br areas Previous imaging studies using either NaF PET

areas [39,40]. Previous imaging studies, using either NaF PET(CT), BS, SPECT, or whole-body MRI found the same thoraco-lumbar and pelvic predominant distribution of bone metastases from PCa [36,41,42].
PCa can theoretically disseminate either through pro-static veins or the lymphatics. The disease may involve the lower spine and pelvis via regional lymph nodes. But the most likely predominant hypothesis to explain the preferential involvement of the lower thoracic and lumbar spine and of the pelvi-femoral area by PCa metastases has been proposed decades ago and is known as the ‘‘Batson 
theory’’ [43]. This theory highlighted the key role played by a valveless venous system extensively developed along the vertebral column up to the skull and featuring rich anasto-moses between the veins of the pelvic viscera, the veins in the spinal canal, and the intercostal veins, communicating with the azygous veins, and from there with the bronchial veins. Hence, this system allows metastatic seeding bypass-ing the caval venous system and pulmonary circulation. The valveless nature of the system explains the retrograde migration of neoplastic Amiloride in the venous flow from the pelvic veins to this system in relation with variations in
Table 3 Proportions of tumors in region 1 when region 2 was free of tumors.
Data are proportions; raw data are followed by % in parentheses. Numbers in brackets are 95% CI.
abdominal pressures (during straining, coughing or lifting for example) [43].
The predominant involvement of the thoraco-lumbo-pelvic area in PCa patients parallels the results of previous studies on MRI screening and coming to the conclusion that a limited MRI marrow screen confined to the axial skele-ton (i.e., the thoraco-lumbar spine and pelvi-femoral area) would be sufficient for bone screening and would not result in any significant loss of accuracy for staging compared with BS or with whole-body MRI surveys, as isolated peripheral skeletal metastases without involvement of the thoraco-lumbar spine and pelvi-femoral area were almost never observed in these studies [25,36].
Promoting MRI surveys limited to the thoraco-lumbo-pelvic area, might be regarded as ‘‘outdated’’, when whole-body MRI is available and is recommended in many studies, allowing for a concurrent bone and lymph node metastatic screening [9,14,44—46]. However, a thoraco-lumbo-pelvic MRI bone survey is limited to 15 minutes of magnet time, compared to the 45—50 minutes required for a whole-body MRI study. MRI surveys might be considered as a reliable triage tool to ‘‘rule in’’ metastatic disease in patients at high risk. The presence of polymetastatic disease in these areas would be sufficient to start therapy [47].
In our study, we found a more homogeneous distribu-tion of MM lesions compared to PCa metastases. This is because MM is a disease that diffusely involves the red marrow-containing skeleton [48]. Extensive infiltration of the bone marrow by abnormal plasma cells is invariably seen at autopsy, including 15% of patients in whom no abnormality was seen on radiographic skeletal surveys [49]. The current study on MRI thus suggests that extensive body coverage is necessary to confidently identify bone involvement. This is in line with the previous demonstration that an MRI survey limited to the spine missed significant lesions compared to more exhaustive MRI coverage [50]. This also parallels the observation that spine and pelvis coverage using MRI was insufficient compared to whole-body radiographic skeletal survey given the high sensitivity of skull and ribs radio-graphs to detect osteolytic MM lesions [51]. Finally, our results support the whole-body MRI approach recommended by authorities in the field of MM and its preference over a limited spine and pelvic survey [19,21].
Our study has several limitations. First, the small number of patients needs further larger scale studies to validate our results. The lack of added value of including the cervical spine in both PCa and MM patients, as well as the ribs, proximal humeri and proximal femurs in PCa patients needs to be further confirmed. Second, we deliberately chose two