By studying biomarkers known to be involved in gastrointestinal cancers, researchers at the Johns Hopkins Kimmel Cancer Center and the Johns Hopkins University School of Medicine have developed a biomarker algorithm that, when combined with a noninvasive method to collect esophageal cells for study, could give clinicians insight into which patients [SJMM1] have esophageal cancer or precancerous conditions such as Barrett's esophagus (BE) or high-grade dysplasia. Typically, such determinations are made by endoscopy, an invasive procedure performed with patients under anesthesia.
The method, if validated in further study, potentially could be used as a reflex test to indicate which patients should have endoscopy for a definitive diagnosis. A description of the work was published Jan. 17 in the American Journal of Gastroenterology.
"Our study used a methodically rigorous process to identify biomarkers, and we believe this represents the first study to select biomarkers for Barrett's esophagus, esophageal adenocarcinoma and high-grade dysplasia in this fashion," says senior study author Stephen Meltzer, M.D., professor of medicine and oncology and the Harry and Betty Myerberg/Thomas R. Hendrix Professor in Gastroenterology at the Johns Hopkins University School of Medicine. He is also an American Cancer Society clinical research professor. "Our algorithm included methylation of the genes USP44, TBC1D30 and NELL1, which have been well studied as diagnostic markers in cancers and in cancer biology."
Methylation is a chemical process that can impact gene expression and may play a role in cancer development. USP44 was previously demonstrated to be a marker of prostate, liver and colorectal cancers. TBC1D30 also has been shown to be highly methylated in colorectal cancer. NELL1 has been shown to be a diagnostic marker of both Barrett's esophagus and colorectal cancer, and hypermethylation of NELL1 has been implicated in the genesis of gastric, kidney and lung cancers.
During the study, investigators queried six datasets within the Gene Expression Omnibus database for biomarkers that were at least 30% methylated in BE but less than 5% methylated in normal tissues. They identified 30 candidate biomarkers for further study. Subsequent laboratory testing using methylation-based polymerase chain reaction selected 12 of these biomarkers as demonstrating significantly higher methylation levels in BE than in normal tissues.
Next, the researchers chose seven of these 12 biomarkers (GRAMD1B, USP44, HOXB13, A1BG, SPX, TBC1D30 and eg00720137), plus an additional five biomarkers (CDH13, FLT3, NELL1, TAC1 and SSTI) identified in their previous studies, for more analyses. First, they evaluated the biomarkers in 21 archived normal-Barrett's tissue pairs from patients with BE. Then, they assessed the biomarkers in 234 nonendoscopic esophageal sponge samples collected from patients with BE as well as control (other) patients having endoscopy at Johns Hopkins, the Allegheny Health Network in Pittsburgh or Erasmus University in the Netherlands. This cohort included 78 people with a normal esophagus, 77 with BE, 12 with high-grade dysplasia, five with low-grade dysplasia, one with indeterminate dysplasia and 61 with esophageal adenocarcinoma.
To obtain these samples, participants were asked to swallow a small sponge compressed in a gelatin capsule that has a string attached to it. After the capsule makes its way to the bottom of the esophagus, its gelatin coating dissolves, allowing the sponge to expand. Then, a clinician pulls gently on the string to retrieve the sponge, which collects cells from the esophagus on the way back up.
Samples were taken immediately before endoscopy or during outpatient follow-up within three months after endoscopy. All participants had a confirmed diagnosis of esophageal cancer, BE or high-grade dysplasia, or a biopsy showing absence of these conditions. The median age of participants was 65. Most were male (66%) and from the United States (82%).
Investigators split the samples collected from the sponges into a training set of 199 samples and a test set of 35 samples. All 12 biomarkers were tested in the 199 samples. Based on the training set results, investigators designed a three-biomarker algorithm using USP44, TBCD1D30, NELL1, age and sex. This algorithm yielded an area under the curve (AUC) of nearly 0.97 in identifying healthy tissue versus esophageal cancer and high-grade dysplasia, meaning that it performed very well. Extending the ability to identify healthy control patients versus patients with Barrett's esophagus and high-grade dysplasia or esophageal cancer combined, the AUC was 0.86, which Meltzer says is also good.
"The goal with this sponge-biomarker test is not to provide a definitive diagnosis," says Meltzer, who directs the GI Early Detection Biomarkers Laboratory at Johns Hopkins "Rather, it's to inform them that they may need an endoscopy, because their methylation test results were abnormal."
The incidence of esophageal cancer has increased fivefold in the Western population during the past 40 years, Meltzer says, and it is now the eighth most common cancer and sixth most frequent cause of cancer-related death worldwide. It is estimated that about 5%–12% of patients with gastroesophageal reflux disease (GERD) have BE. The exact number is not known because most BE goes undiagnosed, he says.
"With esophageal adenocarcinoma being the leading esophageal cancer type in the United States, there is an urgent need to apply these markers in a large-scale screening study to assess whether the test can improve detection of Barrett's esophagus and esophageal adenocarcinoma and ultimately help improve survival of these patients, which this study provides impetus for," Meltzer says.
Study coauthors were Andrew Kalra, Ke Ma, Yulan Cheng, Leslie Cope, Yifan Yang, Simran Jit, Yousra Ahmed, Shayan Gheshlaghi, Vincent Castillo, Russell Hales, Vincent Lam, Kristin Marrone, Ken Hui, Michelle Ma, Robert Hughes, Venkata Akshintala, Kathy Bull-Henry, Jinny Ha, Karim Boudadi, Zachariah Foda, Richard Battafarano, Mouen Khashab, Eun Ji Shin, Olaya Brewer Gutierrez and Saowanee Ngamruengphong of Johns Hopkins. Other researchers contributing to the work were from the Allegheny Health Network, Thomas Jefferson University in Philadelphia, Jefferson Einstein Philadelphia Hospital, Eastern Virginia Medical School in Norfolk and Previse in Baltimore (manufacturer of the sponge test).
The study was supported by grants from the National Institutes of Health (CA211457, DK118250, CA287294) and by gifts from Peter Nicholl and Stuart Israelson. Kalra, Meltzer, Cheng and Yang are inventors of the patented biomarkers. Meltzer is a founder of and holds equity in Previse. The Johns Hopkins University School of Medicine owns equity in Previse, and Ma is an equity holder in Previse. The Johns Hopkins University manages these relationships in accordance with its conflict-of-interest policies.
