🎯 Background & Rationale
Accurate malignancy risk prediction for pulmonary lesions is critical in lung cancer screening and early diagnosis. Widely established risk prediction models—such as the Brock and Herder models—were primarily designed for general screening or incidental nodule cohorts. However, their performance drops significantly when applied to high-risk cohorts specifically referred for bronchoscopic evaluation. The LIONS PREY (lung lesion score predicts malignancy) model was developed to provide a tailored, reliable estimate of malignancy risk in this high-prevalence population, prompting this study to externally validate it against the legacy Brock and Herder models.
👥 Study Design & Population
- Study Type: Retrospective Cohort Study
- Population (PICO):
- Patients: Adults undergoing index navigational bronchoscopy for peripheral pulmonary lesions at a tertiary academic center between December 2019 and March 2024 (193 evaluable lesions analyzed).
- Intervention/Index Test: Malignancy risk estimation using the LIONS PREY model, compared against traditional Herder and Brock models.
- Comparison: Predictive performance (AUC, ROC analysis, and classification accuracy) between the LIONS PREY, Herder, and Brock models.
- Outcome: Histologically confirmed malignancy versus benign status.
📈 Methodology & Rigor
The authors performed a retrospective analysis of 193 evaluable lesions. Malignancy risk was calculated for each patient using LIONS PREY, Brock, and Herder models. Diagnostic performance was evaluated using receiver operating characteristic (ROC) curve analysis (calculating area under the curve [AUC]) and overall classification accuracy. Exact numerical data provided demonstrate a high proportion of malignant lesions (134 of 193, or 69.4%), reflecting the referral bias inherent to a tertiary interventional pulmonology practice.
🔬 Key Findings [or Planned Endpoints]
- Prevalence: 69.4% (134/193) of evaluated lesions were histologically confirmed to be malignant.
- Clinical Characteristics: Malignant lesions were significantly associated with increased growth dynamics (2.7 ± 4.1 mm vs. 1.0 ± 4.3 mm, p < 0.001) and spiculation (56.0% vs. 23.7%, p < 0.001).
- Discriminative Ability (AUC): The LIONS PREY model achieved an AUC of 0.94, significantly outperforming both the Herder model (AUC 0.72; p < 0.001) and the Brock model (AUC 0.62; p < 0.001).
- Classification Accuracy: Correct classification rates were highest for LIONS PREY at 85.0%, compared to Herder (70.3%) and Brock (69.1%) (p < 0.001 for all).
⚖️ Critical Appraisal
- Strengths: Independent external validation of a novel prediction tool designed specifically for the advanced diagnostic bronchoscopy population, addressing a known limitation of screening-derived models.
- Limitations: Retrospective design from a single tertiary academic center introduces potential selection and referral biases. The high pre-test probability of malignancy (69.4%) mirrors specialized interventional practices, meaning these results may not directly extrapolate to outpatient primary care settings managing low-risk incidental nodules.
💡 The Clinical Bottom Line
Legacy risk calculators like Brock and Herder lose calibration when applied to patients already triaged to the interventional pulmonology suite for navigational bronchoscopy. The LIONS PREY model provides superior risk stratification and diagnostic accuracy in this high-risk subset. Clinicians should consider incorporating LIONS PREY into pre-procedural evaluations to better contextualize diagnostic yields and guide shared decision-making regarding advanced bronchoscopic tissue acquisition versus alternative management strategies.