Nearly 2,000 years ago, the Roman physician Aulus Cornelius Celsus was vexed by the resilience of cancer. Celsus had been studying the treatment of various tumors when he noted that, for some patients, moments of therapeutic success would somehow give way to more aggressive tumors. The excised carcinomas, he resigned, “have returned and caused death.” It’s been more than two millennia since Celsus’ studies, and while physicians are far better equipped to treat cancer, the prospect of tumor recurrence remains a significant threat to patient survival.
With many overlapping factors, the recurrence rates among cancers can vary widely such that only about 8% of patients with thyroid cancer will develop a recurrent tumor. In contrast, nearly 100% of patients with glioblastoma will. There is still much that we don’t know about what drives tumor recurrence, but it is clear that the evolution of evasive or resistant phenotypes can lead to subpopulations of tumor cells that endure treatment. When these cells are few in number, they may persist below detection thresholds, enabling prolonged periods of growth and eventual tumor reformation.
Efforts to reduce tumor recurrence have focused heavily on three factors: lowering detection thresholds by developing highly sensitive diagnostic tools, shedding light on the mechanisms of tumor evolution, and innovative treatment regimens.
At the nexus of these efforts is circulating tumor DNA (ctDNA), a key marker for residual or recurrent cancer.
Tracking cancer with ctDNA
By its nature, cancer is an insidious disease that can be difficult to distinguish from healthy tissue, particularly in early stages. Screening and diagnostic assays currently rely on a combination of visual identification (such as through mammography, PET scans, CT scans, etc.,) and analyses of surgically biopsied tissues for telling genetic or cytological features. While effective, the low resolution of imaging technology and the highly invasive nature of tissue biopsies limits our ability to detect nascent tumors and precludes the tracking of tumor evolution over time.
However, advances in DNA sequencing technology have opened the door to new possibilities, including the use of liquid biopsies for both cancer screening and monitoring. A liquid biopsy is the collection of blood, urine, or other biological fluids that are known to contain cancer-related biomarkers. One such biomarker is ctDNA.
Both malignant and healthy cells leak fragments of genomic DNA into the bloodstream where these bits of cell-free DNA can circulate for up to 2 hours before degradation. As they’re derived from genomic DNA, cell-free DNA molecules may bring with them epigenetic and mutational signatures that give researchers brief but telling insights into the DNA’s origin. This, along with its short half-life, has made tumor-derived cell-free DNA (otherwise known as circulating tumor DNA, or ctDNA) particularly interesting as a biomarker for early cancer detection and minimal residual disease (MRD) testing in the post-diagnosis setting.
MRD testing and tumor heterogeneity
MRD testing is an approach intended to measure treatment efficacy by assessing patients for the presence of malignant cells during and after treatment. As mentioned above, current monitoring technologies have a limited detection threshold, meaning they’re often not sensitive enough to detect residual cancer cells until tumor regrowth has significantly progressed. In contrast, ctDNA can be used to detect residual cancer cells before they’re detected by standard of care technologies.
For example, a 2017 study examined the potential benefits of tracking ctDNA in patients with non-small cell lung cancer and found that, on average, ctDNA from residual cancer cells was detected 70 days before CT-scanning could reliably identify tumors. In some patients, the gap between ctDNA and CT-scanning detection was more than 150 days; and, in one case, ctDNA analysis identified a metastatic growth that altogether lacked conclusive clinical presentation. These results suggest that ctDNA can greatly improve MRD testing and save both patients and their physicians critically valuable time.
This study is one of many to come out of a larger project known as TRACERx (TRAcking Cancer Evolution through therapy (Rx))—a multi-institutional effort to advance our understanding of cancer evolution through the use of new technologies, including those that enable ctDNA analysis.
Beyond identifying if cancer is present, evidence from studies like TRACERx show that ctDNA can also be used to build a detailed profile of the cancer cell it came from. Doing so can reveal genomic patterns that allude to the DNA’s tissue of origin or reveal therapeutic weaknesses that may be exploited. And, because ctDNA is released from individual cells, it can also give unique insights into the cellular heterogeneity within the patient’s tumor. Tracking such information is critical to understanding how tumors are likely to respond to therapy and is the primary goal of studies like TRACERx. However, accessing this information is far from trivial.
Overcoming limitations in ctDNA analyses
Accessing the wealth of information carried in ctDNA is a significant challenge. Following treatment, mutation-carrying ctDNA fragments may be less than 0.01% of the cell-free DNA in a sample. The rarity of ctDNA is a challenge in and of itself, one made far more difficult by the fact that somatic tissues may undergo hypermutation. This means that cell-free DNA unrelated to the tumor may harbor mutations. To accurately identify a ctDNA molecule as tumor-derived, it must be carrying a mutation that the sequencing assay recognizes as being tumor-specific. The sensitivity of these assays is thus dependent on several factors, including the breadth of mutations being scanned for.
Currently, most MRD assays narrowly focus on only relatively few mutations that are known to be present in the patient’s tumor (an approach broadly referred to as tumor-informed). This narrow focus allows researchers to conserve sequencing resources but often sacrifices clinical sensitivity. However, relying solely on tumor-informed designs can prevent researchers from detecting new mutations that arise during tumor evolution. As such, there is a strong need for an MRD assay that is both broad and efficient.
To this end, Personalis recently announced a collaboration with the TRACERx consortium to enable the use of Personalis’ NeXT Personal® platform. This advanced sequencing platform provides coverage over clinically relevant loci that are known to drive tumor pathology in addition to personalized, tumor-informed coverage. Put another way, NeXT Personal scans for common cancer mutations while also looking for patient-specific mutations. This combination enables NeXT Personal to improve MRD sensitivity 10 to 100-fold over similar technologies, making it possible to detect mutation carrying ctDNA at concentrations as low as 1-part-per-million.
NeXT Personal was specifically designed to reach this level of sensitivity in order to detect residual disease at the earliest time point possible (see Figure 1).
Figure 1: Sensitivity of NeXT Personal. Extrapolation of data from multiple studies suggests that, following surgical resection, patients who will develop recurrent tumors are likely to have mutation carrying ctDNA fragments at a frequency of 1 part-per-million (represented as variant allele frequency, or VAF).
The substantial leap in MRD sensitivity provided by NeXT Personal greatly improves our ability to develop ctDNA as a clinically impactful biomarker and brings us one step closer to a better standard of care in oncology.
