Analysis of metastatic involvement in bone using anatomical and functional information from 18F-FDG PET/CT

Although the overall incidence of bone metastasis is not known, over one-half of the people who die of cancer in the USA every year are thought to have bone involvement. In this study we have developed a method to quantify the metabolic and anatomic changes induced by different types of bone metastases in cancer patients using PET/CT images.

Seventy-three cancer patients with no previous history of chemotherapy or radiotherapy who had definite bone metastases documented by PET/CT and other conventional modalities were selected for this study. PET and computed tomography (CT) images were resampled to the same pixel size. Thereafter, the bone structure was segmented using thresholding. The 50% of the maximum standardized uptake value within the bone mask was used to identify bone lesions in each slice. Using the final regions of interest defined at 70% of the maximum, the lesion characteristics including the mean Hounsfield Units were computed from the PET/CT images. The lesions were subjected to visual confirmation by an experienced physician who also categorized them on the basis of the appearances in CT as lytic, sclerotic, mixed, or no-change type. The lesion characteristics were compared using statistical methods.

In all, 340 bony lesions in 73 patients with different cancer types were analyzed. The lesions were further categorized into four groups on the basis of their anatomical location. The spine hosts the largest number of lesions. The lumbar bones are the most preferential sites within the spine. Statistical comparison of CT values indicated that the difference between no-change and lytic types was significant. Uptake period did not seem to have a significant impact on no-change and sclerotic types. Quantitatively, maximum standardized uptake value for lytic, no change, mixed, and sclerotic lesions were 7.4, 6.1, 8.2, and 7.2, respectively.

A quantitative method provides a convenient way that may serve as a useful tool in monitoring and assessing the response to therapy.