White cell scans and infected joint replacements. Failure to detect chronic infection.

We report the results of imaging with labelled white cells in 52 patients before the revision of 54 cemented joint prostheses at which the diagnosis of infection was made from biopsies. Twenty-five hips were imaged with 111In-oxine-labelled cells; 20 hips and 11 knees were imaged with 99mTc-hexamethylpropylene-amineoxime-labelled cells. Of these, 13 hips and five knees proved to be infected. The scans taken together had an accuracy of 82%, a sensitivity of 44% and a specificity of 100%. Indium scans gave 37% sensitivity, 99mTc labelling 50% sensitivity. Infected arthroplasties with positive scans had presented significantly earlier than those with negative scans, the time after the original insertion being 1.1 years for the true-positive scans and 6.1 years for the false-negative scans. The value of labelled white-cell scans in the detection of infection in failed joint replacements is dependent on the activity of the infection. There is reduced sensitivity to the more insidious infections which affect arthroplasties and aspiration under controlled conditions remains an important investigation.