A combined model of circulating tumor DNA methylated SHOX2/SCT/HOXA7 and clinical features facilitates the discrimination of malignant from benign pulmonary nodules.

Abstract

Background: Despite the advancements in early lung cancer detection attributed to the widespread use of low-dose computed tomography (LDCT), this technology has also led to an increasing number of pulmonary nodules (PNs) of indeterminate significance being identified. Therefore, this study was aimed to develop a model that leverages plasma methylation biomarkers and clinical characteristics to distinguish between malignant and benign PNs.

Methods: In a training cohort of 210 patients with PNs, we evaluated plasma circulating tumor DNA (ctDNA) for the presence of three lung cancer-specific methylation markers: SHOX2, SCT, and HOXA7. Subsequently, we constructed a combined model utilizing methylated SHOX2/SCT/HOXA7 (mSHOX2/SCT/HOXA7) ctDNA levels, the largest nodule size measured by LDCT, and age, employing the binary logistic regression algorithm. Furthermore, we compared the diagnostic performances of the combined model with the Mayo Clinic model and the single mSHOX2/SCT/HOXA7 model by analyzing the area under the receiver operating characteristic curve (AUC) for each.

Results: The combined model demonstrated an impressive AUC of 0.87 and an accuracy of 0.75 in the training cohort, using pathologic diagnoses as the gold standard. This performance was significantly superior to that of the single mSHOX2/SCT/HOXA7 panel (AUC = 0.81, P < 0.0001) and the Mayo model (AUC = 0.65, P = 0.0005). Further validation in a cohort of 82 patients with PNs confirmed the diagnostic value of the combined model. Additionally, we observed that as the size of the nodule increased, the diagnostic accuracy of the combined model also improved.

Conclusions: A combined model incorporating the ctDNA-based methylation status of SHOX2/SCT/HOXA7 genes, the largest nodule size measured by LDCT, and age can serve as a supplementary approach to LDCT for lung cancer. This model enhances the precision in identifying high-risk individuals and optimizes the clinical management strategies for PNs detected by CT.