CT was performed using a PET-CT system (Discovery PET/CT 690; GE Healthcare, Milwaukee, WI, USA) 60 and 120 min after injection. PET/CT was performed only at the 60 min timepoint in cases without significant accumulation of 18F-FDG.
Image analysis. PET and CT images were reviewed by a physician with experience in nuclear medicine. The maximum standardized uptake value (SUVmax) was determined in each region of interest (ROI). SUVmax was calculated using the PET scanner software, using the following formula: SUVmax=C (μCi/mL)/ID (μCi)/w (kg), whereby C is the activity at a pixel within the tissue defined by an ROI, ID is the injected dose, and w is the patient's body weight in kg. SUVmax values within the selected ROI are reported.
Statistical analyses. Values are expressed as means±standard deviations. Comparisons of SUVmax between 2 groups were analyzed using Student's t-tests. Comparisons of ≥2 groups were assessed using analysis of variance. Sensitivity, specificity, positive predictive values (PPV), negative predictive values (NPV), and accuracy were calculated; and 95% confidence intervals (CIs) were determined for each parameter, based on the binomial distribution. SUVmax for malignant lesions were compared with that of benign lesions. Furthermore, SUVmax of primary, recurrent, or metastatic sarcomas, as well as carcinomas, were compared with those for benign lesions. The optimal cutoff levels for SUVmax were defined as the index values that minimized the number of false results by receiver operator characteristic analyses. Correlations between SUVmax and histological evaluation were analyzed using measure of sensitivity, specificity, accuracy, PPV, and NPV. Statistical significance was defined as p<0.05. All statistical analyses were performed using EZR statistical software (Saitama Medical Center, Jichi Medical University, Saitama, Japan).
Results
Comparisons of SUVmax between benign and malignant lesions. The mean SUVmax for benign lesions and malignant tumors were 4.9±3.4 and 8.8±6.7, respectively (Figure 1). Although there was a significant difference between values for benign and malignant lesions (p=0.001), considerable overlap was observed. For malignant tumors, the mean SUVmax of primary sarcomas, recurrent sarcomas, metastatic sarcomas, and the other malignancies were 13.1±7.6, 6.2±5.7, 6.6±5.1, and 10.8±6.7, respectively (Figure 2). Primary sarcomas and other malignancies had significantly higher SUVmax compared to benign lesions (Figure 2; p<0.01), whereas no significant differences occured between benign lesions and recurrent or metastatic lesions (Figure 2).
Sensitivity, specificity, and accuracy of FDG-PET. The sensitivity, specificity, PPV, NPV, and accuracy of SUVmax for differentiating between benign lesions and malignant tumors were 92.9%, 43.8%, 96.8%, 25.2%, and 52.1%, respectively, with an optimal cutoff SUVmax value of 8.2 (Figure 3A; Table I). The sensitivity, specificity, PPV, NPV, and accuracy of SUVmax for differentiating benign lesions from primary sarcomas were 67.9%, 92.9%, 90.5%, 74.3%, and 80.4%, respectively, with an optimal cutoff SUVmax of 8.6 (Figure 3B; Table I).
False positives and false negatives using PET/CT. Although SUVmax for malignant lesions ROIs were significantly higher compared to those for benign lesion ROIs, considerable overlaps in SUVmax were observed (Table II). There were 2 false positives among the benign lesion cases, including 1 each of GCT and LCH. There were 9 false negative cases among the primary sarcomas, including the following types of sarcomas: 2 extraskeletal myxoid chondrosarcomas, 2 Ewing's sarcomas, 1 chondrosarcoma, 1 leiomyosarcoma, 1 osteosarcoma, 1 epithelioid endothelioma, and 1 undifferentiated pleomorphic sarcoma (UPS). In the present study, 1 case of LCH and 4 cases of GCT were categorized as benign lesions. Therefore, the mean SUVmax for LCH was 14.8 and the mean SUVmax for GCT was 8.4 (range=3.6-13.8) (Figure 4), which was higher than for the other types of benign lesions. Cases of multiple lesions including malignancies also made it difficult to differentiate between benign lesions and metastatic lesions. For example, PET/CT in a 68-year-old patient indicated multiple lesions (Figure 5), and histological examinations of the thyroid, left buttock, and ilium revealed thyroid cancer, undifferentiated pleomorphic sarcoma, and sclerotic bone without tumor, respectively.
Discussion
PET SUVmax has been proposed as diagnostic tools to differentiate between benign and malignant lesions (31, 32). Cho et al. (31). reported that PET/CT is an adjunctive diagnostic method for differentiating malignant lesions from vertebral compression fractures. Additionally, Tian et al. reported that malignant lesions have higher SUVmax than benign lesions (malignant lesions, 6.8; benign lesions, 4.5), although considerable overlap was observed (32). Our data also demonstrate significant differences in SUVmax between benign and malignant lesions, although we also observed considerable overlap in SUVmax between groups. Among malignant lesions, the mean SUVmax for primary tumors, recurrent tumors, and metastatic tumors were 13.1±7.6, 6.2±5.7, 6.6±5.1, and 10.8±6.7, respectively. Most primary sarcomas in our study were large masses at the time of diagnosis. In contrast, the recurrent and metastatic tumors in our study were small masses, primarily because most patients had follow-up examinations such as CT and MRI to detect lesions. Therefore, the differences in SUVmax between primary, recurrent, and metastatic sarcomas may be related to tumor size.
Discrepancies between SUVmax and malignancy determination have been reported for some tumors (33-35). In our study, LCHs had the highest SUVmax among benign lesions, and the mean SUVmax for GCTs was higher (8.4, range=3.6-13.8) than for other benign lesions. Cellular uptake of FDG in human tumor cells is associated with expression of glucose transporter protein (GLUT)-1, hexokinase II, and with gene upregulation for these proteins (34, 36). Ong et al. found significantly greater GLUT-1 and hexokinase II in 5 different human cancer cell lines (37). However, other studies have demonstrated an up-regulation of GLUT-1 in human monocyte-derived macrophages (38-40), and the high monocyte/macrophage content within GCTs may explain the FDG uptake (34, 35). Therefore, GCT and lesions containing active macrophages should be considered in differential diagnoses of intensely FDG-avid neoplasms. In contrast, SUVmax values for liposarcoma and chondrosarcoma are usually lower in PET studies, compared to values for other types of malignancies (41). However, SUVmax cut-offs may enable prediction of malignancies in lipomatous and cartilaginous tumors. In a previous report assessing SUVmax in lipomatous tumors, a cut-off value of 0.81 for FDG-PET significantly differentiated between benign lesions and well-differentiated liposarcoma (41). Feldman et al. reported that the sensitivity, specificity, and accuracy of FDG-PET for distinguishing benign and malignant cartilage neoplasms with a cutoff SUVmax values of 2.0 were 91%, 100%, and 97%, respectively (42). Therefore, cases of suspected lipomatous or cartilaginous tumors indicated by MRI or CT should use a low SUVmax cut-off to differentiate between benign and malignant tumors.
Although our data demonstrate significant differences between benign and malignant lesions, considerable overlaps in SUVmax were observed. Furthermore, the inclusion criteria for the present study were limited. The first inclusion limitation was that PET/CT was performed only for lesions that were suspected as malignancies; most of the benign lesions that were suspected as malignant tumors before PET were aggressive tumors such as GCTs. The second inclusion limitation was that the present study included only lesions that were confirmed by histopathological examinations. Exclusion of conventional benign tumors from the study may have influenced SUVmax.
In conclusion, PET is a useful imaging modality for differentiation between benign lesions and primary sarcomas, with 80.4% accuracy. However, differentiation between benign lesions and recurrent or metastatic lesions is difficult because of low FDG uptake in many types of lesions. Therefore, combinations of PET and other methods such as enhanced MRI are preferable for determining diagnoses.
Footnotes
Conflicts of Interest
The Authors declare that they have no competing interests.
- Received May 21, 2018.
- Revision received June 6, 2018.
- Accepted June 8, 2018.
- Copyright© 2018, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved