Smarter Screening for Early Lung Cancer

Smarter Screening for Early Lung Cancer
A three-way risk assessment to optimize decisions on bronchoscopy and biopsy.

On Easter Day, Vanderbilt University Medical Center lost Pierre Massion, M.D., director of the Cancer Early Detection and Prevention Initiative, to a heart attack while he was on an habitual run. One week before, Massion had received R01 funding from the National Cancer Institute to study a multifactorial approach to lung cancer screening. His hope was that this would save lives by offering a targeted, highly reliable way to assess indications for biopsy.

“Nodules are so frequently observed incidentally on scans that unless the patient is a smoker and has personal or family history of cancer, the patient may be sent home for six months,” said Massion’s mentee, Michael Kammer, Ph.D., a research fellow in the interdisciplinary training program in lung research at Vanderbilt. “Currently, the risks in having a bronchoscopy aren’t typically justified by that five percent that turn out to be malignant. But those five percent can be aggressive, and six months later can be too late.”

“The goal of this trial is not necessarily to show that this specific combination of biomarkers works, but that using good biomarkers improves patient outcomes.”

The team aims to improve early detection of high-risk nodules. “The focus of our research is to develop better non-invasive biomarkers to predict who is and is not likely to have cancer,” Kammer said.

Eric Grogan, M.D., a thoracic surgeon at Vanderbilt, now serves as principal investigator on the study, which is being conducted at Vanderbilt-Ingram Cancer Center. “We think that by using our predictive model we can avoid 20 percent of unnecessary invasive procedures for patients without cancer, while getting patients with cancer to the therapy they need several months sooner,” he said.

Screening Gaps and Confounders

While lung cancer screening is increasingly in the spotlight, it remains focused on people with a heavy smoking history. “The rates of screening and positive survival outcomes are still absolutely abysmal,” Kammer said. “The guidelines screen in only a quarter of all people who get lung cancer, and only about 5 to 15 percent who are eligible to get screened.”

“The focus of our research is to develop better non-invasive biomarkers to predict who is and is not likely to have cancer.”

Complicating this is the plethora of benign lung nodules. “Biopsy is the only means to definitive diagnosis, yet there are costs and risks with bronchoscopy, particularly in peripherally located nodules, where the risk of a collapsed lung is about 20 percent,” Kammer added.

New Predictive Model

Recognizing these shortfalls, Massion and his team came up with a three-way approach for refining the criteria for nodule testing: radiomics, expanded clinical history and a biomarker blood test. Pre-trial testing has indicated that this combination outperforms any approach on its own.

“We think that by using our predictive model we can avoid 20 percent of unnecessary invasive procedures for patients without cancer, while getting patients with cancer to the therapy they need several months sooner.”

Through quantitative radiomics, a software algorithm analyzes the original CT scan for 750 quantitative features of the nodule. “Currently, clinicians look primarily at size and shape, spiculation and lobulation to inform their next step. This looks at a myriad of feature subsets, as well as volume, density, heterogeneity and more,” Kammer said.

The second leg of the tripod is expanded clinical history, which includes the patient’s age, smoking history, family history and history of any other cancers, since all four have strong correlations with lung cancer risk.

A new test for CYFRA 21-1, a cytokeratin fragment and circulating biomarker, is the novel third leg of the tripod. Fragments of CYFRA 21-1 from lung epithelial cells are discarded into the bloodstream in higher quantities in the presence of infection, COPD, asthma – or lung cancer. “Together with the other parameters in the testing, it’s a significant biomarker,” Kammer said.

Trial Features

The team plans to recruit over 400 participants who have had a lung nodule detected incidentally on a CT scan and are at intermediate risk, per the current standards. Risk will be assessed by the predictive model, with patients and providers deciding whether to move forward with biopsy.

“One of the beauties of this trial is that there aren’t any extra procedures, exposures or costs to the patient to be able to do this, since the patient is already getting a blood draw and already has a CT scan,” Kammer said.

“The goal of this trial is not necessarily to show that this specific combination of biomarkers works, but that using good biomarkers improves patient outcomes,” Grogan said. “Our hope is to continue adding or refining these over the course of this trial and into the future.”