Mass spectroscopy (MS) in Biosensor Blog! Let me explain. Biosensors are routinely used in doctor’s clinic or as point-of-care devices to detect biomarkers for diagnosis and prognosis of diseases. There is however increasing demand for better clinical tests to detect diseases at early stage or to identify markers that may predict our predisposition to specific diseases. Hence, instead of testing for a single biomarker the trend is to identify panel of biomarkers that together may be a better predictor of a disease. That’s where MS is playing a very important role and have in fact become a standard tool for biomarker discovery. Advance MS techniques can detect upto 4000 proteins simultaneously in a single sample. In an ideal world we would compare samples from healthy individuals with samples from patients using MS and identify the up and down regulations of specific proteins that can then be used as biomarkers for diagnosis/prognosis of diseases. The now famous study by Petricoin and Liotta in 2002 did exactly that for ovarian cancer and claimed “—result yielded a sensitivity of 100% (95% CI 93–100), specificity of 95% (87–99), and positive predictive value of 94% (84–99)”. The techniques they used for their analysis was SELDI (surface-enhanced laser desorption and ionization) from Ciphergen. The claims from the study were disputed because of serious concerns about the reproducibility and reliability of the results and the Ciphergen has since then gone out of business and SELDI was bought by BioRad. The study however showed the power of MS for profiling the serum proteome and drove researchers to further exploit this powerful technology.
Fast forward 7 years and one would think that the problem of reproducibility and reliability would have been solved by now. But not so, going by couple of reports in past two months in high impact journals. I talked about the first report published in Nature Methods in April 2009 in my previous post. That study identified serious inter-laboratory reproducibility problems and attributed them mainly to environmental contaminations and deficiencies in databases used to identify proteins. Second paper that came last week in Nature Biotechnology had just the opposite conclusions! According to the report
“Using common materials and standardized protocols, we demonstrate that these assays can be highly reproducible within and across laboratories and instrument platforms, and are sensitive to low µg/ml protein concentrations in unfractionated plasma.”
What gives? Turns out that the recent report differ from previous one in key aspects
- Method used by the latest report is a quantitative method called multiple reaction monitoring (MRM) whereas previous one was qualitative method
- Number of samples in recent study were 7 compared to 20 in previous one and
- Number of laboratories in recent study were 8 compared to 27 in previous one
- Protein concentrations in current study were moderately abundant >2ug/ml compared to 5pmol in previous study
- There were Oother differences
But still completely different conclusions from two studies may be considered astounding for somebody looking at this technology to provide better and painless diagnosis during their next doctor’s visit. These studies show we still have some way to go before MS becomes an accepted tool for biomarker discovery. As I was reminded early in my research career that any new technology is 1% inspiration and 99% perspiration.
What this all has to do with the Biosensors? – Plenty! First, as a biomarker discovery tool MS is expected to provide a wealth of well characterized biomarkers that can then be validated and transferred to biosensor format for use in Doctor’s clinic. Second, biosensing is moving full ahead with complex technologies for multiplex sensing be it protein arrays, encoded bead based sensors or micro/nano-cantilever sensors. It is worth learning the lessons from MS that reproducibility/reliability across Laboratories/Operators/Instruments/Samples/Patients/etc. etc. should be front and center if any serious foray in diagnostics is intended.