In Good Company: Have Companion Diagnostics Proven Their Mettle?

Advances in the evolving field of precision medicine have led to the development of therapies that target specific biologic, including molecular, abnormalities. Companion diagnostics – tests that identify patients whose cancers harbor these abnormalities – have, out of necessity, emerged hand-in-hand with these new therapeutic approaches.

Because these tests are designed to be paired with a specific drug, the development of both the drug and device requires close collaboration between the manufacturers and the U.S. Food and Drug Administration’s (FDA) drug and device centers. In August 2014, the FDA issued final guidance on the development, review, and approval or clearance of companion diagnostics. Questions about the practical application of these tests and their correct interpretation, however, remain.1

ASH Clinical News spoke with regulators and device developers about the challenges of using genomic abnormalities to guide the treatment of patients with blood cancers.

What Are Companion Diagnostics?

Companion diagnostics are best described as medical devices or tests that provide information that define the condition of use for another medical product – a drug, biologic product, or other device. These tests include in vitro diagnostics that measure a particular biomarker, or nucleic acid-based tests, which analyze variations in the sequence, structure, or expression of DNA and RNA.

“A companion diagnostic is required as a condition of use to make a medical product safe and effective,” Robert Becker, Jr., MD, PhD, chief medical officer of the FDA’s Office of In Vitro Diagnostics and Radiological Health (OIDR), and Elizabeth Mansfield, PhD, director of the personalized medicine staff at OIDR, explained in a review article.2 “A companion diagnostic is therefore a subset of biomarker-oriented tests, which cover all diagnostic tests.”

Generally, companion diagnostics do one or more of the following:

  • Select patients in whom a particular agent or drug would be effective
  • Identify patients who should not be treated with an agent or drug because of a high risk of adverse events
  • Identify patients who match the drug’s or agent’s indications
  • Determine genetic carrier status

Developing a companion diagnostic requires two important components, according to Alessandra Cesano, MD, PhD, chief medical officer of Nanostring Technologies, a company that uses a digital molecular barcoding diagnostic technology.

“The first is the analytic component,” Dr. Cesano told ASH Clinical News. “When you say, ‘With this test, we measure for A,’ you want to be sure that you actually do measure for A. We are looking for sensitivity and specificity, but that doesn’t necessarily mean that measuring A has a clinical impact.”

That comes in the form of clinical validity, she said. “For any test to be considered a companion diagnostic, we first need to show that we can reproducibly measure A. Then, we have to show that when we use this test, which has been analytically validated, it has validity in a specific clinical context.”

One Piece of the Personalized Care Puzzle

Most hematologic malignancies are caused by a genetic alteration, such as a point mutation, chromosomal aberration, or copy number variation (TABLE 1) – all representing new potential therapeutic targets. The FDA has approved two companion tests and targeted therapies for hematologic indications (SIDEBAR, TABLE 2).3

One example is the tyrosine kinase inhibitor imatinib. When studies showed that patients with PDGFRB rearrangements achieved long-term, durable remissions with imatinib treatment, the FDA approved the PDGFRB fluorescence in situ hybridization (FISH) companion diagnostic test to inform the use of imatinib in patients with myelodysplastic syndromes/myeloproliferative disease.3-4

However, relying solely on the results of companion diagnostics to greenlight a patient for a particular targeted agent would be inadequate, James Zehnder, MD, professor of pathology and medicine (hematology) at Stanford University Medical Center in Stanford, California, told ASH Clinical News.

“One aspect of this discussion that doesn’t get enough attention is that clinical decisions are not made on a single test result, but by integrating all of the information available on a patient – clinical, pathologic, radiologic, and family history factors,” he explained. “That’s an essential part of personalized medical care. It’s not just the test result that people are acting on; all of the available information needs to be put into the hopper.”

Advances in genomics have moved the field forward, but the ability to decode all cancer-associated mutations – or novel mutations that may be discovered – is still far in the future. For instance, blocking a hematologic tumor protein pathway can drive cells to develop compensatory mechanisms, ultimately leading to tumor resistance. The potential for a targeted therapy to non-selectively inhibit both mutant and wild-type proteins raises concerns of excessive toxicity, and patients’ genetic variations can alter responses to therapy and affect drug dosing.

Decoding the Genetic Background

When it comes to delivering important information for patient management and therapeutic planning, cancer genomes are clearly the leaders of the pack. Sequencing efforts have identified about 140 genes known to be drivers of the oncogenic phenotype, and they are closely linked to disease mechanisms that support oncogenesis and determine cell fate, cell survival, and genome maintenance.5

Though companion diagnostics can provide a wealth of data about a patient’s specific genetic alterations, they cannot decipher which genetic alteration will be the most effective target for treatment. In a Science review article, Bert Vogelstein, MD, Johns Hopkins University in Baltimore, Maryland, and colleagues offered the following analogy to describe the challenging genomic complexity: “For most cancer types, [the genomic] landscape consists of a small number of ‘mountains’ (genes altered in a high percentage of tumors) and a much larger number of ‘hills’ (genes altered infrequently).”5 This leaves an overwhelming majority of alterations in tumors – point mutations, copy number alterations, translocations, and epigenetic changes throughout the genome – that are not considered “driver gene” mutations, they continued. “They are simply passenger changes that confer no selective growth advantage.”

While normal cells also undergo genetic alterations as they divide and are programmed to enter cell death in response, they explained, cancer cells have evolved to tolerate genome complexity by acquiring mutations. “Thus, genomic complexity is, in part, the result of cancer, rather than the cause,” Dr. Vogelstein and colleagues added.

Given this genomic complexity, how can clinicians who treat patients with hematologic malignancies interpret genetic data in a reliable, meaningful way for their patients?

“That’s a tough question,” Geoffrey Ginsburg, MD, PhD, professor of medicine at Duke Center for Applied Genomics and Precision Medicine at Duke University in Durham, North Carolina, admitted. “In general, we are still learning how to interpret genetic information from genome sequencing. Much of what we derive from this information and how we use it depends on the reference databases that we use for germline sequencing.”

“When it comes to sequencing in hematologic malignancies, most clinicians are going to get back a report from the lab suggesting that certain variants may be underlying the etiology of the tumor. In many cases, it could be several variants,” he added. “So, if there are three or four potential drivers of a tumor, each with a different targeted therapy, which one should be selected as a target? And which one is more likely to be the underlying cause of a patient’s tumor? That is a really tough area.”

Duke, for instance, is one of many institutions that has set up multidisciplinary tumor boards specifically to discuss these complex cases and make treatment recommendations for these patients. “I’d recommend that hematologists/oncologists partner with tertiary cancer centers, and bring their cases into the mix,” Dr. Ginsburg said. “It would be an opportunity to learn more about interpreting these genetic abnormalities.”

Large-scale, global efforts are underway to develop interpretive algorithms for genetic data, he noted. For instance, the Genomics England program combines genomic sequencing and medical records data to sequence 100,000 genomes from 70,000 participants. This and other projects hopefully will help clinicians answer those tough treatment questions, Dr. Ginsburg added. “Eventually, we can begin to designate genomic data in clear categories, such as ‘definitely pathologic’ or ‘uncertain significance,’” he said.

Having these algorithms would also help in communicating complex genetic data and the significance of these data to patients, Dr. Zehnder added. “There is clearly a need for more health-care professionals, like genetic counselors, with the skills to explain this challenging genetic information to patients in a clear way,” he said. “Physicians who have been in practice for a long time may not necessarily have those skills.”

The FDA and the Future

Companion diagnostics are federally regulated under the same rules as any other diagnostic device, but they are approved together with a targeted therapy. The FDA’s Center for Drug Evaluation and Research or the Center for Biologic Evaluation and Research approves the therapeutic product, while the Center for Devices and Radiological Health okays the companion diagnostic.1

“We need to understand the analytic and clinical performance characteristics of the test,” Drs. Becker and Mansfield wrote, explaining the review and approval process of these products. “We evaluate the measurement quality of the testing system, the actual clinical information that the diagnostic can produce, and the correspondence of that measurement with some physiologic feature that indicates that someone should be treated or will benefit from treatment.”

Most companion diagnostics for guiding blood cancer treatment are categorized as class II or III in the FDA’s approval process, Dr. Cesano explained, based on the level of control necessary to assure the safety and effectiveness of the devices and the risk they pose to the patient and/or user. Class II devices are generally prognostic and are not related to a particular therapeutic drug or agent. “The results tell you, in general, whether this patient’s disease is going to have a good outcome – is it an aggressive disease or a more indolent disease? With a class II test, you have to provide analytic and clinical validation.”

Class III devices, on the other hand, are predictive tests that act as “the gatekeeper for the use of a drug,” she said. These tests require a Premarket Approval Application (PMA) to the FDA and must meet analytic and clinical validity standards and demonstrate clinical utility. “If a class III assay is not analytically and clinically robust, there is a greater chance of false-positive or false-negative results, which can have serious consequences for the patient, either because they don’t receive the appropriate treatment or they undergo unnecessary treatment,” Dr. Cesano explained.

The Next Generation of Companion Diagnostics

Over time, the FDA has approved companion diagnostics under a “one indication, one treatment” paradigm, although, according to Dr. Mansfield, that happened more by default than design. “The one test, one drug situation occurred because that is the way the products were brought to us; it is not a requirement.”

But the advent of next-generation sequencing (NGS) in genomics is changing that approval pathway. NGS, also known as high-throughput sequencing, sequences millions of small fragments of DNA in parallel. Clinicians then use bioinformatics analyses to “piece together” the fragments by mapping the individual reads to the human reference genome. Each of the 3 billion bases in the human genome is sequenced multiple times, providing insight into unexpected DNA variations. NGS can be used to sequence entire genomes or targeted to specific areas of interest, including coding a whole exome (22,000 coding genes) or a specific number of individual genes.6

“NGS will allow for multiple tests to be done on a single sample,” Dr. Mansfield pointed out. “We are looking at possible approval of an NGS test that can produce many different diagnostic claims. The complication with this is that more data will need to be generated to determine that the test works. It will take us more time to review, but we think that patients will benefit from having to only undergo one test, rather than multiple tests, to guide their treatment.”

Rather than a “one test, one drug” scenario, NGS will deliver biomarkers that may not yet have a companion diagnostic available. Once a drug comes along that uses that biomarker, the companion diagnostic can be developed based on NGS data, she explained.

NGS is also expected to strengthen the working relationship between diagnostics developers and therapeutics developers. “The challenge is that you have to get a company to sign a collaborative agreement to invest in the companion diagnostic and the drug at the same time,” Dr. Cesano said. “Unfortunately, the pharmaceutical companies tend to approach the diagnostics companies toward the end of the process – at the end of the phase II stage for the drug – and say, ‘We want to start phase III testing next month, so we’ll need a companion assay.’ That’s very late.”

Ideally, a companion diagnostic assay would be used during the phase II trial of the drug to determine clinical validity, and then in phase III to determine clinical utility, she added. “However, that means that investigators need a fully validated assay available by phase III of a clinical trial – that takes time.”

The Laws of LDTs

Time is one of the points of contention between the FDA and developers of so-called laboratory developed tests (LDT). LDTs are assays developed by and used in a single lab, often one that is affiliated with a major treatment or academic center. They are not sold to other labs or hospitals and do not go through the formal FDA PMA process; instead, they fall under the purview of the Centers for Medicare and Medicaid Services through the Clinical Laboratory Improvement Amendments (CLIA).7

The FDA has the option to regulate LDTs, but the agency has chosen not to do so, explained Pam Bradley, PhD, a staff fellow at the FDA. “We’ve been exercising what we call enforcement discretion,” she said.

That is changing, though. In 2014, the FDA issued draft guidance outlining a framework for regulatory oversight of LDTs, taking the position that regulatory oversight of LDTs under CLIA does not adequately address patient safety concerns because CLIA officials do not assess or confirm the tests’ analytic or clinical validity.8

“We aren’t looking to disrupt the LDT industry that physicians and patients rely on,” Dr. Mansfield emphasized. “Right now, most LDTs are never seen by the FDA, so if there is an FDA-approved companion diagnostic, and there are four different labs that developed their own version of it, we don’t necessarily know how each LDT performs.”

The proposed phase-in for greater scrutiny of LDTs is nine years, starting with tests that would be considered class III under the FDA review process. “The FDA review may add a bit of additional time, but the pay-off is that we will actually understand how the tests work,” she said.

Dr. Zehnder expressed concern that the additional time will make it more difficult for assays to keep up with the rapid changes in medicine. For instance, companion diagnostics for the KRAS mutation have been approved for codon 12 mutations; while those tests were working through the approval process, more up-to-date data have emerged that suggest that codon 2,3,4 mutations in KRAS and NRAS are also important for guiding treatment strategies, he pointed out.

“It’s relatively quick and easy to update an LDT and use it in the clinic,” he said, “but taking something like an LDT through the FDA process, which is not responsive to new knowledge, is a primary concern that many of us have.”

Labs that develop and use LDTs are generally “rigorously inspected” every two years, Dr. Zehnder added, and must participate in proficiency-testing to ensure test quality and accuracy. “The preponderance of data suggests that this mechanism works fairly well,” he said.

Whatever direction regulation takes, there is still one elephant in the room when it comes to measuring the ultimate value of companion diagnostics and therapeutic benefit – does marrying a genomic report to a therapeutic strategy truly lead to better outcomes?

“That hasn’t been proven conclusively so I’d still call companion diagnostics experimental,” Dr. Ginsburg said. “This is going to require clinical trials to generate outcomes data more rapidly. In some aggressive cancers, it may not take very long to determine if the pairing of companion diagnostics and therapies really made a difference in outcomes, but, in more indolent cancers, this will prove more challenging.”—By Shalmali Pal 


  1. U.S. Food and Drug Administration. “In Vitro Companion Diagnostic Devices: Guidance for Industry and Food and Drug Administration Staff,” August 6, 2014.
  2. Becker Jr. RJ, Mansfield E. Advances in drug development: companion diagnostics. Clin Adv Hematol Oncol. 2010;8:478-479.
  3. U.S. Food and Drug Administration. “List of Cleared or Approved Companion Diagnostic Devices (In Vitro and Imaging Tools).” Accessed June 27, 2016 from
  4. Cheah CY, Burbury K, Apperley JF, et al. Patients with myeloid malignancies bearing PDGFRB fusion genes achieve durable long-term remissions with imatinib. Blood. 2014;123:3574-7.
  5. Vogelstein B, Papdopoulos N, Velculescu VE, et al. Cancer Genomic Landscapes. Science. 2016;339:1546-1559.
  6. Shendure J, Hanlee J. Next-generation DNA sequencing. Nature Biotech. 2008;26:1135-45.
  7. Centers for Medicare & Medicaid Services. “Clinical Laboratory Improvement Amendments (CLIA).” Accessed June 27, 2016 from
  8. U.S. Food and Drug Administration. “Draft Guidance for Industry, Food and Drug Administration Staff, and Clinical Laboratories: Framework for Regulatory Oversight of Laboratory Developed Tests (LDTs),” October 3, 2014.

 TABLE 1. Treatment Inhibitors Used in Treatment of Hematologic Malignancies*
Gene Genetic Alternations Tumor Type Targeted Agent
Receptor Tyrosine Kinase
ALK Mutation, CNV Anaplastic Large Cell Lymphoma Crizotinib
FGFR1 Translocation CML, Myelodysplastic Disorders Imatinib Methylase
FGFR3 Translocation, Mutation Multiple Myeloma PKC412, BIBF-1120
FLT3 CNV AML Lestaurtinib, XL999
PDGFRB Translocation, Mutation CML Sunitinib, Sorafenib, Imatinib, Nilotinib
Non-Receptor Tyrosine Kinase
ABL Translocation (BCR-ABL) CML, AML Dasatinib, Nilotinib, Bosutinib
JAK Mutation (V617F) Translocation CML, Myeloproliferative Disorders Lestaurtinib, INCB018424
ERK1/2 Mutation Mantle Cell Lymphoma, CLL Ibrutinib
Serine-Threonin Kinase
Aurora A and B Kinase CNV Leukemia MK5108
BRAFV600E Mutation LCH, ECD, Hairy Cell Leukemia Vemurafenib (PLX4032)
Polio-Like Kinase Mutation Lymphoma B12356
Non-Kinase Targets
PARP Mutation, CNV Advanced Hematologic Malignancies, CLL, Mantle Cell Lymphoma BMN 673
CD20 Hodgkin Lymphoma Rituximab
CD52 B-Cell Chronic Lymphocytic Lymphoma Alemtuzumab
CD20 Non-Hodgkin Lymphoma Ibritumomab Tiuxetan
Apoptotic Agents
Proton Pump Inhibitors Multiple Myeloma, Mantle Cell Lymphoma, Peripheral T-Cell Lymphoma Bortezomib, Pralatrexate
CNV = copy number variations; CML = chronic myeloid leukemia; AML = acute myeloid leukemia; LCH = Langerhans cell histiocytosis; ECD = Erheim Chester disease*This table does not identify companion diagnostic tests, but rather targets for agents.

Source: National Cancer Institute. “Targeted Cancer Therapies.” Accessed June 27, 2016 from

Companion Diagnostics Used in the Treatment of Hematologic Malignancies

As of July 1, 2016, the U.S. Food and Drug Administration has approved two companion diagnostic tests to help guide the treatment of hematologic malignancies:

  • VYSIS CLL FISH Probe Kit and venetoclax: This test detects 17p13/TP53 deletion in peripheral blood specimens from patients with B-cell chronic lymphocytic leukemia (CLL) to identify patients for whom venetoclax treatment would be appropriate.
  • PDGFRB FISH for Gleevec Eligibility in Myelodysplastic Syndrome / Myeloproliferative Disease (MDS/MPD) and imatinib: This in vitro diagnostic test detects PDGFRB gene rearrangement from fresh bone marrow samples of patients with MDS/MPD to select patients for whom imatinib treatment is being considered.

The FDA’s Center for Devices and Radiological Health has also approved a number of nucleic acid-based tests for hematologic indications, including:

  • Acute myeloid leukemia: Vysis D7S486/CEP 7 FISH Probe Kit and Vysis EGR 1 FISH Probe Kit
  • B-cell chronic lymphocytic leukemia: Vysis CLL FISH Probe Kit and CEP 12 SpectrumOrange Direct Labeled Chromosome Enumeration DNA Probe

Source: U.S. Food and Drug Administration. “List of Cleared or Approved Companion Diagnostic Devices (In Vitro and Imaging Tools).” Accessed June 27, 2016 from

TABLE 2. Nucleic Acid–Based Companion Diagnostic Tests Approved by the FDA

Disease Trade Name Manufacturer Submission
Acute Myeloid Leukemia Vysis D7S486/CEP 7 FISH Probe Kit Abbot Molecular K131508
Vysis EGR1 FISH Probe Kit Abbot Molecular K123951, K091960
B-Cell Chronic Lymphocytic Leukemia Vysis CLL FISH Probe Kit Vysis K100015
CEP 12 SpectrumOrange Direct Labeled Chromosome Enumeration DNA Probe Vysis K962873

Source: U.S. Food and Drug Administration. “Nucleic Acid Based Tests.” Accessed June 27, 2016 from