Biomarker-Based Blood Test May Extend Reach of CT Screening for Certain Lung Cancers
Smoking-based risk models used to predict lung cancer do not capture the full picture of who is really at risk for disease, according to researchers from MD Anderson Cancer Center.
Although official guidelines from the US Preventive Service Task Force (USPSTF) only suggest computed tomography (CT) screening for adult smokers between the ages 55 and 80 years who have a history of heavy smoking (either a 30-pack/year habit or who have quit within the last 15 years), results from a study published in JAMA Oncology suggest that a biomarker test for people who have smoked at some point in their lives could uncover more individuals who may be good candidates for CT screening.1
The researchers concocted a biomarker risk score based on the presence of 4 specific circulating biomarkers by examining samples of 108 patients with lung cancer who had smoked at some point in their lives (collected before a confirmed diagnosis of lung cancer) compared with samples from 216 smoking-matched control patients.
They tested the accuracy of their risk scores in a validation study of 63 patients with lung cancer (diagnosed up to 1 year after blood draw) who had smoked at some point in their lives, and compared the results with those of 90 smoking-matched control patients.
Even some smoking exposure, below the limits of established guidelines for screening, could put people at risk of developing lung cancer. In addition, the researchers determined that the risk prediction model that assessed smoking exposure alone (an area under the receiver operating characteristic curve [AUC] of .73) predicted disease risk with similar accuracy to the model that forecasts disease on the basis of a combination of smoking history and biomarker score (an AUC of .83; P =.003 for the difference between the values).
Thus, the authors of the study determined that anecdotal evidence of smoking history alone did not predict future cases of lung cancer with much more accuracy than did their biomarker test. And a key to proper prediction of disease was to examine samples taken from patients who had not yet developed disease, rather than measuring for the proteins of interest after a diagnosis of lung cancer.
The test had "superior sensitivity," according to a press release from the University of Texas MD Cancer Center in Houston, where study co-senior author Samir Hanash, MD, PhD, is a professor in the clinical cancer prevention department.2 This heightened sensitivity did not come at the expense of specificity, however (.86 for the smoking model alone compared with 0.95 with their integrated prediction model); the level of false-positives detected was still in alignment with rates established by USPSTF guidelines.
The investigators also concluded that the biomarker score could be used to reduce CT screenings (false-positives) in patients who would be eligible for testing. However, they wrote, using prediagnostic samples for prediction "is clearly needed before such a risk prediction tool can be used in practice."
Interestingly, Dr Hanash said the team, which is working on this research question under MD Anderson's Lung Cancer Moon Shot program, plans to conduct larger validation studies to test their biomarker-dependent prediction model, and that "consultations with the [US Food and Drug Administration] have begun."2
- Guida F, Sun N, Bantis LE, et al. Assessment of lung cancer risk on the basis of a biomarker panel of circulating proteins [published online July 12, 2018]. JAMA Oncol. doi: 10.1001/jamaoncol.2018.2078
- University of Texas MD Anderson Cancer Center. Study shows biomarker panel boosts lung cancer risk assessment for smokers [news release]. https://www.eurekalert.org/emb_releases/2018-07/uotm-ssb071118.php. Published July 12, 2018. Accessed July 11, 2018.