# Non-invasive prediction of EGFR gene mutations in non-small cell lung cancer by multi-parameter CT perfusion imaging

## TL;DR

This study shows that CT perfusion imaging can non-invasively predict EGFR gene mutations in lung cancer patients, offering a potential alternative to biopsies.

## Contribution

The study introduces a combined CT perfusion imaging model (BV + TTP + PS) for non-invasive prediction of EGFR mutations in non-small cell lung cancer.

## Key findings

- Blood volume (BV), time to peak (TTP), and permeability surface (PS) were higher in the EGFR mutation group.
- A combined model of BV, TTP, and PS achieved an AUC of 0.956, outperforming single parameters in predicting EGFR mutation status.

## Abstract

In current clinical practice, invasive methods such as biopsy are commonly used to obtain tumor tissues for epidermal growth factor receptor (EGFR) mutation detection in patients with non-small cell lung cancer (NSCLC). This study aimed to explore the underlying association between various quantitative parameters of CT perfusion imaging (CTPI) and EGFR mutation, thus providing a new auxiliary diagnosis basis for non-invasive prediction of EGFR mutation status in patients with NSCLC.

Patients with a confirmed NSCLC diagnosis by surgery or biopsy were prospectively enrolled. All patients underwent pulmonary CTPI within 1 week before biopsy, as well as EGFR gene detection after biopsy, and were then divided into the EGFR mutation group and the wild-type group. Differences in quantitative parameters between the two groups were analyzed, and significant variables were identified for further construction of the predictive model. The receiver operating characteristic (ROC) curves were constructed, and the area under curve (AUC) was calculated to assess the predictive performance.

A total of 86 patients were included, including 45 women and 41 men. There were 47 cases in the mutation group and 39 cases in the wild-type group. A univariate analysis showed that compared with the wild-type group, blood volume (BV) (5.56 ± 1.51 vs. 3.04 ± 1.07, p < 0.001), time to peak (TTP) (29.31 ± 5.12 vs. 25.99 ± 5.68, p = 0.006), and permeability surface (PS) (18.98 ± 6.79 vs. 11.77 ± 5.56, p < 0.001) were all higher in the mutation group. No statistical differences were found in the other five quantitative parameters (p > 0.05). A multivariate logistic regression analysis identified BV (p < 0.001), TTP (p = 0.029), and PS (p = 0.014) as independent predictors of EGFR mutation. According to the ROC, the AUC of BV, TTP, and PS were 0.916, 0.739, and 0.788, respectively, and the corresponding cut-off values were 4.69, 23.84, and 12.11, respectively. The AUC of the combined predictive model (BV + TTP + PS) reached 0.956, which was superior to that of any single parameter (p < 0.05).

BV, TTP, and PS were independent predictors of EGFR mutation in patients with NSCLC. The combined CTPI parameter model (BV + TTP + PS) had the highest predictive performance and could be more reliable than any single parameter in clinical auxiliary diagnosis.