An oral rinse could be used to detect head and neck squamous cell carcinoma (HNSCC) by detecting hypermethylated genes in abnormal cells that are shed into the saliva, say researchers in the January 1 issue of Clinical Cancer Research. However, more work is needed to refine the test, they comment.

Lead researcher Joseph Califano, MD, from Johns Hopkins Kimmel Cancer Center, in Baltimore, Maryland, said a saliva test is a better bet than a blood test in these cases, because some head and neck cancers do not shed genetic material into the blood.

A similar proposal, an oral rinse to detect head and neck cancer, was described last year by Elizabeth Franzmann, MD, from the University of Miami, in Coral Gables, Florida, as reported by Medscape Oncology. That test was based on a protein biomarker, soluble CD44, but Dr. Franzmann and colleagues reported that combining soluble CD44 with a test for hypermethylation increased its accuracy.

In the current study, Dr. Califano and colleagues studied a group of 211 patients with head and neck cancers, and compared them with 527 healthy individuals. Everyone took the saliva test, which involved using an exfoliating brush on the inside of the mouth and throat, and then rinsing and gargling with a salt solution. The resultant cell-laden saliva sample was then analyzed with PCR for hypermethylated genes.

The researchers investigated a panel of 21 genes and looked specifically for CpG-rich promoters known to be differentially hypermethylated in HNSCC. They identified 8 genes that were the best predictors of head and neck cancer, and then tested saliva with several different panels containing 3 to 5 genes.

The best result for sensitivity was obtained from a panel containing the genes CCNA1+DCC+DAPK+p16. This was tested in 176 patients and 417 controls, and resulted in 33.5% sensitivity and 91.8% specificity.

Better results for specificity were obtained with other panels — 3 gave a specificity of 97.1%, but these gave lower sensitivity results. A panel containing CCNA1+MINT31+p16 gave a sensitivity of 24%, whereas those containing MINT31+CCNA1 and CCNA1+p16 both resulted in a sensitivity of 22%.

"Improved sensitivity was obtained at the cost of decreased specificity," the researchers comment. "In general, we were able to define a panel for HNSCC detection with a high specificity but accompanied by a low sensitivity," they write. "This combination of characteristics would not be advantageous for population-based screening." However, the panels with high sensitivity and low specificity that they identified in these studies might be useful for surveillance in a high-risk population and could have potential in identifying relapses after treatment for head and neck cancer, they say.

Further work is needed to refine the test by uncovering additional hypermethylated genes that play a role, and to automate the test before multi-institutional clinical trials can begin, the researchers commented in a statement.