Download Regulation, Reimbursement, and the Long Road of

VALUE IN HEALTH 16 (2013) S27–S31
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/jval
Regulation, Reimbursement, and the Long Road of Implementation of
Personalized Medicine—A Perspective from the United States
Felix W. Frueh, PhD*
Opus Three LLC, Gaithersburg, MD, USA
AB STR A CT
There is undisputed evidence that personalized medicine, that is, a more
precise assessment of which medical intervention might best serve an
individual patient on the basis of novel technology, such as molecular
profiling, can have a significant impact on clinical outcomes. The field,
however, is still new, and the demonstration of improved effectiveness
compared with standard of care comes at a cost. How can we be sure
that personalized medicine indeed provides a measurable clinical
benefit, that we will be able to afford it, and that we can provide
adequate access? The risk-benefit evaluation that accompanies each
medical decision requires not only good clinical data but also an
assessment of cost and infrastructure needed to provide access to
technology. Several examples from the last decade illustrate which types
of personalized medicines and diagnostic tests are easily being taken up
in clinical practice and which types are more difficult to introduce. And
as regulators and payers in the United States and elsewhere are taking
on personalized medicine, an interesting convergence can be observed:
better, more complete information for both approval and coverage
decisions could be gained from a coordination of regulatory and
reimbursement questions. Health economics and outcomes research
(HEOR) emerges as an approach that can satisfy both needs. Although
HEOR represents a well-established approach to demonstrate the effectiveness of interventions in many areas of medical practice, few HEOR
studies exist in the field of personalized medicine today. It is reasonable
to expect that this will change over the next few years.
Keywords: health economics, outcomes research, personalized
medicine, regulatory, reimbursement.
A brief search in PubMed [1] for “personalized medicine” reveals
10,249 hits, a search for “health economics and outcomes
research” reveals 15,409 hits, and a combination of both search
terms reveals a mere 48 hits. By combining “personalized medicine” with “economics,” one finds 519 articles; by combining
“personalized medicine” with “outcomes,” one finds 989 articles.
Clearly, the intersection of personalized medicine with outcomeor economic-oriented research is poorly investigated. Also, the
increase (as small as the sample size may be) during the last 3
years is interesting: 5 articles in 2009, 6 in 2010, and 12 in 2011
compared with an average of 1 per year for the last decade. Have
we just realized that outcomes and cost matter for the implementation of personalized medicine?
Ten years ago, Lesko and Woodcock [2] described in a seminal
article how the Food and Drug Administration (FDA) envisions
pharmacogenetics to help guide drug development: it was
becoming increasingly apparent that the regulatory body will
get exposed to such information and that a regulatory path needs
to be developed to appropriately and accurately review the data.
Moreover, the FDA, which sees trends in drug development long
before the final products are used in routine clinical practice,
highlighted the advent of a new era in which patients will be
treated and taken care of on the basis of their own molecular
profile. After the release of the final “Guidance for industry:
pharmacogenomic data submissions” [3] in 2004 (the 2003 draft
guidance was extensively discussed and many public comments
have been incorporated in the final guidance), a rapid increase in
pharmacogenetic- and other biomarker-driven drug development
data submitted to the FDA was observed [4,5]. Assuming an
average of 5-year delay from the time of submission to reaching
the market, we indeed arrive at the 2009 upswing of publications
on outcomes and cost in the field of personalized medicine.
Considering that pharmacogenetic information was part of drug
labels for a much longer period of time [6], it is still surprising
that not much emphasis has been put on evaluating changes in
clinical outcomes and determining cost associated with this field.
Several possible explanations for the paucity of such data exist.
Most of the early pharmacologically relevant biomarkers used
in personalized medicine (or “pharmacogenetics”; the term “personalized medicine” was in fact introduced much later) were
pharmacokinetic markers, such as variations in cytochrome P450
(CYP450) enzymes. In those early days, associations between a
marker and a clinical outcome, particularly one that then could
be affected, for example, via dose adjustment, were identified
after the drug had already reached the market (a notable
exception marks Her2/neu, a pharmacodynamic marker that
was essential for the development of trastuzumab introduced
to the US market in 1998). Because these markers were not
discovered within the context of the actual drug development
effort, the nature of the studies demonstrating the potential
Copyright & 2013, International Society for Pharmacoeconomics and
Outcomes Research (ISPOR). Published by Elsevier Inc.
Conflicts of interest: The author has indicated that he has no conflicts of interest with regard to the content of this article.
* Address correspondence to: Felix W. Frueh, Opus Three LLC, Gaithersburg, MD 20878, USA.
E-mail: [email protected].
1098-3015/$36.00 – see front matter Copyright & 2013, International Society for Pharmacoeconomics and Outcomes Research (ISPOR).
Published by Elsevier Inc.
http://dx.doi.org/10.1016/j.jval.2013.06.009
S28
VALUE IN HEALTH 16 (2013) S27–S31
clinical impact of the markers was markedly different from that
of studies requiring a marker as an integral part of drug development (and its potentially required use as a diagnostic to guide
therapy). To the most part, these studies were limited, with a
small n, and not oriented toward hard clinical outcomes, but
rather using soft or surrogate end points such as pharmacokinetic. Therefore, it was difficult to translate these markers into
clinical practice, and in situations in which diagnostics measuring these markers had been developed, uptake was (and continues to be) slow: the lack of convincing studies that focus on
relevant clinical outcomes poses a significant hurdle for the
acceptance of personalized medicine in the clinic. Studies for
newer markers that have been critical or even required for the
(co-)approval of a drug associated with the marker of interest,
however, are more rigorous, and the demonstration of clinical
effectiveness and cost-effectiveness is significantly streamlined
(see below). Therefore, many markers that are often cited in the
context of personalized medicine have not been studied in
pivotal trials, and although exploratory or smaller studies were
conducted and may point toward clinical effectiveness and costeffectiveness, they were not convincing enough to regulators to,
for example, update the label of a drug and requiring the use of a
test or to payers to cover the payment of a test (or even require a
test before authorizing the reimbursement for a drug). Moreover,
in situations in which the FDA took the initiative to update the
label of a particular drug, for example, warfarin [7], clopidogrel
[8], and irinotecan [9], translation into clinical practice occurs
slowly and reimbursement for these tests remains fragmented.
More recently, in addition to tests that are directly associated
with the use of a particular drug, developed either in conjunction
of the drug or later [10], a third category of tests that are not
associated with a particular drug therapy (although they can
inform about appropriate therapies) has emerged. It is interesting
to take a closer look at these three categories of personalized
medicine tests, and the regulatory and reimbursement pattern
they reveal:
1. Tests developed in association with a drug (drug-test codevelopment, e.g., Her2/neu for trastuzumab [11]). This category
of tests benefits from the rigor of studies needed to bring the
drug to the market, which bears several advantages: the
regulatory pathway requires the test and the drug to be
approved at the same time, reimbursement usually follows
in line with the requirement of the test to demonstrate
appropriate use of (or even eligibility to receive) the drug,
and if the drug fails to gain approval, the test is likely not
needed (at least not in this particular context). Moreover, in
clinical practice, there is a significantly lower burden of
informing and educating health care professionals about the
benefit of the test because in this situation the test will likely
be required to gain access to the drug. The onus of demonstrating the impact of the test is not only on the developer of
the test but also on the manufacturer of the drug because of
the vested interest in making the test available. This category
of tests poses the least challenge with respect to demonstrating the effectiveness of a personalized medicine approach: it
is inherent to the product (which is a personalized medicine
product by definition), and clinical effectiveness and costeffectiveness data encompass both drug and test simultaneously. If approved by regulators, products in this category are
also likely to be covered and reimbursed by payers.
2. Tests developed after the drug has reached the market. The
fundamental difference between this and the aforementioned
scenario is the state of clinical practice: although the previous
example establishes clinical practice for both the test and the
drug simultaneously, here the introduction of the test
requires an adjustment or change in established clinical
practice: this is much harder to achieve. Two different types of
tests in this category exist: tests that are developed specifically for one drug product and tests that are of more general
use. For the former, tests are developed as either improvements in existing tests already marketed (e.g., fluorescence
in situ hybridization testing in lieu of immune histochemistry
testing for trastuzumab) or new tests for drugs that were
marketed without the need for a specific test (e.g., HLA-B*5701
testing for abacavir). The latter include tests such as assays for
drug-metabolizing enzymes (e.g., CYP450s) that are relevant
for the use of various drugs [12]. Although all these tests are
developed after the drugs they are useful for, there are
significant differences with respect to the level of evidence
needed for them to be successfully introduced into clinical
practice: clinical utility for tests that follow tests already on
the market has, by definition, already been established. The
characteristics of the new test (in particular sensitivity,
specificity, and cost) determine its performance compared
with that of its predecessor, and it can be judged relatively
easily whether the cost-benefit profile of the new test is
superior to that of the existing assay. For tests for which no
predicate assays exist, this evaluation is more difficult and
includes demonstration of clinical utility. In addition, costeffectiveness becomes a more critical component: the introduction of a new test will add cost and the demonstration that
such additional cost to the system is warranted is necessary.
This can be achieved by a significant improvement in outcomes, by a demonstration of overall savings to the system, or
ideally by both. For example, HLA-B*5701 testing for abacavir
has seen a rapid uptake: the high sensitivity and specificity of
the test [13] enabled the use of abacavir in a much larger
population owing to the ability of the test to detect patients
who are at risk for a severe adverse event: the clinical benefit
of the test clearly justified the additional expense of a test and
likely also compensated for costs associated with the management of the adverse event, which may occur in patients for
which abacavir poses a risk: clinical utility of HLA-B*5701
testing seemed apparent, and cost-effectiveness has been
demonstrated [14] and also put in perspective by others later
on [15]. In contrast, slow uptake was seen in situations in
which the performance of the test was less clear, differences
in outcomes harder to detect, and/or the information derived
from the test more difficult to translate into precise clinical
actions. It is interesting to note that in many of these latter
cases, the clinical outcome—that is, the benefit of testing—
may not be immediately apparent, but observed only over a
period of months or even years (e.g., CYP2C19 testing for
clopidogrel [16,17]) as opposed to a benefit that is much more
rapidly discernible (e.g., HLA-B*5701 testing for abacavir). This
delayed feedback makes it more difficult to design studies
demonstrating the clinical utility of such tests not only due to
long follow-up periods that may be needed but also due to the
larger sample size required to demonstrate the correlation
between the intervention (e.g., dose adjustment and change in
therapy) and a more distant clinical outcome (e.g., prevention
of a secondary event [16]). Consequently, it is also significantly
more challenging to demonstrate cost-effectiveness in these
situations. It is therefore not surprising that the incentives to
develop such tests and to demonstrate their clinical and
economic impact are misaligned with the interest in realizing
an attractive return on investment because studies needed to
demonstrate the clinical and economic impact can take
several years and are costly. In many cases, it is difficult or
even impossible to turn the development cost of such tests
into profits, given the rates at which such tests are usually
reimbursed. There are some situations, however, in which at
least theoretically this approach appears to be more
VALUE IN HEALTH 16 (2013) S27–S31
interesting: for example, if the decision impacts the selection
of an expensive (usually branded) versus a cheaper drug (e.g.,
a generic) or significantly improves the clinical utility of a
generic drug, the development of such tests may be feasible.
This “genetics for generics” approach has been successfully
demonstrated for the use of CYP450 2C9 and VKORC1 testing
for warfarin [18] and is further described elsewhere [19,20].
3. Tests not directly associated with a particular drug. The
effectiveness of these tests is based on the assumption that
the test can add significant information to the clinical decision making without, however, requiring the test for a
particular intervention. Demonstrating clinical effectiveness
can be challenging, but because many of these tests are
addressing questions in clinical areas of high need and high
cost (e.g., OncotypeDx [Genomic Health, Redwood City, CA]
and Mammaprint [Agendia, Amsterdam, The Netherlands] to
determine whether or not radiation therapy is needed for
patients with breast cancer, or Foundation One (Foundation
Medicine, Cambridge, MA) to determine which therapy might
be most appropriate on the basis of the molecular profile of the
cancer), the level of evidence accepted to use the tests in
clinical practice is reduced: for example, it is not unreasonable
to have such tests marketed without evidence derived from a
prospective trial if convincing evidence can be obtained retrospectively (e.g., from information derived from the literature
or with prospective validation of retrospective data) [21,22].
Because these tests are introduced in areas of high medical
need and high costs associated with the treatments (or failure
of treatments), the economic model for these tests is based on
reducing health care costs if the decision about which treatment to choose or the identification of potential risk can be
done early and unnecessary treatment or downstream complications can be prevented. Still, these technologies are new
and the lack of prospective outcome-focused evidence does
present an additional challenge, particularly for reimbursement decisions because payers often label such tests “experimental,” which usually results in a “no-coverage” assessment.
As a result, reimbursement for these technologies has been
challenging (further discussed below) and slow [23].
How can we accelerate generating the evidence that demonstrates clinical effectiveness and cost-effectiveness of these
tests? Although regulators generally focus on the test by itself,
payers want to know how effective the test is compared with
alternative tests (if available) or interventions and preferably so
in the real world, that is, the environment in which reimbursement decisions are actually being made. Ultimately, clinical
outcomes are affected by more factors than the ones included
in most clinical trials and the long-term benefit of an intervention
may depend on criteria that may not be detected (or even
excluded from detection) in trials such as adherence and adverse
events. This puts developers of personalized medicine tests at
crossroads for developing data needed to satisfy regulatory or
reimbursement needs. Could such evidence be generated simultaneously? Even though significantly different, the two divergent
demands overlap on several aspects and strategies to integrate
one with the other are feasible.
The FDA classifies tests (devices) into three categories on the
basis of risk and intended use: class I are low risk, class II
medium risk, and class III high risk. The rigor of review and
evidence needed for regulatory clearance increases with the risk
category of the device. Recently, the agency issued two documents aimed at providing guidance to industry to clarify the
regulatory thinking around the evidence needed for the codevelopment of the drug and the test [24] and to elucidate “Factors to
consider when making benefit-risk determinations in medical
device premarket approval and de novo classifications” [25]. The
S29
latter document is particularly interesting considering that it
states that “Clinical testing methods for medical devices can
include, when appropriate, randomized clinical trials in the
appropriate target population, well-controlled investigations,
partially controlled studies, studies and objective trials without
matched controls, well-documented case histories conducted by
qualified experts, reports of significant human experience, and
testing on clinically derived human specimens (DNA, tissue,
organ and cadaver studies)” (compare section 3.2 in Food and
Drug Administration [25]). This direct citation from the Federal
Register [26] points to the fact that more than one way exists to
create the necessary evidence for regulatory approval of a device.
Importantly, it indicates that randomized trials are not always
needed, which opens a door to real-world, outcome-based
approaches that may indeed be more aligned with the needs of
payers. The Federal Register also mentions that “the evidence
required may vary according to the characteristics of the device,
its conditions of use, the existence and adequacy of warnings and
other restrictions, and the extent of experience with its use” [26].
The goal of reimbursement decision making, however,
depends less on risk but more on real-world clinical outcomes
as well as the cost associated with reaching these outcomes: it
aims at optimizing health outcomes within a defined insured
population, considering all available treatment options and within
the budgetary constraints of the payer. This results in a different
set of parameters being considered by payers compared with
regulators. For example, payers are more interested in comparative effectiveness information, cost offsets, absolute versus relative risk (e.g., a 20% reduction in risk in the entire population is of
greater interest than a 20% reduction in 1% of the population), the
impact on individuals compared with the impact on the population, and a number of other factors as detailed in Epstein et al.
[27]. In addition, workplace productivity, absenteeism, quality of
life, and other indirect cost metrics are considered by different
types of payers (e.g., health plans, self-insured employers, government, and others) with different levels of importance.
So, how can a regulatory focus aimed predominantly on safety
and efficacy be combined with payers’ needs to address realworld effectiveness and cost? Recently, regulators indicated a
heightened interest in comparative effectiveness data and realworld outcomes combined with an effort to increase postmarket
safety data [28]. Considering such interest in light of new
regulatory guidance on generating evidence for medical devices
[25], this emerging regulatory framework creates a path for a
convergence where the “dimension of reality” and the “dimension of comparativeness” meet in a middle area (Fig. 1) in which
efforts that parallel regulatory and reimbursement needs are
feasible. This area can be defined, for example, by real-world
trials that investigate new medical products or interventions
versus any alternative treatment. In the case of personalized
medicine, information may be shifted more toward the effectiveness side where randomized controlled trials may not be required
(or may not be feasible under some circumstances) and casecontrol or cohort studies may be deemed appropriate. This shift
will, however, at least initially, occur on a case-by-case basis and
depends on various factors, including the ones outlined above
such as a rigorous assessment of risk. Nevertheless, it is now
reasonable to generate strategies for gaining regulatory approval
moving from a currently static to an increasingly dynamic
paradigm (Fig. 2), which addresses the needs of payers at the
same time. This can have important consequences for the
implementation of personalized medicine into clinical practice.
In contrast to such a synergistic convergence, some diagnostic
companies started to look for alternative ways to establish the
evidence needed for regulatory as well as reimbursement decisions. This may be driven by the uncertainty about specific
regulation and reimbursement potential and has led to radically
S30
VALUE IN HEALTH 16 (2013) S27–S31
Fig. 1 – Path of Convergence created by the emerging regulatory framework. The dimension of reality and the dimension of
comparativeness meet where alternative treatments are compared with new interventions and hard clinical outcomes are
measured in a rigorous fashion. Reprinted from the Pharmaceutical Forum - Working Group on Relative Effectiveness,
Availability of data to conduct relative effectiveness assessments, with permission from the European Commission.
different business models: for example, the development and
sale of “laboratory developed tests” has become more frequent
owing to the perception of no or limited regulatory oversight for
such tests. In fact, there appears to be a trend toward these types
Assessor
body
Evidence
focus
of tests from not only a regulatory viewpoint but also a reimbursement perspective. It allows manufacturers to get paid more
directly (including via the patient’s insurance), but it often
requires the creation of a billing and reimbursement department
Current paradigm is static
Future paradigm is dynamic
Approval
Approval
Regulators
Payers
Absolute (safety
and) efficacy
Clinical and cost
effectiveness
• Quality, safety,
efficacy
• Benefit-risk profile
• Relative
efficacy/effectiveness
• Cost vs. health benefit
• Budget impact
Regulators
Payers
New relative safety and efficacy assessment,
including real-world effectiveness assessment
• Quality, safety,
efficacy
• Benefit-risk profile
• Cost vs. health benefit
• Budget impact
• Relative efficacy/effectiveness
Studies /
Data
• Emphasis on RCT
• Placebo-controlled
• Active controlled RCT
• Observational studies
• Cost effectiveness /
utility analyses
• Budget impact analyses
• Emphasis on RCT
• Placebo-controlled
• Cost effectiveness /
utility analyses
• Budget impact analysis
• Active controlled RCT
• Adaptive Ph III-IV trials
• Observational studies
• Meta analysis
Fig. 2 – Shift in strategies to generate (clinical) evidence that is useful for both, regulatory and reimbursement needs. Notably,
a shift from static, to a more dynamic paradigm of evidence generation is observed. Data from Eichler et al. [29].
VALUE IN HEALTH 16 (2013) S27–S31
to directly negotiate and manage the reimbursement process
with payers including the management of appeals in the case of a
no-coverage decision. By choosing this route of entering the
market, manufacturers can reach clinical practice faster and are
more likely to stay in control of pricing. Nevertheless, it prevents
the tests from being sold to third parties, which can be a severe
limitation, and not all manufacturers are willing to establish an
operation that includes noncore functions, such as sales force
and billing and reimbursement divisions, that may be needed to
create, expand, and maintain a solid presence in the market.
It is reasonable to assume that in the future, clinical effectiveness and cost-effectiveness questions will be addressed earlier in
the development of personalized medicines. The answer to the
question about where and how studies demonstrating clinical
benefit combined with cost-effectiveness are being conducted is
evolving, and new, innovative approaches are being explored to
combine regulatory and reimbursement needs. The evolving and
still relatively uncertain regulatory environment as well as the
challenges posed by the reimbursement system can negatively
impact the development of new personalized medicine products:
more clarity is needed from regulators and payers explaining
their expectations for data and evidence in approval and coverage decisions. Nevertheless, manufacturers must accept that
without outcomes and economic information, their products will
be difficult to introduce into clinical practice: health economics
and outcomes research will be the key to opening this market.
Source of financial support: ISPOR provided a modest
honorarium.
R EF E R EN CE S
[1] National Center for Biotechnology Information. Available from: http://
www.ncbi.nlm.nih.gov/pubmed. [Accessed October 26, 2012].
[2] Lesko LJ, Woodcock J. Pharmacogenomic-guided drug development:
regulatory perspective. Pharmacogenomics J 2002;2:20–4.
[3] Food and Drug Administration. FDA guidance: pharmacogenomic data
submissions. Available from: http://www.fda.gov/downloads/
RegulatoryInformation/Guidances/ucm126957.pdf. [Accessed October
31, 2012].
[4] Goodsaid FM, Amur S, Aubrecht J, et al. Voluntary exploratory data
submissions to the US FDA and the EMA: experience and impact. Nat
Rev Drug Discov 2010;9:435–45.
[5] Goodsaid F, Frueh FW. Implementing the U.S. FDA guidance on
pharmacogenomic data submissions. Environ Mol Mutagen 2007;48:
354–8.
[6] Frueh FW, Amur S, Mummaneni P, et al. Pharmacogenomic biomarker
information in drug labels approved by the United States Food and
Drug Administration: prevalence of related drug use. Pharmacotherapy
2008;28:992–8.
[7] Access Data. Prescribing information for COUMADIN® tablets. Available
from: http://www.accessdata.fda.gov/drugsatfda_
docs/label/2010/009218s108lbl.pdf. [Accessed October 31, 2012].
[8] Access Data. Highlights of prescribing information for PLAVIX. Available
from: http://www.accessdata.fda.gov/drugsatfda_docs/label/2011/
020839s052lbl.pdf. [Accessed October 31, 2012].
[9] Access Data. Prescribing Information for Camptosar® (irinotecan
hydrochloride) injection. Available from: http://www.accessdata.fda.gov/
drugsatfda_docs/label/2006/020571s030lbl.pdf. [Accessed October 31, 2012].
S31
[10] Towse A, Garrison, LP. Economic incentives for evidence generation:
promoting an efficient path to personalized medicine. Value Health
2013;16(Suppl.):S39–43.
[11] Baselga J, Tripathy D, Mendelsohn J, et al. Phase II study of weekly
intravenous recombinant humanized anti-p185HER2 monoclonal
antibody in patients with HER2/neu-overexpressing metastatic breast
cancer. J Clin Oncol 1996;14:737–44.
[12] Sim SC, Kacevska M, Ingelman-Sundberg M. Pharmacogenomics
of drug-metabolizing enzymes: a recent update on clinical
implications and endogenous effects. Pharmacogenomics 2013;13:
1–11.
[13] Mallal S, Phillips E, Carosi G, et al. HLA-B*5701 screening for
hypersensitivity to abacavir. N Engl J Med 2008;358:568–79.
[14] Hughes DA, Vilar FJ, Ward CC, et al. Cost-effectiveness analysis of HLA
B*5701 genotyping in preventing abacavir hypersensitivity.
Pharmacogenetics 2004;14:335–42.
[15] Schackman BR, Scott CA, Walensky RP, et al. The cost-effectiveness of
HLA-B*5701 genetic screening to guide initial antiretroviral therapy for
HIV. AIDS 2008;22:2025–33.
[16] CAPRIE Steering Committee. A randomised, blinded, trial of clopidogrel
versus aspirin in patients at risk of ischaemic events (CAPRIE). Lancet
1996;348:1329–39.
[17] Ellis KJ, Stouffer GA, McLeod HL, Lee CR. Clopidogrel
pharmacogenomics and risk of inadequate platelet inhibition: US FDA
recommendations. Pharmacogenomics 2009;10:1799–817.
[18] Epstein RS, Moyer TP, Aubert RE, et al. Warfarin genotyping reduces
hospitalization rates results from the MM-WES (Medco-Mayo Warfarin
Effectiveness study). J Am Coll Cardiol 2010;55:2804–12.
[19] Reese ES, Daniel Mullins C, Beitelshees AL, Onukwugha E. Costeffectiveness of cytochrome P450 2C19 genotype screening for selection
of antiplatelet therapy with clopidogrel or prasugrel. Pharmacotherapy
2012;32:323–32.
[20] Genome Web. Medco’s new idea: PGx program for generics to dispense
‘smarter’ treatments and increase savings. Available from: http://www.
genomeweb.com/dxpgx/
medcos-new-idea-pgx-program-generics-dispense-smarter-treatmentsand-increase-sa. [Accessed October 30, 2012].
[21] Cobleigh MA, Tabesh B, Bitterman P, et al. Tumor gene expression and
prognosis in breast cancer patients with 10 or more positive lymph
nodes. Clin Cancer Res 2005;11(24, Pt 1):8623–31.
[22] Cronin M, Sangli C, Liu ML, et al. Analytical validation of the Oncotype
DX genomic diagnostic test for recurrence prognosis and therapeutic
response prediction in node-negative, estrogen receptor-positive breast
cancer. Clin Chem 2007;53:1084–91.
[23] Gustavsen G, Phillips K, Pothier K. The reimbursement landscape for
novel diagnostics: current limitations, real-world impact, and proposed
solutions. Available from: http://www.healthadvances.com/pdf/
novel_diag_reimbursement.pdf. [Accessed August 19, 2013].
[24] Food and Drug Administration. FDA draft guidance: in vitro companion
diagnostic devices. Available from: http://www.fda.gov/downloads/
MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/
UCM262327.pdf. [Accessed October 31, 2012].
[25] Food and Drug Administration. FDA guidance: factors to consider when
making benefit-risk determinations in medical device premarket
approval and de novo classifications. Available from: http://www.fda.
gov/downloads/MedicalDevices/DeviceRegulationandGuidance/
GuidanceDocuments/UCM296379.pdf. [Accessed October 31, 2012].
[26] 21CFR860.7. Code of federal regulations title 21. Available from: http://
www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfCFR/CFRSearch.cfm?
FR=860.7. [Accessed October 31, 2012].
[27] Epstein RS, Frueh FW, Geren D, et al. Payer perspectives on
pharmacogenomics testing and drug development. Pharmacogenomics
2009;10:149–51.
[28] Giffin RB, Woodcock J. Comparative effectiveness research: who will do
the studies? Health Aff (Millwood) 2010;29:2075–81.
[29] Eichler HG, Bloechl-Daum B, Abadie E, et al. Relative efficacy of drugs:
an emerging issue between regulatory agencies and third-party payers.
Nat Rev Drug Discov 2010;9:277–91.