Imaging strategy for detecting lung metastases at presentation in patients with soft tissue sarcomas.

PURPOSE: To identify the risk of lung metastases at the time of diagnosis in patients with soft tissue sarcomas (STS) and to establish the optimum imaging strategy for the diagnosis of these metastases and whether this affects outcome.

MATERIALS AND METHODS: A retrospective review of an orthopaedic oncology database identified 1170 patients with newly diagnosed STS during a 7.5-year period (1996-2004). The patient demographics, tumour type, size, depth, histology grade and presence of metastatic disease at presentation were studied. The chest radiograph (CXR)/computed tomography of the chest (CT chest) findings, performed as part of the initial staging study, were available in all patients. We estimated the efficacy of CXR in identifying pulmonary metastatic disease compared with CT chest and whether this affected patient survival.

RESULTS: The incidence of metastases at diagnosis was 10% (116 patients), 8.3% (96 patients) had lung metastases on chest CT and 1.7% (20 patients) had metastases elsewhere. The risk of having lung metastases at diagnosis was 11.8% in high grade tumours, 7% in intermediate grade and 1.2% in low grade tumours. CXR alone detected 2/3 of all lung metastases. The positive predictive value of the CXR was 93.3%, the negative predictive value 96.7%, the sensitivity 60.8% and the specificity 99.6%. The accuracy was 96.9%. CT overestimated metastases in 4% with a sensitivity of 100%, specificity of 99.6% and accuracy of 99.6%. Median survival of patients with lung metastases at diagnosis was 11 months and there was no significant difference in survival between those who had metastases detected on CXR or purely on CT.

DISCUSSION: We recommend that all patients with a suspected STS should have a CXR at presentation, prior to histological diagnosis. CT of the chest should then be performed in those patients with an abnormality on the presentation CXR and routinely in those patients who have large, deep seated or high/intermediate grade tumours and in certain histological subtypes where the incidence of lung metastases at diagnosis is known to be high. In our experience, this strategy will detect 93% of all chest metastases. With current treatment strategies for metastases, outcome is not likely to be affected by any delay in diagnosis.