Detection of distant metastases in patients with oesophageal or gastric cardia cancer: a diagnostic decision analysis

Abstract

Computed tomography (CT) is presently a standard procedure for the detection of distant metastases in patients with oesophageal or gastric cardia cancer. We aimed to determine the additional diagnostic value of alternative staging investigations. We included 569 oesophageal or gastric cardia cancer patients who had undergone CT neck/thorax/abdomen, ultrasound (US) abdomen, US neck, endoscopic ultrasonography (EUS), and/or chest X-ray for staging. Sensitivity and specificity were first determined at an organ level (results of investigations, i.e., CT, US abdomen, US neck, EUS, and chest X-ray, per organ), and then at a patient level (results for combinations of investigations), considering that the detection of distant metastases is a contraindication to surgery. For this, we compared three strategies for each organ: CT alone, CT plus another investigation if CT was negative for metastases (one-positive scenario), and CT plus another investigation if CT was positive, but requiring that both were positive for a final positive result (two-positive scenario). In addition, costs, life expectancy and quality adjusted life years (QALYs) were compared between different diagnostic strategies. CT showed sensitivities for detecting metastases in celiac lymph nodes, liver and lung of 69, 73, and 90%, respectively, which was higher than the sensitivities of US abdomen (44% for celiac lymph nodes and 65% for liver metastases), EUS (38% for celiac lymph nodes), and chest X-ray (68% for lung metastases). In contrast, US neck showed a higher sensitivity for the detection of malignant supraclavicular lymph nodes than CT (85 vs 28%). At a patient level, sensitivity for detecting distant metastases was 66% and specificity was 95% if only CT was performed. A higher sensitivity (86%) was achieved when US neck was added to CT (one-positive scenario), at the same specificity (95%). This strategy resulted in lower costs compared to CT only, at an almost similar (quality adjusted) life expectancy. Slightly higher specificities (97–99%) were achieved if liver and/or lung metastases found on CT, were confirmed by US abdomen or chest X-ray, respectively (two-positive scenario). These strategies had only slightly higher QALYs, but substantially higher costs. The combination of CT neck/thorax/abdomen and US neck was most cost-effective for the detection of metastases in patients with oesophageal or gastric cardia cancer, whereas the performance of CT only had a lower sensitivity for metastases detection and higher costs. The role of EUS seems limited, which may be due to the low number of M1b celiac lymph nodes detected in this series. It remains to be determined whether the application of positron emission tomography will further increase sensitivities and specificities of metastases detection without jeopardising costs and QALYs.