Maximum SUV Benchmarks for Distinguishing Physiologic from Pathologic Spinal Cord F-18 FDG Uptake

Introduction: Our objective with this retrospective review was to establish benchmark maximum SUV measurements for the spinal cord’s physiologic F18-FDG uptake at predetermined segments, including mid cervical (C3-4), upper thoracic (T5-6), lower thoracic (T9-10), and mid lumbar/cauda equina (L3). We then established benchmark maximum SUV (mSUV) measurements at which pathologic processes such as leptomeningeal metastases or sarcoidosis should be raised as diagnostic possibilities.

Methods: Twenty F18-FDG PET/CTs were reviewed to establish baseline physiologic FDG cord uptake, excluding patients with (1) marrow hyperplasia (2) severe spinal arthropathy or curvature (3) metallic hardware about the spinal canal (4) motion artifact. Spherical ROI mSUV was recorded at predetermined levels as outlined in the project background.

Twelve F18-FDG PET/CTs for patients with MR findings and/or CSF positive for leptomeningeal metastatic disease or leptomeningeal sarcoidosis were retrospectively identified and the spherical ROI mSUV recorded at the aforementioned predetermined levels.

During the data collection, it was observed that normal patients demonstrate physiologic increased FDG uptake at the level of T12. For reference, the mSUV measurements at T12 were obtained for ten normal subjects.

Subsequently, the data was processed using SPSS and receiver operating curves (ROCs) generated for each level at which measurements were obtained. Utilizing the ROCs, benchmark mSUV measurements were chosen at each level to optimize sensitivity and specificity for detection of pathologic cord uptake.

Results: At the mid-cervical level (C3-4), mean mSUV for the control subjects was 2.3 with standard deviation (SD) 0.45, and for positive cases mSUV was 3.8 with SD 2.0. The point on the ROC which optimizes sensitivity and specificity falls at mSUV 2.95, with sensitivity 75% and specificity 95%.

At the upper thoracic level (T5-6), mean mSUV for controls was 1.61 with SD 0.31, and for positive cases mSUV was 2.34 with SD 0.74. The point on the ROC which optimizes sensitivity and specificity falls at mSUV 1.85, with sensitivity 75% and specificity 75%.

At the lower thoracic level (T9-10), mean mSUV for controls was 2.0 with SD 0.5, and for positive cases mSUV was 2.95 with SD 1.28. The point on the ROC which optimizes sensitivity and specificity falls at mSUV 2.25, with sensitivity 92% and specificity 90%.

For normal subjects, the mean mSUV at T12 was 2.5.

At the mid lumbar level (L3/cauda equina), mean mSUV for controls was 1.46 with SD 0.29, and for positive cases mSUV was 3.08 with SD 1.88. The point on the ROC which optimizes sensitivity and specificity falls at mSUV 1.85, with sensitivity 67% and specificity 90%.

Conclusions: With this study, the authors have attempted to reduce uncertainty in distinguishing normal spinal cord uptake from pathologic uptake by suggesting benchmark mSUV at four representative levels. Distinguishing pathologic uptake can be challenging due to physiologic variations in cord uptake at various levels - particularly at T12, where physiologic uptake is higher than the benchmark for pathologic uptake at other levels.

It should be noted that the suggested benchmark mSUV measurements were chosen at the authors’ discretion, and the ROC tables will be included for the readers’ reference. Sensitivity and specificity can of course be titrated based upon the selected benchmark mSUV level.

• Download figure
• Open in new tab
• Download powerpoint