All laboratories, all pathology laboratories, have to meet certain standards. As part of those standards, what you want to achieve is that when a test is done in one laboratory you get a similar result in another laboratory. So you want to ensure that any test results that come out of your laboratory are in sync with everyone else’s. The way we do that is by a system of internal and external quality assurance to ensure, first of all, that the test is working properly in your own laboratory and, secondly, that it matches the results that you would obtain in other laboratories.
How do the expensive and labour-intensive formal harmonisation exercises in the service of clinical trials fulfil your expectations for the degree of inter-laboratory concordance?
First of all, if a laboratory test is part of a clinical trial, if the trial depends on that test, for instance as a companion diagnostic, then it’s absolutely essential that it is tested within multiple laboratories to make sure that it is robust. So you’re looking for intra- and inter-laboratory variation and checking that actually it’s within the limits you’ve set. You would set those for each individual test separately.
The best way to do that is using an external quality assurance scheme and very often the external quality assurance schemes that are set up within clinical trials are then used formally between clinical laboratories afterwards as a way of ensuring that the tests, when they’re rolled out, are actually done properly. So although it seems expensive at the time when you start doing it, it actually has some knock-on effects which are quite good afterwards.
How should clinical trials be ideally conducted?
You want to conduct research, clinical research or any form of cancer research for that matter, to the same standards that you would in a clinical laboratory, certainly not lower. So at the moment clinical laboratories need or are required in many countries to meet ISO 15189 and that is really setting the standard that clinical laboratories should meet.
One of the things we have done within IARC is to set up a new collaboration to try to improve the evidence levels that are coming through into the WHO classification of tumours for diagnosis. As part of that, The International Collaboration for Cancer Classification and Research has been set up. It’s a bit of a mouthful so we call it IC3R. The IC3R initiative includes discussing with the International Standards Organisation, ISO, whether we should have an international standard for research laboratories. That would be a lighter system, probably, than ISO 15189 but it would achieve, hopefully, much the same things.
So, for instance, when samples were sent to the laboratory they would have to be identified properly. You would need training records for your staff. You would need to ensure that equipment was properly maintained and that maintenance contracts were in place. All of the major things that happen within an ISO 15189 certified laboratory would need to happen within research laboratories as well. That, I think, will lead to a big improvement in the quality of research data that we see.
What important points were discussed regarding this topic at this year’s IBCD 2020?
We discussed quite a lot of this, actually. There was limited opportunity on Zoom, obviously, for trying to get a full discussion going but we did have quite a good discussion of the different aspects of harmonisation. There was also a recognition that certain types of test had had more problems than others. Sometimes this all works very, very well and you find that a particular companion diagnostic is in good shape from a fairly early stage in its development and at other times it is much, much more difficult, either because the biology is difficult or because the test itself is difficult.
I think we all recognise that probably for breast cancer now we’re in a fairly good shape with four tests, or three perhaps, but certainly oestrogen receptor, progesterone receptor and HER2 are reasonably well known and there are very robust tests that are used to guide treatment. Whereas Ki67 is perhaps less robust and is more difficult because the biology is more difficult. So is the pre-analytical side of that.
Then we come to things like PD-L1 where there are clearly much bigger problems and that is down to the fact that there are multiple antibody clones being used and there are multiple drugs being used and the comparison between them, at least initially, was not done. That has now largely been done and I think things are improving but, again, the biology is difficult and it is evident that PD-L1 expression alone is probably not sufficient to predict who will respond and who will not to those agents.
What does the future look like for inter-lab concordance in terms of new clinical trials?
I’d like to think that we’re learning as we go; I think everyone would like to think that. Certainly this is an area that is going to continue to be important for a long time to come. The major issue now is how we amalgamate new companion diagnostics or complementary diagnostics alongside others that have to be done on that same patient. So very often it’s a question of how do you use the limited amount of tissue or material that you’ve got and how do you ensure that you’re getting as much as you can get from that small amount of material.
The way in which that is going to go in the future is that we use multiplex tests, tests that can look at multiple different parameters at once and give you the answers that are still eligible for use in conjunction with other data to guide the patient’s treatment. So if you’re going to guide treatment using multiple tests then you need to ensure that you can do that on a single platform ideally and with a limited amount of tissue.
Is there anything you would like to add?
No, I’d just like to thank the organisers for putting on the conference, despite the problems posed by having to do it with Zoom and doing it electronically. I thought it worked extremely well and I’m very grateful for the opportunity to be able to talk to you.