Evaluation of sensitivity and specificity of upper extremity radionuclide venography in cancer patients with indwelling central venous catheters.

Two hundred twenty upper extremity radionuclide venograms were performed using upper extremity injection of 5 mCi of Tc-99m DTPA in each arm of patients with indwelling central lines. Evidence of obstruction was found in 123 patients, collateral flow without anatomic obstruction in six patients, and a slow-flow pattern in 12 patients. Twenty-six of these also underwent upper extremity contrast venography within 48 hours of the scan. Contrast venograms and radionuclide venograms agreed in 19 patients (16 correctly identified as obstructed, three correctly identified as unobstructed). Six patients showed the slow-flow pattern without collaterals or obstruction. Subsequently, six follow-up contrast studies showed no evidence of obstruction or collaterals. The authors conclude that obstruction with collateral flow on radionuclide venograms correctly predicts obstruction. However, the slow-flow pattern does not and should not be used as the sole criterion to diagnose partial obstruction.