Traditionally, oncologists believed that solid-cancer tumors were largely homogeneous – that their molecular makeup was the same throughout the body—and treated their patients accordingly.
We now know that tumors are far more complex. Not only are they densely heterogeneous1, and vary significantly both within and between primary and metastatic sites, they also change over time in response to treatment2.
A tumor biopsy captures, by necessity, only a small piece of a tumor to be used for histological staining or genomic sequencing. As a result, important disease features may not be present in the specific section sampled by the biopsy needle. This intra-tumor spatial heterogeneity can present real challenges to patients with advanced cancer and to the doctors who are building their treatment plans. If a biopsy needle misses the section of the tumor containing an important drug target, no genomic testing performed on that tissue sample, however robust, will be able to find it.
A landmark paper published in the New England Journal of Medicine (NEJM) in 2012 demonstrated the extent of this complexity. Researchers performed DNA sequencing on nine different biopsies taken from a single tumor and found that each contained an average of 70 somatic genomic alterations – but just 34 percent of those alterations were present in every region.
The diagnostic challenge presented by heterogeneity is amplified when tumors begin to spread and metastasize. Metastatic lesions are often in sites that are difficult or impossible to biopsy such as in the bones or brain, or are present in such numbers as to preclude thorough sampling. It’s usually impractical – and very painful/dangerous for the patient – to perform biopsies on multiple sites. However, these lesions can contain different cellular characteristics, including potentially druggable genomic targets that differ from the primary tumor site. Indeed, in the NEJM study above, the low inter-biopsy concordance in the primary tumor was even lower in the metastases. And these genomic differences matter. In another, more recent study, 53% of patients harbored clinically informative alterations in one or more metastases that were not present in the primary tumor.
In contrast to tissue-based tests, which necessarily sample only a small portion of a single tumor, liquid biopsies sample cell-free DNA shed from tumor cells throughout the body, thus providing the doctor a global summary of the genomic alterations present in the patient. Importantly, the fastest-growing cells, which shed the most cell-free DNA, are the most clinically relevant because metastases and fast-growing tumors are more likely to pose a threat to the patient’s life than a slow-growing primary tumor3.
The genomics of a tumor may vary not only in space, but also over time.
For instance, in non-small cell lung cancer about 15-40 percent of patients’ tumors harbor an alteration in the EGFR gene. Tumors with these alterations can be effectively treated with drugs like erlotinib, gefitinib, and afatinib; however, eventually the tumors evolve, and typically acquire new alterations that subvert the efficacy of the original tyrosine kinase inhibitor. In this specific example, the most common of these acquired resistance alterations is EGFR T790M, in turn, can be targeted by new therapies designed to specifically inhibit tumors with this resistance mutation.
And in multiple studies, researchers have shown that when there is a long delay between the time a tissue sample is acquired and when the liquid biopsy is performed, the tests yield different results. But when the blood and sample are collected concurrently, or when the alterations are so-called truncal driver mutations that are likely to be present before and after treatment, concordance for important oncogenes approaches 85% or higher4,5.
For instance, researchers at the University of Pennsylvania Perelman School of Medicine looked at 50 advanced lung cancer patients for which Guardant360 and tissue sequencing results were available. They found that when the tests were performed within two weeks of each other, concordance for the important EGFR oncogene was 100 percent, but when the blood was tested more than 6 months after tissue testing concordance fell to 60%. The authors concluded that the blood test is a real-time test that can capture newly evolved tumor mutations not present in older tissue biopsies6.
In another example, researchers from the University of California, Davis showed that the somatic genomic landscape detected by Guardant360 in more than 15,000 patients was highly similar to the landscape described in other population-scale studies7, including The Cancer Genome Atlas (TCGA). The exception to this was with resistance alterations, which were largely absent from the treatment naïve TCGA population, but were often found in the Guardant patient cohort of advanced patients.
Tissue acquired at the time of a patient’s diagnosis is often outdated and many times can’t help a doctor determine if their patient is a candidate for a new drug. And a repeat invasive biopsy may be expensive, painful, and dangerous, which is why many doctors order liquid biopsies at progression.8
Temporal heterogeneity can also have important implications for medical testing even in the absence of tumor evolution. For example, anti-cancer treatments including chemotherapy, radiotherapy, and targeted therapies kill cancer cells, which can decrease the amount of tumor in liquid and tissue biopsies alike. Indeed, just as aspirin may temporarily reduce your fever when you have the flu, producing “discordant” results on your thermometer if you take your temperature before and during treatment, liquid and/or tissue biopsies taken after or during cancer therapy may detect fewer variants than tests performed prior to therapy because fewer cancer cells remain.
Heterogeneity and the future of cancer care
Cancer is a complex disease. A single patient’s cancer can harbor dozens of different genomic alterations spread unevenly throughout their body, and these alterations can evolve in response to treatment. As we increase our collective understanding of tumor biology, patients will benefit from more accurate diagnostics that can follow this evolution and guide dynamic precision treatment plans that evolve as the tumor does throughout the course of a patient’s care.
1. Gerlinger et al., N Engl J Med 2012
2. De Mattos-Arruda et al., Annals of Oncology 2014
3. Brastianos, et al., Cancer Discov. 2015
4. Lanman et al. PLoS One 2015 Oct 16
5. Kim et al. Oncotarget. 2015 Nov 24
6. Thompson et al., Clin Cancer Res Dec 1
7. Zill et al. J Clin Oncol 34, 2016
8. T Lokhandwala et al. Clin Lung Cancer. 2016 Jul 21