CT Scan radiomics helps to predict pancreatic cancer metastases

CT radiomics helps to predict pancreatic cancer metastases

By Erik L. Ridley, AuntMinnie.com staff writer

January 7, 2022 — A nomogram that utilizes CT radiomics data can preoperatively predict lymph node metastasis in patients with pancreatic ductal adenocarcinoma (PDAC), according to research published online January 6 in Cancer Imaging.

Researchers from the Naval Medical University in Shanghai, China, developed a nomogram that includes radiomics features as well as CT-reported lymph node status. In testing, the nomogram yielded an area under the curve of 0.81.

“The presented radiomics nomogram that incorporates the radiomics signature and CT-reported [lymph node] status is a noninvasive, preoperative prediction tool with favorable predictive accuracy for [lymph node] metastasis in patients with PDAC,” the authors wrote.

Although accurate preoperative lymph-node staging of pancreatic ductal adenocarcinoma is essential for providing patients with appropriate counsel regarding surgical decisions and prognosis, it’s difficult to provide with currently available methods, according to the researchers led by first author Dr. Yun Bian, PhD, and corresponding author Dr. Jianping Lu.

As a result, they sought to develop a radiomics nomogram to provide preoperative prediction of lymph node metastasis. They first retrospectively gathered 225 consecutive patients with surgically resected and pathologically confirmed PDAC and who had received multislice contrast-enhanced CT exams.

Of these patients, 180 were used in the training group and 45 cases were set aside as a validation cohort. A least absolute shrinkage and selection operator (LASSO) logistic regression algorithm was used on the training set to select the most useful radiomic features. The final radiomics “signature” included 13 features from the arterial CT scans.

Next, a prediction model was built using multivariable logistic regression analysis. A radiomics score is then calculated by the algorithm for each patient and incorporated into a nomogram that also includes the CT-reported lymph node status. The nomogram provides preoperative individualized prediction of metastases.

The researchers found that the nomogram yielded a promising level of performance on the validation cohort, which included 26 cases that were lymph node-negative and 16 cases that were lymph node-positive.

Performance of radiomics nomogram for predicting PDAC metastasis
AUC 0.81
Sensitivity 84.2%
Specificity 69.2%
Accuracy 75.6%

The nomogram that incorporated both radiomics features and CT-reports lymph node status outperformed a model based only on CT-reported lymph node status, which only produced an AUC of 0.63 on the test set.

The researchers also performed decision-curve analysis to estimate the predicted net benefit of the radiomics nomogram in clinical practice. They found that using a nomogram risk threshold probability of between 25 and 75% for lymph node metastases would yield more clinical benefit than utilizing a “treat-all” or “treat-none” scheme.

In the future, the authors plan to incorporate genetic markers into the nomogram.

“A combination of gene marker panels and a radiomics signature may improve the ability to predict [lymph node] metastasis in patients with PDAC,” they wrote.