Dana Farber/Harvard Cancer Center, Cancer Proteomics Core
Name: Towia Libermann
Position: Director, Dana Farber/Harvard Cancer Center, Cancer Proteomics Core, 2004 to present; director, Beth Israel Deaconess Medical Center, Genomics Center 1999 to present; associate professor of medicine, Harvard Medical School, 2000 to present.
Background: PhD in immunology, the Weizmann Institute of Science and Technology, Rehovot, Israel, 1986; postdoc in immunology, Whitehead Institute for Biomedical Research; 1987 to 1990.
In the Jan. 15 issue of Clinical Cancer Research, Towia Libermann and his colleagues report a study in which they evaluated the sensitivity and specificity of the SELDI mass spectrometer for detecting established hepatocellular cancer, and compared it against a trio of commonly used detection markers.
They identified an 11-peak algorithm for HCC detection and found the SELDI had a sensitivity of 79 percent and specificity of 86 percent. In particular, Libermann and his colleagues report that the instrument proved more accurate than alpha-fetoprotein at the cutoff of 20 nanograms per milliliter, Lens culinaris agglutinin-reactive at the cutoff of 10 percent, and probthrombin induced by vitamin k absence at the cutoff of 125 milliabsorbance units for detecting small liver cancer tumors.
In spite of their findings, Libermann is moving away from the SELDI instrument as more advanced MS technology has made even more accurate detection possible.
ProteoMonitor spoke with Libermann this week. Below is an edited version of the conversation.
Why did you choose to use the SELDI platform?
At the time when we started the study [three years ago], this was pretty much the only type of technology where you could run enough samples through in a relatively fast period of time. We pretty much have switched most of our technologies now to either using MALDI-TOF-TOF instruments or … the LTQ Orbitrap instrument for doing biomarker discovery.
At the time, the other instruments were still not performing that well yet, and you didn’t have the labeling technologies such as the iTRAQ reagents and so on to really multiplex samples into a single analysis. So running a lot of sample through a high-end mass spec was really not feasible in a short period of time.
When we at that time … just used a linear mode to get sort of something equivalent to the SELDI-TOF instrument, sensitivity at that time, for instance, on the MALDI-TOF instrument in the linear mode wasn’t as good as [on] the SELDI-TOF instrument. Obviously, the resolution is much higher on the MALDI instruments.
The main thing is [the SELDI] was the fastest and most direct way of … profiling many samples in a very rapid time with replicates and being able to get … a survey of relative quantification of many peaks in many samples.
If you were starting the study now, would you use the SELDI?
What we would use now, and what we’re doing right now, is … using the Applied Biosystems 4800 MALDI-TOF-TOF instruments, and … iTRAQ labeling. We recently started to go from the 4-plex to the 8-plex iTRAQ reagent, so we can label eight samples and mix them together to one single analysis and fractionation on HPLC and then analysis on the mass spec.
We’re taking eight samples, each of them labeled with a different tag, then you mix them together, and you fractionate them and you run them on a mass spec as a single sample, and then you get for eight samples, the relative quantitation for every single peptide.
Can you share any results that you’re getting with the MALDI-TOF-TOF?
We haven’t done [any follow-up] yet to this particular study with regard to the liver cancer, so that one we’re starting right now.
But we’re doing all different kinds of other studies … for example, with liver fibrosis, where we use serum or liver samples. And we do studies for brain tumors with cerebrospinal fluid; we do studies with tissue extracts, for example, [ones] related to cardiovascular disease where we have heart muscle and isolate tissue extracts and do quantitative analysis; [and] we do studies for early detection of biomarkers in ovarian cancer.
We’re at the beginning right now, we’re more testing the qualitative and quantitative aspects in order to see what all the parameters are that you have to take into account when you do these kinds of patient studies where you get plasma from patients from all over the country that may have been stored [at] different times and different temperatures. What is actually the reliability of these kinds of samples?
Are you still testing the SELDI for liver cancer? Are you doing follow-up studies?
We are pretty much more or less stopping with the SELDI-TOF technology, and there are several reasons. First of all, the company Ciphergen [now called Vermillion] sold off the assets in regards to the technology to Bio-Rad, and we don’t really see anything from Bio-Rad coming out [where] they actually are going to further develop the technology … that would make it more useful, particularly with regard to better resolution and eventually being able to identify proteins directly on the SELDI platform.
Therefore, we don’t really see too much of a future with the SELDI-TOF approach. And the main thing, now with the MALDI-TOF-TOF approach, for example, or even taking an Orbitrap instrument, you really see such a dramatic improvement that you immediately get protein IDs together with the quantitation, that you can now run quite a lot of samples through the other high-end technologies. You get much better quality data.
Then what’s the point of your study? It’s pretty focused specifically on the SELDI platform.
I think it’s more of a stepping stone for the next study that we do on the MALDI-TOF-TOF instrument, and we’re developing on the MALDI-TOF-TOF platform something equivalent to the SELDI platform.
With iTRAQ and all these kinds of things, it’s very nice that you can run more samples through, but time-wise it still takes significantly longer if you do tandem mass spectrometry on every peptide to go through a lot of samples.
So we are definitely interested in applying [a] SELDI-TOF approach to a MALDI-TOF instrument on the linear mode and doing more the profiling and first identifying the differentially expressed peaks, either on the protein level or on the peptide level, and then being able to zoom in only on the ones that are differentially expressed.
That definitely would significantly speed up the throughput. So even though the SELDI is an instrument we probably will abandon, the strategy might not change too much on the new instrument.
Your study said that the SELDI was particularly sensitive to small tumors. If that holds up with other platforms, does that mean that mass spec can be used to detect liver cancer early on?
The whole rationale behind the study was really to find biomarkers that can detect liver cancer at early stages, and then the next level that we want to go to is to detect within the population of patients with liver cirrhosis, the ones who are actually at risk for developing liver cancer who may not even have yet any visible tumor.
That would be, obviously, the most interesting study. A lot of it is obviously related to [collecting] enough samples, and to collect from the same patients multiple samples over time, and then to see which of the liver cirrhosis patients convert to liver cancer patients.
And then to do a study where we know what should differentiate early liver cancer from cirrhosis, and then see whether we can find markers in cirrhotic patients that come up at a time when you don’t see any tumor.
The problem right now is that even though you have biomarkers such as alpha-fetoprotein that have some specificity, [they don’t] really detect most of the small tumors. At the stage of the small tumors, you can … pretty much cure the patient. The current way to detect most of the patients with tumors when they are larger, you have very [few] options.
Obviously, there are a variety of different therapies, but typically you cannot cure them. So there is definitely an urgent need to identify markers that can detect liver cancer at a much earlier stage.
Alternatively, there are MRIs, CAT scans, those types of things that can be done, but again, that’s also an issue with regard to [having] to do it with all the patients with liver cirrhosis relatively frequently, so it’s a matter of costs and convenience. And overall, a lot of these technologies … [use] quite a lot of radiation. That’s not really beneficial to do it every three months, every six months.
If the findings of this and other studies hold up, what do you envision happening? Will mass specs be used as a routine screening tool for liver cancer?
I think the current way that most of the mass specs work, I don’t think it’s necessarily the most sensitive and the most cost-effective way of doing something in a clinical lab. I still think it will be more an ELISA type instrument that will do this kind of test.
And obviously, if [there are ] multiple markers, you cannot do just a regular ELISA, but there are a variety of different companies now that offer multiplex ELISAs and micro-titer wells. Then you have Luminex beads and then you have Fluidigm coming up with a multiplex microfluidic system for multiple antibodies and various other companies with different technologies.
So I think it currently will be more cost-effective and faster to do it on other platforms. But you need obviously then specific antibodies to identify proteins, which I think, still from the commercial development point of view, makes sense.
Obviously, if there were some kind of an MRM approach, which on a triple-quad that would be highly sensitive and fast and very specific and very cost-effective, then maybe there would be some kind of competition with some kind of ELISA approach.
But I think mass spectrometry is more for the discovery and maybe validation on the academic level, but not yet from a clinical point of view. I think you would have to redefine how the mass spec works.