Cytel is hosting a webinar, “A Clinician’s Perspective on Cancer Drugs Development”, on April 28, 2020. Our speaker, Professor Martin Fey, Medical Oncologist from Switzerland, will brief us on treatment evolution and give us a deep dive into clinician perspective on endpoints, PRO and patients perspectives, and importance of biomarkers in oncology.
In this interview, we speak to Professor Fey about his experience of over forty years in medical oncology, the evolution of clinical cancer trials, the difference between clinically meaningful and statistically significant results, the debate around patient perspectives and other important topics around cancer drugs development.
Cytel: Could you give us some background about your career and how you came to be involved in medical oncology?
Professor Martin Fey (MF): I was a medical student at the University of Bern in Switzerland. During my postgraduation, I began studying Pathology, which I found very interesting because I liked pictures, morphology and diagnostics. However, in pathology you do not get any opportunities to interact with the patients. So, I decided to take up Haematology which in a way links pathology, pictures, i.e. looking at bone marrow slides and blood films, with clinical problems. Since Haematology has close connections with oncology, I felt I needed some background training in Medical Oncology. Professor Kurt Brunner who was a pioneer in Medical Oncology in our country and elsewhere, had a strong influence on me. After I finished my internship in his department, I received a job offer from him which I immediately accepted, and there has been no looking back since then.
I never regretted my choice because I still believe that Medical Oncology is an incredible field. Over the past 40 years of my career, I have seen tremendous development and progress, which you do not get to see in other fields of medicine.
Cytel: How has design and conduct of clinical cancer trials evolved over the years?
MF: We have seen a lot of changes. The early trials from the 70s and 80s would probably not stand the scientific scrutiny which we now apply when we decide whether to activate a trial or not. Those trials were rather loosely controlled. Most of them are what we would now call Phase 2 trials, where some new drugs or combinations of drugs were given, and the patients were observed to see if there was an effect or not. In those early years, these trials with very simple designs were quite adequate.
In my upcoming Cytel webinar, I am going to talk about several time-honoured treatments, which brought the respective cure rate of specific cancers from zero (no survival) to appreciable levels of cure rates of 50 to 70 percent. For example, Cisplatinum, which is a very old and extremely unpleasant drug, was introduced in the 70s or early 80s. It resulted in an increase of the cure rate of testicular cancer in young men from zero to 50 to 60 percent, and eventually 80 to 90 percent. Over time new drugs did provide advantages in a step-wise fashion, but the benefits kept getting smaller, which is why larger numbers of patients needed to be tested. Eventually randomized Phase 3 trials were introduced, and the statistics had to be refined.
We now have new trial types such as adaptive trial designs, including basket trials, umbrella trials, and the like. We will still need to learn how to handle these to the best advantage.
Cytel: What is the difference between clinically meaningful and statistically significant results, and what consequences would that have on a patient?
MF: In my webinar, I shall begin by presenting the aims of a clinical trial testing a new drug or a new treatment seen with the eyes of a patient. Cancer patients hope to live longer and/or better. To my mind, this is clinical relevance. When we choose a trial design, we should choose clinical endpoints which directly measure the clinically meaningful effects of the treatment in cancer patients. In a trial you might eventually see a difference between the new and an old treatment. The statisticians will of course look, whether that difference is, by their techniques, significant or not. However, their appraisal (i.e. statistical significance) will not as such and automatically imply that the result will be clinically meaningful – an aspect which is quite often forgotten.
Statisticians must come in at an early stage of the trial design, for example, when they calculate how many patients we need to witness a putative difference or anything else we are looking for. A results is “statistically significant” when the significance level is at five percent or lower. If you see a very small difference in a very large number of patients, it may well be statistically significant (at the five percent significance level), but clinically perhaps it is totally irrelevant. That should be kept in mind when we look at trial results, and I think clinically meaningful results should reign supreme. The statistical significance is just a technical tool.
Cytel: What is the debate around patient perspectives about? How would this effect trial designs?
MF: As I mentioned earlier, when you ask a patient what they expect from a new therapy, for them true progress would mean they can live longer and better lives. This should be considered as an important patient perspective. In clinical trials, patient reported outcomes should have more weight than they have at present because they ensure that patients will report, in their own way, how they personally benefit from a new treatment. A good example is quality of life (QoL) assessment in trials.
QoL data are mostly collected as a secondary endpoint. If you look at many clinical trials, you will see that an appreciable percentage of them completely lack ‘quality of life’ assessment, or provide an incomplete QoL data set. Progression-free survival (PFS) is a very frequently used endpoint in Phase 2 trials and in Phase 3 trials. This endpoint, although very popular among clinicians and the regulatory registration agencies, does not at all consider patient reported outcomes, quality of life, or the aims put forward by the patients. PFS just looks at improvement of imaging findings, whether a tumour lesion seen on the CT scan would get smaller, or whether a serum tumour marker decreases, etc. There is good evidence, that often PFS is not a good surrogate marker for patient survival or quality of life.
In my opinion, PFS should be abandoned in clinical trials used for registration of new drugs (chiefly phase III trials) and should be replaced by clinical endpoints.
Cytel: What do statisticians designing clinical trials need to keep in mind about patient perspectives?
MF: We should build trials based on patient needs and not with the aim to only lower tumor marker results or to embellish imaging findings. When I review trials, I often observe that statisticians have increasingly turned into very good clinical oncologists. Although they seldom or never see patients, they are increasingly aware that they should look at clinical relevance and not just at their statistical numbers. They now pay more attention to whether a specific trial design with its primary endpoints, on which the statistics are based, will indeed lead to proof that the patient will derive direct clinical benefit from a new treatment. It is also important that statisticians should be called in at the very early stage of designing trials. A close collaboration between clinical investigators and trial statisticians from the very beginning is essential in order to come up with clinically relevant success.
Cytel: Will you be discussing the relationship between biomarkers and the clinicians’ perspectives? What are the key aspects of this relationship?
MF: Biomarkers have gained increasing importance because our knowledge about the biology of cancer has increased tremendously, thanks to the introduction of molecular biology and tumour immunology in clinical oncology. The progress which we have witnessed over the last few decades is overwhelming.
We need to employ the appropriate diagnostic procedures to find these biomarkers in a tumour biopsy taken from a patient. This diagnostic precision is not trivial. The diagnostic relevance of the marker needs be clear from the trials run to establish its validity. The diagnostic techniques should be precise, sensitive, specific, and their results must be reproducible. Many tests in molecular oncology today use techniques such as next generation sequencing or PCR detection of specific mutations in tumours or in the germline. I think the sensitivity, specificity and reproducibility of these molecular tests are nowadays very good. However, we still sometimes rely on immuno-histochemistry, which gives us a full range between negative results, very strongly positive results and anything in between, and is thus more difficult. Finally, sampling errors are important, as tumours are often biologically heterogeneous.
When we look at treatment options, we should select the patients based on their respective markers present in their tumour, and there must be a close link between the marker present and the treatment effect (this is called a predictive biomarker). A good example is non-small cell lung cancer where the presence of the biomarker (specific gene mutations) shows if the patient is going to benefit from a specific pill (i.e. an oral tyrosine kinase inhibitor) or not. However, not all cancer types show such clear links between a marker and a targeted treatment, and thus, we still have a long way to go in order to link biomarkers to specific cancer treatment, on an individual or personalized medicine basis.
Cytel: What are the shortcomings of current measurements of cost-benefit?
MF: Firstly, if you are studying new drugs in Phase 2 trials and Phase 3 trials, and compare them with established standards, it is quite difficult to come up with an appreciation of true cost benefits. When a drug is still experimental, we do not yet have a price for it. The next problem is that health care systems and reimbursement of medical costs are very different in various countries. This has a direct impact on how patients or their health insurances are billed and what the ”true” treatment costs are.
When we talk about costs, we need to split them into direct and indirect costs. Direct costs comprise the price of the drugs, diagnostic procedures (for example getting a biopsy), imaging, surgery etc. Indirect costs are linked e.g. to the patient being unable to work or not being as productive at work than if he/she were healthy. Measurement of indirect costs is particularly tricky, but assessment of direct costs is also often incomplete.
Register today for Cytel’s complimentary webinar “A Clinician’s Perspective on Cancer Drugs Development” by Professor Martin fey. The presentation will focus on medical oncology and will cover important topics on cancer drugs development, using different examples of oncology trials.
About Professor Martin Fey
Professor Martin Fey is an experienced board-certified Medical Oncologist who provides expertise in the design and conduct of clinical cancer trials. He was a Professor and Head of the University Department of Medical Oncology at the University of Bern and the University Hospital of Bern, Inselspital, from 1994 until 2017. Since then he continues to work as a Senior Consultant in Medical Oncology at the same institution.