As we prepare to step out into 2018 it is time to reflect on the past year and consider what the New Year might bring.
For me the highlight of 2017 which summed up so much was the statement by Dr Janet Woodcock, the FDA’s Director of the Center for Drug Evaluation and Research that the clinical trials system is “broken”. She said this to an academic conference on ‘real world evidence’ in September. She also looked ahead and said that clinical research networks and ‘master protocols’ must be the shape of the coming future. She noted that there has been “very little historical use of real world experience in drug regulatory decisions about effectiveness.”
She is not alone in voicing the comment ‘broken’ about cancer research. The rare cancer community has been saying it for some years and to their credit the regulators have been listening. The recent approval of oloratumab for advanced sarcoma was very welcome. It came on evidence from a randomised Phase 2 study but it is a 2-year interim approval while a Phase 3 RCT is concluded. ‘Real world evidence’ (RWE) will also be gathered.
The challenge of the RWE agenda is gathering data prospectively, facing down the long-standing belief it can only be retrospectively gathered. Those same beliefs support the double-blind randomised controlled trial (RCT) with an endpoint of Overall Survival (OS), as the ‘gold standard’ in cancer research. However, the ways through which we gather evidence are changing. The biggest change is that targeted therapies are having their targets proven in pre-clinical studies using cell-lines and in Phase 1 dose/safety studies. What goes into Phase 2/3 is a drug with confidence that there will be a high level of targeted responses. This can make selection crucial and randomisation ethically difficult. It has meant studies with cross-over from placebo, randomisation between different doses or methods of delivery, and choice of questionably appropriate comparator arm treatments. Because time is needed to reach the OS median and because successive lines of different treatment may be given to patients, so-called surrogate endpoints such as PFS (Progression Free Survival) are used. All these patches on the traditional RCT are making proper analysis and effective regulation difficult. It has already been shown that many approved cancer drugs drop out of standard therapeutic use very quickly because the promise of the study is not realised in the real world.
There is growing agreement that ‘real world evidence’ is an issue which must be addressed and this will mean new methodologies. That is probably the single most pressing agenda in cancer research. We have moved to the point where traditional trials cannot give us the evidence of value which we need without good RWE data to support it. In rarer cancers getting all the evidence in a timely way is also important. We cannot define the value of a treatment without RWE and we are in increasing danger of reaching the point where drug costs will be politically and socially unsustainable unless we can prove value. A research project at Oxford University has demonstrated that arguments about the ‘price of a life’ etc cut little ice with the general public so politicians may find they are on strong ground for limiting access to high cost drugs for cancer.
One of the best sources of RWE data is ‘quality of life’ studies. This is the patient-centred focus on a new treatment which complements the clinical/medical data. There are proven methodologies which give good data. Too often the QoL data gathered in a research study of a new treatment is not analysed or published. It seems to be used if it supports conclusions the researchers want to see and gets hidden away if it doesn’t. This must change. But the QoL tools we have are each unique, their data cannot be compared easily and cannot be aggregated to provide a bigger picture. This must change too. There is work to do.
So what about Dr Janet Woodcock’s comments about networks and master protocols. Formalised networks maximise patient numbers when study inclusion has to be selective, as with targeted therapies. Networks are also the best way of gaining experience in using statistical tools such as Bayesian probability and in developing the skills of investigators in new approaches to clinical research. In rare cancers networks already exist – SARC in the USA, EORTC in Europe for example – but in the more common cancers they are less formalised internationally although in the UK we have the structures of NCRI and NIHR. Master protocols are protocols which cover a range of studies eg one drug in several diseases or targeted indications, or one disease/indication with several drugs. Such protocols are increasingly being employed. Academic networks are also building protocols which allow multiple drugs with variation in the sequence these can be taken following relapse. Such studies need the independence of academic investigators plus the provision of drugs by pharma. They can take time to negotiate but their value in eliminating poor performers is huge.
In non-drug studies the blinded RCT will still dominate the evidence base for clinical decision-making. It is the surest way of reducing, even eliminating, bias although as statistical techniques and practice evolves we will find probability statistics playing a greater part. However if we grow reliance on QoL to inform the definition of value and influence regulatory practice with new drugs we can expect the same standards to apply to non-drug studies. A growth in QoL will create massive volumes of data so I would predict that this is the next background step necessary, a ‘big data’ project bringing together QoL data from across the research spectrum, providing analyses which can inform change in clinical practice and which will give patients better quality information.
We seem to be on the cusp of important change in clinical cancer research. What Dr Janet Woodcock will say in September 2018 may not be massively different from 2017 but I hope she will be identifying the beginnings of change which brings new methodologies in clinical research, together with robust QoL tools which provide reliable (and comparable) data on real world experience, and which can all be brought together to allow us to define value in a practical way which society can accept and pay for.