Prefer to listen to the interview? Head over to our podcast 'Conversations with COSMIC' to hear the full interview with Dr Gianmarco Contino, MD PhD and follow us for new episodes coming soon!

Gianmarco Contino, MD PhD, is an Associate Professor of Cancer Genomic Medicine and an Honorary Upper Gastrointestinal Surgery Consultant at the University of Birmingham. He leads the Cancer Structural Variation and Aneuploidy Research Laboratory at the Institute of Cancer and Genomic Sciences [birmingham.ac.uk] of the University of Birmingham. Previously he was a Clinical lecturer at the University of Cambridge and Addenbrooke’s Hospital. He trained as a surgical oncologist and translational scientist in internationally renowned institutions including the European Institute of Oncology (Milan), Harvard Medical School and Mass General Hospital (Boston), University of Illinois at Chicago and Imperial College (London). He is also a member of the Von Hügel Institute of Cambridge University and a contract Professor at the Campus Biomedico University of Rome. He is member of several scientific and research societies and he is currently serving as the communication chair and gastrointestinal cancer group chair of the EORTC Pathobiology Group.

Gianmarco is the director of the Clinical Oncology MSc and module lead of the Cancer Pathology and treatment module of the BMedSc Clinical Sciences programme at the University of Birmingham. He is a dedicated mentor of the next generation of scientists and oncologists and an advocate of equal access to science promoting the training of students from disadvantaged backgrounds. He actively volunteers for In2scienceUK.

He published extensively on molecular biology and cancer genomics of Pancreatic and Oesophageal Adenocarcinoma in many important science journals. His research has been supported by Cancer Research UK, American Italian Cancer Foundation, the Associazione Italiana per la Ricerca sul Cancro, Fondazione Umberto Veronesi, the National Institute for Health Research and the Imperial College NHS foundation.

We've just described quite an impressive resume of many different roles. Could you talk us through the various hats you've worn and how they relate to one another?

Becoming a clinician scientist has been an exciting journey. When I first started my residency in Italy, I would regularly get the question: what do you want to be, a clinician or a scientist? Fortunately, since then, clinician-scientists have become more and more important in science and in medicine. I'm glad that now there are many paths in the UK, the US and in many other countries that train people to become good scientists and clinicians, bringing together these two sets of skills to advance clinical sciences. In my career, I had to switch between the two “jobs” to complete my training and this brought me from Italy to different places in the United States such as Boston and Chicago, and in the UK including London, Cambridge and now Birmingham. I have done some great ‘academic tourism’! I learned a lot about how medicine is currently done, how it can be done differently, how research is conducted, but also what the major barriers for scientific discoveries to become available to patients are.

I remember once hearing a clinical scientist speaking at a conference, and showing his timetable. He explained how he had to do 10 different things at once, at all times. This is a challenging role that you've taken on, what keeps you motivated?

You have to make some choices.

If you want to run a lab and a good research programme, you have to tailor that in a way that works for you, the patients and the team that you're supervising. It's really exciting! It's really motivating to see where your research is going, what is the clinical context in which you are working and how this can have an impact for patients. I feel really privileged to have this constant reminder of why I'm doing all this.

It's lovely to speak to somebody who clearly loves what they do! Your clinical specialty is upper gastrointestinal (GI), could you tell us a bit about what cancers this includes and if you focus on a particular type of cancer?

In the UK upper GI specialties include oesophagus, stomach and the first part of the small bowel. There are two other main branches of gastrointestinal surgery: hepatobiliary that focuses on liver, then pancreas and colorectal. My main interest is in oesophageal adenocarcinoma which is, for reasons that we don't entirely understand, relatively common in males in the UK and in other Western countries and is linked to gastroesophageal reflux and obesity. The only known precursor of oesophageal adenocarcinoma is Barrett’s oesophagus, which is a premalignant condition with a very low risk of transformation to cancer. This means the management and the follow up of these patients needs to be sensible to avoid the progress to intermediate form we call dysplasia, which carries a high risk of transformation to cancer, but also making sure that we don't overtreat low risk subjects with constant endoscopic exams. Early detection is one of the most important challenges, we know that many patients will come to a diagnosis of oesophageal adenocarcinoma without having had a previous diagnosis of Barrett's oesophagus. The good news is that there are now novel strategies to try to identify these patients early (e.g. cytosponge and other non-endoscopic diagnostic tests). The big challenge for oesophageal adenocarcinoma in the next 5-10 years is to reduce the incidence of advanced disease by diagnosing patients at the stage of dysplasia (i.e. early disease) and to improve the still poor outcome of invasive cancer with more targeted treatments.

Speaking of outcome, we know oesophageal cancer still has very poor survival. What are the current treatments? Why is survival so low? And thinking about all of the work you've done in various different clinics around the world, what do you think the best strategies are for tackling it?

Well, the paradigm around the treatment of oesophageal cancer has changed because of some key trials that have been run in the UK and in Europe. They clearly showed that there was an advantage in treating cancer with neoadjuvant therapy, meaning with a chemo or a chemoradiotherapy before surgery instead of going straight to surgery and then giving chemotherapy (this second strategy is called adjuvant therapy). This approach has shown to have a beneficial impact on the survival of patients. These ‘curative treatments’ are for patients who at the time of diagnosis are within a certain disease stage. The reality is that, unfortunately, we still diagnose too many patients at late stages when the treatment for them is only palliative, meaning that we don’t have chances to take them to a stage where we can also operate on them. Additionally, among the patients in a curative pathway, there are responders and non responders, the most successful trials show a five year survival of about 50%, which in the clinical practice likely translates into something lower than that. For those eligible for treatment, the chances of cure have improved dramatically in the last 10 years, but we still have a long way to go to improve outcomes for non responders and for those with late stage disease.

Our podcast interview with you came about after we met a couple of your PhD students at a Genomics England conference, and they happen to mention that they're using COSMIC in their work. I'm really interested to know how you first came to know about COSMIC and how you've used it through the various stages of your career in research.

COSMIC really was one of the very first useful resources for cancer geneticists. When I first started my lectureship in Prof. Rebecca Fitzgerald’s lab in Cambridge, she had just joined the International Cancer Genome Consortium. She also had established OCCAMS, a consortium of high volume referral centres for Oesophageal Adenocarcinoma that had recruited quite a sizable number of samples from oesophageal cancer patients across the country. That biobank has been really the backbone of the collaboration with the International Cancer Genome Consortium project that led to the sequencing of 500 whole genomes from the most common cancer types, including oesophageal cancer. When I started, the aim was to create a catalogue of mutations of Oesophageal Adenocarcinoma. COSMIC is actually a pancancer catalogues of mutations that we find in somatic tissues, i.e. the mutations that arise in the cancerous cells. I was actually lucky enough to contribute to some of the COSMIC data! In the lab, we had sequenced nine oesophageal cancer cell lines. There are about 10-11 cell lines commonly used as models of oesophageal adenocarcinoma and we generated the whole genome sequences. Sanger had started this project called the Cell Lines Project and they were starting to establish some guidelines, which were very helpful. In particular, for example, what cutoff you use to call a copy number alteration in a cancer cell line, meaning how many copies of a region you need to call that the gene has been amplified or deleted in a cancer and or in a cell line. There are other more sophisticated thresholds too to make sure that what you're calling a mutation is actually a somatic mutation and not just a variant or something that belongs to the germline background. It was really exciting then to see that this small side project I was working on actually contributed to a resource used by millions of scientists.

On a daily basis, the COSMIC catalogue is really the place to go if you want to check that a mutation found in a sample or in a cohort of patients have been found in other cancers, or whether for example, a gene is altered through other mechanisms such as copy number alterations or structural variants. It's really the “Yellow Pages” of somatic mutations. One other application that I found very helpful was the Cancer Gene Census. One major problem we face is: how do you define a driver gene? Meaning a mutation that contributes to the progression of cancer. I mean, it's obvious for major genes, but even then, major genes like P53 or EGFR2 may have variants that are not clearly functionally implicated in the progression of cancer. It's really helpful to have a resource where you can go and see whether the mutation you found has been already listed in that catalogue. Of course this is always a work in progress.

There's still literature with a surprising number of sequencing mistakes, or false positives that are reported as drivers.This shouldn’t happen nowadays because we've got resources to filter for that.

From what we understand, tumour boards occur when a group of experts, like yourself, gather to discuss molecular subtypes of specific cancers and recommendations of targeted therapies. Is this correct? And can you tell us a little bit more about this?

One of the improvements in the standard of care of cancer actually happened when expert clinicians started to talk together. The radiologists, pathologists, the surgeon, the oncologist, the radiotherapist, the specialist nurse, they now all meet together and all look at the case holistically to discuss what's the best path for that patient. Now we've got another piece of information that needs to be integrated into this approach, and these are the cancer genomes. Major efforts like the 100,000 genomes of Genomics England and the creation of a network of NHS Genomic Medicine Hubs fostered the sequencing capacity with all the logistics needed to deliver sequencing data with a quick turnaround. Those results can now be discussed at genomic tumour boards and provide precision oncology opportunities for patients.

More and more we will not just look at the CT scan or the PET scan or the biopsies of a patient, but also scan the genome of that cancer to see whether we can try to treat it in a way that is more tailored around its biology. An easy way to gain insight into a cancer genome is to use a gene panel that allows you to sequence a selected number of genes that are relevant to different cancers.

This information is often very helpful, but it becomes really important especially when patients are in trial. When the more established first line treatments do not work, we often need to find alternative strategies that are supported by less evidence. In these cases, having genetic information might help to point to the right drug or drug combination. So, looking at this information clinicians can now discuss, whether there are specific trials or specific drugs that should be considered for that patient in addition or alternative to the standard treatments. This is happening for many cancers. Oesophageal cancer benefits from this to a certain extent, for example, in order to decide for immunotherapy, there are some markers that can be identified through sequencing and they predict whether the patient is more likely to respond to checkpoint inhibitors. This is just one example of a target though, there is a portfolio of actionable targets that will hopefully expand more and more.

What we have learned in the last few years is that there are different markers that can be extrapolated from sequencing such as mutational signatures (i.e. a pattern of mutation types that indicate a certain exposure or DNA repair defect). For instance, breast cancers with a specific signature that underlie defects in a DNA repair pathway named homologous recombination may be more likely to respond to the category of drugs PARP inhibitors. We are now building up a knowledge base that is really essential to match mutations with drugs and clinical indications. We still need a lot of studies and a lot of experimentation. These biomarkers are going to become much more complex and specific as we accumulate knowledge around the biology of cancer. For instance, instead of looking at mutations individually, we might, at some point, be able to understand the role of a mutation in the context of other mutations.

Are there standardised gene panels that are used clinically? Or does it vary on a case by case basis? And how does a new gene panel come about?

In practice there are some very established panels such as the ones from Foundation Medicine from Illumina that contain a certain number of cancer relevant genes and allow you to identify other metrics including mutational signatures, tumour mutational burden, or copy number alterations. These are pretty much universal. For very common cancers, these panels work very well because they are very cheap, quick to do and likely cover all the actionable alterations. In addition they're very quick to analyse because you have small amounts of data, so they are very manageable and effective.

For rarer types of cancer, for example, paediatric cancers, or patients where you have rare mutations, you may have to escalate the type of sequencing you do: for example whole exome sequencing or whole genome sequencing. For cancers like oesophageal adenocarcinoma we really need to build panels that take into account the biology of the tumour. For chromosomally unstable tumours for example, the integration of panels that uses short reads with orthogonal sequencing techniques such as Oxford Nanopore to look for long stretches of DNA and resolve complex rearrangements can potentially provide additional information although the utility needs to be proved by more research.

Increasingly, we will be able to derive useful information from what we call liquid biopsy, meaning that we can analyse cancer mutations directly from a blood drawing. That's very helpful when we want to monitor the disease or response to treatment. Another evolution of the field will be the integration of complex biomarkers such as DNA, RNA, epigenetic modification and proteins. The field is evolving quite rapidly.

What is the current state of personalised precision medicine in the clinic, are tumour sequences standard? or is this likely to happen in the long term?

Tumour sequencing is not standard in healthcare. For some cancers we are routinely sequencing single genes such as KRAS, EGFR or BRAF, or using hybrid techniques (i.e. looking at the expression of proteins or using fluorescent probes). That’s what really started the field of molecular pathology. However, the cost of sequencing is now becoming so affordable, that it really doesn't make sense to do several single tests, as it is more cost effective to just go for a gene panel. From my perspective, I think that gene panels are going to become the routine in future and we will be doing whole exome or whole genome sequences for more rare cancers. The weak link, at the moment, is having enough evidence to make sure that we can make good use of this information. You may have several potential targets in a patient but you need evidence that targeting them with a specific drug gives a superior outcome over the standard treatment. The other issue is making sure trials are more widely accessible and designed to accommodate the wealth of information that comes from genetic testing.

The way we are reasoning about the biology of cancer is going to change quite dramatically, because we've got to be able to integrate information in ways we cannot even think of at the moment. We've been mostly talking about DNA in this conversation, that we shouldn't forget that it's not just the DNA. There's also the transcriptome (meaning how these genes encoded by DNA are expressed into RNA and finally translated into proteins), or the epigenome (meaning the modification that the genome undergoes and changes the expression of genes). What we do need is more access to targeted treatments, changes in the regulation on how we run trials and some centralised way to enrol patients into trials to ensure equal access and increase the number of patients treated in the frame of clinical trials.

Is COSMIC used as part of the clinical decision support, or in defining these panels that we were talking about previously?

COSMIC enters at different levels, cancer genomes often carry about 5000 or even 200,000 point mutations. Essentially we know that the number of drivers that are really important for progression are a handful, maybe 10 or even less. So we really need filtering, and this is where resources like COSMIC Cancer Gene Census are really important. This process needs to be integrated in what we call bioinformatics pipelines and eventually generate a report, and this report needs to come up with the ‘highlights’ of the cancer and what we think is actionable. Of course, there are other applications of COSMIC, such as Actionability, that informs whether the mutation that we found is actually actionable (i.e. we have a drug designed to target that mutation). This together with many other resources that are becoming available can eventually generate a more complex report that doesn't only give you the results of the test, but also give you some specific indication on what is the priority of the different mutation in terms of treatment (ESMO has generated useful guidelines in this respect), the body of evidence supporting the clinical indication, the trials available and so on.

You mentioned that a lack of trial data, not only in terms of genetic data, is a hugely limiting factor. Could talk a little bit about just how limiting poor representation of, perhaps, different ethnicities and minorities is within those databases.

This is really a concerning problem at the moment and this is something that the scientific community has been realising in the last few years. We're now trying to fix this with projects globally to gain more information about cancer genomes in different populations. The reality is that more than 90% of genomic databases today contain patients of European backgrounds, and that's a hugely limiting factor because we are using this knowledge to treat patients with diverse ethnic backgrounds and this may lead to consequences with respect to efficacy and toxicity of treatment or even prevention strategies. This is just not right and must be corrected. There are some big sequencing projects going on in Africa, India and China. It is critical all ethnicities are represented in mutation databases to deliver rational, targeted treatment. In the UK there are also some specific projects now such as Genomics England. In my lab we are trying to get insights from some African cancer cohorts. We know that Africa has got the highest degree of diversity and that's because we all came out of Africa, that's where the humans have spent most of their time, therefore most of the genetic diversity has been generated there.

Another crucial aspect is the accuracy of clinical annotation of sequencing data. To make sense of genetic data we need to know the risk factors and exposures, the treatment response and toxicities, the clinical history and so on. If we want to have an impact on future generations of patients we need a global effort to standardise the way we collect and share data. Patients need to be aware and make sure that they consented specifically for these uses. The paradox is that, at the moment, we are sequencing and generating incredibly useful data and we cannot analyse them because of regulatory constraints. On one hand we need to preserve the confidentiality of genetic information, but, with the right change of attitude, we could easily put in place a safe system to make these data safely available with the right procedures. Patients are generally very keen to have their data used for research if they are correctly informed. We need to build tighter connections and trust between researchers and patients.

What would you say is the next big thing for you? Could we get perhaps a sneak peek into any upcoming publications or work?

We are studying a series of interesting aneuploidy models of oesophageal adenocarcinoma that we have generated in the lab, and we are starting to unravel the data. We really hope that we’re going to have some insight into the specific mechanisms of how oesophageal adenocarcinoma thrive on chromosomal instability and aneuploidy. We hope this will help to identify targets or mechanisms of dependence that we can take advantage of for treatment purposes. We're also looking at the germline side of copy number and structural variation. We all come with about 3 million different point mutations that make us a little different, but there's also a background of structural variation that is different from person to person that we inherit from our parents. This may also contribute to our risk of developing cancer, so we are actively working on that to understand whether there is a heritability link to structural variation, we hope to have some interesting results on that soon too.