A Duke University-led lung cancer study presented last month at the American Society of Clinical Oncology meeting suggests that MALDI mass spectrometry's usefulness as a protein biomarker discovery platform is currently being held back by technological limitations.
In particular, the study indicates that the technology's limit of detection is not yet sufficient to detect key low-abundance proteins and that it suffers from reproducibility issues that, while ultimately resolvable, could present significant practical obstacles in a clinical environment.
Between February 2004 and May 2006, the researchers enrolled 1,074 patients with a clinically suspicious stage I lung lesion with the aim of developing a proteomic profile that could distinguish between benign and malignant lesions. Ultimately 913 of these patients proved eligible – 723 of which had non-small cell lung cancer and 190 of which had benign nodules.
Using a training set of roughly 100 known benign and 200 known cancer samples, the scientists ran patient serum on a PerkinElmer prOTOF 2000 MALDI-TOF instrument in an effort to identify differential peak patterns that could be used to distinguish the cancer cases from the benign. The remaining samples were reserved for use as a validation set.
According to David Harpole, vice chief in the division of surgical services at Duke, the researchers were able to identify a number of potentially distinguishing peaks, but weren't ultimately able to develop a model promising enough to pursue further.
"The bottom line was that we ran [the samples], and we were able to create several models, but none of them had the fidelity that we felt would be enough to move it forward," Harpole told ProteoMonitor. "We had around 20 or 25 peaks that we were able to ascertain, but a lot of those were in the noise range, and using several different normalization platforms we just couldn't make something stable enough to move forward."
"The sensitivity and specificity were just nowhere near what I would want," he said. "You would really need like a 99 percent sensitivity and a specificity of 80 to 90 percent, and we just didn't have that."
Harpole added that, while MALDI instruments have improved since the study's mass spec analyses were done, he and his collaborators at the Mayo Clinic think the technology still isn't ready for the sort of serum-based discovery work they attempted.
"It's close," he said. "It's much better, and the good news is that we have all this material banked, so it's available when the right machine comes around. But in 2011, the answer is no. Hopefully in 2013 or 2014 the answer will be yes."
The study also found reproducibility to be a significant challenge, said Gordon Whiteley, director of the National Cancer Institute's Gaithersburg facility, which performed the mass spec analysis.
"We spent a lot of time getting the method reproducible," he told ProteoMonitor. "We had to control both the temperature and humidity of the lab very tightly or we would get drifting. We optimized the buffers, the timing all of that stuff. It was a fairly long program to do that."
Achieving such tight control "was very expensive," Whiteley said. "Like anything, if you have enough time and money you can do it, but in terms of being a practical solution, at the time it definitely wasn't, and it probably still isn't. In terms of the practicality of taking it to a clinical lab, basically our conclusion was that it was possible but totally impractical."
Despite these difficulties, MALDI-MS has found uses in clinical proteomics – perhaps most notably as the platform used by Biodesix in its VeriStrat test for guiding treatment of NSCLC patients. The company performs the test out of its CLIA-approved laboratory, which, Harpole said, is the same strategy that the Duke researchers would have adopted had their efforts yielded a useful diagnostic.
Moving forward, he said, his team plans to try a more targeted method, first attempting to identify some novel tumor-specific proteins in lung tumor tissue and then seeing if they can identify them in serum to be used in a test.
Whiteley agreed that this sort of targeted discovery approach would probably be necessary.
"If you're just looking at the entire proteome, you need to beef it up somehow in a controlled fashion so that you see things," he said. He suggested that targeted mass spec approaches like SISCAPA might prove useful in this respect.
One difficulty in developing a lung cancer diagnostic to identify cancerous lesions following CT scans is the relatively high reliability of such scans, Harpole said.
"It's going to be very difficult, because if you take patients with suspicious nodules who are smokers, that's about 85 to 90 percent predictive, so you're going to have to improve on that, and that's going to be a very high bar," he said.
Harpole noted that such a test will likely be in high demand, however, given that recently released results from the NCI's National Lung Screening Trial showed that CT scans reduced lung cancer mortality by 20 percent compared to standard chest X-rays.
"There are going to be a lot of people getting chest CTs and showing abnormal nodules [and] it would be really great to have a proteomic test that would allow you to discern malignant potential in these things," he said.
This echoes comments made by Charlie Birse, associate director of product development at Celera – which has since been acquired by Quest Diagnostics – in an April interview with ProteoMonitor discussing that company's work on a similar lung cancer diagnostic (04/15/2011).
The NLST results are "likely to prompt some form of nationwide screening," Birse said. "If that does get implemented then there's going to be a major need for a test that can distinguish the status of pulmonary nodules" detected during CT scans.
Celera's test, which in a study examining 40 cases and 40 controls was able to distinguish between benign and malignant lesions with a sensitivity and specificity of 83 percent, consists of a six-protein panel identified by analyzing tumor tissue samples with an Applied Biosystems QSTAR XL.
Proteomics firm Somalogic is also developing a protein biomarker diagnostic for NSCLC, though that test has been developed using the company's Somamer technology as opposed to mass spectrometry (PM 12/10/2010).
The NCI's Early Detection Research Network plans this summer to launch a validation trial comparing lung cancer diagnostics including Celera and Somalogic's (PM 04/22/2011).
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.