Click here to view this Outlook as an Adobe Acrobat PDF.

No. 3, October 2010

Molecular diagnostics are undergoing increasing scrutiny by regulators as they grow in complexity and clinical importance. The Food and Drug Administration (FDA) has long left its sister agency, the Centers for Medicare and Medicaid Services (CMS), to oversee labs that run the tests. Now the FDA is widely expected to exert new regulatory control over most of these tests. Legislation pending in the Senate, and likely to be incorporated into the reauthorization of the Medical Device User Fee and Modernization Act next year, would do the same. Instead of replacing the current regulatory framework and dictating how tests can be used, the FDA should develop complementary policies that reflect the fact that the tests are only one part of the clinical mosaic used to make medical decisions. If the FDA imposes regulations that are needlessly broad or burdensome, the result could be higher costs, fewer new tests, and additional obstacles to developing safer and more personalized drug-delivery systems.

Key points in this Outlook:

• The FDA plans to impose more regulation on molecular diagnostics, which will create uncertainty and could discourage innovation in personalized medicine.

• Instead, we should encourage reform that uses a staged approval strategy, makes it easier for developers to update their diagnostic tests and pair them with drugs, and does not require costly trials and submissions.

• The most prominent legislation on this issue would allow regulatory requirements to be more closely matched to the perceived risk of the tests.

The FDA recently announced that it will be exerting more regulatory scrutiny over molecular diagnostics.[1] Molecular diagnostics are lab tests that look at individuals' DNA. These tests can measure the expression of individual genes, and they are being used to determine everything from the presence of a disease to the risk of developing one. They can also identify patients likely to benefit from a drug and those most likely to develop a side effect.[2] Molecular diagnostics have become an essential part of clinical practice and of efforts to improve the development and prescription of drugs to make treatments safer and more targeted.

The diagnostic field has undergone rapid innovation over the last decade, in part because labs were able to incorporate new information into tests quickly and offer them to patients. The FDA wants to regulate the vast majority of these molecular diagnostics as medical devices, regardless of who is performing the tests and how they are being sold to doctors and patients. Yet when it comes to safety, the industry has a fairly good track record. If the FDA steps in to impose regulations that are too costly, time consuming, or burdensome, it could stall much of the investment and entrepreneurship that have fueled recent innovations. The regulatory regime needs to be matched to the risks and mindful of the opportunities that hang in the balance.

The strategy that the FDA is considering for increasing its oversight of these tests is based on a template the agency first drafted almost ten years ago. Yet the science for developing and validating the tests has evolved measurably since then. There are opportunities for the FDA to leverage advances in science and test development to create a more flexible, more modern, and less costly regulatory pathway. These strategies will be discussed in the second half of this Outlook.

The FDA's Current and Anticipated Role

Long at issue for the FDA has been a subset of molecular diagnostics that are developed and performed by independent labs, known as "laboratory-developed tests" (LDTs). The FDA recognizes that a growing proportion of important medical decisions are driven by these increasingly sophisticated tests, many of which undergo no formal FDA premarket review.

The LDTs are sometimes referred to by the more pejorative term "homebrews." Companies that develop LDTs are typically operating as labs, performing the tests on behalf of doctors (and sometimes patients) who ship blood and tissue samples to them. As such, the lab companies have long argued that they are selling a lab "service," rather than a medical device. Thus, they contend that the FDA, which regulates medical devices for safety and effectiveness, has no jurisdiction over them. Instead, the labs are already subject to oversight by the Clinical Laboratory Improvement Amendments (CLIA) of 1988, a set of federal regulatory standards enforced by CMS inspectors.[3] These CLIA standards apply to any clinical lab operating in the United States.[4]

CLIA was enacted in 1967 to regulate labs engaging in interstate commerce as well as those based in hospitals. A 1988 amendment extended CLIA's oversight to most labs regardless of their location and gave CMS the authority to require that labs have quality systems in place to ensure the accuracy, reliability, and timeliness of test results.[5] Today, all clinical labs that provide information to patients and doctors must be certified by CMS and comply with the requirements of CLIA. These regulations address the labs themselves, such as the training of personnel, not the tests.

CLIA is not the only regulatory regime overseeing LDTs; the FDA already plays a significant role. But FDA regulation of LDTs is mostly confined to the reagents used in tests, under the terms of the FDA's Analyte Specific Reagent (ASR) rule.[6] Analytes (reagents) are the substrates in a diagnostic test that bind to a particular target to help identify it. But the vast majority of labs, using these reagents in tests they developed inside their own facilities, need only validate the accuracy of the resulting diagnostic tests and disclose in a test report that FDA approval was not required for the diagnostic itself.

Under the ASR rule, the analytes can only be sold to manufacturers of in vitro diagnostic devices (IVDs), labs certified under CLIA for high-complexity testing, and organizations that use the tests for research or other nondiagnostic purposes. Labs using ASRs in their own tests must comply with CLIA quality-control requirements. In this way, the FDA's ASR rule and the CLIA requirements are highly complementary. But the FDA does not regulate the tests that labs develop in their own facilities using the reagents--the LDTs overseen by CLIA. The labs often state that the tests are for "research purposes only" to avoid the kind of implied treatment claims that could invite FDA scrutiny. Still, the FDA knows full well that many of these tests are being used for clinical purposes anyway.

Molecular diagnostics become subject to FDA regulation when they are packaged as a "kit" to be sold to other labs, with all the components needed to run them, such as analytes and instructions. In these cases, the FDA reviews the entire kit as an IVD. The IVD must undergo premarket review to demonstrate that it meets certain standards for reliability. This usually involves a 510K submission, which requires test developers to meet various technical specifications. Sometimes IVDs must meet a higher standard that requires labs to demonstrate how they improve clinical decision making, usually through the submission of a premarket approval application (PMA). But here again, the FDA has been irked by companies that sell most of the pieces of a test as an assembled package, but try to skirt regulation by leaving out some of the components that would formally make it a "kit" subject to FDA oversight.

In summary, despite the existing regulatory framework, the FDA believes there are gaps in oversight. While the FDA currently regulates many of the components used in assembling LDTs, it does not review many of the finished tests. Of the tens of thousands of molecular tests clinically available, the FDA has reviewed only a few hundred. The FDA fears that unreliable tests could be generating false results and leading patients to erroneous decisions and potentially bad medical outcomes. Moreover, the FDA believes that regulation is not being determined by a test's parameters, risks, or intended use, but merely by how it is marketed. Consumer groups have lobbied the agency to exert more oversight.[7] Now, with legislation pending in Congress that would increase FDA jurisdiction over these tests, the FDA is on the cusp of gaining long-sought leverage.

A New Regulatory Paradigm Takes Shape

Regulators generally evaluate diagnostics according to three parameters: their analytic validity, clinical validity, and clinical utility. The accuracy of a molecular diagnostic, as with any lab test, depends on its analytic validity: whether the test is consistently able to measure the specific gene or protein that it is designed to evaluate. A test's reliability is a function of its clinical validity, which means that the test results consistently and accurately predict the clinical target or outcome that the manufacturer claims.[8] The analytic and clinical validity of a diagnostic depend on the intrinsic properties of the test and the abilities of the lab performing it.

The final parameter on which FDA regulators sometimes evaluate a diagnostic test is its clinical utility. This refers to whether or not using the test has a positive impact on clinical outcomes. Evidence demonstrating clinical utility can take years to establish. CLIA generally requires analytic validity, and its review process for labs ensures clinical validity. But CLIA does not require that tests demonstrate clinical utility; doctors and patients are left to determine whether a test improves their decision making.

This paradigm is what stands to undergo the most change if the FDA exerts greater purview over these tests. Greater FDA regulation of molecular diagnostics would provide more oversight of a test's analytic and clinical validity. But it would also require premarket data to demonstrate the clinical utility of tests being marketed to influence treatment decisions, especially the decision to use certain drugs. The lack of assurance that a test has clinical utility causes some of the greatest concern inside the FDA.

The Origins of FDA Regulation

Risk-Based Regulation. The FDA's efforts to regulate LDTs started in the mid-1990s. The agency began to make a series of policy statements showing that it viewed labs as manufacturers and the LDTs themselves as medical devices, not services.[9] The FDA said it intended to require some measure of premarket review for these tests.

These efforts picked up momentum in early 2000, when the FDA began to publicly contemplate a comprehensive, risk-based regime for overseeing the entire field of LDTs. Around 2002, the FDA's Office of In Vitro Diagnostics--which regulates diagnostic tests inside the FDA's Center for Devices and Radiological Health--released a first draft of that policy. It was a regulatory scheme that divided the universe of LDTs into three categories based on their potential for "risk," which would depend, in part, on the claims being made about a diagnostic test's intended use.

Under the proposed framework, low-risk tests (those that tested for benign conditions where there were little or no consequences from a false positive or false negative outcome) would be exempt from premarket review. Intermediate-risk tests would require a 510K clearance, which would likely include some clinical evidence to verify their clinical validity and, in some cases, clinical utility. This might include a test that made a prognostic claim, such as predicting the likelihood or severity of disease, rather than a treatment claim, such as selecting a particular medicine. For example, a test that identified a patient as having a more aggressive form of breast cancer without recommending a particular course of treatment might require a 510K application. Under a 510K, the clinical data could come from retrospective studies that used banked tissue samples; labs would not be required to run new, prospective clinical trials on patients. Finally, many high-risk tests (those that recommend a particular course of treatment, for example) would require a PMA, the most stringent level of oversight--for example, a test that not only identified a patient with a more aggressive form of breast cancer, but also recommended a particular form of chemotherapy.

Yet precisely which tests would require a PMA remains an open question. PMAs would certainly be required for many of the LDTs that propose to guide treatment decisions, such as those that recommend a certain cancer drug. For these tests to come to market, the FDA would require a clinical trial in which some patients were randomized to treatment based on the results of the diagnostic test, while other patients were assigned treatment through traditional clinical decision making. The FDA would only approve such a diagnostic test if it improved the assignment of patients to a drug and improved outcomes over conventional decision making (in other words, had clinical utility). But this is an expensive and lengthy hurdle; it could cost $10 million to $50 million. This is often financially unfeasible for a low-margin, generally low-profit diagnostic test.

Nonetheless, the leadership in the Office of In Vitro Diagnostics began to preview its proposed regulatory framework at public meetings, creating angst in the diagnostics industry. Ultimately, the policy was stymied by concerns inside the FDA that the agency lacked the clear legal authority to regulate LDTs. The FDA was torn over whether LDTs met the definition of a medical device, thus making them subject to FDA review, or were indeed a service exempt from FDA regulation. The FDA was also concerned that it did not have sufficient resources to take on broader regulation of these tests. Moreover, there was no clear evidence that existing LDTs were flawed or that CLIA was not fulfilling the regulatory goals that the FDA sought to pursue.

Rather than expanding its reach through new regulation, the FDA decided to try to use its existing rules in a more robust fashion to provide tighter oversight of the industry. This led to a series of "untitled" warning letters to some diagnostic-test makers whose products and claims put them outside the agency's existing regulations.[10]

Incremental Regulation. By 2006, however, the agency's point of view changed. A consensus formed inside the FDA that LDTs indeed met the definition of a medical device, regardless of whether the companies were marketing them as laboratory services.[11] Meanwhile, bipartisan legislation was introduced by the late Senator Edward Kennedy (D-Mass.) that would have classified LDTs as medical devices and made them subject to FDA regulation. While the FDA did not change its legal interpretation simply to preclude the legislation, the agency's new position made the Kennedy bill a moot point. The bill was never introduced to the full Senate. The question turned instead to how the FDA would approach its new, self-ascribed mandate.

The FDA decided to take on regulation of LDTs in an incremental fashion, starting with the tests that it felt both posed the most significant public-health risk and were sufficiently complex that they exhausted the capabilities and assumptions of the CLIA regulatory framework. The CLIA framework assumed that tests were based on fairly straightforward technology. The FDA intended to demonstrate its ability to apply oversight to this narrow segment of tests, and then gradually expand its jurisdiction over other tests. The agency recognized that inserting itself in the entire field of LDTs would have caused a lot of uncertainty regarding which tests would ultimately fall under its purview. Some worried that excess uncertainty could stall investment until the FDA worked out how it would draw lines around different categories of tests, a process that could take years.

So the FDA defined a category of tests that had not been previously segmented from the larger group of LDTs and dubbed them "in vitro diagnostic multivariate index assays" (IVDMIAs). The IVDMIAs were diagnostics that often tested for dozens of different molecular markers simultaneously and then generated a single clinical recommendation by aggregating the different outputs from each marker. In this way, the results were based on previous experimentation; the patterns of gene expression that these tests measured were not meaningful unless correlated with previous experiments on known patient samples, at which point they became useful in guiding future clinical decisions.

In this respect, the IVDMIAs had two features that the FDA believed put them squarely outside the capacity or assumptions of the CLIA framework. First, doctors using them could not independently decouple the results of the tests. Because the tests generated a clinical recommendation based on the collective measurement of dozens or hundreds of different genetic targets, doctors could not look behind all the markers that drove the clinical recommendation. They had to rely on the overall output, which itself was based on the results of the previous experiments that were typically not available to them.

Second, the test results were based on previous experiments, in many cases done on banked tissue samples held by the test developer. This meant that a CLIA lab performing the same test could not independently verify its reliability, unless it could get access to the samples used in the initial validation of the test. Finally, while CLIA oversaw how well the lab performed the test, it did not assess the development of the test. So if that previous experimentation were faulty, CLIA could not uncover this flaw. The FDA knew of at least one instance in which that validation had gone awry, undermining the reliability of a prominent IVDMIA that was intended to help doctors guide cancer-treatment decisions. Although the error was uncovered before that test was commercialized, this episode stirred the agency's concerns. So in 2006, the FDA issued a draft guidance document explaining how it proposed to regulate the IVDMIA tests.[12]

Comprehensive Regulation. But in summer 2010, that policy was shelved, largely because the FDA seems to have rejected the incremental approach to regulating LDTs in favor of a comprehensive solution. The FDA's recent decision to rescind the IVDMIA guidance--and hold a public meeting to discuss its approach to LDTs--sends a clear message that it is reverting to its 2002 posture. Sometime next year, the FDA intends to craft regulations or guidance that will address LDTs in one sweep, while trying to establish boundaries as to when it will apply its various premarket requirements to each category of tests.

The resulting scheme, like the FDA's 2002 draft proposals, is likely to segment LDTs into fairly broad categories based on their intended use and their underlying technology. In many cases, FDA regulation will only mean applying Quality System Regulations (QSRs) to the labs. QSRs require labs to meet certain standards and specifications. But in other cases, especially for more sophisticated tests or tests being used to guide treatment decisions involving other regulated therapeutics (mainly prescription drugs), the FDA will require clinical data for premarket review to validate reliability and clinical utility. While aiming at a comprehensive scheme that sweeps in the vast majority of LDTs, the FDA has given no indication as to how its new policy can be more granular than its 2002 effort.

The result will be hazy guidelines assigning different tests to different regulatory buckets. Uncertainty, along with the increased regulatory costs and time involved in satisfying the PMA process, will discourage investors from making new bets on diagnostics. It will also keep test developers guessing about how they will need to develop and validate new tests. Diagnostic tests are generally lower-margin medical products; they do not earn the kinds of high returns that a successful drug generates. If the costs of developing diagnostics continue to rise, and if the FDA regulates diagnostics using a scheme meant to regulate drugs, investing in many of these endeavors simply becomes unviable.

In addition, the FDA will likely require that diagnostics used in guiding prescribing decisions be validated by prospective, randomized trials. These trials would form the basis of a PMA-type application to the FDA. It could take sponsors two years or more to complete the necessary clinical trials and then more than a year for the FDA to review an application. These trials cost at least $10 million, and often much more.

The requirement for a PMA trial will also lead to a tightly controlled development process. It will push tests into the hands of a single lab or platform rather than favor the development of a test kit that can be distributed widely and run in multiple labs--inadvertently creating testing monopolies. This precludes the kind of broad distribution of a test that supports its use by multiple labs and its distribution by many different lab sales forces. The market reality is that, in many cases, doctors require training on how to incorporate a new diagnostic test into their clinical practice. By requiring a PMA trial, the FDA will stifle the market framework that helps drive education and the adoption of tests among clinicians.

While the FDA has long wanted to assert greater jurisdiction over molecular diagnostics, the agency was nonetheless cognizant of its narrow authority, its limited resources, and the lack of political support for placing new burdens on the lab industry. Now things are different. The FDA's resources (and perhaps also its legal authority) remain in doubt, but the agency's internal resolve, and the political support it has for taking on a more active role, are no longer in question.

Where Should the FDA Play a Role?

The FDA's public-health interest in asserting more oversight over some LDTs is not without merit. But that does not mean that CLIA cannot continue to serve a meaningful purpose, or that FDA regulation is broadly required. There are steps the FDA can take to lessen the potentially harmful impact of its evolving policies on continued innovation in the diagnostics industry.

The pace of development and innovation in the industry has proceeded quickly largely because test developers were free to adapt their tests without the need for lengthy and expensive FDA review. This fact should not be lost on policymakers. Indeed, there have already been cases in which FDA requirements delayed tests from incorporating the most up-to-date information, rendering them commercially unviable and, worse still, clinically obsolete.

This happened with a diagnostic test used for profiling the genome of HIV to determine the resistance that a particular strain of the virus has to the usual complement of AIDS medications. A number of years ago, the company Visible Genetics came in for full FDA approval of their HIV genotyping test. Yet their test quickly became obsolete as a result of FDA regulation. Doctors knew that because the company had to get a supplemental approval from the FDA each time it wanted to adapt its collection of gene markers for changes in HIV-resistance patterns, the Visible Genetics test was not kept up to date. Meanwhile, most of these tests were also available as LDTs, which were not subject to the same regulation and could be regularly adapted as the HIV genome mutated. Visible Genetics had paid more than $5 million for a PMA trial--expecting labs, under pressure from the FDA, to switch from their LDTs to its approved kit. But that never happened. Labs favored the LDTs because they were not only more current but also cheaper, since they lacked the high development costs of the Visible Genetics test.[13]

Use Retrospective Trials to Validate Tests. Recognizing the barriers caused by new regulation, the FDA should focus on the areas where its resources can have the most public-health impact. The FDA can address many of its public-health concerns by improving the science for developing diagnostic tests and providing sponsors with regulatory incentives to invest in the kind of validation strategies that the FDA wants them to pursue--the creation of policy carrots rather than regulatory sticks. Instead of requiring prospective, randomized trials and PMA submissions, tests can be validated in retrospective (backward-looking) trials using tissues banked from known patient samples. This kind of regulatory approach would impose a shorter and less-costly burden on manufacturers.

However, one impediment to this proposal is that the FDA does not trust the reliability of banked tissue samples. The FDA has some valid concerns; samples are sometimes poorly collected and stored. But the FDA could help improve the quality and reliability of these "bio-banks" by issuing voluntary standards on how tissues should be collected and stored if they are going to be used as the basis of regulatory filings. Then industry members might collaborate on the development of shared bio-banks. The FDA should be willing to rely on these kinds of retrospective studies if it gained better assurance that the banked tissues were collected reliably.

Develop FDA Regulations That Complement CLIA. Any new FDA policy should also be developed complementary to efforts that reinvigorate CLIA. In March 2007, then-senator Barack Obama, along with Senator Richard Burr (R-N.C.), cosponsored the Genomics and Personalized Medicine Act (S 976). The legislation was a broad initiative that, among other provisions, created a CLIA specialty area for molecular and genetics clinical tests. Instead of supplanting CLIA with the FDA, the bill called on the Institute of Medicine to conduct a study looking at how FDA regulations could be leveraged to complement the current CLIA regime. Such a complementary and comprehensive approach should form the foundation of the FDA's new effort.

Make Updating Tests Easier. Furthermore, the FDA should create a flexible framework for manufacturers to update test markers as new information becomes available. This is especially true for multivariate tests. The FDA argues that it could use "bridging" studies, allowing diagnostic makers to file shorter applications when they want to upgrade panels of markers rather than requiring a full revalidation of their tests. But the FDA has not explained how this would work. Moreover, when challenged by innovators, the FDA has typically required lengthy new trials for tests that update panels of markers, despite its pledge to allow less-burdensome bridging studies.

Offer Incentives for Tests Paired with Drugs. The FDA should also offer new incentives for developing diagnostic tests paired with particular drugs, which would enable more effective prescribing and offset some of the burden created by additional regulation. Most of these drug-diagnostic combinations are being developed for oncology, so cancer should be a focus of policymaking. One idea would be to allow drug developers to apply for accelerated approval in frontline (early-stage) cancer uses by showing superior efficacy over available therapy in a subset of frontline patients. The frontline patients would be chosen based on their expression of a marker revealed by the diagnostic test.

Use a Staged Approval Strategy. Finally, the FDA should implement a staged approval strategy, in which the full validation of the diagnostic (specifically, the components of the diagnostic that deal with manufacturing and sales, as opposed to the clinical aspects) can take place after the drug is approved. This is important because it would allow drug developers to pursue strategies that link the prescribing of their drug to the identification of a marker that selects patients more likely to respond to the treatment. Right now, many drug developers are worried that the increasingly onerous and uncertain regulation of diagnostic tests can slow down the drug review when they try to couple prescribing their new drugs with a diagnostic, so they shun these promising drug-development strategies altogether. In a staged approval, the diagnostic could still be incorporated as a component of the drug label while it continues to undergo additional validation after the drug is approved.

It is noteworthy that the Europeans have a regulatory approach to molecular tests that is much closer to the CLIA process than the new scheme contemplated by the FDA. The European IVD directive requires labs performing tests to meet certain technical specifications. In addition, IVD manufacturers are obligated to systematically review the experience gained from their IVDs on the market and report problems to regulators.[14] This more flexible regulation has its own practical benefits. For example, European regulators can more quickly incorporate important new diagnostic tests into treatment paradigms.

One widely cited example is a test used to guide the prescription of a new class of colon-cancer drugs, based on a molecular marker found in tumors that helps identify patients who are most likely to benefit from the drugs. In this case, the European drug regulators incorporated the test into the drug labeling and treatment guidelines almost a year before the FDA grudgingly accepted the diagnostic test. In large measure, the FDA was reluctant to adopt the test because it did not regulate it. In other words, the FDA presumed that the molecular test (which is a relatively straightforward diagnostic) was unreliable because it was being performed in labs regulated by CLIA. The data on which the test was validated were also largely derived from retrospective analyses of large clinical trials that examined the outcomes of patients with and without the marker, rather than from the FDA's preferred construct: clinical trials in which patients are randomized to treatment based on the diagnostic and which are conducted specifically to validate the test.

Implementing Change through Legislation

The way to change LDT regulation may be the reauthorization of the Medical Device User Fee and Modernization Act next year, rather than the FDA's internal policy process. Legislation is likely to get coupled to the act that would give the FDA new authority over LDTs.

The bill getting the most attention is the Better Evaluation and Treatment through Essential Regulatory Reform for Patient Care Act, under development by Senator Orrin Hatch (R-Utah). It would create inside the FDA a new Center for Advanced Diagnostics Evaluation and Research that would be responsible for ensuring the safety and efficacy of a new category of tests called "advanced personalized diagnostics" (APDx), comprising both test kits and LDTs. The bill defines an APDx as a regulated product "distinct from a device" that is used for diagnosing or treating disease and provides analysis of DNA, RNA, a chromosome, a protein, or a metabolite. This includes components of a diagnostic needed to perform a test and generate test results, such as instruments, reagents, and the lab-report template. However, the APDx category would not include lab services, test reports to individual customers, or consultations by a lab to treating doctors.[15]

The Hatch bill would provide a basic level of oversight for all APDxs by requiring sponsors of such tests to register their products in a database managed by the new FDA center. Similar to the FDA's earlier proposal, tests would be slotted into one of three categories, based on whether they have a low, moderate, or high health impact. The FDA would require premarket clearance for high-risk tests. But instead of submitting a 510K clearance or PMA, developers would submit a "premarket claim statement" that lays out the category and clinical features of the test. In this way, the regulatory regime would be more clearly mapped to the unique claims made by diagnostic tests, reducing uncertainty about where tests would fall within different regulatory buckets. CMS would still enforce CLIA to certify lab services, operations, and test performances of LDTs, and APDx developers that market some types of diagnostics would still need their lab to be accredited under CLIA.

The legislation, by all accounts, appears promising. Among other strengths, it would recognize the uniqueness of diagnostic tests. Rather than try to superimpose the existing device regulatory scheme on diagnostic tests, it would establish a new framework and allow regulatory requirements to be more closely matched to the perceived risk of these LDT products.

In the end, doctors make clinical decisions all the time on the basis of instruments that provide only a limited degree of precision, from interpretations of an MRI to listening to a heart with a stethoscope. It is reasonable to expect a molecular diagnostic to be analytically valid, but there is no reason why an LDT should be held to a much higher standard--clinical utility--than the many other tests that doctors perform.

The FDA has often preferred to expand its regulatory reach broadly and then exercise its enforcement discretion to exempt certain product categories. That seems to be its approach toward regulating molecular diagnostics. But this approach to regulation always creates maximal uncertainty, discouraging investment and scientific risk taking--which the FDA has failed to recognize. The question is how much innovation we are willing to trade in exchange for greater certainty about the performance of the resulting tests. We need to design regulatory approaches to ensure that the tests are reliable while not overburdening those seeking to develop these important new technologies.

Click here to view this Outlook as an Adobe Acrobat PDF.

Notes

1. U.S. Food and Drug Administration, "FDA/CDRH Public Meeting: Oversight of Laboratory Developed Tests (LDTs), Date July 19-20, 2010," available at www.fda.gov/MedicalDevices/NewsEvents/WorkshopsConferences/ucm212830.htm (accessed September 30, 2010).

2. Many unpredictable, "idiosyncratic" drug reactions are probably based, in part, on rare immune responses to drugs that we do not fully understand. Gene and expression testing can help stratify patients based on their risk for these problems. For example, an immune-system gene called HLA-B 1502 is linked to dangerous skin reactions following treatment with a certain antiseizure medicine (carbamazepine). Tests for that gene have improved drug safety by identifying patients at risk for the rare, sometimes-fatal side effect.

3. Inside CMS, CLIA is implemented by the Center for Medicaid and State Operations, Survey and Certification Group, Division of Laboratory Services. The program is funded by user fees collected from labs.

4. Centers for Medicare and Medicaid Services, "Clinical Laboratory Improvement Amendments (CLIA) Overview," available at www.cms.hhs.gov/clia (accessed September 30, 2010). CLIA defines a clinical lab as any facility that performs lab testing on specimens derived from humans for the purpose of providing information for the "diagnosis, prevention, or treatment of disease or impairment, and for the assessment of health."

5. Patrick A. Rivers, Aram Dobalian, and Francesco A. Germinario, "A Review and Analysis of the Clinical Laboratory Improvement Amendment of 1988: Compliance Plans and Enforcement Policy," Health Care Management Review 30, no. 2 (April-June 2005): 93-102.

6. Classification information for ASRs can be found in Code of Federal Regulations title 21, §864.4020(a). They can be antibodies, specific receptor proteins, or even sequences of DNA. By binding to or reacting with microscopic elements of a blood or tissue sample, the analyte allows a lab test to identify or measure a particular molecular target. Most ASRs do not require premarket review. The FDA requires mandatory preapproval only for ASRs used to diagnose contagious diseases such as HIV or tuberculosis, or tests used to screen donated blood. Recently, the FDA also started to classify ASRs as diagnostics subject to its regulation when the ASRs are combined with specific performance claims. Additionally, another recent FDA policy change requires that genetic tests using ASRs also provide premarket notification.

7. In April 2010, the Genetics and Public Policy Center at Johns Hopkins University, the Coalition for 21st Century Medicine (a group representing genetic-testing companies), and more than one hundred patient and advocacy groups, genetic-test manufacturers, and other parties sent a joint letter to Department of Health and Human Services Secretary Kathleen Sebelius, calling on her to revamp regulation of genetic tests and arguing that "accurate, reliable, and timely advanced diagnostics offer enormous promise, but poor quality tests can harm patients and waste scarce resources."

8. A definition of analytic validity, clinical validity, and clinical utility can be found at the Centers for Disease Control, "Genomic Translation," available at www.cdc.gov/genomics/gtesting/ACCE (accessed September 30, 2010).

9. Lawrence M. Killingsworth, "FDA Ignores CLIA and Moves to Regulate Clinical Laboratories," Clinical Chemistry 39, no. 2 (1993): 179-80.

10. "FDA Tells Roche Its New Diagnostic Cannot Bypass Premarket Approval," Journal of Clinical Engineering 29, no. 1 (2004): 21-22.

11. That definition stated that a device was, among other things, any product that was "intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease."

12. A copy of the guidance can be found at U.S. Food and Drug Administration, "Draft Guidance for Industry, Clinical Laboratories, and FDA Staff--In Vitro Diagnostic Multivariate Index Assays," available at www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm079148.htm (accessed September 30, 2010).

13. Wendy Diller, "In Diagnostics, Is FDA Approval Superfluous?" In Vivo 21, no. 4 (April 2003).

14. "Directive 98/79/EC of the European Parliament and of the Council of 27 October 1998 on In Vitro Diagnostic Medical Devices," Official Journal of the European Communities L 331 (1998): 1-37.

15. Turna Ray, "Draft Bill Proposes FDA Create New Division to Review ‘Advanced Personalized Diagnostics,'" Pharmacogenomics Reporter, June 23, 2010.