Cancer genomics, a young and evolving field, seeks to comprehensively examine all gene alterations and expression differences across the entire genetic landscape. With their increasing sophistication, genomic technologies are revolutionizing the way we detect, evaluate, and treat cancer. For example, since cancer genomes change over time, longitudinal profiling incorporating artificial intelligence (AI) promises to provide more accurate assessments of changes and to thereby improve therapeutic methods. Additionally, improvements in liquid biopsy analyses allow a simple blood draw to characterize circulating tumor DNA (ctDNA) for comparison to panels of key cancer genes.
Advances in optical genome mapping (OGM), an enhanced cytogenomic technology, are enabling higher resolution and increasing sensitivity for more precise measurements and curation of genomic changes. In addition, there are new short- and long-read sequencing technologies that are accelerating and improving the accuracy with which variants across the genome, transcriptome, and epigenome are identified.
We spoke to several scientists who attended the 14th Annual Cancer Genomics Consortium held in St. Louis, MO, last August. They offered some key perspectives on the field’s current position and future direction.
Longitudinal testing
Profiling alterations in the genomes of cancer patients may provide a more dynamic rather than static perspective on critical therapeutic information. “Understanding genomic changes over time requires longitudinal profiling of the tumor or the ctDNA,” notes Kate Sasser, PhD, CSO, Tempus. “In the past, repeat sampling was cost prohibitive, but as sequencing costs come down and analytical platforms get smarter and faster, this is being addressed.”
Sasser says that Tempus has built a comprehensive portfolio of oncology NGS assays. “This includes matched tumor/normal, and both DNA (panels or whole exome) alongside RNA (whole transcriptome),” she details. “We also run liquid ctDNA profiling, and our R&D teams are working on new features for these tests, including using our ctDNA assays (xF and xF+) to quantify molecular tumor response or minimal residual disease (MRD) measurement, both of which we believe will be important tools for clinical researchers and physicians.”
The company continues to improve analytical sensitivity. “We need ways to differentiate noise and signal, and to get to more sensitive levels of detection,” Sasser reports. “Fortunately, that technology is also emerging, with some platforms now in the range of 3–10 ppm. Further, advances in machine learning and AI are helping to increase signal-to-noise ratios.”
Decreasing the sample size required for analysis is also being addressed. “We continue to look for ways that we can run genomic profiling with smaller sample volumes, more sensitive technology, and faster analytics to bundle longitudinal testing,” Sasser says. “We are working on analytical methods to achieve ultrasensitive MRD monitoring. We are also applying machine learning and AI applications to contextualize the genomic information, which is an exciting emerging area of R&D.”
According to Sasser, the field of cancer genomics is still young and full of possibilities: “I think we’ll see greater adoption of genomic profiling for all cancer patients, at much earlier stages, with assays that provide the whole genome and contextualize the information with AI and machine learning. Sequencing will occur at multiple timepoints in the patient’s journey, including for risk prediction and early detection, as well as therapy matching and molecular response monitoring. Therefore, comprehensive genomic profiling will be more ubiquitous and woven across clinical cancer care.”
A liquid biopsy test for genomic profiling
Generally speaking, liquid biopsy
describes the interrogation of biomarkers in bodily fluids to gain insights into an individual’s health condition, susceptibility to disease, or response to treatment. “Among different technologies, the analysis of circulating tumor DNA obtained from the blood plasma has demonstrated the greatest clinical utility,” says Florian Klemm, MD, senior NGS scientist, Sophia Genetics. “A simple blood draw not only allows for genomic profiling in cases where tissue sampling might be too invasive or tissue quality insufficient, but also enables the representation of a tumor’s heterogeneity in a single assay.”
While many studies have documented the clinical benefits of liquid biopsy, it is gaining recognition on another front. Klemm reports, “The increased sensitivity required to analyze cfDNA also concurrently leads to the inclusion of biological signal that requires careful interpretation. For example, mutations derived from clonal hematopoiesis not only inform about the underlying expansion of nonmalignant white blood stem cells harboring cancer-associated variants, but also constitute a confounder when mutations in patients with solid tumors are examined.”
Klemm says the company has built a streamlined, DNA-only, sample-to-report NGS workflow that addresses common challenges in cfDNA analysis such as limited input amount and low ctDNA variant fraction. “We have combined an optimized library preparation protocol that incorporates our proprietary CUMIN molecular identifier technology with our state-of-the-art, DDM-driven, analytical tools such as duplex-aware variant calling,” he details. “This enables laboratories to obtain reliable and reproducible signal detection in their cfDNA analyses.”
According to Klemm, Sophia Genetics has launched a collaboration with Memorial Sloan Kettering Cancer Center (MSK) to decentralize “MSK-ACCESS,” a comprehensive liquid biopsy test developed by MSK covering 146 key cancer-associated genes from MSK’s solid tumor genomic profiling assay MSK-IMPACT. “This is a truly democratizing effort,” he declares. “It gives laboratories worldwide access to the expertise and clinical knowledge of a leading cancer center in tandem with the analytical DDM platform of Sophia Genetics.
“Importantly, the ‘MSK-ACCESS’ powered with our DDM (research use only) solution incorporates a matched white blood cell DNA analysis to identify clonal hematopoiesis of indeterminate potential (CHIP) variants, greatly increasing the power to detect tumor-specific variants in cfDNA. This offers the opportunity to employ a sophisticated tool over multiple timepoints to monitor response to therapy and reveal the emergence of drug resistance.”
Optical genome mapping
Cancer genomics faces two technological challenges, says Alex Hastie, PhD, vice president of clinical and scientific affairs, Bionano Genomics. He describes them as follows: “The first challenge is the detection of structural variations, a task that has been addressed with technologies such as karyotyping and DNA sequencing. However, with karyotyping, sensitivity and resolution are very low. And many SVs elude conventional DNA sequencing methods. The second challenge is the detection of variants that exist in only a fraction of the tumor but could represent the most malignant cells—the cells that would need to be targeted for treatment.”
Enter optical genome mapping (OGM), a next-generation cytogenomic technology that enables higher resolution than traditional cytogenetic/cytogenomic technologies. It also provides a comprehensive evaluation of structural variations and copy number variations.
“OGM is a technique that can detect all classes of structural variation at high resolution and sensitivity, including when the variant is in a very low variant allele fraction,” Hastie explains. “As a result of these advantages, OGM can detect known and novel structural variations related to the biology of the tumor, and these variants may be or may become biomarkers for diagnosis, prognosis, and treatment.”
How does it work? Hastie summarizes, “OGM images extremely long molecules of genomic DNA extracted from tumor cells and linearized in a nanochannel. This linearization and imaging enables precise measurement of the DNA across the genome, and when changes in measurements are observed, they can be itemized, curated, and classified and correlated to the biology, often identifying clinically relevant variants. Optical genome mapping is able to detect copy number variations, inversions, translocations, repeat array expansions/contractions, aneuploidies, chromothrypsis, allelic imbalances, loss of heterozygosity, absence of heterozygosity, and other complex rearrangements.”
Hastie notes that the sensitive technology can be especially helpful for detection of homologous recombination deficiency signatures, a cancer phenotype resulting from defects in normal cell division and DNA repair mechanisms that can damage chromosomal DNA. “This damage,” he continues, “can make some cancer cells highly malignant, but interestingly, when cells have this class of defect, they become good targets for drugs that can kill these cells in a specific way. To understand which drugs might be most effective against a tumor, it is necessary to determine if it has homologous recombination deficiency, and this involves measuring the level of structural variation of that particular tumor.”
Bionano Genomics continues to refine the technology. Hastie discloses, “We are working on many upgrades to the OGM workflow, including improvements to sample preparation, higher throughput, and more powerful bioinformatics.”
Copy number variations
Other tools for analyzing copy number variations include Thermo Fisher Scientific’s OncoScan CNV assays. According to the company, evaluating and understanding copy number variations and allelic imbalances across the entire genome is a key aspect of comprehensively profiling solid tumors as well as identifying predictive biomarkers.
The assay utilizes molecular inversion probe (MIP) technology for identifying copy number variations, as well as loss of heterozygosity, copy neutral loss of heterozygosity, and somatic mutations.
For the assay, molecular inversion probes targeting specific loci are hybridized to genomic DNA. The company reports that molecular inversion probes are designed such that successful annealing results in a single-stranded gap that can be filled in by a specific gap-filling enzyme. This produces a circular molecule and removes the nonhybridized targets. Next, a nuclease linearizes the circularized molecular inversion probes that are subsequently amplified and hybridized to the microarray. The system also works on degraded DNA samples, such as formalin-fixed, paraffin-embedded samples.
The long and short of sequencing
“Current technologies detect only some and not all of the DNA, RNA, and epigenetic variants that are important for detecting and improving the understanding of cancer,” points out Jonathan Bibliowicz, PhD, associate director, Cancer Genomics, PacBio. “Sequencing technologies are needed that can discover all variants across the genome, transcriptome, and epigenome.”
Two types of sequencing are employed to articulate the genome, long- and short-read sequencing. Each has its strengths and limitations. For example, as Bibliowicz notes, there are two tasks better addressed by long-read sequencing than by short-read sequencing: “Most structural variations, a class of variations that drive the development of most cancer types, are missed by standard short-read sequencing. And most of the RNA isoforms that have been shown to drive cancer are completely missed by standard short-read sequencing.”
Thus, to resolve large variants (including tandem repeat expansions and structural variations) and to map challenging repetitive regions of the genome, fast and accurate long-read sequencing technology is critical. To facilitate long-read sequencing, PacBio provides Revio, a HiFi system capable of delivering the equivalent of 1,300 genomes per year. “We are continuing to enable new applications on our platforms, most recently whole genome sequencing of somatic variants in cancer using Revio,” Bibliowicz adds. “This allows the most comprehensive cancer whole-genome interrogation currently available.”
Short-read sequencing can also be problematic. Bibliowicz elaborates, “A limitation of standard short-read sequencing is that these technologies are error-prone, which limits our ability to detect rare variants due to the noise of the system. In other words, the noise (errors) overcomes the signal (the variants).”
To help solve these challenges, PacBio has developed a new short-read technology called the Onso system. “This highly accurate short-read sequencing technology can detect cancer variants at unprecedented sensitivity for ‘needle in a haystack’ scenarios, like liquid biopsy,” Bibliowicz asserts.
According to Bibliowicz, PacBio’s technologies can shed light on a wide range of cancers. “Our technologies,” he asserts, “have been applied to both solid and hematological tumor samples, as well as both pediatric and adult cancers.”
As to the future of the field, Bibliowicz shares some insights: “Because of cancer’s complexity, cancer genomics will become multiomic. While cancer is a disease of the genome, mutations at the level of DNA that drive cancer result in dysregulation at all levels of cell biology. Besides dysregulation at the genomic level, there is dysregulation at the transcriptomic, epigenomic, and proteomic levels. Different types of information will be combined to unravel this complexity and create a new generation of assays and clinical tests that do a better job at diagnosing and monitoring cancer.”
