Patients presenting pain in the spine are suspected to have lesions compressing the spinal cord, and often recommended to receive MRI for diagnosis. For patients who have been diagnosed with cancer before, spinal pain could be a sign of metastasis. For patients who do not have a known disease, a correct diagnosis of the detected lesion is critical for guiding subsequent workup procedures (such as additional test, imaging or biopsy). Very often the morphological appearance of bone erosion and soft tissue masses that compress the spinal cord are similar on pre- and post-contrast MRI, and difficult to be differentiated. DCE-MRI may provide additional information to further characterize the detected lesion for diagnosis, staging or therapy monitoring. For patients who do not have history of cancer, if the type of metastatic cancer can be predicted, it will be very helpful to decide the imaging examination that need to be done next to find the primary cancer.In a retrospective review of spinal MRI that included a DCE sequence, a total of 76 patients (41 male, 35 female, mean age 56) who were confirmed to have metastatic cancer were found. The primary cancer types were: 35 Lung (mean age 55), 11 Thyroid (mean age 49), 12 Breast (mean age 56), 7 Prostate (mean age 72), 7 Liver (mean age 53), 4 Kidney (mean age 64). MR scans were performed on a 3T Siemens system. After the abnormal region was identified, DCE-MRI was performed using the 3D VIBE sequence, with TR=4.1ms, TE=1.5 ms, flip angle=10°, matrix=256×192 and FOV=250×250 mm. Approximately 30 slices with 3-mm thickness were prescribed to cover the abnormal vertebrae. The temporal resolution was 13 seconds, and a total of 12 frames were acquired. The contrast agents, 0.2 [mmol/kg] Gd-DTPA, was injected for measuring the signal intensity time course. Three heuristic parameters were measured: the maximum signal enhancement (SE) ratio normalized to the pre-contrast intensity [(Smax-S0)/S0]; the steepest wash-in enhancement ratio [(S2-S1)/S0] (S1 and S2 were two adjacent time points that showed the greatest signal enhancement); the wash-out slope [(Slast-Speak)/Speak x 100%], or if no peak using the signal intensity at 67 seconds as the reference [(Slast-S67s)/S67s x 100%]. Two compartmental pharmacokinetic analysis was applied to obtain the in-flux transport constant Ktrans and the out-flux rate constant kep ( [1/min]), by using the fast and medium population-based blood curves as references. The correlation between heuristic and pharmacokinetic parameters was investigated, and the differences among all cancer types were also investigated.Of all 76 cases, 51 showed a peak intensity during the DCE period and had a negative washout slope. There was a high correlation between the maximum SE ratio with the in-flux Ktrans rate (Figure 1), and the steepest wash-in SE ratio with the in-flux Ktrans rate (Figure 2). For the wash-out patterns, there was a high correlation between the wash-out slope calculated based on the intensity at the peak or at 67 seconds with the out-flux kep rate (Figure 3). The correlation with Ktrans and kep analyzed using the fast and medium blood curves was comparable, slightly better for the fast blood curve. For the differences among different cancer types, the correlation of wash-in SE ratio with Ktrans is shown in Figure 4, and the correlation of wash-out slope with kep is shown in Figure 5. The lung cancer had a wide distribution. The breast and thyroid cancer had the most prominent wash-out DCE patterns (mean wash-out slope: -16% for thyroid, and -11% for breast). Consistent with the greatest wash-out slope, the thyroid group had the highest mean kep. The breast cancer group had the highest wash-in SE ratio of 1.83, and they also had the highest Ktrans among all cancers.Our results show that a simple heuristic analysis to calculate wash-in and peak enhancements and the wash-out slope from the measured DCE kinetics can yield parameters that are highly correlated with fitted pharmacokinetic parameters, with the correlation coefficient r > 0.9. There is no significant difference between different primary cancer types, and it will be extremely difficult to predict metastatic cancer in the spine. Nonetheless, all lesions were clearly visible, and that can be used for guiding biopsy. Instead of using the quantitative parameters for differential diagnosis, since they were related to vascular and cellular properties they may serve as prognostic factors to predict treatment response and overall outcome. The prognosis of patients analyzed in this study is being closely followed for future analysis.