Introduction
Breast cancer is the second-most deadly malignancy in women [1] and accounts for one-fourth of all expected new female cancer cases. In 2009, about 92,000 American women were diagnosed with breast cancer and over 40,000 are expected to die of the disease [2]. Many women with early-stage breast cancer are offered adjuvant chemotherapy to prevent recurrence. While new genetic-based technologies that can help predict cancer recurrence or treatment response present exciting opportunities for improving outcomes, little is known about how such technologies are being incorporated into practice and whether they are being used appropriately to make treatment recommendations [3,4].
Human Epidermal growth factor Receptor 2 (HER2) is a gene that influences cell growth, division and repair; a normal cell has 2 copies of HER2. About 18–20% of breast cancers have amplification of this gene [5,6,7], which is associated with rapid tumor proliferation, shorter disease-free survival and poorer overall survival [8,9,10]. Trastuzumab (Herceptin®, Genentech Inc., San Francisco, Calif., USA) acts by targeting production of the HER2 protein to prevent the growth of HER2-positive cancer cells, thereby reducing recurrence of disease and reducing mortality [11,12,13,14,15]. However, trastuzumab only benefits women with HER2 gene amplification, is expensive (USD 44,000–65,000 per year [16,17]) and can be cardiotoxic [18]. Therefore, selecting appropriate patients to receive trastuzumab is vital.
Currently, 2 types of tests are approved by the U.S. Food and Drug Administration for determining HER2 status. The immunohistochemistry (IHC)-based test (e.g. DAKO HercepTest; Ventana Pathway) measures production of HER2 protein by the tumor. Test results are ranked as 0, 1+ (negative), 2+ (equivocal), or 3+ (positive). The fluorescence in situ hybridization (FISH) test (e.g. Vysis PathVysion; Ventana INFORM HER2 probe) quantifies the number of copies of the HER2 gene in tumor cells. A positive HER2 test is defined as IHC 3+ and, to a lesser extent, IHC 2+ [19,20,21] or a HER2:CEP17 ratio >2 [22,23]. While some reports suggest that FISH technology more accurately predicts response to trastuzumab than IHC technology [24], a recent summary report indicates that the 2 tests are comparable if careful validation testing is performed [25]. HER2 testing may also predict response to several systemic therapies, including anthracyclines and resistance to endocrine therapy, although the evidence is not always consistent [10,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45].
The Food and Drug Administration initially approved trastuzumab in 1998 for use in patients with metastatic breast cancer. In November 2006, it approved trastuzumab as an adjuvant therapy for women with lymph node-positive and HER2-positive breast cancer [12,13,14]. Trastuzumab is not recommended for patients with a positive HER2 test result if they have cardiovascular risk factors. A joint guideline from the American Society of Clinical Oncologists (ASCO) and the College of American Pathologists (CAP) state that HER2 testing should be performed for all invasive breast cancers regardless of lymph node disease status [25], and the National Comprehensive Cancer Network (NCCN) also endorsed HER2 testing [46,47,48,49,50].
Despite these well-developed, evidence-based practice guidelines, however, little research has been done on this test in real-world settings. In particular, Phillips [3] indicated that little is known about what percentage of patients are tested for HER2, which testing methods are used, whether patients are retested to confirm indeterminate results, and how many patients with negative or equivocal results receive trastuzumab.
Our study addressed this knowledge gap by evaluating utilization and treatment patterns associated with HER2 testing for patients with breast cancer in an integrated healthcare delivery system. We studied a cohort from this health plan with more than 12 years of electronic medical records and other data sources. We documented the uptake and use of HER2 testing and evaluated whether testing was being done appropriately according to professional guidelines. We considered the use of IHC versus FISH testing and trastuzumab prescriptions in the context of the HER2 test result. This study is one of the largest and most comprehensive studies illustrating the real-world use and impact of HER2 testing.
Subjects and Methods
Study Population
Study participants were patients at Kaiser Permanente Northwest (KPNW), an integrated healthcare delivery system serving more than 470,000 members in Oregon and Southwest Washington. KPNW’s members are demographically representative of the coverage area in terms of the age, gender and racial or ethnic distribution, and include about 20% of the area’s population. Medicare members represent about 12% of KPNW’s total membership. Members over 65 represent 12.8% of total membership, 2% are below 200% of the federal poverty level and 13% are minorities. We identified women with a primary diagnosis of breast cancer through KPNW’s tumor registry. Women were eligible for the study if they were diagnosed with their first primary breast cancer between January 1, 1998 and December 31, 2007 and did not have missing data for tumor stage at diagnosis. We required that participants receive their diagnosis and initial treatment at KPNW. KPNW patients are treated at 5 area hospitals, and there are currently 10 oncologists on staff, although there were changes in staff over the 10-year study period.
The Institutional Review Board at KPNW approved this study and did not require written informed consent. The Oregon Genetic Privacy law requires health care providers to notify patients that any specimens or health information will be available for anonymous or coded genetic research unless the person ‘opts out’. About 13% of KPNW’s membership has opted-out, and these individuals were excluded from this study.
Data Collection
We abstracted electronic data on patient characteristics, tumor characteristics, HER2 testing status, test results for FISH and IHC separately, and trastuzumab treatment. The centralized tumor registry contains information on all cancers diagnosed at KPNW since 1960, and survival data is continuously updated. A trained abstractor keys items directly into the registry for each identified tumor. The pharmacy database records all prescriptions dispensed by KPNW outpatient pharmacies and includes date of dispensing, dose, prescribing physician, and unique codes using standard nomenclature to identify each drug.
HER2 Genetic Testing
All data used in this study are derived from testing that occurred as part of routine medical care provided by KPNW clinicians. Between 1998 and 2000, KPNW implemented an internal practice guideline of systematic screening for all women diagnosed with invasive breast cancer. According to this protocol, IHC is used as the initial HER2 test, followed by FISH testing to clarify or confirm equivocal or positive IHC findings.
IHC testing for HER2 was conducted by the KPNW regional laboratory using the HercepTest (Dako, Carpinteria, Calif., USA) and was classified as negative (0 or 1+), equivocal (2+) or positive (3+). FISH testing for HER2 was performed by Quest Diagnostics using the Vysis PathVysion test (Abbott Molecular, Abbott Park, Ill., USA) and was classified as negative (HER2/CEP 17 ratio <1.8), equivocal (HER2/CEP 17 ratio between 1.8 and 2.2) or positive (HER2/CEP 17 ratio >2.2). The standard protocol at KPNW changed in October 2007 to make FISH testing (performed at Quest Diagnostics) the initial HER2 test. For all cases with equivocal (1.8–2.2) FISH results, and for known grade 3 tumors with negative FISH results, IHC testing was also performed by Quest Diagnostics.
Chart Abstraction
Following the initial analysis of tumor-registry data, a single abstractor manually checked a sample of data points in the categories described below against the electronic medical record using standard data collection forms. Abstracted variables included IHC and FISH test results, date of test, stage of disease at diagnosis, lymph node involvement, tumor size, and trastuzumab use. For training, we developed instructions and a set of ‘practice’ charts that were scored by 2 study abstractors and compared [51]. Abstraction forms were entered into an electronic database using double data entry to ensure accuracy. Two reviewers discussed unexpected values to resolve issues.
We abstracted charts in 7 categories: (1) patients with a diagnosis of noninvasive breast cancer who nevertheless received HER2 testing (n = 11) after 1999; (2) patients with a diagnosis of invasive breast cancer who did not receive HER2 testing (n = 154) after 1999; (3) patients who received FISH testing, but not IHC testing, after 2004 (n = 93); (4) patients with a negative IHC test result that was confirmed by FISH after 2004 (n = 87); (5) patients with an equivocal IHC test result that was not confirmed with FISH after 2004 (n = 69); (6) patients who received trastuzumab, but did not have a positive HER2 result (n = 26); and (7) HER2-positive patients with distant metastatic or regional (after 2004) breast cancer who did not receive trastuzumab (n = 13 and 21, respectively). We verified findings in a random subset of patients for each category (n = 50; except categories with fewer than 50 observations).
Statistical Methods
Patients were classified as receiving trastuzumab if any of the following national drug codes were in the pharmacy records after their date of diagnosis: 50242013460, 50242013468, 50242005656, 63552047001, or if procedure code J9355 was in the procedures database. Lymph node status was dichotomized into positive (one or more positive nodes) or negative. We used SEER staging criteria [52] to define noninvasive breast cancer as patients diagnosed with ductal carcinoma in situ (DCIS) and invasive breast cancer as patients diagnosed with localized, regional or distant-metastasis breast cancer. All analyses, including descriptive statistics and summaries, were produced using R (version 2.6.2; R Foundation for Statistical Computing, www.r-project.org).
Results
Pattern of HER2 Test Utilization
There were 3,623 women who met the criterion of a primary breast cancer diagnosis during 1998–2007 (table 1). We excluded 31 women because of missing tumor registry data on cancer stage at diagnosis, a critical variable. Of the remaining women, about 538 (15%) had a diagnosis of DCIS or noninvasive breast cancer, and 3,054 (85%) had invasive breast cancer.
We defined women as ‘evaluated’ for HER2 if they received either IHC or FISH testing. Overall, 69% of the study population was evaluated for HER2. However, this summary statistic obscures critical differences over time and by disease stage (fig. 1). First, according to clinical guidelines, the 15% of women with noninvasive breast cancer are not recommended for HER2 testing. In this group, only 2.5% received HER2 testing, and the proportion remained consistently low over the entire study period. Of those with invasive breast cancer, 81% received HER2 testing, a proportion that substantially increased between 1998 and 2000 from 12% to over 94%. After 2000, more than 94% of invasive breast cancer patients received HER2 testing over all years combined. Women with both invasive and noninvasive breast cancer were more likely to be evaluated for HER2 if they were also evaluated for other tumor markers including ER status (p < 0.0001) and nodal involvement (p ≤ 0.007) (table 1).
Fig. 1
Rate of HER2 testing at KPNW from 1998 to 2007. The solid lines correspond to patients diagnosed with invasive breast cancer with Medicare/Medicaid insurance (diamonds) or with other insurance (circles), and the dashed lines correspond to patients diagnosed with ductal carcinoma in situ or noninvasive breast cancer.
We manually abstracted medical charts to verify findings from the tumor registry. We estimated the KPNW protocol was not followed for <3% of patients diagnosed since 2000, after correcting for errors. For the 11 patients diagnosed since 2000 with noninvasive breast cancer who received HER2 testing according to the tumor registry, 10 patients had a diagnosis of DCIS. The remaining patient had no tumor stage in the medical record. We were unable to find evidence of HER2 testing in the medical chart for 3 of the 10 DCIS patients. As such, 7 of these 10 patients were correctly identified as evaluated for HER2. Thus, the tumor registry correctly identified a small number of patients with noninvasive breast cancer who received HER2 testing.
There were 154 patients diagnosed with invasive breast cancer after 1999 who did not receive HER2 testing according to the tumor registry, although they should have received this testing according to the KPNW protocol. We conducted a chart review for a random subset of 50 of these cases and discovered that 63% of the chart-reviewed cases actually did receive an IHC or FISH test. Thus, after accounting for these errors in the tumor registry, we estimate that only about 2–3% of patients with invasive breast cancer did not receive HER2 testing.
Fewer invasive breast cancer patients with Medicare/Medicaid insurance were evaluated for HER2 compared with other insurance products (79% vs. 83%; p = 0.001) (fig. 1).Most patients (92%) diagnosed with noninvasive breast cancer and evaluated for HER2 were treated inside KPNW. A slightly higher proportion of patients diagnosed with invasive breast cancer and not evaluated for HER2 were treated outside KPNW (22% vs. 18%; p = 0.03). It is possible that this difference is a result of the fact that documentation for HER2 testing was not available for some patients treated outside KPNW.
Use of IHC and FISH Tests
The majority of patients who underwent HER2 testing received the IHC test (96%). FISH results were only recorded in the tumor registry database after 2004 (table 2). Forty-two percent of those diagnosed with invasive breast cancer between 2004 and 2007 (n = 1,232) received FISH testing, and 82% (433) of these also received IHC testing. About 81% of patients who received FISH testing, but not IHC testing, were diagnosed after October 2007, when KPNW adopted FISH testing as the primary HER2 test.
Table 2
Comparison of FISH and IHC test results for invasive breast cancer patients diagnosed after 2004
Confirmation of IHC Results
As table 2 shows, tumor registry data indicate inconsistent adherence to KPNW policy on FISH testing following an equivocal or positive IHC test result. However, chart review reveals better compliance. For 50 patients with an equivocal IHC test result who did not have a FISH test result in the tumor registry, chart review revealed that 88% did have a FISH test result in the medical record. Furthermore, although the KPNW protocol does not include confirmation of negative IHC test results by FISH, chart review confirmed that 88% of patients with a negative IHC test result and a FISH test result in the tumor registry did receive FISH testing. However, 12 (27%) patients did not actually have a negative IHC test result. These findings indicate that negative IHC test results are occasionally confirmed by FISH in this setting, but also highlight some apparent discrepancies in the tumor registry.
About 40% of patients who received both tests and had a positive IHC test result were found to have the opposite finding (negative result) for the FISH test (table 2). In contrast, nearly everyone who received both tests and had a negative IHC test result was found to have a consistent negative FISH result. As discussed below, these discrepancies can impact decisions about whether to treat with trastuzumab.
Does Treatment with Trastuzumab Depend on HER2 Status?
In the entire population of patients evaluated for HER2, 14% had a positive test result (using IHC and/or FISH) (table 3). The majority (81%) of patients who received trastuzumab had a positive HER2 test result using IHC, FISH or both in the tumor registry. Subsequent chart review of the remaining 26 patients indicated that, in nearly all cases (95%), patients who received trastuzumab were appropriate candidates for this therapy based on their HER2 genetic test result (table 4). Overall, <1% of patients who did not have a positive HER2 test received trastuzumab.
Table 4
Reasons for treatment with trastuzumab for 26 patients without a positive HER2 test result in the electronic tumor registry file
Table 3 shows the proportion of patients who received trastuzumab by tumor stage and HER2 status. Prior to 2005, only 9% of HER2-positive patients received trastuzumab; the majority of treated patients had regional or metastatic disease (88%), consistent with professional recommendations at that time. After 2004, trastuzumab use increased for all stages of disease, with an overall frequency of 55% among HER2-positive patients, and increasing use among patients with more advanced disease (up to 80% for those with distant metastatic disease). A few patients had discordant HER2 test results for IHC and FISH, and these patients generally did not receive trastuzumab (4% received treatment). There were several reasons why women with a positive HER2 test did not receive trastuzumab (table 5), although chart review found that 5 out of 26 women who did not receive trastuzumab according to the tumor registry actually did receive treatment outside of KPNW.
Table 5
Reasons for no treatment with trastuzumab among HER2-positive women with distant metastatic or regional (after 2004) breast cancer
Discussion
We conducted a retrospective analysis of patients diagnosed with breast cancer between 1998 and 2007 to evaluate the utilization and treatment patterns associated with HER2 testing in an integrated healthcare delivery system. The prevalence of HER2 testing in appropriate candidates with breast cancer was very high (>94%) following an initial 2-year period. Trastuzumab treatment was guided by HER2 test results in most instances (95%), and <1% of patients with a negative or equivocal HER2 test result were treated with trastuzumab.
These results are important for several reasons. First, there is still a great need for research describing how sophisticated genetic tests, with potentially confusing treatment ramifications, are taken up by clinicians and whether such tests are being appropriately used in treatment. This study shows that in a large integrated managed care setting, HER2 testing is being performed on the appropriate patients, and the results are being interpreted correctly in terms of treatment implications. Only 2 other similar studies [53,54] have been done in managed care settings. Although the sample sizes for these studies were smaller, their findings were similar to ours. Stark et al. [53] conducted a study at Henry Ford Health System (Detroit, Mich., USA) between 1999 and 2000. They reported that 51.9% of women diagnosed with primary breast cancer were evaluated for HER2. Barron et al. [54] reviewed the charts of 380 patients in commercial health plans diagnosed in 2005 through mid-2006. HER2 testing occurred in 98.1% of patients with invasive breast cancer, and only one patient (out of 52) who received trastuzumab did not have a documented positive HER2 test result. For HER2-positive women diagnosed with stage 2 or higher breast cancer (n = 45), 87% received trastuzumab.
Two studies conducted in the UK have recently been reported [55,56], which address similar questions in the context of a national healthcare system. In the UK, the National Institute for Health and Clinical Excellence guidance has recommended adjuvant trastuzumab for women with positive HER2 test results with normal left ventricular ejection fraction and without cardiac contraindications to trastuzumab therapy since 2005 [57]. Coulson et al. [55] evaluated patients who received HER2 testing between September 2007 and August 2008 in the North Trent Cancer Network. In this study, 15.1% of tested subjects were HER2 positive, and 67% of HER2-positive subjects were treated with trastuzumab. The primary reasons that patients did not receive treatment were (1) age >75 years with or without general frailty or poor performance status, (2) patient refusal or (3) high cardiac risk. This study did not include subjects who did not receive HER2 testing. Webster et al. [56] evaluated patients diagnosed with early breast cancer during 2006–2007 in the South West Wales Cancer Network. In this study, 10.4% of tested subjects were HER2 positive, and 13.5% did not receive HER2 testing. Patients who were not assessed for HER2 status were primarily elderly women who did not receive surgical intervention after the initial biopsy. Among the HER2-positive subjects, 72.3% received trastuzumab.
In contrast, Tong et al. [58] evaluated Medicare data for patients diagnosed with breast cancer in 2005. Only 22% of patients in that study were evaluated for HER2, and 94% of those received IHC alone, 1% received FISH alone, and 5% received both tests. Furthermore, 61% of patients who received trastuzumab were not evaluated for HER2. Stark et al. [53] also indicate that the type of health insurance (capitated insurance vs. fee-for-service [FFS]) influenced the probability of receiving HER2 testing, with an increased likelihood of testing for those with capitated insurance (OR = 1.59; p = 0.027).
This study provides a crucial stepping stone to further research in the complex field of genomic medicine. At present, very few healthcare systems have the ability to examine the uptake and use of genetic tests on a scale large enough to evaluate them systematically. Without reliable data on how genetic tests are used to inform medical decisions, we cannot achieve the next level of genomic research. In the context of HER2 testing and treatment decisions, for example, we can now build upon the existing data to evaluate patient outcomes and adverse events that patients experience from treatment in the context of their HER2 test result.
Nevertheless, the electronic data sources employed for this study had a few limitations. For example, we did not confirm the validity of HER2 test results because this study focused on how clinician treatment decisions are influenced by evaluation of HER2, and the reported test result was the most relevant data for our research question. Additionally, we did not limit the study population based on membership criteria, a potential limitation since some individuals may have incomplete treatment information if they left the health plan before their treatment was complete. About 8% (n = 299) of subjects in the study population had 2 years or less of membership following breast cancer diagnosis. The possibility of incomplete treatment history is particularly concerning for this group. However, when we restricted the data to only these subjects, the conclusions do not change regarding the impact of HER2 status and disease stage on trastuzumab treatment status (data not shown). About 70% of subjects have 10 years or more of membership following breast cancer diagnosis. Finally, we identified inconsistencies between the tumor registry and the results from manual chart review, which are primarily instances of missing data in the tumor registry. Since it was not feasible to manually abstract all the records in this study, we only performed targeted chart review in situations with unexpected findings.
Our findings indicate that KPNW is systematically performing HER2 evaluation on patients with invasive breast cancers, and the information is used to make treatment decisions. The presence of an integrated, highly-utilized and well-established electronic medical record has likely improved communication of test results between pathologists (who order/perform the test) and oncologists, who make treatment recommendations. While it is unknown whether the findings of this study are widely applicable beyond managed care settings, about 25% of Americans receive healthcare in a managed care setting [59], and these findings are directly relevant to this substantial minority of the population.These questions should be addressed in other settings for comparison, as part of an assessment for implementation research.
Acknowledgements
The authors would like to thank Dr. Mona Ishag from the KPNW Department of Pathology and Dr. Mark Hornbrook from the Center for Health Research for their careful review and comments during the preparation of the manuscript. This work was partially supported by resources developed through a grant from the National Cancer Institute (U19 CA 79689). This research was conducted at a single site of the HMO Cancer Research Network (CRN). The CRN consists of the research programs, enrollee populations and databases of 14 HMO members of the HMO Research Network. The overall goal of the CRN is to conduct collaborative research to determine the effectiveness of preventive, curative and supportive interventions for major cancers that span the natural history of those cancers among diverse populations and health systems. The 14 health plans, with nearly 11 million enrollees are distinguished by their longstanding commitment to prevention and research, and collaboration among themselves and with affiliated academic institutions.
Get Permission
PubMed ID
