A Probability-Based Approach to Selecting Patients with Lung Cancer for Pretreatment Invasive Staging

RATIONALE: There is an opportunity to better select patients with non-small cell lung cancer (NSCLC) for pretreatment invasive staging-reducing the overall number of procedures without missing nodal metastases any more often than guideline-recommended criterion-based strategies. Doing so may avoid unnecessary testing, procedure-related risks, procedure-related costs, and treatment delays, and it may increase guideline-concordant invasive staging. OBJECTIVE: We compared a probability-based selection strategy for invasive staging to criterion-based strategies recommended by the National Comprehensive Cancer Network (NCCN) and American College of Chest Physicians (ACCP). METHODS: A retrospective cohort study was conducted of patients with NSCLC, without evidence of stage IV disease, who underwent invasive staging, resection, or both. The probability-based strategy stratified patients into high- and low-probability groups-analogous to criterion-based strategies that steer high-probability patients to invasive staging. We selected a threshold for stratification that matches the frequency by which criterion-based strategies select patients with nodal disease for invasive staging. We compared these selection strategies in terms of the expected frequency of invasive staging and associated confidence intervals (CIs). RESULTS: Among 3,080 patients, 783 (25%, 95%CI 24-27%) had pathologically confirmed nodal disease. The NCCN criterion-based strategy had a 79% (95%CI 78-80%) overall expected frequency of invasive staging. This criterion-based strategy selected 94% (95%CI 92-95%) of patients with true nodal disease to undergo invasive staging and selected 74% (95%CI 72-76%) of patients without true nodal disease to undergo invasive staging. The probability-based strategy had a fixed rate (94%) of selecting patients with true nodal disease for invasive staging-matching the performance of the criterion-based strategy. The probability-based strategy had a 74% (95%CI 73-76%) overall expected frequency of invasive staging and selected 68% (95%CI 66-70%) of patients without true nodal disease to undergo invasive staging. The probability-based strategy performed no differently than the ACCP criterion-based strategy. CONCLUSIONS: A probability-based selection strategy for pretreatment invasive staging may result in fewer overall procedures, without omitting them in patients with true nodal disease any more often than a NCCN criterion-based strategy. If externally validated, a probability-based selection strategy could offer an alternative approach to pretreatment invasive staging.