Prognostic utility of the breast cancer index and comparison to Adjuvant! Online in a clinical case series of early breast cancer

Article (PDF Available)inBreast cancer research: BCR 13(5):R98 · October 2011with56 Reads
DOI: 10.1186/bcr3038 · Source: PubMed
Abstract
Breast Cancer Index (BCI) combines two independent biomarkers, HOXB13:IL17BR (H:I) and the 5-gene molecular grade index (MGI), that assess estrogen-mediated signalling and tumor grade, respectively. BCI stratifies early-stage estrogen-receptor positive (ER+), lymph-node negative (LN-) breast cancer patients into three risk groups and provides a continuous assessment of individual risk of distant recurrence. Objectives of the current study were to validate BCI in a clinical case series and to compare the prognostic utility of BCI and Adjuvant!Online (AO). Tumor samples from 265 ER+LN- tamoxifen-treated patients were identified from a single academic institution's cancer research registry. The BCI assay was performed and scores were assigned based on a pre-determined risk model. Risk was assessed by BCI and AO and correlated to clinical outcomes in the patient cohort. BCI was a significant predictor of outcome in a cohort of 265 ER+LN- patients (median age: 56-y; median follow-up: 10.3-y), treated with adjuvant tamoxifen alone or tamoxifen with chemotherapy (32%). BCI categorized 55%, 21%, and 24% of patients as low, intermediate and high-risk, respectively. The 10-year rates of distant recurrence were 6.6%, 12.1% and 31.9% and of breast cancer-specific mortality were 3.8%, 3.6% and 22.1% in low, intermediate, and high-risk groups, respectively. In a multivariate analysis including clinicopathological factors, BCI was a significant predictor of distant recurrence (HR for 5-unit increase = 5.32 [CI 2.18-13.01; P = 0.0002]) and breast cancer-specific mortality (HR for a 5-unit increase = 9.60 [CI 3.20-28.80; P < 0.0001]). AO was significantly associated with risk of recurrence. In a separate multivariate analysis, both BCI and AO were significantly predictive of outcome. In a time-dependent (10-y) ROC curve accuracy analysis of recurrence risk, the addition of BCI+AO increased predictive accuracy in all patients from 66% (AO only) to 76% (AO+BCI) and in tamoxifen-only treated patients from 65% to 81%. This study validates the prognostic performance of BCI in ER+LN- patients. In this characteristically low-risk cohort, BCI classified high versus low-risk groups with ~5-fold difference in 10-year risk of distant recurrence and breast cancer-specific death. BCI and AO are independent predictors with BCI having additive utility beyond standard of care parameters that are encompassed in AO.

Figures

RESEARCH ARTICLE Open Access
Prognostic utility of the breast cancer index and
comparison to Adjuvant! Online in a clinical case
series of early breast cancer
Rachel C Jankowitz
1*
, Kristine Cooper
2
, Mark G Erlander
3
, Xiao-Jun Ma
3
, Nicole C Kesty
3
, Hongying Li
3
,
Mamatha Chivukula
4
and Adam Brufsky
1
Abstract
Introduction: Breast Cancer Index (BCI) combines two independent biom arkers, HOXB13:IL17BR (H:I) and the 5-gene
molecular grade index (MGI), that assess estrogen-mediated signalling and tumor grade, respectively. BCI stratifies
early-stage estrogen-receptor positive (ER+), lymph-node negative (LN-) breast cancer patients into three risk groups
and provides a continuous assessment of individual risk of distant recurrence. Objectives of the current study were to
validate BCI in a clinical case series and to compare the prognostic utility of BCI and Adjuvant!Online (AO).
Methods: Tumor samples from 265 ER+LN- tamoxifen-treated patients were identified from a single academic
institutions cancer research registry. The BCI assay was performed and scores were assigned based on a pre-
determined risk model. Risk was assessed by BCI and AO and correlated to clinical out comes in the patient cohort.
Results: BCI was a significant predictor of outcome in a cohort of 265 ER+LN- patients (median age: 56-y; median
follow-up: 10.3-y), treated with adjuvant tamoxifen alone or tamoxifen with chemotherapy (32%). BCI categorized
55%, 21%, and 24% of patients as low, intermediate and high-risk, respectively. The 10-year rates of distant
recurrence were 6.6%, 12.1% and 31.9% and of breast cancer-specific mortality were 3.8%, 3.6% and 22.1% in low,
intermediate, and high-risk groups, respectively. In a multivariate analysis including clinic opathological factors, BCI
was a significant predictor of distant recurrence (HR for 5-u nit increase = 5.32 [CI 2.18-13.01; P = 0.0002]) and
breast cancer-specific mortality (HR for a 5-unit increase = 9.60 [CI 3.20-28.80; P < 0.0001]). AO was significantly
associated with risk of recurrence. In a separate multivariate analysis, both BCI and AO were significantly predictive
of outcome. In a time-dependent (10-y) ROC curve accuracy analysis of recurrence risk, the addition of BCI+AO
increased predictive accuracy in all patients from 66% (AO only) to 76% (AO+BCI) and in tamoxifen-only treated
patients from 65% to 81%.
Conclusions: This study validates the prognostic performance of BCI in ER+LN- patients. In this characteristically
low-risk cohort, BCI classified high versus low-risk groups with ~5-fold difference in 10-year risk of distant
recurrence and breast cancer-specific death. BCI and AO are independent predicto rs with BCI having additive utility
beyond standard of care parameters that are encompassed in AO.
Introduction
Approximately 75% of wo men with early-stage, estrogen
receptor-positive (ER
+
), lymph node-negative (LN
-
)
breast cancer treated with endocrine therapy do not
develop distant relapse of disease [1]. Adjuvant
endocrine therapy alone may be sufficient for the major-
ity of ER
+
LN
-
patients. However, because it is difficult
to identify tho se E R
+
LN
-
patients with a low risk for
recurrence, many patients are subjected to unnecessary
chemotherapy and the potential for increased toxicity
without added benefit [1]. Currently, within the early-
stage hormone receptor-positive patient population,
treating physicians rely heavily on traditional clinico-
pathological criteria such as age, tumor size, and tumor
* Correspondence: jankowitzr@upmc.edu
1
Department of Medicine, Division of Hematology/Oncology, UPMC,
University of Pittsburgh Cancer Institute, 300 Halket Street, Suite 4628,
Pittsburgh, PA 15213, USA
Full list of author information is available at the end of the article
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
© 2011 Jankowitz et al.; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative
Commons Attribution License (http://creativecom mons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
grade to better quantify individual risk and direct sys-
temic therapy.
Adjuvant! Online (AO) is a web-based actuarial tool
that incorporates such criteria in order to predict
patient o utcome at 10 years on the basis of data from
the Surveillance, Epidemiology, and End Results (SEER)
registry and therapeutic benefit on the basis of the
Oxford overviews of randomized clinical trials [2]. The
AO model was developed in the US and validated with
a Canadian cohort [2] but is subject to a number of lim-
itations. For example, individualized AO estimates of
recurrence risk are sensitive to variability in comorbidity
assessment [3], and its (Adjuvant! Online) estimates of
recurrence are not truly individualized, because they are
based on data incorporated into binned categories
(tumor size, nodal status, and so on) [4].
Genomic-based assays are standardized, reproducible
prognostic tools that have the potential to assess recur-
rence risk beyond standard clinicopathological para-
meters [5]. The Breast Cancer Index (BCI), a gene
expression-based signature, is a continuous risk predic-
tion model that combines the gene expression profiles
of the HOXB13/IL17B R ratio (H:I) and the Molecular
Grade Index (MGI) [6,7]. BCI was developed and tested
within the p reviously conducted prospective Stockholm
trial of low-risk, ER
+
LN
-
women rando mly assigned to
tamoxifen v ersus no therapy [6,8]. Within this cohort,
BCI i s a highly significant prognostic tool, identifying
more than 50% of patient s as low-risk [6]. Although H:I
and MGI both predict clinical outcome in patients with
ER
+
LN
-
breast cancer, these biomarkers measure differ-
ent components of the underlying tumor biology. H:I is
linked to dysfunctional estrogen signalling in breast can-
cer [9], and MGI is a fi ve-gene signature that recapitu-
lates tumor grade [7]. In a previous study, the
combination of H:I and MGI was demonstrated t o b e
superior to either index a lone in predicting brea st can-
cer recurrence risk in ER
+
LN
-
breast cancer [7]. The
aims of the present study were to validate BCI in a clini-
cal case series and to compare the prognostic utility of
BCI with that of AO.
Materials and methods
Study population and tumor samples
Patients were identified by a search of the Cancer Regis-
try Data at the UPMC University of Pittsburgh Cancer
Institute, a large database that includes comprehensive
demogra phic and clinical outcome data on patients with
breast cancer. Eligible patients included al l subject s who
had ER
+
(10% or more nuclear stai ning of the tumor
cells), LN
-
invasive breast cancer diagnosed between
1990 and 1999 and who had received adjuvant tamoxi-
fen. In addition, patients were included in the study
only if there wa s an associated formalin-fixed, paraffin-
embedded (FFPE) tissue block (or blocks) from the time
of original diagnosis in order to confirm ER status and
tumor grade. This study was approved by the Institu-
tional Review Board at the University of Pittsburgh. In
accordance with the approval, informed consent from
the patients was not required.
A h ematoxylin and eosin (H&E) slide for each case
was reviewed to confirm tumor grade by using the Not-
tingham grading system to include tubules, nuclear
grad e, and mitosis. To confirm ER status, one represen-
tative tumor block was selected for immunohistochemis-
try (IHC) analysis. IHC was performed on the s elected
FFPE tissue blocks with pre-dilute rabbit monoclonal
antibodies directed against ER alpha (SP1; Ventana,
Tucson, AZ, USA), and the recommended protocols o f
the manufacturers were followed as specified. A tumor
was conside red ER-positive if there w ere 10% or more
nuclear staining of the tumor cells.
Breast Cancer Index and Adjuvant! Online calculation
A representative block for each case was processed at
UPMC, and unstained sections were sent to bioTheranos-
tics, Inc. (San Diego, CA, USA) for BCI analysis. All
laboratory and raw data analyses were subsequently com-
pleted at bioTheranostics, Inc. without knowledge of clini-
cal outcome. For each case, an H&E slide was generated at
bioTheranostics, Inc. and examined to confirm 40 % con-
tent of invasive cancer. Macrodissection of invasive tumor
cells was completed. RNA was extracted and a real-time
reverse transcription-polymerase chain reaction assay for
BCI was performed as previously described [7,10]. The
genes analyzed were HOXB13, IL17BR (HOXB13:IL17BR
(H:I) index), BUB1B, CENPA, NEK2, RACGAP1, RRM2
(Molecular Grade Index), ACTB, HMBS, SDHA, and UBC
(reference genes). Primer and probe sequences for these
genes were the same as previously described [7,11]. Raw
polymerase chain reaction data were collected and subse-
quently the BCI algo rithm was calculated and categorical
risk was assigned for each case: low risk, BCI of less than
5; intermediate risk, BCI of from at least 5 to less than 6.4;
high risk, BCI of at least 6.4 [6]. Predefined BCI scores
were sent to UPMC, and, once they were submitted, asso-
ciated clinical outcome for each patient was linke d with
the respective BCI risk score. AO scores for risk of recur-
renceandmortalitywerecalculatedonthebasisofthe
online tool Adjuvant! for Br east Cancer version 8.0 [12].
Clinicopathological data, which included age, tumor size,
tumor grade, ER status, and nodal status, were collected
for each patient by a third-party, honest broker, who then
provided the de-identified outcome information to the
clinical investigators. The clinicopathological data and
treatment for individual patients were then inputted
online, and AO 10-year predictions for recurrence and
mortality were obtained.
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
Page 2 of 8
Statistical analysis
Breast Cancer Index validation
The primary endpoint was distant recurrence-free survi-
val (defined as the time from diagnosis to the time of
first distant metastasis). Secondary endpoints included
breast cancer-specific survi val (defined as the time from
diagnosis to the time of a breast cancer-related death),
overall su rvival (defined as the time from di agnosis to
the time of death from any cause), and 10-year rates of
distant metastasis, all-cause mortality, and breast can-
cer-specific mortality. For time to distant recurrence,
event times were censored at the time of a second pri-
mary or at the time of a local recurrence or at the last
follow-up time for those with no event. For time to
breast cancer death, event times were censored at the
time of a second primary or at the time of non-breast
cancer death or at t he last follow-up time for those
alive. For time to death by any cause, the event times
were censored at the last follow-up time for those alive.
Chi-square test s fo r ind ependence were performed to
assess the association between categorical variables. Dis-
tant recurrence-free survival, overall survival, and breast
cancer-specific surviva l and the 10-year rates of distant
rec urrence, all-cause mortality, and breast cancer-spe ci-
fic mortality w ere e stimated for the population from
Kaplan-Meier survival curves. Cox models were used to
assess whether BCI as a continuous risk model pro vided
prognostic information independently of traditional clin-
ical and histological parameters (age, tumor size, tumor
grade, and so on ). The hazard ratio (HR) for the contin-
uous BCI score was calc ulated relative to an increment
of 5 BCI units.
Breast Cancer Index comparison with Adjuvant! Online
To compare BCI with AO, the continuous risk model of
BCI was used as previously described [6]. AO predicts
rec urrence defined as the reappearance of breast cancer
at any site (local, contralateral, or distant) and breast
cancer mortality (10-year follow-up), both of w hich are
estimated and derived from total survival analyses of
SEER data [13]. To compare the prognostic abilities of
AO and BCI, the 10-year predicted risk of any recur-
rence or mortality was calculated from AO and evalu-
ated in conjunction with the predicted risk of distant
recurrence by BCI. Given the differences in defining
breast cancer-specific death between BCI and AO,
breast cancer survival with AO was compared with
breast cancer-specific survival and overall survival with
BCI. BCI and AO are scored on dif ferent numerical
scales: 0 to 10 for BCI and 0 to 100 for AO. To more
accurately compare BCI with AO, HRs were calculated
relative to an increment of their interquartile ranges
(2.936 for BCI and 5 for AO). All statistical procedures
- with the except ion of the receiver operating character-
istic (R OC) analysis, which was per formed in R (version
2.13.0) - were conducted with the statistical software
SAS (version 9.2; SAS Institute Inc., Cary, NC, USA).
The predictive accuracy of the BCI score was deter-
mined by using a time-dependent ROC curve method
for censored surviva l data [14,15]. A Cox model with
selected covariates was fit, and the predicted values were
used to generat e time-dependent sensitivity and specifi-
city and the corresponding ROC curve at each o bserved
event time. The area under the curve was plotted over
time to assess the predictive accuracy of the model in
dis tinguishing subjects who have an event before time t
from those who do not. Models co ntaining only the AO
score versus models containing the AO plus the BCI
scores were generated to quantify the impact that BCI
had on t he predictive accuracy for the outcomes of time
to distant recurrence a nd time to breast cancer-specific
death for all subjects and for subjects treated with
tamoxifen alone.
To de termine the accuracy of risk assessment by BCI
and AO over a period of 10 years, we completed a glo-
bal concordance summary (integrated area under the
curve, or iAUC), which is a measure of agreement
between survival time and predicted risk over a period
of 10 years of follow-up [14,15]. The concordance
measure estimates the probability that, for two ran-
domly chosen individuals, the subject with the shorter
survival time also has the larger risk score. A model
with perfect agreement would have a value close to 1,
whereas a value of 0.5 is no better than random
chance.
Results
Patient characteristics and Breast Cancer Index
distribution
Of the 386 patients who were identified in the UPMC
database for analysis and who met the eligibility criteria
(ER
+
LN
-
tamoxifen-treated), 265 had paraffin tissue
blocks available for assessing BCI. Patient characteristics
are shown in Table 1. Overall, the median time of fol-
low-up for all patients was 10.3 years, the rate of distant
recurrence was 15%, and mortality rates were 11% and
15% for breast cancer-specific and all-cause mortality,
respecti vely. As this was a clinical case series, treatment
was not un iform. Of the 265 patients, 85 (32%) received
adjuvant chemotherapy in addition to adjuvant tamoxi-
fen, whereas 180 (68%) were treated with tamoxifen
alone. Non-recurrence and recurrence rates between
these two treatment groups were not significantly differ-
ent (P ~0.2), and the chemotherapy-treated group con-
sisted of patients who were younger (51% were younger
than 50 years old; P < 0.0001) and who had higher
tumor grade (P = 0.04) and larger tumors (P < 0.0001)
in comparison with the group treated with tamoxifen
alone.
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
Page 3 of 8
For all 265 patients, the pre-defined categorical risk
stratification by BCI resulted in more than half of the
patients (55%) stratified as low risk with 21% and 24%
as intermediate and high risk, respectively (Table 2). For
the 180 patients treated with tamoxifen alone, BCI stra-
tified 59%, 19% and 22% of patients as low, intermediate
and high risk respectively (Table 2).
Association of Breast Cancer Index with patient outcome
The 10-year rates of distant metastasis, all-cause mor-
tality, and breast cancer-specific mortality are shown
in Tabl e 2. For all patients, the 10-year rates of distant
metastasis-free survival were 93.4%, 87.9%, and 68.1%
for low-, intermediate-, and high-risk BCI categories,
respectively (Table 2). The overall survival rates were
93.3%, 92.0%, and 71.8% and the breast cancer-specific
survival rates were 96.2%, 96.4%, a nd 77.9% for the
low-, intermediate-, and high-risk BCI groups,
respectively.
For patients treated with tamoxifen alone, the 10-year
rates for re currence and mortal ity were similar. The 10-
year rates of dista nt metastasis-free survival were 94.7%,
90.9%, and 66.8% for low-, intermediate-, and high-risk
groups, respectively. The rates for ove rall survival were
92.8%, 90.0%, and 72.2% and the rates for breast cancer-
specific survival were 96.7%, 97.1%, and 77.4% for the
low-, intermediate-, and high-risk groups, respectively.
Proportional hazards models for outcome w ere per-
formed with and without BCI and inc luded the known
prognostic cl inical variables of age, tumor size, tumor
grade, and treatment (Tables 3, 4, 5). BCI was evalua ted
as a continuous variable as previously developed in a
retrospective analysis of the Stockhol m trial [6]; the HR
for recurrence and mortality was calcul ated relative to
an increment of 5 BCI units, which is half the range of
BCI.
When BCI was included in the model of known prog-
nostic factors, BCI w as highly significant and was
Table 1 Patient characteristics
Description Patients without chemotherapy Patients with chemotherapy P value
a
All patients
Number 180 85 265
Age, years
Median (range) 60 (34, 81) 49 (25, 70) 56 (25, 81)
50 143 (79%) 42 (49%) < 0.0001
Tumor grade 0.040
1 53 (29%) 13 (15%) 66 (25%)
2 101 (56%) 55 (65%) 156 (59%)
3 26 (14%) 17 (20%) 43 (16%)
Tumor size
Mean (standard deviation) 1.4 (0.9) 2.2 (1.4) 1.7 (1.2)
2 cm 32 (18%) 45 (53%) < 0.0001 77 (29%)
Radiation treatment 0.5073
Yes 150 (83%) 68 (80%) 218 (82%)
No 30 (17%) 17 (20%) 47 (18%)
Recurrence event 0.199
No recurrence 137 (76%) 66 (78%) 203 (77%)
Locoregional/Contralateral recurrence 12 (7%) 1 (1%) 13 (5%)
Distant recurrence 25 (14%) 16 (19%) 41 (15%)
Second primary 6 (3%) 2 (2%) 8 (3%)
Median (range) follow-up, years 10.2 (2.3, 18.0) 10.3 (2.9, 17.3) 10.3 (2.3, 18.0)
Breast cancer-specific death events 21 (12%) 9 (11%) 30 (11%)
All-cause death events 29 (16%) 11 (13%) 40 (15%)
a
Comparing tamoxifen alone versus tamoxifen plus chemotherapy groups.
Table 2 Kaplan-Meier estimates of the rate of event for Breast Cancer Index risk groups
Rate at 10 years (standard error)
Risk category Percentage of patients
(n = 265)
Distant metastasis All-cause mortality Breast cancer-specific mortality
Low 54.7% 6.6% (2.2%) 6.7% (2.2%) 3.8% (1.7%)
Intermediate 21.1% 12.1% (4.8%) 8.0% (3.9%) 3.6% (2.5%)
High 24.2% 31.9% (6.1%) 28.2% (5.9%) 22.1% (5.4%)
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
Page 4 of 8
associated with recurrence risk (HR = 5.32; 95% confi-
dence interval (CI) 2.18 to 13.01; P = 0.0002; Table 2),
all-cause mortality (HR = 6.77; 95% C I 2.80 to 16.41; P
< 0.0001; Table 3), and breast cancer-specific mortality
(HR = 9.60; 95% CI 3.20 to 28.80; P < 0.0001; Table 4).
Comparison of Breast Cancer Index with Adjuvant! Online
Univariate Cox models were fit separately for both BCI
and AO. AO was significantly a ssociated with risk of
recurrence (HR = 1.53; 95% CI 1.25 to 1.86; P < 0.0001),
all-cause mortality (HR = 1.58; 95% CI 1.22 to 2.06; P =
0.0006), and breast cancer -specific mortality (HR = 1.68;
95% CI 1.24 to 2.23; P = 0 .0001). BCI was also signifi-
cantly associated with risk of recurrence (HR = 2.77;
95% CI 1.71 to 4.51; P < 0.0001), all-cause mortality
(HR = 3.03; 95% CI 1.86 to 4.94; P < 0.0001), and breast
cancer-specific mortality (HR = 3 .53; 95% CI 1.98 to
6.31; P < 0.0001). A combined multivariate analysis with
only BCI and AO showed that both independently
remained significantly associated with risk of recurrence
(Table 3), all-cause mortality (Table 4), and breast can-
cer-specific mortality (Table 5).
To determine the accuracy of risk assessment by BCI
and AO over the course of a 10-year period, we com-
pleted a global concordance summary, which is a mea-
sure of agreement between survival time and predicted
risk over the course of a 10-year period of follow-up
(iAUC). The concordance measure estimates the prob-
ability that, for two randomly chosen individuals, the
subject with the shorter surv ival time also has the larger
risk score. A model with perfect agreement would have
avaluecloseto1,whereasavalueof0.5isnobetter
than chance. For time to distant recurrence for all
patients, iAUC values were 0.642 (0.586, 0.716) for mod-
els with AO only and 0.717 (0.656, 0.800) for mode ls
with AO+BCI. For the patients treated with tamoxifen
alone, these probability values increased to an iAUC of
0.671 (0.589, 0.754) and 0.750 (0.659, 0.847) for models
with AO only and AO+BCI, respectively.
Time- dependent ROC cur ve analyses over the 10-year
period of outcome demonstrated that, for early time
points (< 4 years), the model scores provided limited
differential ability (between AO versus AO+BCI) in dis-
tinguishing those patients who had the event before
time t from those who did not. However, for time points
approximately 4 years after diagnosis, the predictive
accuracy for recurrence increased for models including
BCI compared with models with AO only. For all
patients, the minimum and maximum predictive accura-
cies from 4 t o 10 years were 61.4% to 66.2% for AO
only and 71.1% to 75.7% for AO+BCI. For patients
receiving tamoxifen a lone, the ranges for this time per-
iod were 54.6% to 65.0% for AO and 73.4% to 80.7% for
AO+BCI (Figure 1).
Table 3 Cox proportional hazards model for likelihood of
distant recurrence
Hazard ratio
(95% CI)
P value
Multivariate without BCI
Age, 50 versus < 50 1.46 (0.70-3.05) 0.317
Tumor size, 2 cm versus < 2 cm 1.84 (0.94-3.58) 0.074
Tumor grade, reference grade 1 0.006
Grade 2 3.60 (1.08-12.03) 0.038
Grade 3 7.04 (2.00-24.80) 0.002
Treatment
a
1.13 (0.56-2.28) 0.742
Multivariate with BCI
Age, 50 versus < 50 1.34 (0.64-2.82) 0.444
Tumor size, 2 cm versus < 2 cm 1.69 (0.87-3.28) 0.124
Tumor grade, reference grade 1 0.009
Grade 2 3.73 (1.11-12.54) 0.033
Grade 3 6.77 (1.92-23.81) 0.003
Treatment
a
0.99 (0.49-2.03) 0.985
BCI
b
5.32 (2.18-13.01) 0.0002
Multivariate with only BCI and AO
Adjuvant! Online
c
1.42 (1.16-1.74) 0.0007
BCI
c
2.47 (1.50-4.07) 0.0004
a
Tamoxifen alone versus tamoxifen plus chemotherapy groups.
b
Hazard ratio
calculated for Breast Cancer Index (BCI) relative to an increment of 5 units.
c
Hazard ratio calculated for BCI and Adjuvant! Online (AO) relative to
increments of their interquartile range (2.936 for BCI and 5 for AO). CI,
confidence interval.
Table 4 Cox proportional hazards model for all-cause
mortality
Hazard ratio
(95% CI)
P value
Multivariate without BCI
Age, 50 versus < 50 2.77 (1.16-6.58) 0.021
Tumor size, 2 cm versus < 2 cm 1.24 (0.62-2.48) 0.541
Tumor grade, reference grade 1 0.004
Grade 2 1.96 (0.73-5.21) 0.180
Grade 3 4.81 (1.71-13.57) 0.003
Treatment
a
0.99 (0.46-2.14) 0.981
Multivariate with BCI
Age, 50 versus < 50 2.62 (1.10-6.28) 0.031
Tumor size, 2 cm versus < 2 cm 1.13 (0.56-2.27) 0.728
Tumor grade, reference grade 1 0.004
Grade 2 2.02 (0.75-5.38) 0.162
Grade 3 4.86 (1.73-13.66) 0.003
Treatment
a
0.90 (0.42-1.96) 0.796
BCI
b
6.77 (2.80-16.41) < 0.0001
Multivariate with only BCI and AO
Adjuvant! Online
c
1.45 (1.10-1.91) 0.009
BCI
c
2.82 (1.70-4.67) < 0.0001
a
Tamoxifen alone versus tamoxifen plus chemotherapy groups.
b
Hazard ratio
calculated for Breast Cancer Index (BCI) relative to an increment of 5 units.
c
Hazard ratio calculated for BCI and Adjuvant! Online (AO) relative to
increments of their interquartile range (2.936 for BCI and 5 for AO). CI,
confidence interval.
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
Page 5 of 8
Discussion
Prognostication of individual risk for distant recurrence and
death for patients with ER
+
LN
-
breast cancer treated only
with adjuvant endocrine therapy continues to be a challenge
given that approximately 75% of these patients remain dis-
ease-free at 15 years [1]. In addition, the Oxford overview
of a meta-analysis of 34,873 women reports the low recur-
rence rate of 2.2% per year for ER
+
LN
-
patients receiving
tamoxifen alone [16]. Furthermore, 12-year follow-up of
1,536 patients who had ER
+
LN
-
breast cancer and who
were randomly assigned to tamoxifen versus tamoxifen plus
chemotherapy had r ecurrence-free survival rates of 79%
and 89%, respectively, and overall survival rates of 83% and
87%, re spectively, and absolute chemotherapy benefits of
10% for recurrence-free survival and 4% for overall survival.
A large proportion of ER
+
LN
-
patients are disease-free
more than 10 years after diagnosis when treated with endo-
crine therapy alone and, therefore, could forgo chemother-
apy and the associated toxicities. Prognostic indices, which
identify low-risk patients, can be used to generate informa-
tion to aid in the treatment decision process for determin-
ing optimal adjuvant therapy for ER
+
LN
-
patients.
Early breast cancer assessment tools for prognosis such
as the St Gallen breast cancer consensus guidelines and
the AO assess risks and benefits associated with adjuvant
therapy [13,17,18]. For AO, the selection criteria for with-
holding chemotherapy is b ased primarily on the
integration of clinicopathological correlates (tumor grade,
nodal status, tumor size, and ER status) and comorbidities,
all of which have inherent limitations in assessment or
measurement [3,4]; however, correct assessment is of
major importance in order to avoid unnecessary toxic side
effects associated with chemotherapy [18].
Here, w e report the prognostic performance of the
gene expression-based BCI within a clinical case series
of patients with ER
+
LN
-
breast cancer and demonstrate
that BCI is a highly significant predictor of distant
metastasis and death in patients treated with adjuvant
tamoxife n, with or without chem otherapy. With catego-
rical stratification, BCI identified more than 50% of the
patients with low risk with a 10-year rate of recurrence
of 6.6% and breast cancer-specific mortality rate of 3.8%.
In a multivariate model that includes clinicopathological
covariates, BCI remained a significant factor associated
with recurrence risk and mortality.
Prognostic gene expression signatures have the cap-
ability to offer objective and reproducible predictive risk
ass essmen ts beyond the traditional cri teria used for AO
to guide adjuvant treatment selection. The overall corr e-
lation between BCI and AO for individual 10-year risk
of recurrence assessment was low (r = 0.2), and AO risk
was higher for most patients. For example, of the studys
226 patients who were defined by AO as having at least
a 10% risk of recurrence at 10 years, 51% were classified
by BCI as low-risk, 22% as intermediate-risk, and 27% as
high-risk. Actual recurrence rates for these three groups
were 10%, 14%, and 34%, respectively.
Proportional hazards models that include both AO
and BCI demonstrate that both factors are highly signifi-
cant predictors for outcome. Furthermore, iAUC ana-
lyses demonstrate that the addition of BCI to AO
increases the probability for concordance between survi-
val time and predicted risks from 67% to 75% globally
for 10-year outcome for ER
+
LN
-
patients receiving
tamoxifen only. Time-dependent ROC analysis enables
further granulation of the additive effect of BCI to AO
for predicting recurrence. The addition of BCI has its
greatest accuracy benefit 4 to 10 years after diagnosis;
for patients treated with tamoxifen only, BCI + AO had
a recurrence prediction of 81% versus 65% for AO only.
A potential reason for the observed additive accuracy
effect of BCI after 4 years may be that tumor grade, a clini-
copathological covariate used in calculating AO, has its
greatest prognostic strength 0 to 5 years after diagn osis;
specifically, high-grade t umors are associated with
increased recurrence risk. In contrast, a retrospective
study (n = 2,838 patients) designed to identify factors asso-
ciated with residual risk of recurrence (> 5 years) reported
that only low-grade tumors were significantly associated
with recurrence for patients receiving adjuvant therapy
and d isease-free for 5 years [19]. This suggests that
Table 5 Cox proportional hazards model for breast
cancer-specific mortality
Hazard ratio
(95% CI)
P value
Multivariate without BCI
Age, 50 versus < 50 2.34 (0.90-6.13) 0.082
Tumor size, 2 cm versus < 2 cm 1.36 (0.61-3.02) 0.451
Tumor grade, reference grade 1 0.008
Grade 2 8.08 (1.07-60.88) 0.043
Grade 3 18.22 (2.33-142.53) 0.006
Treatment
a
1.09 (0.45-2.61) 0.850
Multivariate with BCI
Age, 50 versus < 50 2.23 (0.85-5.87) 0.103
Tumor size, 2 cm versus < 2 cm 1.28 (0.57-2.87) 0.554
Tumor grade, reference grade 1 0.008
Grade 2 8.67 (1.14-65.73) 0.037
Grade 3 18.95 (2.42-148.48) 0.005
Treatment
a
0.96 (0.40-2.33) 0.931
BCI
b
9.60 (3.20-28.80) < 0.0001
Multivariate with only BCI and AO
Adjuvant! Online
c
1.52 (1.11-2.08) 0.009
BCI
c
3.27 (1.79-5.98) 0.0001
a
Tamoxifen alone versus tamoxifen plus chemotherapy groups.
b
Hazard ratio
calculated for Breast Cancer Index (BCI) relative to an increment of 5 units.
c
Hazard ratio calculated for BCI and Adjuvant! Online (AO) relative to
increments of their interquartile range (2.936 for BCI and 5 for AO). CI,
confidence interval.
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
Page 6 of 8
accurate prediction of recurrence throughout the entire 10
years requires not only assessment of tumor differentiation
and tumor proliferation for pr edicting early recurre nces
(that is, 0 to 5 years) but also prediction of recurrences
more than 5 years by i dentifying the su bset of low-grade
or indolent appearing tumors with high potential to
metastasize. BCI is a score that integrates two prognostics:
MGI, an index hig hly correlative with tumor grade [7],
and H:I, which has a low correlation with tumor grade
[11]. Studies are ongoing to further examine the potential
differing prognostic contributions of MGI and H:I for
early and late recurrences, respectively.
Current gene expression-based signatures such as the
21-gene, 70-gene, 76-gene, and 97-gene genomic grade
derive a significant amount of their prognostic and predic-
tive strength fro m the expression measurement of genes
associated with tumor differentiation and proliferation
[20-22]. For example, a multi-institutional study of 198
LN
-
patients without systemic adjuvant therapy demon-
strated that the 70-gene, 76-gene, and 97-gene genomic
grade signatures had similar prognostic performance; in all
three signatures, the genes controlli ng tumor differentia-
tion and proliferation had the greatest prognostic discrimi-
natory strength [20,23]. In addition, among the four
different gene groups of the 21-gene signature, only the
proliferative gene group is significantly associated with
chemot herapy benefit in the ER
+
LN
-
patient coho rt ran-
domly assigned to tamoxifen or tamoxifen plus che-
motherapy [21]. Furthermo re, time dependence of the
prognostic performance of the 76-gene was examined over
a period of 15 years to predict distant metastasis and over-
all survival and was compared with AO; for both risk
assessments, the HRs decline over time [24]. The prognos-
tic performance of the 21-gene signature has been com-
pared with AO and is consistent with our findings that
prediction of the risk of recurrence with the assay is inde-
pendent of that with AO [25,26].
Conclusions
Overall, this study is an independent validation of the
strong prognostic performa nce of BCI. This study is lim-
ited by the fact that it was a retrospec tive, single-institu-
tion study and that results may have been biased on the
basis of specimen availability and patterns of referral to
the tertiary academic center. Additional studies are
ongoi ng to validate BCI in a randomized tr ial population
and to examine the utility of BCI to predict chemosensitiv-
ity and chemot herapy benefit. However, results from this
validation study indicate that BCI identifies a large propor-
tion of low-risk patients and is additive to AO for predict-
ing the risk of recurrences.
Abbreviations
AO: Adjuvant! Online; BCI: Breast Cancer Index; CI: confidence interval; ER:
estrogen receptor; FFPE: formalin-fixed , paraffin-embedded; H&E:
hematoxylin and eosin; H:I: HOXB13:IL17BR; HR: hazard ratio; iAUC: integrated
area under the curve; IHC: immunohistochemistry; LN: lymph node; MGI:
molecular grade index; ROC: receiver operating characteristic; SEER:
Surveillance, Epidemiology, and End Results.
AB
Figure 1 Time-dependent receiver operating characteristic curve analysis of Adjuvant! Online (A O) and AO with Breast Cancer Index
(BCI). Linear predictors are derived from a Cox model with AO only (red circles) and from a model with AO and BCI (blue plus) for (a) all
subjects and for (b) subjects treated with tamoxifen (Tam) alone. Curves plot the area under the curve (AUC) over time and compare the
accuracy of the model score to distinguish patients who will develop a distant metastasis from those who will not. The separation of the curves
demonstrates the gain in predictive accuracy of including BCI in the model.
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
Page 7 of 8
Acknowledgements
The authors would like to thank Ranelle Salunga and Yen Tran for their
technical assistance and Victor Vogel and Cathy Schnabel for critical review.
This study and manuscript preparation were funded by bioTheranostics, Inc.
Author details
1
Department of Medicine, Division of Hematology/Oncology, UPMC,
University of Pittsburgh Cancer Institute, 300 Halket Street, Suite 4628,
Pittsburgh, PA 15213, USA.
2
Department of Biostatistics, UPMC, University of
Pittsburgh Cancer Institute, 300 Halket Street, Suite 4628, Pittsburgh, PA
15213, USA.
3
bioTheranostics, Inc., 9640 Towne Center Drive, Suite 200, San
Diego, CA 92121, USA.
4
Department of Pathology, UPMC, 300 Halket Street,
Suite 4628, Pittsburgh, PA 15213, USA.
Authors contributions
AB developed the concept and contributed to the interpretation of the
data. ME developed the concept, wrote the manuscript, and contributed to
the interpretation of the data. X-JM developed the concept. RCJ developed
the concept, collected the clinical data, wrote the manuscript, and
contributed to the interpretation of the data. MC collected the clinical data.
HL performed the analysis and contributed to the interpretation of the data.
KC performed the analysis, wrote the manuscript, and contributed to the
interpretation of the data. NCK wrote the manuscript and contributed to the
interpretation of the data. All authors read and approved the final
manuscript for publication.
Competing interests
ME and X-JM are employees and stockholders of bioTheranostics, Inc. and
are named inventors of the HOXB13:IL17BR gene signature within an issued
US patent (assignee is bioTheranostics, Inc.). NCK and HL are employees and
stockholders of bioTheranostics, Inc. The other authors declare that they
have no competing interests.
Received: 25 March 2011 Revised: 12 July 2011
Accepted: 14 October 2011 Publi shed: 14 October 2011
References
1. Fisher B, Jeong JH, Bryant J, Anderson S, Dignam J, Fisher ER, Wolmark N:
Treatment of lymph-node-negative, oestrogen-receptor-positive breast
cancer: long-term findings from National Surgical Adjuvant Breast and
Bowel Project randomised clinical trials. Lancet 2004, 364:858-868.
2. Olivotto IA, Bajdik CD, Ravdin PM, Speers CH, Coldman AJ, Norris BD,
Davis GJ, Chia SK, Gelmon KA: Population-based validation of the
prognostic model ADJUVANT! for early breast cancer. J Clin Oncol 2005,
23:2716-2725.
3. Ozanne EM, Braithwaite D, Sepucha K, Moore D, Esserman L, Belkora J:
Sensitivity to input variability of the Adjuvant! Online breast cancer
prognostic model. J Clin Oncol 2009, 27:214-219.
4. Paik S, Tang G, Fumagalli D: An ideal prognostic test for estrogen
receptor-positive breast cancer? J Clin Oncol 2008, 26:4058-4059.
5. Bao T, Davidson NE: Gene expression profiling of breast cancer. Adv Surg
2008, 42:249-260.
6. Stal O, Jerevall PL, Ma XJ, Li H, Salunga R, Erlander M, Sgroi D, Holmlund B,
Skoog L, Fornander T: Validation of prognostic utility of HOXB13:IL17BR
and Molecular Grade Index in early stage breast cancer. Cancer Res 2009,
69:504s, (Abstract 577).
7. Ma XJ, Salunga R, Dahiya S, Wang W, Carney E, Durbecq V, Harris A, Goss P,
Sotiriou C, Erlander M, Sgroi D: A five-gene molecular grade index and
HOXB13:IL17BR are complementary prognostic factors in early stage
breast cancer. Clin Cancer Res 2008, 14:2601-2608.
8. Rutqvist LE, Johansson H: Long-term follow-up of the randomized
Stockholm trial on adjuvant tamoxifen among postmenopausal patients
with early stage breast cancer. Acta Oncol 2007, 46:133-145.
9. Wang Z, Dahiya S, Provencher H, Muir B, Carney E, Coser K, Shioda T,
Ma XJ, Sgroi DC: The prognostic biomarkers HOXB13, IL17BR, and CHDH
are regulated by estrogen in breast cancer. Clin Cancer Res 2007,
13:6327-6334.
10. Jerevall PL, Ma XJ, Li H, Salunga R, Kesty NC, Erlander MG, Sgroi DC,
Holmlund B, Skoog L, Fornander T, Nordenskjold B, Stal O: Prognostic
utility of HOXB13: IL17BR and molecular grade index in early-stage
breast cancer patients from the Stockholm trial. Br J Cancer 2011,
104:1762-1769.
11. Ma XJ, Hilsenbeck SG, Wang W, Ding L, Sgroi DC, Bender RA, Osborne CK,
Allred DC, Erlander MG: The HOXB13:IL17BR expression index is a
prognostic factor in early-stage breast cancer. J Clin Oncol 2006,
24:4611-4619.
12. Adjuvant! Online homepage. [http://www.adjuvantonline.com].
13. Ravdin PM, Siminoff LA, Davis GJ, Mercer MB, Hewlett J, Gerson N,
Parker HL: Computer program to assist in making decisions about
adjuvant therapy for women with early breast cancer. J Clin Oncol 2001,
19:980-991.
14. Heagerty PJ, Zheng Y: Survival model predictive accuracy and ROC
curves. Biometrics 2005, 61:92-105.
15. Heagerty PJ, Zeger SL:
Multivariate continuation ratio models:
connections and caveats. Biometrics 2000, 56:719-732.
16. Early Breast Cancer Trialists Collaborative Group (EBCTCG): Effects of
chemotherapy and hormonal therapy for early breast cancer on
recurrence and 15-year survival: an overview of the randomised trials.
Lancet 2005, 365:1687-1717.
17. Goldhirsch A, Coates AS, Gelber RD, Glick JH, Thurlimann B, Senn HJ: First
select the target: better choice of adjuvant treatments for breast cancer
patients. Ann Oncol 2006, 17:1772-1776.
18. Goldhirsch A, Ingle JN, Gelber RD, Coates AS, Thurlimann B, Senn HJ:
Thresholds for therapies: highlights of the St Gallen International Expert
Consensus on the primary therapy of early breast cancer 2009. Ann
Oncol 2009, 20:1319-1329.
19. Brewster AM, Hortobagyi GN, Broglio KR, Kau SW, Santa-Maria CA, Arun B,
Buzdar AU, Booser DJ, Valero V, Bondy M, Esteva FJ: Residual risk of breast
cancer recurrence 5 years after adjuvant therapy. J Natl Cancer Inst 2008,
100:1179-1183.
20. Sotiriou C, Pusztai L: Gene-expression signatures in breast cancer. N Engl J
Med 2009, 360:790-800.
21. Paik S, Tang G, Shak S, Kim C, Baker J, Kim W, Cronin M, Baehner FL,
Watson D, Bryant J, Costantino JP, Geyer CE Jr, Wickerham DL, Wolmark N:
Gene expression and benefit of chemotherapy in women with node-
negative, estrogen receptor-positive breast cancer. J Clin Oncol 2006,
24:3726-3734.
22. Piccart-Gebhart MJ: New developments in hormone receptor-positive
disease. Oncologist 2011, 16(Suppl 1):40-50.
23. Haibe-Kains B, Desmedt C, Piette F, Buyse M, Cardoso F, Vant Veer L,
Piccart M, Bontempi G, Sotiriou C: Comparison of prognostic gene
expression signatures for breast cancer. BMC Genomics 2008, 9:394.
24. Desmedt C, Piette F, Loi S, Wang Y, Lallemand F, Haibe-Kains B, Viale G,
Delorenzi M, Zhang Y, dAssignies MS, Bergh J, Lidereau R, Ellis P, Harris AL,
Klijn JG, Foekens JA, Cardoso F, Piccart MJ, Buyse M, Sotiriou C: Strong time
dependence of the 76-gene prognostic signature for node-negative
breast cancer patients in the TRANSBIG multicenter independent
validation series. Clin Cancer Res 2007, 13:3207-3214.
25. Goldstein LJ, Gray R, Badve S, Childs BH, Yoshizawa C, Rowley S, Shak S,
Baehner FL, Ravdin PM, Davidson NE, Sledge GW Jr, Perez EA, Shulman LN,
Martino S, Sparano JA: Prognostic utility of the 21-gene assay in
hormone receptor-positive operable breast cancer compared with
classical clinicopathologic features. J Clin Oncol 2008, 26:4063-4071.
26. Dowsett M, Cuzick J, Wale C, Forbes J, Mallon EA, Salter J, Quinn E,
Dunbier A, Baum M, Buzdar A, Howell A, Bugarini R, Baehner FL, Shak S:
Prediction of risk of distant recurrence using the 21-gene recurrence
score in node-negative and node-positive postmenopausal patients with
breast cancer treated with anastrozole or tamoxifen: a TransATAC study.
J Clin Oncol 2010, 28:1829-1834.
doi:10.1186/bcr3038
Cite this article as: Jankowitz et al.: Prognostic utility of the breast
cancer index and comparison to Adjuvant! Online in a clinical case
series of early breast cancer. Breast Cancer Research 2011 13:R98.
Jankowitz et al. Breast Cancer Research 2011, 13:R98
http://breast-cancer-research.com/content/13/5/R98
Page 8 of 8
    • "The Breast Cancer Index (BioTheranostics, San Diego, CA, USA) is an RT-PCR-based assay that can be applied to FFPE tissues and is performed by a central laboratory to predict the risk of distant recurrence in ER-positive, lymph node-negative breast cancers. This assay includes two independent biomarkers, the HOXB13:IL17BR ratio and a five-gene molecular grade index that primarily consists of proliferation-related genes [16,17]. The two signatures together comprise the Breast Cancer Index score. "
    [Show abstract] [Hide abstract] ABSTRACT: There is growing consensus that multigene prognostic tests provide useful complementary information to tumor size and grade in estrogen receptor (ER)-positive breast cancers. The tests primarily rely on quantification of ER and proliferation-related genes and combine these into multivariate prediction models. Since ER-negative cancers tend to have higher proliferation rates, the prognostic value of current multigene tests in these cancers is limited. First-generation prognostic signatures (Oncotype DX, MammaPrint, Genomic Grade Index) are substantially more accurate to predict recurrence within the first 5 years than in later years. This has become a limitation with the availability of effective extended adjuvant endocrine therapies. Newer tests (Prosigna, EndoPredict, Breast Cancer Index) appear to possess better prognostic value for late recurrences while also remaining predictive of early relapse. Some clinical prediction problems are more difficult to solve than others: there are no clinically useful prognostic signatures for ER-negative cancers, and drug-specific treatment response predictors also remain elusive. Emerging areas of research involve the development of immune gene signatures that carry modest but significant prognostic value independent of proliferation and ER status and represent candidate predictive markers for immune-targeted therapies. Overall metrics of tumor heterogeneity and genome integrity (for example, homologue recombination deficiency score) are emerging as potential new predictive markers for platinum agents. The recent expansion of high-throughput technology platforms including low-cost sequencing of circulating and tumor-derived DNA and RNA and rapid reliable quantification of microRNA offers new opportunities to build extended prediction models across multiplatform data.
    Full-text · Article · Dec 2015
    • "In an independent validation study, BCI classified 265 ER-positive node-negative patients administered tamoxifen alone or tamoxifen with chemotherapy into low-risk and high-risk cohorts. High-risk patients had a fivefold increase compared to the low-risk cohort in 10-year risk of distant recurrence [23]. "
    [Show abstract] [Hide abstract] ABSTRACT: Genomic assays measuring the expression of multiple genes have made their way into clinical practice and their utilization is now recommended by major international guidelines. A basic property of these tests is their capability to sub-divide patients into high- and low-risk cohorts thereby providing prognostic, and in certain settings, predictive decision support. Here, we summarize commercially available assays for breast cancer including RT-PCR and gene chip-based tests. Given the relative uncertainty in cancer treatment, multigene tests have the potential for a significant cost reduction as they can pinpoint those patients for whom chemotherapy proves to be unnecessary. However, concordance of risk assessment for an individual patient is still far from optimal. Additionally, emerging multigene approaches focus on predicting therapy response, which is a black spot of current tests. Promising techniques include the homologous recombination deficiency score, utilization of massive parallel sequencing to identify driver genes, employment of internet-based meta-analysis tools and investigation of miRNA expression signatures. Combination of multiple simultaneous analyses at diagnosis, including classical histopathological diagnostics, monogenic markers, genomic signatures and clinical parameters will most likely bring maximal benefit for patients. As the main driving force behind such genomic tests is the power to achieve cost reduction due to avoiding unnecessary systemic treatment, the future is most likely to hold a further proliferation of such assays.
    Full-text · Article · Feb 2015
    • "For tamoxifen-treated and untreated ER-positive patients classified as low risk by the BCI, absolute risks of breast cancer death at 10 years post-diagnosis were 1.1% and 5.1% in their study and 3.5% and 5.1% in ours. In the only other study to evaluate the BCI to date [17], corresponding estimates for low-risk patients receiving adjuvant tamoxifen with or without chemotherapy or adjuvant tamoxifen alone were 3.8% and 7.2%, respectively. "
    [Show abstract] [Hide abstract] ABSTRACT: Introduction Studies have shown that a two-gene ratio (HOXB13:IL17BR) and a five-gene (BUB1B, CENPA, NEK2, RACGAP1, RRM2) molecular grade index (MGI) are predictive of clinical outcomes among early-stage breast cancer patients. In an independent population of lymph node-negative breast cancer patients from a community hospital setting, we evaluated the performance of two risk classifiers that have been derived from these gene signatures combined, MGI+HOXB13:IL17BR and the Breast Cancer Index (BCI). Methods A case-control study was conducted among 4,964 Kaiser Permanente patients diagnosed with node-negative invasive breast cancer from 1985 to 1994 who did not receive adjuvant chemotherapy. For 191 cases (breast cancer deaths) and 417 matched controls, archived tumor tissues were available and analyzed for expression levels of the seven genes of interest and four normalization genes by RT-PCR. Logistic regression methods were used to estimate the relative risk (RR) and 10-year absolute risk of breast cancer death associated with prespecified risk categories for MGI+HOXB13:IL17BR and BCI. Results Both MGI+HOXB13:IL17BR and BCI classified over half of all ER-positive patients as low risk. The 10-year absolute risks of breast cancer death for ER-positive, tamoxifen-treated patients classified in the low-, intermediate-, and high-risk groups were 3.7% (95% confidence interval (CI) 1.9% to 5.4%), 5.9% (95% CI 3.0% to 8.6%), and 12.9% (95% CI 7.9% to 17.6%) by MGI+HOXB13:IL17BR and 3.5% (95% CI 1.9% to 5.1%), 7.0% (95% CI 3.8% to 10.1%), and 12.9% (95% CI 7.1% to 18.3%) by BCI. Those for ER-positive, tamoxifen-untreated patients were 5.7% (95% CI 4.0% to 7.4%), 13.8% (95% CI 8.4% to 18.9%), and 15.2% (95% CI 9.4% to 20.5%) by MGI+HOXB13:IL17BR and 5.1% (95% CI 3.6% to 6.6%), 18.6% (95% CI 10.8% to 25.7%), and 17.5% (95% CI 11.1% to 23.5%) by BCI. After adjusting for tumor size and grade, the RRs of breast cancer death comparing high- versus low-risk categories of both classifiers remained elevated but were attenuated for tamoxifen-treated and tamoxifen-untreated patients. Conclusion Among ER-positive, lymph node-negative patients not treated with adjuvant chemotherapy, MGI+HOXB13:IL17BR and BCI were associated with risk of breast cancer death. Both risk classifiers appeared to provide risk information beyond standard prognostic factors.
    Full-text · Article · Mar 2013
Show more