Computed tomographic scan of the chest underestimates the number of metastatic lesions in osteosarcoma

Purpose

Survival in osteosarcoma correlates with complete resection of primary and metastatic disease. The feasibility of complete pulmonary metastasectomy using thoracoscopy has been raised. Because palpation is not possible, minimally invasive techniques require preoperative radiological enumeration and localization of metastases not presenting at the lung surface. We hypothesized that computed tomographic (CT) scanning underestimated the number of pulmonary metastases in these patients.

Methods

Institutional review board approval was obtained. We determined the association between the number of lesions identified by CT scanning and the number of metastases found at thoracotomies for metastatic osteosarcoma from May 1996 to October 2004. Correlations between CT findings and pathology results were computed using the Kendall τ-b correlation coefficient. Depth, in millimeters, from the pleural surface was measured for those lesions seen on CT scan.

Results

We analyzed 54 consecutive thoracotomies performed in 28 patients for whom complete imaging was available. Computed tomographic scanning was performed a median of 20 days before thoracotomy (range, 1-85 days). Correlation between the number of lesions identified by CT and the number of metastases resected at surgery was poor, with a Kendall τ-b correlation coefficient of 0.45 (P < .001). In 19 (35%) of 54 thoracotomies, CT scanning underestimated the number of pathologically proven, viable and nonviable metastases found by the surgeon. Accounting for viable metastases only, correlation between the number of lesions identified by CT and the number of metastases resected at surgery was 0.50 (P < .001), and CT scanning underestimated the number of viable metastases present in 14 (26%) of 54 thoracotomies. Many lesions (32%) were pleural-based, but nearly half (47%) were 5 mm or deeper from the pleural surface of the lung.

Conclusions

Even in the era of modern CT scanning, only a very rough correlation exists between CT findings and the number of lesions identified at thoracotomy. In more than one third of thoracotomies in our series, metastases would have been missed by any tactic besides manual palpation of the lung during open thoracotomy. Minimal access procedures should not be the approach of choice if the goal is resection of all pulmonary metastases in osteosarcoma.