Liquid biopsies, virtually unknown even a year or two ago, are becoming common tools in precision diagnostics for cancer. Here, I will try to explain some of the more important differences between liquid and “traditional” tumor biopsies.
Biopsies of solid tumors (e.g., lung, breast, or brain tumors) involve surgically removing a small part of a tumor and sending it to pathology lab. In the last few years, doctors have also started to send some tumor samples to special service labs that analyze tumor DNA for the presence of cancer-related mutations.
By definition, regular biopsies can be intrusive and are sometimes associated with side effects, such as bleeding or infection. However, they provide some really essential information; i.e., the histology and grade of the tumor and other tumor characteristics necessary to determine the best choice of treatment. For lung cancer, for example, a biopsy determines the type of tumor—adenocarcinoma, squamous cancer, small-cell lung cancer, or another, less common type. For breast cancer, a routine test will determine if the tumor expresses estrogen, progesterone receptors, and a protein called HER2. These tests are critically important in guiding treatment choices. If mutational analysis of cancer-related genes is also performed (which doesn’t always happen, unfortunately), it may guide treatment with targeted drugs.
Liquid biopsies are so-named because they are performed with blood samples, thus obviating the need for invasive biopsies. They are based on the discovery that metastatic tumors “shed” cells into the bloodstream, where most of these cells—not being accustomed to survive long in blood—die and release their DNA (known as cell-free DNA, or cfDNA). This cfDNA can be analyzed for the presence of mutations in cancer-causing genes, and the amount of cfDNA with these mutations can be monitored over time with multiple blood tests.
These days, liquid biopsies are not designed to determine the precise histological type of tumors; i.e., the type of tissue from which a tumor originated. This could be potentially developed in the future, but for now, the available tests sequence a limited number of genes in cfDNA. This defines one of the current limitations of liquid biopsies.
If (hopefully) a solid biopsy is sent for mutational analysis, this will reveal a set of mutations that are found in the particular sampled part of the tumor only. This is of key significance because tumors are heterogeneous. Tumors, or at least many tumors, continuously evolve (develop new mutations), either because of the inherent genomic instability, or because they are under pressure to provide more space and nutrients for themselves or are trying to survive treatment with drugs or radiation. One part of a tumor may contain a set of mutations that only partially overlaps with mutations found in a different part of that same tumor. When it comes to metastatic tumors, they may be even more divergent from the primary tumor that seeded them. This obviously implies that mutational analysis of solid tumors is imperfect, presenting a molecular snapshot of a small tumor sample rather than a complete representation of all mutations found in all tumors in one patient.
Is the liquid biopsy better? Since it analyzes cfDNA from cells that are shed into the bloodstream by all tumors present in a patient with metastatic cancer, it is expected to present a complete picture of all mutations found in all tumors. However, is it correct to assume that a primary tumor in the breast, for example, releases as many cells into the bloodstream as a lung metastasis present in the same patient? Most likely not. This means that not all mutations can be expected to be represented by cfDNA equally well. Because of this, mutations present only in one metastatic tumor out of several in a patient may not be picked up by liquid biopsies.
So the major difference between the two types of biopsies is clear: a solid biopsy defines which mutations are present in one particular area of one tumor only. A liquid biopsy, while it promises to find all mutations in all tumors in a patient, may not pick up mutations from tumors that do not shed enough cells into bloodstream, or mutations that are found only in a small area of a tumor.
Which test is best?
Recently, a paper published in the Journal of the American Medical Association (JAMA) revealed just how large the differences really are between the two most popular commercially available cancer mutation tests. The FoundationOne (F1, Foundation Medicine) solid biopsy test provides analysis of well over 300 genes for mutations and other types of aberrations. Guardant360 (G360, Guardant Health) is a liquid biopsy that analyzes 73 genes in cfDNA. The study in JAMA compared the results of the two tests performed for just nine patients, and found that there was little concordance between F1 and G360. Of course, the authors only looked at genes that were analyzed in both tests; i.e., 70 of the genes analyzed in the G360 test. Only 22% of mutations in these 70 genes were concordant between both tests. As a consequence, recommendation for specific targeted drugs was concordant in just 25% of the cases.
This story garnered a lot of attention and was featured in The Atlantic and by NPR, where reporters noted the existence of other studies that point out large differences in the results of solid versus liquid biopsies.
On one hand, Guardant Health presented at ASCO 2016 (a major annual oncology conference), reporting that “the overall accuracy of ctDNA sequencing in comparison with matched tissue tests was 87% (336/386). The accuracy increased to 98% when blood and tumor were collected less than six months apart.” (The nine patients in the JAMA publication had both tests performed no more than 2.5 months apart.)
However, in one study the concordance of test results in 28 patients was only 17%. In another study specifically in pancreatic cancer, liquid biopsies did not detect any mutations in a quarter of the patients. Moreover, they missed the most frequent mutation found in pancreatic cancer (in the KRAS gene) in 29% of blood samples, whereas solid tumor sequencing detected it in 87% of biopsies. The authors concluded that liquid biopsies are not yet adequate in pancreatic cancer. This, in my mind, is a valid conclusion in this particular type of cancer. It would be most useful to conduct a large-scale comparison between all the results of the two types of biopsies available for patients who had both. Unfortunately, these tests are very expensive, and few patients have both performed. It is does not help that the two major providers of the tests are involved in litigation over the new liquid biopsy test (ACT) from Foundation Medicine.
In a better world, the two companies would have conducted comparison studies. It really makes sense that tests produce comparable results only in certain types of cancer. These types of cancer should be defined, and the choice of test should be based on the results of large comparison studies. Well, this is not a better world yet. However, Foundation Medicine plans to compare the results of its F1 and ACT tests in patients who have received both.
The future of liquid biopsies
Meanwhile, Guardant Health intends to extend its gene panel to 500 genes. Moreover, it plans to develop a method for early detection of cancer; i.e., expand the use of its platform to what is basically screening of healthy people for signs of cancer. In this endeavor, Guardant Health faces steep competition from the company largely responsible for developing next-generation sequencing (rapid, high-volume DNA sequencing) as a clinical tool—Illumina. Illumina has recently spun off a start-up, Grail, which also aims to perform deep sequencing analysis of cfDNA for early detection of cancer (cancer screening, essentially).
This is all very interesting, but before undertaking the development of liquid biopsies as a cancer screening tool, it would be really great if all cancer patients, or at least those with advanced or metastatic cancers, could get molecular testing (either solid or liquid biopsy) that is affordable and covered by insurance plans.