Tumour fraction in circulating free DNA as a biomarker of disease dynamics in metastatic prostate cancer

Bookmark and Share
Published: 9 Feb 2018
Views: 1764
Dr Atish Choudhury - Dana-Farber Cancer Institute, Boston, USA

Dr Choudhury speaks with ecancer at the 2018 ASCO Genitourinary Cancers Symposium about furthering our understanding of the genomics and genetics in metastatic disease from blood samples. 

Sequencing cfDNA (circulating free DNA) deduces where the DNA originates from (tumour or not).

This quantification can be used for a lot of different applications, and correlated with clinical features.

Examples of these features are PSA, anaemia and alkaline phosphatase. 

These are all increased with and increased level of tumour derived cfDNA.

The exciting part of Dr Choudhury's research is that it may be possible to correlate response to treatment with levels of tumour derived cfDNA in the blood.

This may provide a very cheap, and early biomarker to deduce whether treatment is working. 

In the future, a more homogenous study would be beneficial, and with larger cohorts.

Although there are still areas of this research which need deeper examination, it could be a very useful biomarker in the early stages of treatment. 

In all metastatic disease types it's very, very hard to actually understand the genomics and the genetics of the cancer because metastatic biopsies are uncommonly performed in patients with metastatic disease. So we're very interested in understanding can we learn something about the cancer from what we would call blood biopsies, just can we learn about the genetics of the cancer from actually a blood sample?

So I've been working for many years with a team at the Broad Institute to try to understand the features of genetics and biology that we can understand from the blood. Certainly one of the most important pieces from that is DNA. So we've been doing sequencing from circulating free DNA from patients with metastatic cancer for several years at this point and there are many different pieces that we've been able to put together over this period of time. One of the most important things that we've been able to find and innovate in this field that has been very popular all over the world is to actually quantify how much of the DNA in the blood actually originates from the cancer rather than non-cancerous tissues. We've used that quantification for a lot of different other downstream applications to really understand which patients' DNA from their blood is actually amenable to other downstream applications, more rigorous sequencing called whole exome sequencing. But it turns out in my studies that I presented at the poster yesterday was about trying to correlate the tumour fraction that we detect in the blood with clinical features - so which patients actually have more DNA derived from the tumour or not.

Basically once we were able to quantify how much DNA was derived from tumour versus non-cancerous tissues, and this is based on an algorithm that was derived by bioinformaticians at the Broad Institute which we call ichorCNA and it's based on what we call ultra-low pass whole genome sequencing where you sequence across the entire genome but at a very low depth of sequencing called 0.1x to have a very inexpensive way to determine this, about $150 per patient. So once we derived those numbers across 722 plasma samples across several patients we identified 669 of those samples that were amenable to analysis from about 140 patients. We were basically able to correlate the amount of DNA that we saw that was derived from the tumour with different clinical features.

So it turns out that we know some clinical features that predict survival in patients with prostate cancer and these include their level of anaemia, so the haemoglobin, the level of a parameter called alkaline phosphatase in their blood, and the alkaline phosphatase is derived from liver and bone, and their PSA level and the location of their metastases. It turns out that patients with more metastases obviously have more DNA that's derived from tumour in their circulation. Patients who are more anaemic seem to have more tumour DNA in their circulation. People who have a higher alkaline phosphatase also have more tumour DNA in their circulation. So that proves the obvious but it's good to validate that in our set.
The more interesting part is how this changes with time and with therapy. There are a few interesting observations that we made. One is that when a patient starts an effective therapy, and an effective therapy means their PSA went down by at least 30% over the course of the next six weeks or so, their tumour fraction also went down. So in prostate cancer, for example, we have a really good biomarker, which is PSA, that tells you if a treatment is working but a lot of other cancers don't have a biomarker that you can get from the blood. So if we could prove that a decrease in tumour fraction after starting treatment actually correlates with a treatment response then you could use that to understand early on if a treatment that you're starting is actually working or not. So that's one of the important findings that we made.

After that what happens is actually very variable. So patients who have prostate cancer and are on hormonal treatments and they are on these hormonal treatments for a long time their PSA might be creeping upwards but you don't see any changes on scans. But what we see is that the tumour content in the blood goes up and down and up and down over this course of time and we really don't have a great explanation of why that's the case. We think that there might be bursts of activity, whether that's proliferation or metabolism or inflammation we just don't know, that might correlate with these bursts of tumour DNA that we are detecting in the circulation. So while a decrease in tumour fraction might be a good marker of whether a treatment is working at the beginning, a rise in tumour fraction while you're on therapy doesn't mean that it has necessarily stopped working.

Is there a lot more work to be done?

Absolutely, yes, across larger cohorts and across a more homogeneous cohort because these were patients who were treated with a lot of different types of treatment. So what we're looking at now is for patients who participated in clinical trials, and we have banked samples from before they started and for different time points on the trial, do we actually see this as consistent results across many patients participating in clinical trials of different agents.

What are the potential implications of this?

What we're hoping to understand, certainly, is is this a really good biomarker of initial response. Because once you can prove that that's the case then it will help clinically in a lot of different ways. One is that there's this phenomenon that we see in prostate cancer and other cancer types where the first set of imaging tests that you get after you start a treatment doesn't always represent whether the cancer is responding or not. These are phenomena that we call flare. So on a bone scan, for example, you can see more spots when you start an effective treatment and we don't know right now whether that represents a progression of the cancer or just an effect of the treatment. But if you can see that it correlates with an increase or decrease in the tumour burden then that could be very helpful clinically to know earlier on if your treatment is working or not.

The second piece, and this is potentially more relevant in other types of cancer that respond to immune therapies, is that immune therapies can sometimes be accompanied by something called pseudo-progression where you get growth in different spots that you see on the scans. It turns out that that growth is very temporary and related to an immune infiltration rather than progression of the cancer itself. So it might help understand, again in other cancer types, whether a patient is responding to therapy or not.

If you know, for example, that a new treatment is working but only modestly well based on this tumour fraction then it might make you decide earlier to actually intensify the treatment, add a second agent to make it work better or switch it if the therapy is causing side effects and it's just not working.