18F–FDG PET/CT is widespread used to characterise tumor glycolytic activity, which is a valuable marker of tumor biological behavior [16-17]. In this study, we assessed the usefulness of 18F–FDG PET/CT in differentiating soft tissue sarcoma (STS) and bone sarcoma (BS) from benign lesions. Numerous studies investigated that PET-derived semiquantitative estimation parameters such as SUVmax, MTV and TLG are considered for meaningful indicators. SUVmax represents the glucose metabolism of single integrin in the tumor. MTV and TLG can reflect the overall tumor burden. Nonetheless, the ability of individual parameters for discriminate malignant from benign in soft tissue and bone tumors are not always feasibly. Soft tissue and bone tumors are highly heterogeneous group of tumors, delay in diagnosis will form a negative impact on patients outcome [18].Finding a simple and reliably Imaging model to characterize the biological behavior is critical to overcome many overlapping features. HF as an additional parameter obtained through PET image was reported to be a method to reflect the intratumorally structure heterogeneity of 18F-FDG uptake [19-20]. In this study, we comprehensively evaluate the feature parameters of 18F–FDG PET/CT imaging and finally constructed a well-established model based on SUVmax and HF for differential diagnosis between malignant and benign in soft tissue and bone tumors.
Alipour R, et al. [21] research showed that HF in malignant parotid tumors were higher than benign ones, concluded that HF is a reliable value in distinguishing benign from malignant parotid. Kim SJ, et al. [22] research claimed that HF could be a predictor for characterization of thyroid nodule. But these studies only relied on the univariate analysis and neither further performed the multivariate analysis to remove the interaction among variables nor deeply explore the combined application value of parameters. The present study, the significant feature parameters between malignant and benign group were screened according to the results of univariate analysis, including the tumor size, SUVmax, MTV, TLG and HF (all P value were<0.01). Furthermore, multivariate logistic regression analysis result revealed that the SUVmax and HF were both identified as independent risk factors for malignant tumor and can be implemented to established regression prediction model, the odds radios of SUVmax and HF were 1.135(95%CI: 1.026~1.256, P =0.014), 7.869(95%CI: 2.119~29.230, P =0.002), respectively. The results demonstrated that the prediction function of the model was accurate and feasible. Nakajo M, et al. [23] also conduct a univariate analysis in 63 cases of musculoskeletal tumors by using cumulative SUV-volume histogram (CSH) method [24], and the results showed the area under curve of CSH in malignant tumors was higher than benign ones, the conclusion is similar to our results. However, this method is equivalent to the concept of dose-volume histogram for evaluating radiotherapy regimen which applied to PET/CT functional imaging data, the clinical practicality is extremely limited.
Regarding the different growth rate, vascular distribution, and necrosis characteristics of each tumor cell population, are found to be different in biological behaviors [25]. In our study, the area under the curve (AUC)for regression model was 0.860 (95%CI: 0.771~0.948, P =0.000) was higher than SUVmax (AUC: 0.744, 95%CI: 0.628~0.860, P =0.000) and HF (AUC: 0.790,95%CI:0.684~0.896, P =0.000).The cutoff value to discriminate malignant group tumors from benign ones were 0.47, 5.95, 0.46 for regression model P value, SUVmax and HF, with 6/33, 9/33,10/33 false-positive benign lesions and 6/37, 9 /37, 10/37 false-negative malignant lesions, respectively. It follows that the regression prediction model combined with SUVmax and HF has higher diagnostic performance. In general, 18F-FDG uptake is not homogeneous within tumors, the biological characteristics of tumors are determined not only by tumor cells but also tumor microenvironment, including immune cells, endothelial cells and tumor-related fibroblasts [26-27]. SUVmax reflects the highest glucose metabolism in tumor cells, HF reflects the intratumorally spatial heterogeneity of glucose metabolism, the combination of SUVmax and HF can be considered as an incorporation of intertumoral structures from point to surface, which can more comprehensively reflect the glucose metabolism inside the tumor and characterize the biological behavior of tumors, so as to more accurately characterize malignancy and benign classification of tumors and further reducing the overlap of differential diagnosis.
Despite in previous research tumor size and volume are often considered as an indicator of malignancy tumor [27-28]. However, in our study multivariate logistic analysis results showed that when SUVmax and HF were introduced simultaneously to the regression model , the tumor size, MTV and TLG became no statistical significance, demonstrates there were some overlap and existed interactions among the parameters. Or perhaps the predictive value of tumor size and volume in a single space is limited, the tumor develop rapidly along with period incremental moving forward, such as tumor doubling time maybe higher likelihood of malignancy [29].
Of course, this study has certain limitations, First, the sample number of this study is not big enough, such phenomenon is due to some pathological classification of STTs are relatively rare. Second, this paper is a retrospective study, the collection of histopathology data was limited, minority of the pathological classification of the samples were confirmed only by biopsy pathology and the histological subtype of several cases were not clearly defined. Nevertheless, we believe that our results are qualified to be utilized for reference. This paper proposes a new concept, which can effectively integrate the metabolic information of 18F-FDG PET/CT imaging, and help to the clinical standardized management of soft tissue and bone tumors. A large sample of prospective cohort studies that involves imaging characteristic parameters and histopathology factors is recommended.