Next-Generation Sequencing Problematic, If Promising

Genomic profiling in cancer may open the door to the use of unproven therapies

Targeted cancer therapies have been one of the most important steps forward in personalized medicine — also called precision medicine. The value of these drugs is that they work in a high proportion of cancer patients who have a specific genetic alteration. Conversely, they may not work in patients without that mutation.

These medications are made possible by the existence of genetic biomarkers for the presence of a specific cancer-causing pathway. Pharmacogenetic tests for the biomarkers help clinicians know when to use a targeted therapy, thus reducing trial-and-error therapy and producing great benefits for patients.

Rapid advances in gene sequencing, specifically genomic profiling using next-generation sequencing, are providing new insight into the genetic heterogeneity of cancers by identifying additional alterations in cancers where they were not previously known.

Genome sequencing’s new low cost

“The cost of gene sequencing has plummeted, opening the door to many new applications,” says William Nelson, MD, PhD, director of the cancer institute at Johns Hopkins Kimmel Cancer Center. One of those applications is next-generation sequencing of the genomes for solid tumor cancers.

The table shows the extent of this cost reduction. An entire human genome includes about 30,000 genes.

Cost per human genome (log scale)


However, genomic profiling is also leading to claims that targeted therapies, even conventional medications, may be candidates for off-label use for cancers for which they are not approved by the FDA. So while technology is propelling cancer diagnosis to the next level of understanding, it also potentially opens the door to unsupported therapy.

Genomic profiling defined

Genomic profiling looks at the alterations of all of the protein producing genes in the entire human genome, or at specific genomes like that of a cancer tumor. This spectrum analysis is powered by next-generation sequencing, NGS, which is a technologic leap into the future of precision medicine. Instead of looking for genetic alterations serially, gene by gene, NGS relies on high speed parallel processing of a library of small DNA segments to “read” different types of alterations.

The NIH says it would take years to sequence a person’s entire genome with earlier technology, compared to days or weeks with NGS. In addition, NGS is significantly more sensitive than previous technologies, so it also has the capability to find low frequency mutations that escaped earlier technologies.

The power of NGS has allowed research laboratories and upstart direct-to-consumer genetics labs to create panels of 200+ genes that they use to identify all types of genetic alterations.

“Next-generation sequencing simplifies and reduces the cost of drug development…,” says William Nelson, MD, PhD, an expert on translational genetics.

In doing so they have created a new buzzword — clinically actionable targets — to claim that the additional alterations they find may be candidates for drugs outside of the cancers they are approved for, in spite of the lack of solid documentation of their effectiveness.

Technology outpaces understanding

“Most solid organ tumors have 60 mutations and 100 or so translocations [a chromosome breaks off and attaches to a different chromosome], most of which we do not understand the significance of,” says William Nelson, MD, PhD, director of the Johns Hopkins Kimmel Cancer Center and an expert on translational genetics. “The technology now is such that we can find all of the changes before we know what they mean.

“The big concern is not that genomic profiling will lead to increasing costs. The more important worry is that it will get less expensive and people will go off half cocked with information that isn’t truly worth acting on. Some labs are going to say that they have found one alteration among 200 others and that one small study said this particular drug might be related to this alteration.”

This is already happening. The web site of a genetics lab, GeneKey, says its multi-gene panel “provides information to identify potential treatments that would not be considered otherwise. We search for options not only among known lymphoma treatments, but among the more than 2,000 FDA-approved drugs.”

Academic and research labs also use the term clinically actionable targets, although in that context it could mean there are opportunities to investigate new indications for existing drugs.

There is a very positive side to genomic profiling. “Drug development companies are totally enamored of this technology,” says Nelson. “There are hundreds of drug compounds in discovery. Next-generation sequencing simplifies and reduces the cost of drug development because it eliminates many of the steps in the former empiric drug development process. For example, it simplifies and reduces the cost of clinical trials.”

Genomic profiling via NGS is still in the process of proving its analytic validity, clinical validity, and clinical utility.

Analytic validity is the ability of profiling to accurately identify the intended genetic alteration target and to exclude false positive results. Clinical validity relates to how closely a genetic alteration is related to the presence or risk for a specific cancer. Clinical utility exists when a treatment or therapy is useful in decision making and improves patient outcomes.

The analytic validity of NGS is affected by complexity that does not exist for other tests. The starting point is the need to have high quality tumor samples that accurately represent the alterations that may be found. In addition, the traditional standards of sensitivity, specificity, and accuracy for the millions of repeated reads of genetic alterations come into play.

FDA’s perplexed

Genomic profiling is not regulated by the FDA, which has been wrestling with how it might regulate genomic tests. It has not figured out how to get it right — what will be regulated, how evaluations will take place, or when this oversight might kick in.

Each lab may freely develop tests and market them. Labs are usually certified by CMS under the Clinical Laboratory Improvement Act (CLIA), and by the College of American Pathologists (CAP). However, these certifications deal primarily with proficiency and reproducibility and do not encompass clinical validity.

Standards have been developed by the Next-Generation Sequencing: Standardization of Clinical Testing workgroup and the American College of Medical Genetics. The elements in these standards include specimen preparation, the method for “reading” alterations, and the software that interprets the significance of the reads of possible mutations.

“The analytic validity of genetic testing is extremely high. Everyone who has looked at it in any way has said that technical analysis is very good; when we find a mutation it is usually confirmed with an alternate sequencing technology, and the result is usually correct,” says Andrew Faucett, a director at Geisinger Health System’s genomic medicine institute. He served as a reviewer of an Institute of Medicine report on incorporating genomic information into clinical practice.

“We are interested in tests that can clearly impact treatment de­cisions,” says Steve Perkins, MD, vice president for medical affairs at UMP Health Plan.

There is at least one potential limitation in the procedures for confirming the analytical validity of NGS. “Next-generation sequencing can pick up alterations that Sanger sequencing, the former gold standard, would miss,” says Mark Rubin, MD, director of the institute of precision medicine at the Weill Cornell Medical College. “When we go to validate next-generation sequencing for regulatory purposes, we are asked to use older technologies which are less sensitive.”

A validation study

An October online article in the journal Nature Biotechnology describes the work of Foundation Medicine in demonstrating the analytical validity of its cancer profiling test. That test is a 287-cancer gene panel that examines 4,557 alterations. The validation study tested 2,221 samples of breast, lung, colorectal, head and neck, prostate, and other cancers. The studies validated the Foundation Medicine test against other testing methods.

Alterations were reported in 174 of 189 (92%) tested genes. Testing found 1,579 unique alterations with an average of 3.06 alterations per sample (range, 0–23). A tumor can harbor multiple alterations at any position in the DNA with a wide range of frequencies for each alteration. The validation study found clinically actionable alterations in 76% of the tumors with an average of 1.57 actionable alterations per patient sample, three times the number of actionable alterations detected by current diagnostic tests.

In summary, the profiling test found numerous alterations that would not have been identified by previous technologies. This information in the hands of desperate patients could create false hopes for additional therapy options.

The sheer volume of additional alterations identified through genomic profiling makes it difficult to determine the clinical validity of an individual alteration.

Clinical validity

“Clinical validity depends on the disease and the genetic alteration,” says Faucett. “We have a good understanding of the types of mutations in some diseases and there’s a much broader group of diseases and alterations that we’re learning about.”

Cancer genomes and entire human genomes contain many genetic alterations that are insignificant, not necessarily disease producing. “One category of genetic alterations is research-grade information; the alternation is read accurately so it is a true mutation, but its frequency is very limited and its impact is unknown,” says Rubin.

“These represent the largest percentage of what we find. Right now we’re struggling for a systematic approach to determine what are the important mutations that we should focus on developing the next treatment targets.”

The identification of previously unknown alterations complicates knowing if they are clinically valid.

“Herceptin for HER2 amplification in breast cancer works part of the time, which means we do not fully understand all of the cancer-causing alterations,” says Faucett. “But that biomarker was based on previous testing capabilities, and became the standard of care for Herceptin therapy. Today, though, health plans, as the result of finding additional alterations through next-generation sequencing, could say that alteration as a clinically actionable target is not supported by evidence.”


Experts say that databases of genomic profiles serve as key references to support efforts to determine the clinical validity and the utility of genomic profiling.

Genomic profiling via NGS is still in the process of proving its analytic validity, clinical validity, and clinical utility.

“One of the most important concepts for next generation sequencing and its relationship to precision medication is that we are developing a knowledge base,” says Rubin. “That knowledge base today includes information that is not actionable because we do not fully understand the identified alteration, but in five years some of that information will be found actionable.” There are many different types of reference databases — publicly funded cooperative databases, databases operated by various consortia, international databases, and databases at hospitals and research institutions. One challenge is the differences in information stored in the databases and finding an effective means for a researcher to share or easily access information.

“It would be interesting to develop something similar to crowdsourcing where questions or developments could be shared widely and quickly, especially on truly new or rare situations,” says Rubin.

Genetic variants

The 1,000 Genomes Project is the first project to sequence the genomes of a large number of people to provide a comprehensive resource on genetic variations. Its goal is to find most genetic variants that have frequencies of at least 1% in the populations studied.

The project provides information on populationwide variations that may or may not be related to specific diseases. It serves as a resource to determine the extent of natural variations in the population.

The Cancer Genome Atlas is specific to cancer. “It provides a static [one time] image of mutations in common tumors,” says Rubin.

A common limitation of genomic databases is that they do not provide information on subsequent development of diseases or the effects of treatment, says Nelson.

“Where we are really making inroads is in projects like the Stand Up to Cancer work in prostate cancer,” says Rubin. “Five institutions are participating in a study that tracks advanced castrate resistant patients. Mutations are sequenced before, during, and after treatment.

“Patients in this study are in a well-defined protocol so there’s the potential to learn about the development of mutations that lead to resistance, the weaknesses of existing drugs, and what additional drugs are needed to deal with resistance.”

Plans reluctant to cover next-generation sequencing

False hope for many patients

Genomic profiling is still in its infancy. Several health plans have said that they are not seeing claims for a large number of profiling tests. Some claims for genomic testing may be paid, although these tests are considered experimental.

“Our coding system does not allow us to identify specific types of genetic tests because tests are broken down into many components and the codes are very general,” says Donald Liss, MD, a medical director at Independence Blue Cross.

Liss explains how health plans will handle coverage decisions for profiling. “With new technologies like genomic profiling, we, and many other health plans consult with technology assessment firms,” says Liss. “We will look at the validity and utility of these tests, evidence from controlled trials, and other authorities such as the incorporation of the technology into clinical guidelines from places like the National Comprehensive Cancer Network.”

Rapidly evolving technologies such as genomic profiling may be easier to handle at health plans with integrated provider organizations such as the University of Pittsburgh Medical Center. “When it comes to the development of medical policy for new technologies at the health plan, we have the ability to create a dialogue with our clinicians and researchers to get a perspective on the value of a new technology that may not otherwise be available,” says Steve Perkins, MD, vice president for medical affairs at UMP Health Plan.

“We are interested in tests that can clearly impact treatment decisions. We ask if this particular test in this particular case is going to impact treatment decisions. If the answer is yes then coverage is feasible. The onus is on us to determine if the test is responsible,” says Liss. He adds, “The other area that is of interest to insurers is what is on the horizon. We need to be thinking about how technologies will impact the standard of care as we design our plans and project future medical claims costs.”

Over time the use of and coverage for genomic profiling may develop a twist in comparison to how diagnostic tests are used.

Profiling is likely to be considered experimental until its utility is proven and then prior authorization is likely to ensure its appropriate use.

After a while, profiling may evolve to limited use. “Genomic profiling will not be used universally. There are situations when you want to do a simple focused genetic test such as the companion diagnostics being approved for new targeted therapies,” says Rubin. “Then there will be other situations where it may make sense to do sequencing; it can come into play if the patient becomes resistant to first-line therapy, such as in lung cancer, because you are looking for the presence of several other targetable mutations.”