The Value of Personalizing Medicine: Medical Oncologists’ Views on Gene Expression Profiling in Breast Cancer Treatment

This study explored medical oncologists’ views of gene-expression profiling (GEP) and factors impacting its use in clinical practice. Oncologists valued GEP as a treatment-decision support tool despite concerns about its cost and over-reliance, overuse, and inappropriate use by other oncologists, as well as patients’ limited understanding of GEP. The results identify a need for decision aids to support patients’ understanding and clinical practice guidelines to facilitate standardized use of the test.

Abstract

Objectives.

Guidelines recommend gene-expression profiling (GEP) tests to identify early-stage breast cancer patients who may benefit from chemotherapy. However, variation exists in oncologists’ use of GEP. We explored medical oncologists’ views of GEP tests and factors impacting its use in clinical practice.

Methods.

We used a qualitative design, comprising telephone interviews with medical oncologists (n = 14; 10 academic, 4 in the community) recruited through oncology clinics, professional advertisements, and referrals. Interviews were analyzed for anticipated and emergent themes using the constant comparative method including searches for disconfirming evidence.

Results.

Some oncologists considered GEP to be a tool that enhanced confidence in their established approach to risk assessments, whereas others described it as “critical” to resolving their uncertainty about whether to recommend chemotherapy. Some community oncologists also valued the test in interpreting what they considered variable practice and accuracy across pathology reports and testing facilities. However, concerns were also raised about GEP’s cost, overuse, inappropriate use, and over-reliance on the results within the medical community. In addition, although many oncologists said it was simple to explain the test to patients, paradoxically, they remained uncertain about patients’ understanding of the test results and their treatment implications.

Conclusion.

Oncologists valued the test as a treatment-decision support tool despite their concerns about its cost, over-reliance, overuse, and inappropriate use by other oncologists, as well as patients’ limited understanding of GEP. The results identify a need for decision aids to support patients’ understanding and clinical practice guidelines to facilitate standardized use of the test.

Implications for Practice:

Oncologists value gene-expression profiling (GEP) as a treatment decision support tool, despite expressing concerns about its cost, over-reliance, overuse, and inappropriate use by other oncologists. Although many oncologists said it was simple to explain GEP to patients, paradoxically, they remained uncertain about patients' understanding of the test results and their treatment implications. These results identify a need for decision aids to support patients' understanding and clinical practice guidelines to facilitate standardized use of GEP tests.

Introduction

Approximately 22,600 new cases of lymph node-negative (LN−), estrogen receptor-positive (ER+) breast cancer are diagnosed each year in Canada, and 100,000 are diagnosed in the United States [1, 2]. Adjuvant treatment for these patients may include chemotherapy, hormonal therapy, combined chemotherapy and hormonal therapy, or observation alone. Historically, the decision to treat ER+, LN− breast cancer patients with adjuvant chemotherapy has been guided by clinical and pathological factors along with clinician and patient preferences [3]. Clinical practice guidelines recommend adjuvant chemotherapy for the majority of women with tumors more than 1 cm in size to improve disease-free and overall survival [3]. However, many of these women would remain disease-free as ∼15% of patients have distant recurrence within 5 years without chemotherapy [4]. This suggests that ∼85% of these patients may be overtreated and exposed to toxicity with little or no clinical benefit [5]. More sensitive predictive factors are needed to refine treatment selection and mitigate risk for patients unlikely to derive benefit from adjuvant chemotherapy [6].

Gene-expression profiling (GEP) is intended to meet this need. Oncotype DX, for example, has been validated as an independent prognostic measure of the risk of recurrence in tamoxifen-treated patients with LN−, ER+ breast cancer [7, 8] and as predictive of the magnitude of chemotherapy benefit [9]. Although Oncotype DX is the most commonly used GEP test, there are others, such as MammaPrint, Prosigna, and the 50-gene PAM50 test. Recent studies have described the prognostic similarity of standard immunohistochemical assays to Oncotype DX and comparison with the PAM50 test [10, 11]. However, the validation studies for Prosigna and the PAM50 test were based on endocrine-treated postmenopausal women only, whereas Oncotype DX has been found to be prognostic and predictive of chemotherapy benefit in pre- and postmenopausal patients [12]. Despite these variable levels of evidence to support GEP tests’ predictive and prognostic validity [5, 13–15] and their statistical issues [16], guidelines recommend GEP tests for clinical practice as a complement to conventional risk recurrence assessments [17–19].

GEP tests are increasingly being used in clinical practice. Studies demonstrate that oncologists use GEP results to determine when and how much chemotherapy to administer [20–23]. For example, results from GEP testing have been associated with lower chemotherapy administration in one prospective study [22]. Similarly, a small retrospective analysis showed that GEP results changed treatment recommendations in 14 of 68 patients, with half of these changes from chemotherapy to hormonal therapy alone [21]. GEP results changed treatment recommendations and actual adjuvant therapy received, consistent with the recurrence risk score suggested by the GEP test [21]. Another study found that GEP results changed treatment decisions in 31% of patients, with more changes made against than for chemotherapy.

These studies demonstrate the impact of GEP tests on clinical practice. However, variation exists in the use of GEP results among oncologists, and factors influencing this variation have not yet been explored. We sought to explore medical oncologists’ views of GEP tests and the factors impacting their use in clinical practice.

Materials and Methods

Study Design

This qualitative study, along with a discrete choice experiment survey that aims to estimate the utility of the GEP test relative to other factors, is part of a mixed methods study to examine the perceived value of GEP tests among breast cancer patients and medical oncologists. The Research Ethics Boards at St. Joseph’s Hospital, Sunnybrook Health Sciences Centre, and Princess Margaret Hospital approved the study.

Sample Recruitment

We recruited medical oncologists through two academic oncology clinics (Princess Margaret Cancer Centre and Sunnybrook Health Sciences Centre), advertisements on professional societies’ websites, and referrals from the research team. Those practicing in community hospitals were recruited through referrals from the research team.

Data Collection

We conducted telephone interviews with oncologists using a semistructured discussion guide based on literature review and clinical consultation. Interviews explored oncologists’ use of and reservations about GEP in treatment decision making. Demographic data were collected before the interviews.

Data Analysis

Interview data were digitally recorded for verbatim transcription. All transcripts were checked by the researcher against the sound files for accuracy and corrected where necessary. Corrected transcripts were entered into HyperResearch software and coded for both anticipated and emergent themes using the constant comparative method including searches for disconfirming evidence [24]. Analyses were validated through peer debriefing, during which developing themes were identified and discussed with the study team.

Results

Participant Demographics

In total, 14 oncologists participated in interviews (n = 10 academic, 4 in the community). A majority of oncologists were young (64% ≤39 years old, range: 32–67) and had practiced in academic hospitals (71%) for an average of 10 years (Table 1).

Table 1.

Participants’ characteristics

Perceived Value of the GEP Test as a Decision Support Tool

Oncologists considered GEP to be a decision-support tool that enhanced confidence in their established approach to risk assessment in cases in which the best course of action was unclear to both patients and physicians:

The biggest thing is that it’s given some additional confidence to the pathology and to trying to identify these women who have relatively low-risk disease who can really avoid treatment. . . . It gives me . . . more confidence in making that recommendation, and I think it also gives patients some added confidence as well in terms of allaying their anxieties that really they’re not going to be forgoing a small potential benefit by avoiding chemotherapy if they're really, truly low-risk. [GEP01]

It was also seen as an added “comfort measure” to confirm baseline prognostic risk assessments.

The test was particularly valued in situations in which the oncologist was truly undecided about the best course of action because traditional risk assessment left them with intermediate risk or concern about risk accuracy. In these cases, it was seen as diminishing the level of indeterminacy:

There are a lot of patients where the clinical decision for chemo is really borderline. . . . So this tool allows us to actually hone in on which group of patients is most likely to benefit. It still doesn’t tell us 100% who does or doesn’t benefit, but at least it narrows it down a lot more. [GEP06]

The guidance that the test offers oncologists in situations of clinical equipoise was especially important for some. One oncologist described the test as a “tie-breaker”:

There are those scenarios where we really, really are on the fence and . . . the test kind of gives us a little bit of a tie-breaker that’s more concrete than just saying, “Well, I kind of feel like she needs chemo,” or “I kind of feel like she doesn’t.” [GEP10]

Impact of the GEP Test on Practice

Despite valuing the test as a decision-support tool, most oncologists did not feel that it fundamentally altered their practice:

I don’t think the overall process has changed for me that I feel like my decisions for adjuvant therapy now are hanging on this test coming back. . . . It’s just added to it. [GEP07]

Most participants felt that the test results had not altered the amount of chemotherapy they ordered but rather allowed them to order or forego chemotherapy, more often, in the right cases:

I think on balance I’m probably prescribing just as much chemotherapy because . . . there are patients that I may have recommended chemotherapy to that then come back lower risk and don’t need it, and then there are those patients that I would not have recommended chemotherapy to . . . who then come back with a test where I’m more inclined to offer them chemotherapy. [GEP10]

Oncologists practicing in community hospitals raised an additional issue arising from a perceived variability of practice between testing facilities, which led some to question the accuracy of pathology reports. In such cases, the test allowed them to interpret the reports with greater confidence:

Maybe . . . this is just making up for inconsistent pathology reports across the region. . . . Even if that was all it was doing, I would still argue that’s a benefit. [GEP11]

Concerns About the GEP Test

Despite providing additional confidence and more appropriate use of chemotherapy, oncologists raised concerns about the test’s overuse, inappropriate use, over-reliance on results within their medical community, its proprietary nature, and high cost. These concerns dampened oncologists’ overall enthusiasm for the test.

Overuse of the GEP Test

Because of its high cost, oncologists emphasized the need to determine patients’ willingness to act on the results in advance of ordering the GEP test:

I’m very clear with them that if we’re going to spend $4,000 on a test and if the test comes back and it says you need chemotherapy, that they need to take chemotherapy. [GEP03]

Some oncologists described colleagues ordering the test before they’d had the opportunity to determine whether the patient would be willing to act on the results:

I’ve encountered people who use it just kind of second nature, so they do it on everybody. . . . There are some oncologists that make up their mind before they even see the patient that, “You know what? I’m going to order this test.” And if you don’t take into account what your patient wants you’re going to be over-ordering that test. [GEP10]

There was widespread agreement that on the basis of both public expense and clinical utility, the test should only be ordered if it was likely to add value to the decision-making process. Ordering the test when it would be unlikely to impact treatment decisions would otherwise amount to overuse.

Inappropriate Use of the GEP Test

Oncologists identified circumstances, which they had either observed or were anecdotally aware of, in which they felt that colleagues were using the test inappropriately, especially in unvalidated groups:

Occasionally I’ve even seen the test being requested for hormone receptor negative patients. . . . I suspect that there’s insufficient knowledge on account of the requesting physician. [GEP03]

In fact, a number of participants described themselves as using the test in a wider group of patients, acknowledging that this likely diverged from the practices of some of their colleagues:

If it’s a grade 3 tumour, I’ll use it because I would appreciate a low result on that for not having chemo, and all the grade 2’s I think I send regardless. . . . So for patients with a micromet here and there I’ve used it, for patients who are node-positive. . . . Maybe there’s some reason why I’m really sitting on the fence about chemotherapy with them. . . . They may have co-morbid illnesses or something else that I would like a reason not to give them chemotherapy. [GEP13]

Over-reliance on the GEP Test

Oncologists also expressed concerns about over-reliance on the test relative to other, more established pathology indicators. Concerns that the results of the GEP test were being weighted more heavily than classic pathology were widely shared:

The only misgiving I would have is for people who tend to rely on it too much. . . . I think we forget that it’s still a test, it’s still not infallible and it really is only one factor in the big decision that we’re making. [GEP10]

This concern was partly based on uncertainty about the technical proficiency of the company’s pathologists given the inherent difficulties of dealing with tumor heterogeneity and of testing very small tumors:

Some people probably will put more weight on the results of Oncotype DX than on regular pathology, which I think you should do with caution on the account that the pathologists that the company uses, we don’t really know how good they are, we don’t know which part of the tumour they selected, so if the initial pathology suggests high-risk disease and the Oncotype says low-risk, I would be very concerned by that. [GEP03]

Proprietary Nature of the GEP Test

The most consistently voiced concern about the GEP test was about the proprietary nature of the technology and the associated high cost. This concern was further exacerbated by the fact that testing currently takes place in only one laboratory in California:

It’s expensive and it would be nice if there were some competition to make it cheaper. . . . It is entirely possible that you could get the same information . . . from a much, much cheaper test that hospitals could do locally. [GEP06]

The aggressive marketing of the product was an additional concern:

I think part of the appeal of Oncotype is their tremendous marketing campaign. . . . I’m always uncomfortable with something that’s hyped so much directly to patients. [GEP08]

Challenge of Communicating About the GEP Test

Oncologists’ approaches to explanation of the test varied markedly. Some framed their explanations in terms of personalized medicine, some emphasized the predictive value of the test, deliberately de-emphasizing its prognostic element, whereas others reduced their explanation to a dichotomous chemo/no-chemo bottom line.

For example, this oncologist emphasized the personalized nature of the test, contrasting it with Adjuvant Online, which was described as providing a population-based derivative rather than offering an individualized result from GEP:

I tell them that when we do the Adjuvant Online, it’s a good tool to start with because it gives some perspective, but this is based on all the trials and it is not specific for her, right? But we do have a tool, there are genes that have been tested that have been validated by studies to help predict the actual relapse pattern and their prognosis. [GEP14]

Others positioned the test as being primarily of predictive rather than prognostic value. This was due to concern that patients would focus on the likelihood of a recurrence (prognostic), as indicated by the test, rather than the potential value of chemotherapy (predictive).

Another interesting contrast emerged around the presentation of the test results as either continuous or dichotomous variables. A number of oncologists presented the results in terms of a chemo versus no chemo framework:

So I give them a cutoff for chemo and no chemo, I don’t really go into the high, medium or low, and when we discuss the actual test itself I will say to them, “If you fall above this number we will give you chemo,” so I basically dichotomize it into chemo and no chemo rather than, “This is your risk,” because I think the test is useful as an indicator of whether you need chemotherapy or not. [GEP03]

Others were more fluid in their explanations and tended to avoid numerical indicators, instead emphasizing the need for decision-making based on ongoing discussion.

Interestingly, although many oncologists asserted that explaining the test to patients was not difficult, it was also not uncommon for them to share a complex or lengthy explanation of the test or subsequently to indicate doubt that patients understood it:

You’re talking about risks which I think most people . . . have a very hard time conceiving of and really defining their behaviour based on it, and then you add in the additional complexity of what a molecular test is . . . and how that might add to the tests that have already been done. . . . So . . . even though you try to explain it carefully and you make yourself available to ask questions and go over it again, I’m really not sure that most of patients really understand what the test is all about. [GEP01]

Discussion

Amid growing use of GEP tests [20–23], our study provides timely insight into their varied use and perceived value in clinical practice. Oncologists generally valued GEP as a decision support tool that provided additional confidence in situations of uncertainty and allowed them to order chemotherapy for the right patients on a more consistent basis. However, the test’s perceived value was tempered by oncologists’ reservations about its use. Many were concerned about an over-reliance on the test as the definitive indicator of a patient’s prognosis. There was also widespread agreement that the test should only be ordered if it was likely to add value to the decision-making process.

These reservations are consistent with general concerns about overuse of laboratory testing, which has increased rapidly, contributing to the financial strain on the health care system [25]. Estimates suggest that 10%–50% of laboratory testing in Canada might be unnecessary [26], and one-third of United States health care spending results from overuse or misuse of tests, procedures, and therapies [27]. Assessments of overuse rest on balancing benefits and harms, benefits and costs with patient preferences [28]. These tradeoffs are inherent in GEP testing, placing GEP tests in the vanguard for potential overuse. The results from our study reveal potential overuse and divergent practice patterns, and identify a need for clinical practice guidelines to support standardized ordering, application, and interpretation of the test.

Interestingly, oncologists’ discomfort with GEP tests’ proprietary nature and high cost may mitigate the potential for overuse. They stressed the need to determine patients’ willingness to act on the results in advance of ordering the test. However, such practice might be challenging in two ways. First, assessing patients’ willingness to follow the course of action suggested by the results may create a situation that is unintentionally coercive. Second, as reported in our previous work [29], oncologists are positioned as gatekeepers of GEP, tasked with providing access in medically appropriate cases. However, oncologists’ perceptions of appropriateness can vary, and this led some patients to perceive inequities in access to GEP. Such gatekeeping efforts by individual clinicians create inherent tensions underlying access to and utilization of genetic services [29, 30]. Ultimately, oncologists’ concerns about the proprietary nature of the test reflect systemic challenges of integrating proprietary technologies in a socialized health care system, which may be best served by standardized approaches in the allocation of funds for, and use of, new technologies [29, 30].

The most striking feature of the results was the contradiction between many oncologists’ assertions that the GEP test was relatively easy to explain to patients, the often lengthy and complex explanations they then provided, and their beliefs that patients’ expectations and understanding of the test were often unrealistic and limited. These results confirm our related work on patients’ perceptions of GEP, which revealed patients’ misunderstanding of the test and “magical thinking” about the test’s true value and validity [31]. This trend is not specific to GEP; literature demonstrates that the public possesses variable knowledge and health literacy [32–38], challenging risk comprehension and health behaviors [39]. Nonetheless, the challenges in communication highlighted in this study substantiate the need for informational or decision aids to support communication and comprehension of the test.

There are limitations to our study. The study did not record oncologists’ explanations of GEP tests to patients. Nor was it designed to test patient comprehension or to explore provider efforts to explain the GEP test. For those reasons, it is not possible to ascribe a causal relationship between oncologists’ explanation of GEP testing and what patients understood. Many participants trained at the hospitals participating in this study, leading to greater homogeneity of attitudes and approaches than would be evident in a more varied participant sample.

Conclusion

Our study reveals timely insights into GEP’s impact on clinical practice, oncologists’ reservations about the test, and the challenges inherent in communication about the test. Oncologists valued the test as an additional decision support tool, despite their concerns about its reliability, cost, inappropriate use and over-use by other oncologists, and patients’ limited understanding of the test. The results identify a need for decision aids to support patients’ understanding and for clinical practice guidelines to facilitate standardized use of the test by oncologists.

Author Contributions

Conception/Design: Yvonne Bombard, Linda Rozmovits, Maureen Trudeau, Natasha B. Leighl, Ken Deal, Deborah A. Marshall

Provision of study material or patients: Yvonne Bombard, Maureen Trudeau, Natasha B. Leighl

Collection and/or assembly of data: Yvonne Bombard, Linda Rozmovits, Ken Deal, Deborah A. Marshall

Data analysis and interpretation: Yvonne Bombard, Linda Rozmovits, Maureen Trudeau, Natasha B. Leighl, Ken Deal, Deborah A. Marshall

Manuscript writing: Yvonne Bombard, Linda Rozmovits, Maureen Trudeau, Natasha B. Leighl, Ken Deal, Deborah A. Marshall

Final approval of manuscript: Yvonne Bombard, Linda Rozmovits, Maureen Trudeau, Natasha B. Leighl, Ken Deal, Deborah A. Marshall

Disclosures

Maureen E. Trudeau: RNA Diagnostics (C/A, OI), Roche, Amgen, Novartis, AstraZeneca, Essai, Sanofi, Pfizer (RF); Deborah Marshall: Optum Insight (C/A). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board