Prognostic factors used in breast cancer can broadly be divided into those that determine the extent of disease (i.e., tumor staging) and those that determine biological tumor characteristics. The latter are now crucial not only in predicting prognosis but also in selecting an appropriate treatment regimen
[7–9]. In addition, the usefulness of PET parameters in breast cancer has also been reported currently
[5, 10, 11]. However, although many authors have advocated that the SUVmax of FDG PET is an important predictor of progression after treatment, the clinical importance of SUVmax as a prognostic factor for malignancy is still controversial
[12, 13]. Because SUVmax is a single-voxel-based value, it may not be an adequate surrogate marker for the true biology of the whole tumor. In addition, SUVmax is susceptible to statistical noise and thus may be an unstable parameter. On the other hand, it has been reported that volume-based parameters (e.g., metabolic tumor volume, MTV, TLG, etc.) were significantly associated with an increased risk of recurrence and death in patients with surgically resected non-small cell lung cancer
Volume-based parameters of FDG PET represent the metabolic tumor burden of disease, reflecting both tumor volume and glucose utilization rate. These indices have been considered to be potentially reliable parameters for providing more details about the status of diseases in various types of cancers, especially lung tumors, oro/nasopharyngeal, and rectal cancers
[15–18]. Most recently, some authors have reported that TLG was superior to SUVmax as a predictor of prognosis in malignancies
[19, 20]. Our results showed that the WTLG of MBC in the initial presentation (group A) was correlated with OS not only in the univariate analysis but also in the multivariate analysis after adjustments for tumor phenotype and HSUVmax. Since ER/PR status was not a significant prognostic factor in our analyses, it could be said that WTLG is a better predictor of prognosis than ER/PR status in MBC patients at the initial presentation. Furthermore, SUVmax of primary breast cancers was not a significant predictor of OS in the multiple variable analysis as well as in the univariate analysis. Thus, WTLG may be considered a better prognostic factor than SUVmax in primary breast cancer patients who show metastatic disease at the initial presentation.
Conversely, the univariate and multivariate analyses for group B identified only ER/PR status as an independent prognostic factor. This result might suggest that WTLG does not accurately represent the extent of disease in recurrent breast cancers. This result may be explained by considering the finding that a similar number of patients in group B whose tumors were ER/PR positive had received endocrine therapy (14/20 [70%], aromatase inhibitor or antiestrogen). At the time of FDG PET/CT, FDG accumulation in MBC may have been diminished by the effects of these hormonal drugs, when compared with FDG accumulation in MBC at the first presentation. FDG PET/CT may be inappropriate for estimating the inherent malignancy or the extent of MBC when patients are receiving chemotherapy, including endocrine therapy. Morris et al.
 reported that the SUVmax of the newly diagnosed MBC to bone was significantly correlated with OS, and with the same group of patients, Ulaner GA et al.
 also have recently reported that the TLG tertile in the multivariate analysis was significantly associated with OS in patients with bone metastasis. We did not perform individual analyses for each metastatic site because of the insufficient sample size in our study. While their observations were inconsistent with our results that TLG was not a significant prognostic factor in patients with recurrence after a surgery (group B), this difference might be caused by the smaller sample size.
There were a few limitations to our study. First, because of the retrospective nature of this study, the patient population in our study was heterogeneous in terms of follow-up strategy (e.g., follow-up period and choice of imaging modality) and treatment regimens. Second, the number of the study subjects was relatively small, as we mentioned before.