Personalized healthcare: the right test for the right patient
Mercogliano is one of these places where time seems to have stood still. Visitors who come to the small Italian village with a population of 12,000, located just 45 kilometers outside of the hectic and crowded Naples, find themselves all of a sudden in a quiet refuge full of old houses, narrow paths and astonishing views of surrounding landscape.
And yet, the village is also home to one of the most advanced laboratories in Italy. Located in a former school building with walls that still carry colorful comic-like paintings reminiscent of its origin is the Centre Ricerche Oncologiche di Mercogliano, or CROM – a satellite institute of the National Cancer Institute of Naples (Istituto Nazionale dei Tumori “Fondazione Pascale” – IRCCS, Napoli).
We visited CROM to meet Dr. Nicola Normanno, Chairman of the Research Department of the National Cancer Institute and Head of the Laboratory of Pharmacogenomics at CROM, to discuss the factors driving the choice of various diagnostic technologies and the growing importance of next-generation sequencing in clinical applications.
Dr. Normanno, what is CROM’s function within the National Cancer Institute?
CROM is a satellite institute of the main center that we have in Naples. We opened this institute in 2007 and immediately started a program for biomarker assessment in solid tumors. Today, we are deeply involved in clinical diagnostics – particularly EGFR testing in non-small cell lung cancer, RAS testing in colorectal cancer, and BRAF testing in melanoma – as well as a range of various research projects.
How many patient samples do you process on a monthly basis?
If we focus on molecular biomarker analysis only, we have approximately 100 patients per month. About 50% of the samples are from the main institute in Naples, and 50% from other hospitals and healthcare institutions in Southern Italy.
"We believe that a reference laboratory like CROM should have more than one technique at hand and be in a position to vary the testing approach."
Dr. Nicola Normanno, Laboratory of Pharmacogenomics at CROM, Naples, Italy
Have you been offering the same scope of service since the institute was established in 2007?
No, it all started with the European Medicine Agency's decision to restrict the use of Cetuximab and Panitumumab in colorectal cancer patients depending on their KRAS status. At that time, all the sudden KRAS testing became mandatory to prescribe these drugs in metastatic colorectal carcinoma patients, and we as a lab had to start offering this service to patients in our region. The other tests were then added to our portfolio over time.
Are you performing all of these tests using the same technology?
We believe that a reference laboratory like CROM should have more than one technique at hand and be in a position to vary the testing approach depending on the characteristics of the tumor specimen and the level of information we need to provide. For instance, we knew that the percentage of tumor cells in many samples would be too low to use sequencing, particularly in lung cancer specimens. So we've established a two-tier approach with Sanger sequencing and real-time PCR tests right from the start.
So is the concentration of tumor cells in the sample the main factor driving the choice of the diagnostic technology?
It is one of the factors, but not the only one. But in general it is fair to say that we'd use highly sensitive technologies such as QIAGEN's therascreen tests when the concentration of tumor cells in a sample is low or when sequencing returns ambiguous results.
The pharmacogenomics laboratory at CROM employs a wide range of modern diagnostic techniques, including QIAGEN's real-time PCR based therascreen assays, Pyrosequencing and next-generation sequencing approaches.
Do other issues influence the choice of the right diagnostic technology?
Another important factor is the number of biomarkers or mutations that you need to analyze. We experienced this in metastatic colorectal cancer. Initially, we were only looking for certain mutations in defined regions of the KRAS gene. Now we also need to test for other rare mutations and alterations in the NRAS gene, since they also drive the response of patients to targeted therapies. In fact, this has completely changed our approach to detecting these mutations.
To what extent?
We had two choices, two technologies we could use: Pyrosequencing and next-generation sequencing. We moved on to develop two different test methods, but in the end decided to use the Pyrosequencing platform for the routine clinical diagnostics. The reason for this is quite simple: We believe that if a patient sample comes to our lab with a request to assess the KRAS and NRAS status, we are only allowed to test those biomarkers. The NGS panels, in contrast, are much broader and cover at least 10 or 12 genes – in case of our research panel, even 22 genes.
So it comes down to the cost-effectiveness of the two methods?
This is an important consideration as well, but not the main point I'd like to make. In routine diagnostics, you can't just go on and analyze additional genes not strictly related to a certain disease without the patient's explicit written consent. This is because a lot of the data you will get back might be difficult to handle. You can end up with germ line mutations, with rare mutations of unknown significance or with alterations that might have some significance in an experimental trial, but are not yet useful in a routine clinical setting. It is a different story if you work with an ethics committee and enroll patients for a research study, but not in the routine day-to-day operations.
Are there any specific differences between the tumor types you have to analyze?
One difference is obviously the number of biomarkers you're looking at. Another important factor –which has a significant impact on the workflow – is the amount of sample material available. In EGFR testing for non-small cell lung cancer, for example, we're mostly dealing with cytology specimens or very small biopsies with a tiny fraction of tumor cells, since most of the patients in an advanced stage of the disease do not undergo a surgery. That's why about 70% of the lung-cancer tests we run are therascreen tests. In colon cancer, in contrast, you usually have huge amounts of tissue.
Where do you see the industry evolving over the next five years? Will NGS become more commonplace in the clinic?
At least in cancer testing, I think the future will be dominated by NGS. There is no other choice because the number of biomarkers we need to assess will increase, as additional biomarkers will offer more opportunities for patients and might really change their prognosis. But if you're not a combined research-diagnostic lab right now, the incentive to do NGS is rather small.
Is the increase in biomarkers the only factor driving NGS?
It's interrelated with other factors as well, namely cost, the time-to-result and the amount of sample you get to work with. Let's talk about the costs first. It's true that NGS is more expensive than other technologies, but if you add several biomarkers then it gets more cost-effective. Then there's the time factor. If you use other technologies to analyze four or five different biomarkers, you will most likely follow a separation approach, starting with the most frequent mutation and then moving towards the less frequent. But in lung cancer, for instance, we know that there are 2-3% of patients with NRAS mutations, 3-5% with BRAF mutations, and so on. Testing this one-by-one would take too long. Finally, there is the amount of tissue. In some cases there is simply not enough material to test for all biomarkers using individual tests, and you can't simply flip the coin to decide that's maybe it’s an EGFR, KRAS or BRAF patient. When targeted drugs will become available, we will need to test all of those biomarkers. So in general, we will need technologies that are cost-effective, quick and can perform all the tests with a smaller input of DNA and tissue available. This is really the challenge for the next few years.
Why aren't we quite there yet – what is hampering the dissemination of NGS in the routine setting?
A wider screening definitely opens up new opportunities for patients, even now. The problem is that NGS is still too complex for routine clinical labs. The bioinformaticians need to have experience with the technology, you must be aware of its challenges and limits, and you need time and reference material to standardize it in your lab. We're not ready yet to bring this scenario to routine clinical labs.
Going forward, Dr. Normanno expects that the number of biomarkers to be analyzed in cancer and other patients will increase - posing challenges for diagnostic workflows that exist today.
You mentioned bioinformatics – how are you dealing with this aspect of your workflow right now?
I believe that bioinformatics is really the limiting point in our workflow today. The bioinformatics pipeline is more important than the wet part in NGS, and I'm not sure that all labs that try to implement the technology are aware of this. With NGS, you start from raw data and if you use the wrong algorithms for analysis this can give you completely wrong results. You can end up with mutations that are not really present in your patient. You need to be aware of this. Otherwise, it can be really dangerous for patients.
What's the reason for this?
These types of errors are present in any kind of analysis. But with NGS, the errors are in the technology, so you really need to have a reliable bioinformatics tool that helps you to reduce them significantly.
One of the last things we wanted to ask is your opinion on the scope of the NGS test approaches that exist today. Are gene panels sufficient for diagnostic applications or will we move on to exomes or even full genome scans?
I think that this is further down the road. I can't imagine what we would do with a whole genome or what it would solve today. My view on this is that we will most likely have two levels of screening. Initially, patients will be screened for the most frequent actionable mutations. If this screen yields no results, then specific subsets of patients – such as young patients or patients that fail any kind of therapy – will be tested more deeply with wider panels up to full exomes or even whole genomes. Maybe in 20 to 30 years we'll do full exomes or genomes for all patients, but certainly not in five years. I'm quite sure of that.