At A Glance
Name: Timothy Yip.
Position: Scientific officer, oncology, Queen Elizabeth Hospital, Hong Kong.
How did you get involved with doing biomarker profiling?
The first time I was in touch with proteomics was when I attended a seminar [in Hong Kong] by a guy who is working at [Ciphergen]. This is the first time I learned about it. Then I started to write to him to start the collaboration.
We are one of the largest oncology centers in HK, and we have a lot of patient resources. So that’s why I wrote to him to say, your technology seems to be very innovative, and we would like to explore using this technology to work on our patient samples. With the support of my boss — my department head — we started this collaboration [to profile for cancer].
Tell me about your first profiling project.
We are very interested in one cancer called nasopharyngeal cancer. This is a head and neck cancer that is very pervasive among southern Chinese. It’s pretty rare among Caucasians in Western countries, but somehow we have a large number of these patients here every year. And we do have a lot of tumor and also serum samples of these kinds of patients. So we started off using the serum from these patients to do the profiling. With the support of my head of department and also consultants we found [a] marker which can predict for relapse in this kind of patient. This is published in the January  issue of Clinical Cancer Research.
Did you discover one marker or several markers?
We discovered several markers, but the fact is that one marker was found to be more related with relapse, and actually the paper is focusing on that marker.
We are working together with another scientist and also our clinicians in our department. So the clinicians [have] provide[d] all the samples to us, and my associate actually went to work at Ciphergen for a brief period of time. Then later on, they put the machine in our institute and we started to work more extensively.
We used SELDI mass spec and then screened out a large number of patient ser[a] against a large batch of normal ser[a]. Then we found some markers which are elevated in the patient at the relapse time. Then we started working with Ciphergen to do protein identification using peptide mapping. After that, followed by tandem mass spectrometry analysis, we identified it to be an acute phase protein which is called serum amyloid A. Although this is an acute phase protein, somehow it is tremendously elevated at relapse of the patient when the patient has distant metastasis of the nasopharyngeal cancer into bone, liver, or lung. So that’s how we found it.
After we found this marker, we made a search and we found there are monoclonal antibodies against this marker. These are commercially available. So we can use that as the marker for something like [the] ELISA immunoassay, which can be used for the serum test.
What is the next step for your group?
After nasopharyngeal cancer, we have also embarked on a proteomics study of TNK cell lymphoma, which is also one of the pervading cancers in Hong Kong too. Also, we work on lung cancer, which is of course on top of the list of pervading cancers, together with our senior medical officer, who is coordinating all the collections. So then after working with all these, we did find some markers which can [be a] signature [for] lung cancer, and Ciphergen is very interested in this issue, so together with our clinician who is collecting the lung cancer specimens, [and] with Ciphergen, we put up a patent which is now initially established, signifying all the biomarkers which can be used potentially for lung cancer diagnosis.
What kind of specificity and sensitivity were you finding?
For the lung cancer markers actually the specificity and sensitivity were all about 95 percent. So that is pretty good. But we do have to further investigate the variation and also the consistency of it in a larger batch of patients. This is ongoing at present.
Where do you get your funding?
Generally mostly from the [government] and also a research grant from our hospital, and from a body called the Hong Kong Anti-Cancer Society.
Now all the hospitals and universities in Hong Kong are paying a lot of attention to [biomarker discovery] and now they are investing in different clinical departments using this technique to find biomarkers for different diseases. So it’s in a booming stage.
What are some of the difficulties you’ve encountered using the SELDI approach?
We find there is some technical difficulty initially in binding, because in binding you have to make a very tight surface so that when you put in your samples there’s no leakage whatsoever. But initially there is some kind of leakage. This is being addressed by first the template which is clamping the ProteinChips — sometimes the edge is damaged to a certain extent and that will have sample leakage. So we have to ensure there is no damage to the edge. The second thing is, when you incubate the samples together with the ProteinChips, there are bubbles, which can prevent the binding. So that also creates variability.
[Also], very often, you find a lot of [disease-] ubiquitous proteins being associated with certain disease serums or disease tissues, simply because you’re comparing [those tissues] with [those of] normal individuals. Because in normal individuals you don’t have any other kind of symptoms. So in order to be more objective and also more specific to certain diseases, you need to expand your control individuals to include patients with inflammation, patients with bacterial infections, and also viral diseases, so you can identify those markers which are present in cancer specifically. Some markers do cross-react in patients with infectious disease or inflammation disease. So I think initially when people started off with all these proteomics studies, they just compared the [presence of markers to that of the] normal individual which of course is very weak, or down-regulated, or even absent. So you find a large batch of all these markers — some are specific to cancer, but some are not. I don’t mean that those markers which cross-react with other diseases are not useful. They do have a use for mechanistic study. But for clinical diagnosis and prognosis, we need the more specific markers. So that’s why I have emphasized including a large number of controls from many other kinds of diseases as very important.
In the future, we will come up with a lot of biomarkers which are present in a cancer but also cross-react with others. This [is becoming] the reality. So we will be incorporating different markers of less specificity but by combining the numbers of it and putting it into an algorithm, perhaps we can make a diagnosis even better than [by using] one single marker of specificity. But we have to define all these markers and their cross-reactivities.
Tell me about your SARS biomarker discovery efforts.
That was a coincidence. Our pathologist was working on SARS and it happened that one day we met and talked about it and then we thought, ‘well, why not try it out using proteomics?’ [I talked] with my head of department, and he said it was a very good idea. So we started off a collaboration with the pathologists and later on with the people who were taking care of the patients in the hospital. It happened that at that time that they had a whole bunch of patients who were infected in the private hospital and being admitted to our hospital. The pathologists were starting to collect follow-up serum from these patients — at time of onset, during treatment, and after treatment, during the recovery period. So we used this batch of serum from this cohort of patients and started off the proteomics study. We found at least 12 markers which were all downregulated in SARS patients as compared with normal and also compared with other upper respiratory diseases like influenza A and B, and adenoviruses. Then we collaborated with the radiologists, and they derived a method to check the extent of the pneumonia of the patients by serial chest X-ray. They have a certain score. We tried to correlate the protein level with the serial chest X-ray score, and we did find that that [some proteins are] also very correlated with the extent of the pneumonia. So this is the manuscript that we are drafting and about to submit.
So the idea is to develop a serum test for SARS and for determining the stages of SARS?
Exactly. Because at the moment, there are two kinds of laboratory tests for SARS: one is the anti-coronavirus antibody test and the other is the RT-PCR. The antibody test is not very sensitive — it can only detect 80 percent of the disease after 1 to 2 weeks. The RT-PCR is good for diagnosis of about 70 to 90 percent of the patients at early onset, so it’s useful for diagnosis, but not very useful for monitoring disease extent, because after 1 to 2 weeks, usually the RT-PCR positive rate would drop to pretty low — less than 50 percent. So it’s only at that stage that the patient develops severe pneumonia. We suspect that actually pneumonia extent is not too related with the viral load, but it’s very correlated with the whole immune response. That is, the virus triggers a severe host immune response, and through some interleukin effects gives a lot of severe inflammation within a very short period of time. So it’s very interesting to study all those acute phase proteins signifying those responses at that period. We are filling the empty gap which cannot be [filled] by conventional antibody or RT-PCR tests.
So you’re using the SELDI system for the SARS work as well?
That’s right. And Ciphergen is also very interested in that, so we’re collaborating on that too.