Accuracy of radiography in grading and tissue-specific diagnosis : a study of 200 consecutive bone tumors of the hand

Objective: To determine the usefulness of radiography and magnetic resonance imaging in differentiating benign from malignant bony tumors of the hand and in making a tissue-specific diagnosis. Design: Two hundred consecutive bony tumors of the hand, the details of which originated from a national databank, were studied in a prospective way by radiography (100%) and by MRI (25%). All tumors were graded on a five-point scale, from certainly benign to certainly malignant, using location and morphology as diagnostic parameters. For all tumors a tissue-specific diagnosis was made, by the proposal of three possibilities in decreasing order of probability. Histological diagnosis was made by peer review, according to the WHO classification. Results: By the combining of "certainly" and "probably" benign (grades I and II) and "certainly" and "probably" malignant (grades IV and V), a correct grading was obtained in 165 (82.5%) of the cases (154 of the 173 benign and 11 of the 27 malignant tumors). A correct tissue-specific diagnosis was included in the three proposed differentials in 87.5%. MRI confirmed a correct diagnosis made on radiography in 72% and improved the grading capability by correctly upgrading malignant tumors and downgrading benign tumors in, respectively, 8% and 12%. The capability to obtain a tissue-specific diagnosis improved with change of an incorrect diagnosis on radiography to a correct one on MRI in 12 cases (24%). Conclusion: Subjective (semiquantitative) grading on radiography by an expert group proved to be excellent when compared with the results of a quantitative analysis of individual grading parameters. Multiple logistic regression analysis of these parameters resulted in a grading formula containing only six variables. The additional value of MRI in grading was amply demonstrated. Already high accuracy of radiography, in making a tissue-specific diagnosis, improved substantially after the performance of MRI