Personalised Medicine And Predicting Chemotherapy Toxicity

Karol Sikora, PhD, FRCR, FRCP, FFPM


March 06, 2019

Hello. This is Professor Karol Sikora talking about personalised medicine and predicting chemotherapy toxicity.

Last month in Paris, four families of patients who died following the administration of the fluoropyrimidine - 5 Fluorouracil (5-FU) took legal action against the French health authorities. Their lawyers have claimed through the Office of Public Prosecutor in Paris that there has been 'involuntary homicide, unintentional injury and endangering the lives of others'. Mr Jule Parade, the plaintiffs' lawyer is claiming that the medical establishment – that's us - knew that looking for certain polymorphisms of the dihydropyrimidine dehydrogenase (DPD) gene could predict toxicity, and asked why was DPD testing not systematically proposed to all patients prior to receiving the drug.

I strongly believe that a court of law is no place for clinical judgement and clinical policy making. What would your view be on this? Let's explore the whole field in a bit more depth.

Precision Medicine

We all agree that personalised or precision medicine is the direction of travel in cancer care. So, let's look at the facts behind this case. Risk banding for the population for personalised screening programmes, making decisions on the need for adjuvant chemotherapy, hormonal therapy, or choosing the correct combination of drugs for patients with metastatic disease is clearly going to rely heavily on the genomic revolution. Whole genome sequencing can now be done cheaply, and increasingly, as the data banks mount up, we'll be prescribing drugs based on logic, rather than past history, clinical trials, and empiricism.

From oestrogen receptor status in breast cancer for hormonal therapy, to ALK mutations in lung cancer for crizotinib, the big hope is that by giving the right drug to the right patient, we will get greater clinical benefit and also save money in the long run.

Molecular signatures of response are high profile. Everyone talks about it. A much less prominent area of research is the use of pharmacogenomics to predict anti-cancer drug toxicity. We know our drugs cause profound side effects and we have developed ways to reduce them. We also have a whole bag of secondary medication we can give to reduce the toxicity – bone marrow growth factors, anti-diarrhoeals, anti-mucositis, analgesics for headaches and so on. But we've not actually paid that much attention to predictive testing to try and find out what dose of an anti-cancer agent to give to a patient to reduce toxicity. And, of course if we do that, we have to address the fact that if we reduce the dose, or prolong the interval, because we suspect toxicity, will we get the same tumour response?

5-Fluorouracil was first used widely for colon cancer in 1962: over 50 years ago. Capecitabine [(Xeloda, Genentech) and others], its oral pro-drug, was licensed in 1998: 20 years ago. So these two fluoropyrimidines are amongst the most widely used chemotherapy agents, and are both on the World Health Organisation's essential drug list. They must be the commonest drugs I've ever given in my career as an oncologist. I'm sure many of you will feel the same. We use them first line for breast, colon, head and neck – indeed for all squamous carcinomas. And we use them in all sorts of situations - neoadjuvant, adjuvant, second line, with or without radiotherapy.

No Routine UK Testing

So, has the French prosecutor actually got a point? Should we be doing DPD testing, and if not, why not?

In the UK, outside of a clinical research setting, DPD genomics is not routinely carried out. The cost of the relatively straightforward screen is about £50. It can be done in a laboratory at St Thomas' Hospital, London, within the NHS but privately, for somewhat more than that, a nice generous mark up. But should we be recommending it for everybody?

I've had one memorable patient, actually an eminent professor of medicine, who nearly died after his first shot of adjuvant 5-FU for a relatively low-risk colon cancer in the adjuvant setting. He was fuming with me. He got all the literature out and said why didn’t you do the DPD screen? I said we don’t normally do it.

There is no consensus around Europe about this.

In the brave new world of personalised medicine, why not?

To me the answer is simple: why bother? DPD polymorphisms are not the only cause of increased sensitivity to fluoropyrimidines. There are many other factors involved that we don't fully understand.

Indeed, even with the genomics, not all patients with an abnormal polymorphism will actually develop increased toxicity.

So, in my experience less than 10% of patients get significant problems and in only around 1% do we have to stop. We may have to reduce the dose, increase the interval, change the way in which we give it. But we don’t really have to stop that often.

There's also an antidote available – little used in my experience - uridine triacetate – which floods cells with uridine and so prevents the toxicity. The trade name is Vistagard (Wellstat Therapeutics). But, as I say, it's rarely used in the UK.

French Guidelines

Against my rather nihilistic views it's interesting to consider the three recommendations of the French Group on Clinical Pharmacology in Oncology (GPCO). They've been published and are as follows:

  1. To screen for DPD deficiency by genotyping in all patients prior to fluoropyrimidine administration.

  2. To phenotype by measuring plasma uracil, and possibly the ratio of dihydrouracil/uracil. The ratio reflects the metabolism of fluoropyrimidine after the administration of the drug.

  3. To reduce the 5-FU dose, and presumably the capecitabine dose, in the first cycle, and to increase gradually through treatment checking for tolerance.

To me, this all sounds like jobs for the boys in the lab. France, like the UK, has over 100,000 people a year starting on either 5-FU [or a] capecitabine regimen.

I suspect our French oncology colleagues are also less keen to get embroiled in this sort of recommendation. It sounds like an awful lot of work for very little actual gain.

Capecitabine is given orally and is easy to stop if there's toxicity. 5-FU, especially with a 5-day loading dose and then that's it, they're loaded up, it's not possible to stop, so we don’t have a problem we're able to solve there.

Overtaken By Events

So, where's it all going? I guess all this is going to go into history as whole genome sequencing is coming. It currently costs around £850 a patient. This is going to drop to less than £100 over the next 5 years. DPD will just be one of many genes analysed. Minor polymorphisms will be identified. It will give us a much more holistic picture of drug toxicity, not just for this class of molecules, but a whole range of drugs.

More importantly it will spawn a huge data bank which will give us information that we can only dream about today.

The choice of anticancer drug, the optimal dose, the timing, will all be personalised - not by the boys in the lab - but by the computer, learning, with artificial intelligence and deep learning, from precedents over the past and moving forward into the future.  

I'm afraid in the new world there'll be no room for lawyers.

This is Karol Sikora, talking about predictive testing for fluoropyrimidine administration. I'd really like to hear your views. What do you think? Should we be screening everybody for DPD before we give them 5-FP or capecitabine? Thank you for listening


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.