Thread by @drbobphillips

It’s still dark. This is no way to start a conference #SIOP2019

At least they gave provided the worlds smallest coffee as compensation #SIOP2019

I’m here to learn about the use and benefits of “liquid biopsies” (biopsy coming from BIOS - life and OPSY - sight)

This makes the description of chunks of solid tissue as a “Biteopsy” even better I think

The main focus here is looking at tiny bits of DNA or particularly proteins that are floating in the blood. Those tiny chunks are the one which, after the cancer cell has died, has unwrapped from its holding stone^ and started to be cut up for waste removal.

^not a real stone

The use can be thought of as looking for response to treatment, confirmation of absence of detectable disease even at really low levels, and surveillance for relapse. (“Cell free DNA”)

Something like this gets done in leukaemia already looking for MRD

There are some technical aspects which need to be made sure are right. A blood sample needs dealing with quickly to make sure the liquidy (plasma) bit of blood is separated from the cell parts. This makes sure the cDNA you spot hasn’t leaked from the cells dying in the blood tube

You could instead buy extremely expensive tubes and not spin the blood sample.

Don’t.

Now. The amount of cfDNA varies between tumour types. Also, those with metastatic disease have more than those without. This isn’t terribly surprising. BUT it is useful as it keeps with the understanding

How to find the stuff though? Well there are “can I spot this exact gene signature” ones - which are fast and widely available - but they require you to know what signature you’re looking for (“FaceID for tumours”)

They work for neuroblastoma with MYCN and ALK #SIOP2019

Other versions draw all the cfDNA and look at it - it will be a mixture of normal DNA from naturally dying cells and tumour DNA. It then needs to be sorted to look for genetic normality and abnormalities in the little snippets. This then gives a hint, or confirmation of diagnosis

It’s not as sensitive (ie it can’t pick up stuff at as low a level of disease as the FaceID version). But you don’t quite need to know what you’re looking for

However. The technique still finds gene abnormalities which may be found in different types of tumours, so doesn’t always tell you exactly what you’ve got.

It might tell you, using a geographical analogy, your tumour is from Leeds but not if it’s from Seacroft or Scarcroft

An area of extreme rareness but very high consequence would be in some retinoblastoma cases. Very very very occasionally there is a child whose eye has disease which is probably RB, but may be a non cancerous alternative. RB treatment would be surgical removal of the eye

To know it was RB is clearly very important.

Using this approach of seeking genetic signature before treatment would be v helpful. It can sort of be done in blood, and maybe with the fluid within the eye. It’s not quite perfect yet but looking promising @ChectUK

There are other questions which can be asked too ... more sciencey ones which are a step or two away from use ... like “how do the tumour mutations change over time in neuroblastoma”

Knowing this without a way of picking a different treatment for the changes we are seeing doesn’t mean we can change what’s going to happen though. What it does is confirm some theories we had and allow us to keep going on the paths of therapeutic development.

Just to put the quantities of cell free DNA into context between time it types and leukaemia ... this graph is log scale on the up-axis ... leukaemia (Hugh purple blob) has 100 to 1000 fold MORE to access to test and follow and examine