News and events

Using molecular vibration to diagnose cancer

Following the recent STAND UP TO CANCER campaign Emma Sandon-Hesketh, Research Development Manager, and Alison Naylor, went to meet with Dr Matt Baker to talk about his research and work on cancer diagnosis.

Can you tell us about your primary current research interests? What is going on for you at the moment?

I am interested in cancer diagnosis, for which I am using vibrational spectroscopy such as Raman and infrared. Raman spectroscopy uses inelastic scattering of monochromatic light (i.e. single wavelength) from samples under irradiation. The scattered light from the sample has a changed wavelength due to the monochromatic light exciting an electron into a virtual excited state and this electron then returns to an energy level one above or below the original energy level. The resulting spectra are characteristic of the type of molecular bonds and hence of the molecular structure of the materials in the sample. Infrared spectroscopy looks directly at the vibration of molecular bonds and thereby can identify them. If you imagine that you have a guitar, if you pluck a guitar string it gives a certain note; if you have a steel or a nylon string then they give different notes. Steel and nylon have different strengths associated with them.


Matt Baker

So essentially based upon the strength of the molecular bond, and the atoms on either end, then you can have different vibrational frequencies which are observable in the vibrational spectrum. Biomolecules are massive molecules with lots and lots of vibrating bonds leading to complex spectra. What we look for in a sample is a signature or fingerprint which is indicative of cancer. There is a region of the infrared spectrum which is called the fingerprint region. Whilst we don’t completely understand where all the vibrations come from, a signature within the spectrum enables us to diagnose the presence of certain cancers.

So presumably similar kinds of cancers have got similar signatures?

Diagnosis using vibrational spectroscopy on blood serum samples has been shown on only a couple of cancers. Before my current work, I showed this to be the case on high grade and low grade prostate cancer using tissue spectroscopy. Currently I am using serum to diagnose between high grade brain cancer and normal patients. Other groups such as Nick Stone down in Exeter, Peter Gardner at Manchester, and Frank Martin at Lancaster are all working on different cancers. One of the main pushes, for me, was the PSA test (Prostate Specific Antigen) which, when it came out was heralded as the answer to all problems for prostate cancer diagnosis but upon later investigation was found to be wanting. Instead of looking for a single marker, which is elusive, multiple markers are thought to be better due to the heterogeneity of cancer. We are looking at the vibrational combinations of multiple markers, from that we can obtain our sample signature for the diagnosis of cancer. One of the interesting things that no-one really has done is to take several different types of cancer, prostate, brain and skin, and put them all into one model, and that is something I want to do with these blood serum tests.

What is the purpose of putting them all into the one model?

First of all so that we can discriminate between them, but also to distinguish between cancer and non-cancer, or differing severities of cancer high cancer versus low cancer – we don’t know if there is a specific signature for cancers that are metastatic for example. I believe there are not, but that work hasn’t been done.

When you talk about high cancer versus low cancer are you talking about the grading system? If so, from a diagnostic perspective, you could then tell the extent of the cancer?

Yes ! For brain cancer there are grade one and two, which are counted as low grade, and grades three and four which are counted as high grade. Gliomas are one of the better cancers to grade like that; this sort of grading is very subjective. It depends on the person viewing the architecture of the tissue in the sample. If you have someone who is experienced they can determine the architecture more easily than an inexperienced person. A study by Latouff and Saad showed only 29.2 % of tumours were graded the same, thus showing the potential for differing interpretations.

You talk in your 2009 article about the potential of having algorithm based technology in clinical settings to increase the rate of efficiencies in grading. Has this happened as yet?

For the prostate cancer study you are referring to, which is a good example, I first looked at a model based upon the Gleason grading (the subjective architecture of the cells), but there is another system for prostate cancer, which is the tumour node metastasis (TNM) staging scale. Basically, the tumours are either T1 or T2 – and that’s contained within the prostatic capsule, so all inside the organ. T3 and T4 tumours are outside the prostatic capsule, which isn’t a subjective measure; it’s either inside or outside. I took the same spectra and built a diagnostic model based upon whether the cancer was inside or outside the prostatic capsule (with outside showing a more severe cancer). We can easily discriminate these to a high sensitivity and specificity which gives the basis of a spectroscopic grading scale based on non-human opinion, and objective. If we can develop this to become a spectroscopic grading scale, then that would be excellent, removing human error. However, there are many things that we have to research first, before introducing it into a clinical environment. Brain Tumour North West is helping us and we are very grateful to them. We have just got a longitudinal study running to follow people throughout their cancer progression taking samples at each point.

I have presented my serum work to the clinical staff and we are discussing putting this technology into the clinic so that it can provide information on cancer progression at their weekly consultation meetings. We are going to try and work this into a research bid and get a PhD student working between UCLan and the hospital to monitor the patient data.

In terms of collaborations with the NHS, how are these going? Have you seen a difference? Is this research making an impact in real terms to real lives?

There are several people who are trialling new technologies, for example a company called River Diagnostics, which has developed a spectroscopic test for pathogenic bacteria and microorganisms. Also, there are people who have managed to do clinical, bed side trials with Raman probes for skin cancer detection and monitoring. Here working with Brain Tumour North West there is a mix of clinicians and academics working together. It is the close working relationship I have had with clinical partners. We meet regularly with Professors Dawson (Consultant Neuropathologist) and Davies (Consultant Neurosurgeon) from Royal Preston Hospital. When we presented the work to them, they encouraged us to go for the grant we are working on with the aim of getting the instrument in the clinic.

At the moment, the techniques for the diagnosis of cancer are fairly invasive or intimidating. What you are proposing seems a lot less so?

Yes. The existing techniques are invasive and can be quite frightening, and especially for brain cancer. The main problems are the cancer symptoms, which, by the time they manifest, mean that there is a good chance that the therapeutic window is smaller. Also, the symptoms may not be specific. You may feel drowsy; your vision may go. For prostate cancer you may have a pain in your lower back and be urinating frequently. It’s not as if you can tell from the symptoms that you have got a cancer. I have put a bid into the EPSRC to study pre-symptomatic diagnosis using these spectroscopic methods. The only current way to diagnose cancer is from a biopsy, and, for brain cancer, that involves surgery to get a biopsy of the brain. The growth may be benign, making complex treatment unnecessary. Similarly with the PSA test for prostate cancer, 75% of individuals with a raised level of PSA actually have no cancer. Only 25% have an invasive procedure to take a tissue biopsy from which there is the potential diagnosis of cancer.

I see three types of technologies developing. One is a screening technology using serum. We have done this for 74 patients, from which 49 showed high grade cancer and 25 who were normal. To improve the study, I would like 500 samples, and that is just going to take time to get. Screening is the key issue; there is no screening at the moment for brain cancer, and recently, there has been news that the screening for breast cancer is more dangerous than people would want it to be – and we have spoken about the problems with PSA, so the only current way to really diagnose cancer is by a biopsy. For brain tumours, that is at least a few days in a hospital, then tumour processing and analysis before being sent to the consultants for a weekly meeting. Spectroscopic screening from serum would be an ideal way to avoid unnecessary biopsies.

It is also simple and it could be done in the doctor’s surgery. The training to perform the spectroscopy is very simple and, moreover, you can build an automated programme that takes the spectrum, compares it to a prognostic model, and then gives a positive or negative result. It could be as simple as that. You could even offer it as part of the Blood Collection services, as they currently screen for HIV and HEP C and use it in any of the normal routine systems, which would be ideal.

What kind of costs are these machines if you are talking about putting them into a clinical setting?

One machine at the moment is approximately £15k. That would come down a lot with a block order, but there are no consumables, apart from pipette tips. When set against the cost of treating a cancer patient, and the fact that one machine will do thousands of samples, it renders the price of the machine almost negligible.

So in terms of future development, you are looking at the efficacy of this kit to the stage of the cancer, being able to differentiate between different cancers and to implementation in a clinical setting?

Yes, that would be great! From the meeting I had with consultants at Royal Preston, we could start preliminary trials, if we get the funding, next year.

Do you believe there will ever be a cure for cancer?

Cancer is essentially a cell that is growing in an uncontrolled way and we can definitely kill those cells but the problem is selectivity. There has been a big increase in certain cancers, but there are some that have a very high rate of success with treatment and are essentially curable nowadays such as childhood leukaemia. So I think that there will be a drug out there that will be able to cure cancer. It depends when you give it. The effectiveness of therapeutic intervention decreases rapidly with the onset of symptoms. The earlier we have a diagnosis of cancer the earlier treatment can be started and the greater the likelihood of the cancer being cured.