ORIGINAL ARTICLE
ER and PR Immunohistochemistry and HER2 FISH
versus Oncotype DX: Implications for Breast Cancer
Treatment
MiHee M. Park, BS, Joshua J. Ebel, BS, Weiquiang Zhao, MD, PhD, and
Debra L. Zynger, MD
Department of Pathology, The Ohio State University Medical Center, Columbus, Ohio
n Abstract: Estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor (HER2) concordance between immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH), and Oncotype DX, a commercially available RT-PCR-based assay which recently began reporting biomarker results was assessed. ER concordance
was 98.9% (262/265), Pearson correlation coefficient (r) = 0.42, and Spearman’s rank correlation (q) = 0.25. Positive percent agreement for ER was 98.9% (262/265). One patient with discordant ER results was not offered hormone therapy
based on the preferential use of Oncotype DX. PR was concordant in 91.3% (242/265), r = 0.80, q = 0.75, and Cohen’s
kappa (j) = 0.63. Positive percent agreement for PR was 90.5% (218/241) and negative percent agreement was 100% (24/
24). HER2 concordance was 99.2% (245/247), r = 0.35, q = 0.28, and j = 0.12. Positive percent agreement for HER2 was
0% (0/2) and negative percent agreement was 100% (245/245). Of the three FISH HER2-amplified cases, two were negative and one was equivocal, and all FISH HER2-equivocal cases (n = 3) were negative by Oncotype DX. Patients that were
FISH HER2-amplified, Oncotype DX HER2-negative did not receive trastuzumab. Although our results demonstrated high
concordance between IHC and Oncotype DX for ER and PR, our data showed poor positive percent agreement for HER2.
Compared to FISH, Oncotype DX does not identify HER2-positive breast carcinomas. The preferential use of Oncotype DX
biomarker results over IHC and FISH is discouraged. n
Key Words: breast cancer, estrogen receptor, human epidermal growth factor, immunohistochemistry, oncotype DX,
RT-PCR
E
strogen receptor (ER), progesterone receptor (PR),
and human epidermal growth factor receptor-2
(HER2) are clinically useful in guiding treatment
options such as tamoxifen, aromatase inhibitors, or
the monoclonal antibody, trastuzumab (1). According
to the American Society of Clinical Oncology/College
of American Pathologists, validated techniques to
assess these biomarkers are immunohistochemistry
(IHC) and fluorescence in situ hybridization (FISH)
(2,3). Recently, molecular testing by reverse transcriptase polymerase chain reaction (RT-PCR) has garnered
attention. Oncotype DX, an RT-PCR-based assay, is a
commercially available test that utilizes extracted
mRNA from formalin-fixed, paraffin-embedded tumors
and is based on the expression of 16 cancer-related
genes and five reference genes (4,5). Since 2008 ER,
PR, and HER2 qualitative and quantitative results
have been included in the Oncotype DX report.
In light of this alternate method of testing, our aim
was to perform a quality control study which evaluates ER, PR, and HER2 concordance between IHC
and/or FISH and Oncotype DX and to interpret our
findings within the context of the current literature.
METHODS
Address correspondence and reprint requests to: Debra L. Zynger,
Department of Pathology, The Ohio State University, 410 W 10th Ave., 401
Doan Hall, Columbus, OH 43210, USA, or e-mail: debra.zynger@osumc.edu
A portion of this data was presented at the 18th Annual Multidisciplinary
Symposium on Breast Disease in Amelia Island, FL on February 16, 2013
and at the United States and Canadian Academy of Pathology annual meeting in Baltimore, MD on March 6, 2013.
DOI: 10.1111/tbj.12223
© 2013 Wiley Periodicals, Inc., 1075-122X/14
The Breast Journal, Volume 20 Number 1, 2014 37–45
A retrospective review of breast carcinoma resected
from August 2008 to July 2012 was performed to isolate cases in which an Oncotype DX assay (Genomic
Health, Redwood City, CA) was ordered at The Ohio
State University Medical Center. From the Oncotype
DX report, recurrence score (RS) and ER, PR, and
HER2 qualitative and quantitative unit scores were
38 • park
ET AL.
obtained. Oncotype DX is requested on a case-by-case
basis by the patient’s breast oncologist or surgeon.
The RS is reported to represent the likelihood of
breast cancer relapse within 10 years in patients who
have been treated with tamoxifen for 5 years (4,5).
RS of 0–17 is low risk, 18–30 intermediate risk, and
31–100 high risk (5). A tumor is ER-negative with
expression units <6.5, ER-positive ≥6.5, PR-negative
<5.5, PR-positive ≥5.5, HER2-negative <10.7, HER2equivocal ≥10.7–11.4, or HER2-positive ≥11.5.
Estrogen receptor and PR were evaluated by IHC
on formalin-fixed paraffin-embedded tissue using clone
1 D5 for ER and PgR 636 for PR (Dako, Carpenteria,
CA). Percentage of positive nuclei was determined by
visual microscopic estimation: <1% negative, 1–9%
low positive, and ≥10% positive. HER2 was evaluated
by IHC using clone 4B5 (rabbit monoclonal; Ventana,
Tucson, AZ). Membrane staining was evaluated by
visual microscopic estimation and semiquantitatively
graded: 0, 1+ negative; 2+ equivocal; 3+ positive.
Amplification of HER2 was evaluated with FISH
using PathVysion HER2 DNA Probe Kit (Abbott
Molecular, Abbott Park, IL), duet scanning imaging
workstation (BioView, Billerica, MA), and accompanying software. A positive result was defined as a ratio
of HER2/chromosome 17 centromeric probe (CEP17)
>2.2, negative <1.8, and equivocal 1.8–2.2. Clinically,
an equivocal result with a ratio of ≥2.0 is eligible for
trastuzumab (3).
Biomarker testing is typically conducted on the
biopsy specimen. Markers can be repeated in the
resection specimen for various reasons including
pathologist preference, equivocal biopsy results, case
testing performed at an outside institution, clinician
requests for repeat testing, administration of neo-adjuvant therapy, or prior negative results. When biomarkers were performed on both the biopsy and
resection, data comparison was made using resection
results as this was the tissue block utilized for Oncotype DX testing. For the purpose of this quality control study, all cases with discordant biomarker data
between IHC/FISH and Oncotype DX were repeated
in the same block sent for Oncotype DX testing if not
previously performed. The patient’s surgeon and
oncologist were notified of all discrepant results
impacting patient management.
Statistical analysis, including the Pearson correlation coefficient (r), Spearman’s rank correlation (q),
and Cohen’s kappa (j), was performed to compare
IHC/FISH and Oncotype DX. ER and PR IHC and
HER2 FISH were used as the reference standard.
Spearman’s rank correlation was used to compare RS
and Oncotype DX ER, PR, and HER2 results. The
Fleiss, Cohen, and Everitt method was used to calculate confidence intervals around Cohen’s kappa. The
Fisher transformation was used to calculate confidence
intervals for the Pearson correlation coefficient. In lieu
of sensitivity and specificity, positive percent and negative percent agreements were calculated, according to
the United States Food and Drug Administration
reporting guidelines (6). Clopper-Pearson confidence
intervals were used for positive and negative percent
agreement. Positive and negative percent agreement
and Pearson correlation coefficient calculations
excluded HER2 equivocal results (n = 5). For concordance, cases with only qualitative data (ER n = 3, PR
n = 3), those lacking a numerical FISH ratio (n = 14),
or those with only HER2 IHC (n = 13), were
excluded. For Oncotype DX, ER scores reported as
≥12.5 were counted as 12.5 (n = 1), PR reported as
≥3.2 as 3.2 (n = 5), PR ≥10.0 as 10.0 (n = 10), and
HER2 reported as <7.6 as 7.6 (n = 5). All analyses
were performed in the R statistical package, version
2.15.1. A P < 0.05 was considered significant.
RESULTS
We identified 265 breast carcinoma cases in which
Oncotype DX tests were performed (Table 1). Procedures were performed at The Ohio State University
Medical Center and 42 referring institutions. 75% of
cases had internal biomarker testing. Oncotype DX
testing was requested in 20.2% of resection specimens. Most cases were pT1c, pN0, and ER-positive,
HER2-not amplified.
Estrogen receptor IHC and Oncotype DX results
were concordant in 98.9% (Table 2). IHC ER-positive, Oncotype DX ER-positive cases had a much
higher mean IHC% of positive staining cells and
Oncotype DX mean score than discordant cases
(Table 3, Table 4, Fig. 1). The latter subset had invasive tumor comprising >50% of the epithelium and
one contained a biopsy cavity. All three discordant ER
patients were reported by Oncotype DX as high risk
and were advised to be treated with adjuvant chemotherapy. One patient was diagnosed with triple negative breast cancer and was not offered hormonal
therapy based on Oncotype DX results, despite IHC
positivity. Comparing IHC and Oncotype DX scores,
a weak to moderate positive correlation was identified
IHC/FISH versus Oncotype DX • 39
Table 1. Clinicopathologic Features of Patients in
which Oncotype DX was Performed
Mean age (median, range)
Oncotype DX usage
2008
2009
2010
2011
2012
pT
pT1a
pT1b
pT1c
pT2
pT3
Indeterminate
pN
pN0(i )
pN0(i+)
pN1mi
pN1a
pNX
ER IHC/HER2 FISH
ER+ HER2 Not amplified
ER+ HER2 Amplified
ER+ HER2 Equivocal
57.9 years (58, 31–85)
22/103
61/262
62/320
73/408
45/209
(21.4%)
(23.3%)
(19.4%)
(17.9%)
(21.5%)
2/265
34/265
150/265
76/265
2/265
1/265
(0.8%)
(12.8%)
(56.6%)
(28.7%)
(0.8%)
(0.4%)
204/264
16/264
18/264
16/264
10/264
(77.3%)
(6.1%)
(6.8%)
(6.1%)
(3.8%)
246/252 (97.6%)
3/252 (1.2%)
3/252 (1.2%)
(Table 2, Fig. 4a). The Spearman’s rank correlation
between RS and ER Oncotype DX was
0.36
(P < 0.001).
Progesterone receptor was concordant between
IHC and Oncotype DX in 91.3% of cases (Table 2).
The 23 discordant cases that were IHC PR-positive,
Oncotype DX PR-negative had a mean IHC% of positive staining cells and Oncotype DX mean score much
lower than IHC PR-positive, Oncotype DX PR-positive cases (Table 3, Fig. 2). All discordant cases had
invasive tumor comprising >50% of the epithelium
and seven contained a biopsy cavity. Comparing IHC
and Oncotype DX scores, a moderate positive relationship was observed (Table 2, Fig. 4b). The relationship between RS and PR Oncotype DX results was
calculated to be 0.70 (P < 0.001) using Spearman’s
rank correlation. Changing the cutoff for PR positivity
by IHC to be 10% yielded a concordance of 95.1%,
Cohen’s kappa of 0.82, positive percent agreement of
96.0%, negative percent agreement of 90.5%.
Human epidermal growth factor receptor-2 results
were concordant in 97.2% between FISH and Oncotype DX with equivocal cases included and 99.2%
with equivocal cases excluded (Table 2). Cases that
were FISH HER2-not amplified, Oncotype DX HER2negative had a lower mean FISH ratio and Oncotype
DX mean score compared to discordant cases
(Table 3, Fig. 3). Of the three FISH HER2-amplified
cases, no cases were positive by Oncotype DX
(Table 5). All three had invasive tumor comprising
>50% of the epithelium and two contained a biopsy
cavity. Upon repeat FISH testing and HER2 IHC,
results remained unchanged. None of these patients
had documented trastuzumab use. Comparing FISH
and Oncotype DX scores revealed a weak positive
correlation (Table 2, Fig. 4c). The relationship
between RS and HER2 Oncotype DX results was calculated to be 0.30 (P < 0.001) using Spearman’s
rank correlation.
DISCUSSION
Determining ER, PR, and HER2 status is crucial to
optimizing treatment outcomes in breast cancer
patients. With supplemental reporting of these biomarkers by Genomic Health, the Oncotype DX assay may
have a more prominent role in oncologic management.
Conflicting views on the utility of Oncotype DX as a test
to accurately measure these biomarkers have been
recently published (7–10). This quality assurance study
aims to analyze ER, PR, and HER2 concordance of
Oncotype DX compared to IHC and/or FISH. At our
institution, clinicians ordered Oncotype DX in ~20% of
Table 2. Concordance between Oncotype DX Results and ER (IHC), PR (IHC), and HER2 (FISH) Results
Concordance
Positive percent agreement
Negative percent agreement
Pearson correlation coefficient, r (CI)
Spearman’s rank correlation, q
Cohen’s kappa, j (CI)
ER
PR
262/265 (98.9%)
262/265 (98.9%)
NA
0.42 (0.31–0.51)
P < 0.001
0.25
P < 0.001
NA
242/265 (91.3%)
218/241 (90.5%)
24/24 (100%)
0.80 (0.75–0.84)
P < 0.001
0.75
P < 0.001
0.63 (0.50–0.77)
P < 0.001
HER2 FISH (Equivocals included)
245/252 (97.2%)
NA
NA
0.35 (0.23–0.46)
P < 0.001
0.28
P < 0.001
0.12 ( 0.08 to 0.31)
P < 0.001
HER2 FISH (Equivocals excluded)
245/247 (99.2%)
0/2 (0%)
245/245 (100%)
NA
NA
NA
CI, 95% confidence interval; ER, Estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor-2; FISH, fluorescence in situ hybridization; IHC, immunohistochemistry; NA, not applicable.
40 • park
ET AL.
Table 3. Oncotype DX Compared to IHC and FISH
Oncotype DX ER
Positive
IHC ER
Positive
n
Mean Oncotype DX score (median, range)
Mean IHC% positive (median, range)
262
10.0 (10.1, 7.1–12.5)
92.3 (95, 30–100)
Negative
3
6.1 (6.0, 5.8–6.4)
16.7 (20, 10–20)
Oncotype DX PR
Positive
IHC PR
Positive
Negative
n
Mean
Mean
n
Mean
Mean
Oncotype DX score (median, range)
IHC% positive (median, range)
Oncotype DX score (median, range)
IHC% positive (median, range)
218
7.8 (8.0, 4.1–10.0)
78.7 (95, 3–100)
0
NA
NA
Negative
23
4.5 (4.6, 3.2–5.3)
6.9 (5, 1–20)
24
3.9 (3.6, 3.2–5.4)
0 or <1
Oncotype DX HER2
Positive
FISH HER2
Amplified
Not amplified
n
Mean
Mean
n
Mean
Mean
Oncotype DX score (median, range)
FISH ratio (median, range)
Oncotype DX score (median, range)
FISH ratio (median, range)
0
NA
NA
0
NA
NA
Negative
2
9.75
3.22
245
8.95
1.05
(9.75, 9.4–10.1)
(3.22, 2.59–3.84)
(9.0, 7.6–10.5)
(1.02, 0.7–1.61)
Equivocal
1
11.0
3.00
1
10.8
1.10
NA, not applicable.
Table 4. ER Discordant Cases: IHC ER-positive, Oncotype DX ER-negative
Case
1
2
3
Biopsy IHC
Resection IHC
Resection Oncotype DX
Oncotype
DX Recurrence Score
Treatment
P (50%)
LP (3%)
P (30%)
P (20%)
P (20%)
P (10%)
N (5.8)
N (6.4)
N (6.0)
47
32
65
Aromatase inhibitor, chemotherapy, radiation
Tamoxifen, chemotherapy (doxorubicin, cyclophosphamide)
Radiation, chemotherapy refused, hormone therapy not offered
LP, low positive; N, negative; P, positive.
cases each year, a rate that remained steady over the
4 year study period. Most cases were appropriately
ordered in ER-positive, HER2-negative patients, and
the majority were node-negative.
Estrogen receptor exhibited a very high concordance
(98.9%) between IHC and Oncotype DX. Similarly
high rates (93–100%) have been reported (7–9,10).
Positive percent agreement was also high at 98.9%,
consistent with other authors (98.9% and 99%) (8,9).
Opposing claims regarding the ability of IHC to identify ER positive cells compared to Oncotype DX have
been published. Kraus et al. concluded that IHC was
more sensitive as all ER discordant cases were positive
by IHC but negative by Oncotype DX. In contrast, a
publication supported by Genomic Health identified
five times more IHC ER-negative, Oncotype DX ERpositive cases than its counterpart (9). Discordant cases
in our study exhibited ≤20% ER-positive cells while
concordant cases had ≥30% and a very high mean
(92.3%). The Oncotype DX mean score for discordant
cases was close to the cutoff for positivity. Other comparative studies have used different measures of IHC
expression such as the modified H-score and the Allred
score (8,9). For discordant cases, Kraus et al. reported
a modified H-score between 10 and 225 (8). Thus,
lower ER expression may account for some but not all
of the false-negative results by Oncotype DX.
The Pearson’s correlation coefficient has been used
to compare IHC and Oncotype DX values provided
for ER. In analyses by this study (0.42) and by Kraus
et al. (0.58), varying degrees of a positive correlation
were detected (8). The appropriateness of Pearson’s
correlation coefficient to describe the underlying relationship between IHC and Oncotype DX is questionable as it is used to measure linear dependence in
normally distributed samples. In the case of ER, a lin-
IHC/FISH versus Oncotype DX • 41
Figure 1. Hematoxylin & eosin photomicrographs with corresponding estrogen receptor
(ER) immunohistochemistry (IHC). a1/a2,
IHC-positive, Oncotype DX-positive with diffuse, strong nuclear positivity. 409. b1/b2,
IHC-positive, Oncotype DX-negative showing
occasional cells having strong expression.
409. c1/c2, IHC-positive, Oncotype DXnegative in which most of the tumor was
negative, yet areas such as this were
strongly positive. 409. d1/d2, IHC-positive,
Oncotype DX-negative with occasional cells
moderately positive. 409.
(a1)
(a2)
(b1)
(b2)
(c1)
(c2)
(d1)
(d2)
ear relationship between IHC and ER Oncotype DX
expression units is not readily apparent. The paucity
of low level ER IHC and corresponding low Oncotype
DX results are problematic. In addition, high IHC
readings tend to occur at a relatively low value for
Oncotype DX, leading to an early IHC ceiling. The
more robust Spearman’s rank correlation showed a
weaker relationship with a low q of 0.25, contrasting
with Badve et al.’s value of 0.85. Although the precise
nature of this difference remains unclear, numerous
ER-negative cases in the data set used by Badve et al.
may be responsible (9).
Based on the RS algorithm presented by Paik et al.,
it can be estimated that a lower Oncotype DX ER score
from a negative result will correlate with a higher RS
(5). This is reflected in our ER discordant cases – all
three patients reported as Oncotype DX ER-negative
were classified in the high-risk RS group and were
advised to be treated with adjuvant chemotherapy.
There is some level of uncertainty as to whether these
RS are spuriously inflated or if these tumors are truly
high-risk. Previous studies have not addressed how RS
has affected the management of breast cancer in ER
discordant cases. Badve et al. commented on a modest
42 • park
ET AL.
(a1)
(a2)
(b1)
(b2)
Figure 2. Hematoxylin & eosin photomicrographs with corresponding progesterone
receptor (PR) immunohistochemistry (IHC).
(a1/a2) IHC PR-positive, Oncotype DX PRpositive. Strong, diffuse reactivity is seen.
409. (b1/b2) IHC PR-positive, Oncotype DX
PR-negative. Clusters of tumor demonstrate
strong expression. 409.
Table 5. HER2 Discordant Cases
Case
Biopsy IHC
Biopsy FISH
(ratio)
FISH Amplified, Oncotype DX Negative
1
E (2+)
A (3.10)
2
P (3+)
A (12.84)
FISH Amplified, Oncotype DX Equivocal
3
P (3+)
A (3.52)
FISH Not amplified, Oncotype DX Equivocal
4
N (1+)
N (1.00)
FISH Equivocal, Oncotype DX Negative
5
E (2+)
E (1.89)
6
No invasive tumor No invasive tumor
7
E (2+)
NA
Resection
IHC
Resection
FISH (ratio)
Resection
Oncotype DX
Oncotype DX
Recurrence Score
Treatment
E (2+)
P (3+)
A (3.84)
A (2.59)
N (10.1)
N (9.4)
43
35
Deceased (non-breast related cause)
Aromatase inhibitor, chemotherapy, radiation
P (3+)
A (3.00)
E (11.0)
31
Trastuzumab, chemotherapy (doxorubicin,
cyclophosphamide, paclitaxel)
N (1+)
N (1.10)
E (10.8)
22
No trastuzumab
E (2+)
E (2+)
E (2+)
E (1.84)
E (2.2)
E (1.86)
N (10.5)
N (10.4)
N (9.9)
0
22
18
No trastuzumab
No trastuzumab
Aromatase inhibitor, no trastuzumab
A, amplified; E, equivocal; N, negative; NA, not available; P, positive.
correlation between the ER Oncotype DX score and
RS (Spearman’s rank correlation = 0.47) which was
similar to this study ( 0.36). A negative Spearman’s
rank correlation shows that a lower ER result by Oncotype DX will result in a higher RS, as expected. However, Badve et al. did not document the RS risk
categories of the discordant ER cases. Future research
with larger studies of RS in tumors with ER IHC positivity between 1% and 20% and negative Oncotype
DX results is warranted.
The preferential use of Oncotype DX for ER determination raises concern. In one of our discordant
cases, the Oncotype DX result was used over IHC and
hormone therapy was not offered. In this application,
Oncotype DX results led clinicians to deny a treatment that would otherwise be indicated. Prior articles
have not discussed the use of hormone therapy in
patients with IHC and Oncotype DX ER discrepancies. It is clinically salient to emphasize that Oncotype
DX is not validated in ER-negative breast tumors and
therefore should not be ordered if prior knowledge of
ER negativity is evident (5). As instructed in the Oncotype DX report, RS from the test should be disregarded in ER-negative patients.
Progesterone receptor findings between IHC and
Oncotype DX revealed a slightly lower concordance
(91.3%) than for ER. Similar concordance rates
(86–94.2%) were reported in previous studies (7–9,10).
IHC/FISH versus Oncotype DX • 43
Figure 3. Human epidermal growth factor
receptor-2 (HER2) fluorescence in situ
hybridization (FISH) and immunohistochemistry (IHC) of discrepant cases. HER2 red signal/CEP17 green signal. (a) FISH-not
amplified, Oncotype DX-equivocal. (a1)
HER2 IHC is 1+ with focal weak membranous reactivity. 409. (a2) No HER2 amplification is seen (ratio 1.0). 609. (b) FISHamplified, Oncotype DX-negative. (b1) HER2
IHC shows 3+ intense, complete expression.
409. (b2) FISH detects amplified HER2 (ratio
>2.3). 609. (c) FISH-amplified, Oncotype
DX-equivocal. (c1) HER2 IHC demonstrates
3+ positivity. 409. (c2) FISH with amplified
HER2 (ratio >2.3). 609. (d) FISH-equivocal,
Oncotype DX-negative. (d1) HER2 IHC is 2+
with some tumor cells showing complete but
not strong membranous expression. 409.
(d2) Equivocal HER2 amplification was identified (ratio 1.9). 609.
(a1)
(a2)
(b1)
(b2)
(c1)
(c2)
(d1)
(d2)
Positive percent agreement was high (90.5%), comparable to other authors (85% and 93.9%) (8,9). Negative
percent agreement (100%) was in accordance with the
literature (96% and 97.2%). IHC identified more positive cases with all discordant IHC PR-positive, Oncotype DX PR-negative cases. All discordant cases
contained ≤20% of PR-positive tumor cells, similar to
ER. Kraus et al. reported that discordant IHC PR-positive, Oncotype DX PR-negative tumors had a modified
H-score between 1 and 110. It is uncertain how discrepant PR results impact the RS. Both this study and that
by Kraus et al. identified a positive Pearson’s correlation coefficient (0.80 and 0.69, respectively).
Spearman’s rank correlation revealed an increased
dependence (0.75) as compared to ER and paralleled
that by Badve et al. (0.85).
Analysis of HER2 results comparing FISH and
Oncotype DX revealed a very high concordance of
97.2%. High concordance rates (96–98%) have been
published (11–13). A closer subset examination demonstrates a striking difference between the two assays.
Albeit with only a few cases positive by FISH, positive
percent agreement for HER2 was 0%, and negative
percent agreement was 100%. Dabbs et al. and
Dvorak et al. reported similar findings of a very low
positive percent agreement (42% and 50%) and extra-
80
60
40
r² =
0.17
20
ER IHC (% cells positive)
(a)
ET AL.
100
44 • park
6
7
8
9
10
11
12
100
20
40
60
80
(b)
PR IHC (% cells positive)
ER Oncotype DX (expression units)
0
r² =
0.64
3
4
5
6
7
8
9
10
PR Oncotype DX (expression units)
r² =
0.06
3.0
2.5
2.0
1.5
1.0
HER2 FISH (ratio)
3.5
(c)
7.5
8.0
8.5
9.0
9.5
10.0 10.5 11.0
HER2 Oncotype DX (expression units)
Figure 4. Immunohistochemistry (IHC)/fluorescence in situ hybridization (FISH) versus Oncotype DX demonstrating varying positive
correlations. (a) Estrogen receptor (ER) IHC versus Oncotype DX.
(b) Progesterone receptor (PR) IHC versus Oncotype DX. (c)
Human epidermal growth factor receptor-2 (HER2) FISH versus
Oncotype DX.
ordinarily high negative percent agreement (100% and
100%) (13,14). These results are in contrast to that of
Baehner et al., funded by Genomic Health, who
detected high positive percent agreement (98%) and
high negative percent agreement (97%) for HER2
(11). The underlying reason for these inconsistencies is
unclear.
Various correlation and agreement statistics have
been utilized in other publications to compare FISH
and Oncotype DX for HER2. In our study, Pearson’s
correlation coefficient for HER2 was 0.35, Spearman’s
rank correlation was 0.28, and Cohen’s kappa was
0.12. Baehner et al. identified a Spearman’s rank correlation of 0.45, moderately similar to our results.
Dvorak et al. and Dabbs et al. calculated a higher
Cohen’s kappa of 0.49 and 0.35, respectively (11–13).
Although the reason for this difference is not clearly
understood, the relative lack of HER2-amplified cases
compared to HER2-negative cases in our data set may
account for lower values of correlation and agreement
using these statistical measures. Pearson’s correlation
coefficient for HER2 has not been previously reported
in the literature.
Although some clinicians may have previously utilized Oncotype DX testing as a “tiebreaker” in HER2equivocal cases, our results strongly advise against its
use in determining HER2 status, especially in this scenario, as our data and previous studies demonstrate
that Oncotype DX consistently reports FISH-equivocal
and FISH-amplified cases as negative or equivocal
(11,13). Taken together, these data do not support
Genomic Health’s marketing of Oncotype DX as an
assay that “provide(s) further clarification especially in
cases of equivocal IHC and/or FISH results or discordance between FISH and IHC,” stated in its online brochure and internet portal website for health care
professionals (14,15). In fact, previous studies funded
by Genomic Health explicitly emphasize that IHC/FISH
tests should continue to serve as methods by which to
base treatment decisions (4,12). Among our FISH
HER2-amplified cases that were identified as negative
or equivocal by Oncotype DX, only one of the three
patients was documented to have received trastuzumab.
Dabbs et al. reported that five patients were not offered
trastuzumab therapy based on HER-negative Oncotype
DX results (10). Dvorak et al. and Baehner et al. did
not report on the type of treatments given to HER2 discordant case patients (11,12). One of our three patients
who was FISH HER2-equivocal, Oncotype DX HER2negative was eligible for but did not receive trast-
IHC/FISH versus Oncotype DX • 45
uzumab therapy. Previous studies have not described
the clinical consequences to this subset of patients.
A possible explanation for the discrepancy between
IHC/FISH and Oncotype DX is that Oncotype DX
utilizes RT-PCR, a molecular technique that disregards tissue morphology. As a result, contamination
of tumor mRNA with non-neoplastic tissue or biopsy
cavity material may occur (8,16,17). Prior studies
have documented that cellular stroma, inflammatory
cells, or the presence of a biopsy cavity can influence
Oncotype DX results (16,17). Baehner et al. has made
the recommendation that microdissection to isolate
the invasive tumor be performed. It is uncertain if and
how microdissection is performed by Genomic Health.
In our series, all discordant cases were tested in blocks
with a predominance of invasive tumor. Several discrepant cases contained a biopsy cavity, although the
majority did not.
Although our results exhibited high to moderate
concordance rates between IHC and Oncotype DX for
ER and PR, our data showed poor positive percent
agreement by Oncotype DX to accurately identify
HER2-amplified carcinomas in comparison to FISH.
Using Oncotype DX to determine HER2 status for
HER2-amplified or HER2-equivocal cases is contraindicated. Rarely Oncotype DX incorrectly determined
ER status. We conclude that Oncotype DX may
potentially lead to false-negative ER, PR, or HER2
results and we advise physicians to exercise caution in
the interpretation of discordant cases.
REFERENCES
1. American Cancer Society. Cancer Facts & Figures 2013.
Available at: http://www.cancer.org/acs/groups/content/@epidemio
logysurveilance/documents/document/acspc-036845.pdf
(accessed
March 20, 2013).
2. Hammond ME, Hayes DF, Dowsett M, et al. American Society of Clinical Oncology/College Of American Pathologists guideline
recommendations for immunohistochemical testing of estrogen and
progesterone receptors in breast cancer. J Clin Oncol 2010;28:
2784–95.
3. Wolff AC, Hammond ME, Schwartz JN, et al. American
Society of Clinical Oncology/College of American Pathologists
guideline recommendations for human epidermal growth factor
receptor 2 testing in breast cancer. Arch Pathol Lab Med
2007;131:18–43.
4. Paik S, Tang G, Shak S, et al. Gene expression and benefit of
chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol 2006;24:3726–34.
5. Paik S, Shak S, Tang G, et al. A multigene assay to predict
recurrence of tamoxifen-treated, node-negative breast cancer. N
Engl J Med 2004;351:2817–26.
6. United States Food and Drug Administration. Guidance for
industry and FDA staff: Statistical guidance on reporting results
from studies evaluating diagnostic tests. Available at: http://medical.
cms.itri.org.tw/pdf/u14.pdf (accessed December 20, 2012).
7. O’Connor SM, Beriwal S, Dabbs DJ, Bhargava R. Concordance between semiquantitative immunohistochemical assay and
oncotype DX RT-PCR assay for estrogen and progesterone receptors. Appl Immunohistochem Mol Morphol 2010;18:268–72.
8. Kraus JA, Dabbs DJ, Beriwal S, Bhargava R. Semi-quantitative immunohistochemical assay versus oncotype DX(â) qRT-PCR
assay for estrogen and progesterone receptors: an independent quality assurance study. Mod Pathol 2012;25:869–76.
9. Badve SS, Baehner FL, Gray RP, et al. Estrogen- and progesterone-receptor status in ECOG 2197: comparison of immunohistochemistry by local and central laboratories and quantitative reverse
transcription polymerase chain reaction by central laboratory. J Clin
Oncol 2008;26:2473–81.
10. Allison KH, Kandalaft PL, Sitlani CM, Dintzis SM, Gown
AM. Routine pathologic parameters can predict Oncotype DX
recurrence scores in subsets of ER positive patients: who does
not always need testing? Breast Cancer Res Treat 2012;131:
413–24.
11. Dabbs DJ, Klein ME, Mohsin SK, Tubbs RR, Shuai Y,
Bhargava R. High false-negative rate of HER2 quantitative reverse
transcription polymerase chain reaction of the Oncotype DX test:
an independent quality assurance study. J Clin Oncol 2011;29:
4279–85.
12. Baehner FL, Achacoso N, Maddala T, et al. Human epidermal growth factor receptor 2 assessment in a case-control study:
comparison of fluorescence in situ hybridization and quantitative
reverse transcription polymerase chain reaction performed by central
laboratories. J Clin Oncol 2010;28:4300–6.
13. Dvorak L, Dolan M, Fink J, Varghese L, Henriksen J, Gulbahce HE. Correlation between HER2 determined by fluorescence
in situ hybridization and reverse transcription-polymerase chain
reaction of the Oncotype DX test. Appl Immunohistochem Mol
Morphol 2013;21:196–9.
14. Genomic Health. The Oncotype DXâ Breast Cancer Assay
Report Includes a Quantitative HER2 Score brochure. Available at:
http://www.oncotypedx.com/en-US/Breast/HealthcareProfessionalsIn
vasive/Overview/~/media/157A0C5B39DD461F915CEA5CE05D5E27.
ashx (accessed September 11, 2012).
15. Genomic Health. Quantitative Single Gene Scores. Available
at: http://www.oncotypedx.com/en-US/Breast/HealthcareProfession
alsInvasive/Overview/Scores (accessed September 11, 2012).
16. Acs G, Esposito NN, Kiluk J, Loftus L, Laronga C. A mitotically active, cellular tumor stroma and/or inflammatory cells associated with tumor cells may contribute to intermediate or high
Oncotype DX Recurrence Scores in low-grade invasive breast carcinomas. Mod Pathol 2012;25:556–66.
17. Baehner FL, Pomeroy C, Cherbavaz C, Shak S. Biopsy cavities in breast cancer specimens: their impact on quantitative
RT-PCR gene expression profiles and recurrence risk assessment.
Mod Pathol 2009;22(1s):28A–9A.