Abstract
1559
Objectives NaF PET/CT has high sensitivity for the detection of bone metastases. However, decreased specificity can be caused frequently by false positive lesions with high radiotracer uptake seen even in benign lesions. The purpose of the study was to evaluate the diagnostic strategy for inconclusive bone lesions in patients with metastatic prostate cancer.
Methods Fifteen patients with metastatic castration-resistant prostate cancer were enrolled, who underwent baseline NaF PET/CT, follow up NaF PET/CT (F/U NaF PET/CT) after 2 or 3 months of chemotherapy or androgen deprivation therapy and dedicated CT at the time of progression (F/U CT). All sites of abnormal uptake on baseline NaF PET/CT were identified. Each lesion was scored by three expert nuclear medicine physicians independently with a final convened consensus score determined using a five-point scale as following: 1 as definitely metastasis, 2 as likely metastasis, 3 as equivocal, 4 as likely benign, 5 as definitely benign. Maximum standardized uptake value (SUVmax) was measured for all lesions. These baseline lesions were reassessed using a reference standard consisting of both F/U NaF PET/CT and F/U CT and determined to be either metastasis, equivocal or benign. The diagnostic performance of the consensus baseline assessment was evaluated firstly for all lesions, and secondly for the more inconclusive lesions (score 2, 3, and 4). ROC analysis was done using SUVmax. Statistical significance of diagnostic performance was assessed by logistic regression analysis.
Results In fifteen patients, 643 lesions with focal uptake were identified. Inter-rater agreements were good or very good between three readers with Kappa values of 0.803, 0.792, and 0.821. For diagnostic performance analysis, 3 lesions were excluded which were equivocal in the final reassessment. Among 640 lesions, the numbers of lesions the initial baseline score of 1, 2, 3, 4, and 5 were 445, 47, 8, 49, and 51 respectively. When all score 1 to 5 lesions were included, the diagnostic performance of the baseline NaF PET/CT consensus score was excellent with accuracy of 97.8%, and SUVmax was also useful for differentiating metastasis and benign. However, only inconclusive lesions were used for the analysis, diagnostic performance was reduced with accuracy of 86.5%, and SUVmax was found to be not helpful in these diagnostically inconclusive lesions (Table 1). 10 were false positive lesions, and among them, 8 lesions were associated with joints (6 at one side of joint, 2 crossing the joint) and 9 showed no sclerosis (Figure 1). Among all inconclusive lesions, there were 20 lesions with uptake crossing the joint, and those were all benign regardless of presence of sclerosis (Figure 2). Among lesions with one side of joint involvement, only 17.1% (6/29) were metastasis if there is no sclerosis, in contrast, 91.7% (11/12) were metastasis if there is sclerosis (Figure 3, Table 2). When we consider lesions with uptake crossing the joint or one side joint involvement without sclerosis as benign, we could make correct decision on 8 lesions from initial 10 false positive lesions.
Conclusions The diagnostic performance of NaF PET/CT was excellent in patients with advanced castration-resistant prostate cancer. However, for lesions considered inconclusive lesions, accuracy of consensus score was limited and SUVmax was not useful. We observed that consideration of the pattern of joint involvement and bone sclerosis may reduce false positive decision.