Skip to main content
Premium Trial:

Request an Annual Quote

Hopkins-Led Team IDs Autoantibody Panel for Early Detection of Lung Cancer


NEW YORK (GenomeWeb) – A team led by researchers at Johns Hopkins University has identified a panel of autoantibody-based protein markers that could aid in the early detection of lung cancer.

Described in a study published last week in Molecular & Cellular Proteomics, the panel consists of autoantibodies to the proteins p53, HRas, and ETHE1. In a cohort of 353 lung cancer patients, 101 patients with benign lung lesions, and 93 healthy individuals, these markers detected lung cancer with sensitivity of 50 percent and specificity of greater than 90 percent.

Performance was roughly equivalent across both high- and normal-risk populations, said Heng Zhu, co-director of Hopkins' High Throughput Biology Center and an author on the study. The work, he noted, was a collaboration between Hopkins, China's Fujian Medical University, and protein microarray firm CDI Laboratories.

He said that Hopkins and CDI, which share the rights to the markers, are now working with their Chinese collaborators to establish clinical studies of the markers in China for lung cancer detection.

"We are trying to license the markers to a Chinese company with the hope that it could be used for early-stage lung cancer screening," Zhu said. "We hope that whatever Chinese company licenses them will get approval from the Chinese Food and Drug Administration to run a large clinical trial to evaluate whether the test could be valuable for [that purpose]. That is the ultimate goal."

While much protein biomarker work relies on either mass spec or immunoassays to measure proteins directly in samples like patient plasma or tissue, Zhu and his colleagues looked at patient autoantibody profiles instead. This, he noted, has potential advantages in that the immune cell replication involved in the immune response serves as a natural amplification of the antibody signal, which could make it possible to detect disease earlier than with conventional protein markers, while at the same time easing the challenges involved in detecting moderate- to low-abundance protein markers.

Additionally, Zhu noted, autoantibodies tend to be more stable in patient blood, which makes them potentially useful as biomarkers.

"Proteins circulating in blood are not very stable," he said. "They fluctuate. So it's not a very reliable system to depend on. Antibody-based biomarkers are very stable and they have long-lasting memories. That's why we and others in the field would prefer auto-antigen-based biomarkers for diagnosis, and even for prognosis."

Indeed, a number of other research groups and clinical diagnostic firms are pursuing auto-antibody-based markers. For instance, Arizona State University spinout HealthTell and German diagnostics firm Protagen both use antibody markers, primarily for testing in the autoimmune space. UK proteomic diagnostics firm Oncimmune offers an auto-antibody-based lung cancer test called EarlyCDT-Lung for evaluating pulmonary nodules picked up on CT scans and assessing a patient's risk of lung cancer prior to CT screening.

Zhu said he and his colleagues have been pursuing autoantibody markers for around a decade. The development of human protein microarrays like that offered by CDI has significantly increased the power of the approach, he said.

For the MCP study, the researchers used CDI's HuProt 3.0 array, which contains 20,240 full-length human proteins, covering roughly 75 percent of the human proteome. Running their patient samples of interest on this array, they were able to identify autoantibodies that were present in greater levels in cases compared to controls, indicating that they and their antigens might be associated with lung cancer.

In the first phase of the study, they screened serum samples from 80 lung cancer patients and 20 healthy individuals against the HuProt arrays, identifying 170 candidate proteins. They then constructed smaller arrays targeting the 170 candidates and screened these against serum samples from a separate cohort of 131 early-stage lung cancer patients (30 early-stage small cell lung cancer, 55 stage I/II adenocarcinoma, and 46 stage I/II squamous-cell carcinoma), 93 healthy controls, and 101 patients with non-cancerous lung conditions (55 pneumonia, 26 COPD, and 20 pulmonary TB). They randomly split the samples from this second phase, using two-thirds to develop their model and the remaining third for validation. The three-marker panel consisting of p53, HRas, and ETHE1 emerged from this analysis.

This panel distinguished lung cancer from controls with sensitivity of 50.7 percent and specificity of 90.7 percent. Looking only at high-risk patients (smokers 55 years or older with 20-pack years or more of smoking history) the performance was slightly lower, at 50 percent sensitivity and 84.8 percent specificity. Converting the array-based panel to an ELISA format, they found the test performed with 49.6 percent sensitivity and 87.9 percent specificity.

Given the test's equivalent performance in both high and normal risk populations, Zhu suggested in could prove useful as a general population screening test, with patients who test positive then passed on for imaging with, for instance, low-dose or high-dose CT scans. However, even at 90 percent specificity, the three-marker panel would be expected to result in a large number of false positives, while the test's sensitivity is well below that of low-dose CT, which in large trials has demonstrated sensitivity of around 94 percent and specificity of around 73 percent.

The higher specificity of the three-marker panel could make it useful for evaluating nodules identified via CT, which has a high-false positive rate. This has been an area of significant interest for proteomic researchers and companies, with the aforementioned Oncimmune as well as Quest Diagnostics and Integrated Diagnostics pursuing such tests. The latter launched its Xpresys lung test for this purpose in 2013 and plans next year to launch a second-generation version of that test that the company said can rule out lung nodules as benign with a negative predictive value of around 97 percent.

Zhu said he and his colleagues are looking for additional markers that could improve the panel's performance. One possibility they noted in the paper is including in their discovery arrays proteins harboring mutations linked to lung cancer. Another route they might pursue is looking at other immunoglobulin types.

"We've gotten some promising [additional] candidates," he said, "but, again, they have to be validated in a much larger cohort. We're in the process of doing that at the moment."