ARTICLE IN PRESS

The Breast (2005) 14, 493–499

THE
BREAST
www.elsevier.com/locate/breast

ORIGINAL ARTICLE

Primary therapy of early breast cancer: 9th International St. Gallen consensus conference

Prognostic and predictive factors revisited

Daniel F. Hayes

Breast Oncology Program, University of Michigan Comprehensive Cancer Center, CCGC 6312,
1500 E. Medical Center Drive, Ann Arbor, MI 48109, USA

KEYWORDS Summary Standard prognostic factors include clinical and pathological staging,
Breast cancer; especially lymph node status and tumor size. Tumor grade and estimates of
Prognostic factors; lymphatic invasion appear to be moderately strong predictive factors, but
Predictive factors; reproducibility is poor, especially for grade 2 tumors. Standard predictive factors
Multi-gene include hormone receptor status and HER-2 amplification and/or over-expression for
expression arrays selection of endocrine therapy and, at least for clinical trials and in the metastatic
setting, of trastuzumab, respectively. Three new markers appear particularly
promising: detection of bone marrow metastases, either at baseline or after 2–4
years of follow-up; expression of UPA/PAI-1 by the primary cancer; and recognition
of simultaneous multiple gene expression patterns, or ‘‘signatures.’’ Important
caveats exist for each of these. Although new technologies offer exciting and
promising new approaches to determining a patient’s prognosis and whether she will
or will not benefit from specific therapies, few have been validated in well-designed,
Level of Evidence I studies. In particular, available data are often confounded by
patient selection and the effects of systemic therapy, which are often not
determined prospectively, not included in analyses, and not reported adequately.
& 2005 Elsevier Ltd. All rights reserved.

Introduction tality reduction? Unfortunately, the side effects,

Mortality from breast cancer is dropping in the
Western world, principally due to the application of
early systemic therapy.1–3 These encouraging data
raise the question: should all patients with newly
diagnosed breast cancer be treated with all
available therapies to ensure breast cancer mor-
Tel.: +1 734 615 6725; fax: +1 734 615 3947.
E-mail address: hayesdf@umich.edu.
toxicities, and cost reduce the appeal of such a
strategy. Therefore, judicial application of accu-
rate, validated, and powerful prognostic and
predictive factors is imperative to increase effi-
ciency of administration of adjuvant systemic
therapy.4
In this regard, standard prognostic factors in-
clude clinical and pathological staging, especially
lymph node status and tumor size. Tumor grade and
estimates of lymphatic invasion appear to be

0960-9776/$ - see front matter & 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.breast.2005.08.023
 ARTICLE IN PRESS
494
moderately strong predictive factors, but reprodu-
cibility is poor, especially for grade 2 tumors.5
Standard predictive factors include hormone re-
ceptor status for selection of endocrine therapy
and HER-2 amplification and/or over-expression for
selection of trastuzumab in the metastatic setting
and at least for clinical trials in the adjuvant
setting.6
Newer factors
More than 100 putative breast cancer prognostic
factors have been reported over the last 10–15
years.4 Of these, three appear particularly promis-
ing for routine clinical use:
(1) Detection of bone marrow metastases, either at
baseline or after 2–4 years of follow-up;
(2) Expression of UPA/PAI-1 by the primary cancer;
(3) recognition of simultaneous multiple gene ex-
pression patterns, or ‘‘signatures’’. Important
caveats exist for each of these.
Bone marrow micrometastases
Several single institution studies have suggested
that patients with bone marrow metastases have a
worse prognosis than those who do not.7 An analysis
of pooled results from several selected centers that
have monitored outcomes in newly diagnosed
breast cancer according to bone marrow status
has confirmed these observations with high statis-
tical significance.8 However, both in the pooled
analysis and in a single study from Norway, these
differences do not appear to be as striking in a sub-
group analysis of patients with favorable standard
prognostic features who did not receive adjuvant
systemic therapy (Fig. 1).8–10 Recently, the Norwe-
gian group has published an interesting exploratory
analysis suggesting that in addition to baseline bone
marrow status, micrometastases 3–4 years after
primary therapy are associated with a high rate of
recurrence during the following few years.9 Taken
together, these results do suggest that bone
marrow metastases is associated with a poorer
prognosis, but it is not clear how these data should
be used clinically. For example, it is unlikely that
patients who have positive nodes but negative bone
marrow would forego the benefits of adjuvant
systemic therapy, and likewise, as noted, it is not
clear that patients with a favorable prognosis by
routine clinical and pathological features should
receive adjuvant chemotherapy based on bone
D.F. Hayes
marrow positivity. Similarly, the clinical utility of
monitoring bone marrow or circulating markers to
identify patients with a high risk of recurrence
after 2–3 years of follow-up has not been demon-
strated.6,11 However, the latter might be used to
elect to change hormone therapies in an otherwise
asymptomatic patient with estrogen receptor posi-
tive disease. A prospective randomized trial to
address this question is being considered.
UPA/PAI-1
UPA/PAI-1 over-expression, as determined by a
highly validated and accurate ELISA, appears to
be strongly prognostic.12,13 However, with longer
follow-up, the initial differences between positive
and negative patients may decrease, although a
small study with ten years’ follow-up suggested
otherwise.14 Regardless, with decreasing size of
primary tumors due to wide-spread screening,15 the
ELISA assay for UPA/PAI-1, which requires 4300 mg
fresh/frozen tissue may not be logistically practical
in many centers. Although a micro-ELISA has been
developed for smaller specimens, such as core
needle biopsies, the prognostic utility of this assay
has not been validated.
Multi-gene analysis
Preliminary level of evidence 3 studies suggest that
multiple gene expression array analysis is an
exciting new approach, with technologies that
work in fresh/frozen or formalin-fixed, paraffin-
embedded tissues.16–21 The most widely studied
such technologies so far are based on DNA segments
representing different genes arrayed on chips or
glass slides to which RNA from a given specimen is
hybridized. Using such technology developed by
Rosetta Inpharmatics (Seattle, WA, USA), investi-
gators from the Netherlands Cancer Institute
reported remarkable results in two separately
published manuscripts.19,20 They initially screened
thousands of genes to develop a 70 gene prognostic
signature that appears to distinguish patients with
favorable from those with unfavorable prognosis
with astonishing accuracy.19 In their most recent
publication, among 295 young patients with breast
cancer for whom tissue had been archived, 180 had
a poor-prognosis signature and 115 had a good-
prognosis signature (Fig. 2).19,20 At 10 years, the
probability of remaining free of distant metastases
was 50.674.5% in the group with a poor prognosis
signature and 85.274.3% in the group with a good
prognosis signature. The estimated hazard ratio for
 ARTICLE IN PRESS
Prognostic and predictive factors 495

Figure 1 Distant disease-free survival (DDFS) and breast cancer-specific survival (BCSS) according to bone marrow
positivity for different prognostic subgroups. (A), (C), (E) DDFS and (B), (D), (F) BCSS in bone marrow-positive (BM+) and
bone marrow-negative (BM
) patients. Separate analysis for (A), (B) node-positive (N+), (C), (D) node-negative (N0),
(E), (F) and T1N0 patients (from Wiedswang et al.10 with permission).

distant metastases in the group with a poor
prognosis signature, as compared with the group
with the good-prognosis signature, was 5.1 (95%
confidence interval, 2.9–9.0; Po0.001). More
recently, a multi-center external validation study
of this 70 gene signature was reported by the
TRANSBIG Consortium. Tissue from 301 patients
followed for at least 10 years in one of six non-
Dutch European centers and one US center were
studied.22 Although the 70-gene profile signature
was again strongly prognostic, outperforming
classic prognostic criteria such as those used by
 ARTICLE IN PRESS
496 D.F. Hayes

Figure 2 (A–F) Survival analysis using Amsterdam 70 gene signature for breast cancer patients. Kaplan–Meier analysis of
the probability that patients would remain free of distant metastases and the probability of overall survival according
to whether they had a 70-gene good prognosis or a poor prognosis signature. The P values were calculated with use of
the log-rank test (from Van de Vijver et al.20 with permission).

the St. Gallen consensus panel,23 the magnitude of
effect was much less than what was previously
reported in the Dutch series, with hazard ratios for
time to distant metastases ¼ 1.85 [1.14–3.0] and
for overall survival ¼ 2.5 [1.4–4.5]. The European
Organization for Research and Treatment of Cancer
(EORTC) and the Breast International Group (BIG)
are now designing a large, prospective clinical trial
to validate the utility of this assay in patients with
newly diagnosed breast cancer (the ‘‘MINDACT’’
study).
In the era of screening, the phenomenon of
smaller tumors at diagnosis has decreased the
availability of frozen tissue for marker analysis.24
Therefore, a multi-gene assay that can be applied
to formalin fixed paraffin embedded (FFPE) tissue
would be of great value. Such an assay has been
developed by Genomics Health Inc (Redwood City,
CA, USA). This assay, designated ‘‘Oncotype DXTM,’’
is based on performance of multiple real time
reverse transcriptase polymerase chain reaction
(RT-PCR) assays to quantify expression of several
selected genes in FFPE tissue. Two hundred and
fifty candidate genes were selected from the
published literature, genomic databases, and in-
house experiments performed on frozen tissue.
Three separate sets of archived tissue from
different clinical data sets were then used to test
the relation between quantitative analysis of
expression of these 250 candidate genes and breast
cancer recurrence. From these data, a panel of 16
cancer-related genes and five reference genes
were used to develop an algorithm to compute a
‘‘recurrence score,’’ ranging from 0 to 100, that
can be used to estimate the odds of recurrence
over ten years from diagnosis.21 Of these three
datasets, the most influential was an important
National Surgical Adjuvant Breast and Bowel
Project (NSABP) study: clinical trial B20, in which
node negative patients with ER-positive tumors
were randomly assigned to tamoxifen alone or with
MF (methotrexate and 5-fluorouracil) or CMF (with
cyclophosphamide) chemotherapy. Only those who
took tamoxifen alone, without chemotherapy, were
studied in this initial evaluation.
Using these data, patients were categorized into
three recurrence score groups using the results
from the three separate data sets, including B20:
 ARTICLE IN PRESS
Prognostic and predictive factors
low risk (recurrence score o18), intermediate risk
(recurrence score 18–30), and high risk (recurrence
score 31–100). These data were then validated
using archival FFPE tissue samples from patients
with node-negative, ER-positive breast cancer who
were randomly assigned to tamoxifen in a different
NSABP trial: B14. Kaplan–Meier estimates of rate of
distant recurrence at 10 years for patients in B14
who took tamoxifen were found to be remarkably
close to those predicted from the training algo-
rithm: 6.8% (4.0–9.6%), 14.3% (8.3–20.3%), and
30.5% (23.6–37.4%; Fig. 3).21 Results from a
separate case control study from the Northern
California Kaiser system are very similar.25 These
results are highly suggestive that the Oncotype
DXTM assay is a reproducible and accurate prog-
nostic test in node negative patients with ER-
positive breast cancers who received tamoxifen.
The NSABP has recently updated their analysis to
include patients in the other arms of B14 and B20.26
In this regard, the investigators were now analyzing
the 21-gene signature as a predictive factor as well
as a prognostic factor. The 16 cancer-related genes
were not randomly chosen. They include ER and
PgR, which are known positive predictive factors
for endocrine therapy and probably negative pre-
dictive factors for chemotherapy; HER2, which is
likely a negative predictive factor for tamoxifen, if
not all endocrine therapies, and may be a negative
Figure 3 Likelihood of distant recurrence, according to
recurrence-score categories for patients with ER-posi-
tive, node-negative breast cancer treated with tamox-
ifen. A low risk was defined as a recurrence score of less
than 18, an intermediate risk as a score of 18 or higher
but less than 31, and a high risk as a score of 31 or higher.
There were 28 recurrences in the low-risk group, 25 in
the intermediate-risk group, and 56 in the high-risk
group. The difference among the groups is significant
(Po0.001) (from Paik et al.21 with permission).
497
predictive factor for benefit from CMF; and various
markers of proliferation, which also probably
predict for response to chemotherapy. Indeed, in
a study of neoadjuvant chemotherapy, Italian
investigators have observed that pathologic com-
plete response to doxorubicin and paclitaxel was
statistically more likely in those with higher
Oncotype DXTM recurrence scores.27 Not surpris-
ingly, the NSABP investigators found that the
Oncotype DXTM assay was a strong predictor of
benefit from MF or CMF in B20.26 There was little or
no benefit from the addition of chemotherapy to
tamoxifen for patients with low or intermediate
recurrence scores, but substantial benefit for those
with high recurrence scores. In the high recurrence
score group, 10-year distant recurrence free
survival for tamoxifen-only patients was approxi-
mately 65% vs. approximately 90% for those who
received CMF (P ¼ 0.001; with a test for interaction
p value of 0.037). Conversely, in B14, the benefit
from tamoxifen vs. observation was almost totally
confined to the low and intermediate risk cate-
gories, with a P value for interaction of 0.001.
Taken together, these data suggest that in patients
who have an apparently favorable prognosis based
on clinical features (negative nodes, positive ER),
the Oncotype DXTM assay helps determine those
most likely to benefit from tamoxifen only (low
recurrence scores) vs. those less likely to benefit
from tamoxifen, but likely to benefit from che-
motherapy (high recurrence scores). The benefits
of chemotherapy in the 25% of patients who have
intermediate recurrence scores remain uncertain.
These results are the basis of a planned prospective
randomized trial that will be performed by the
Breast Cancer Intergroup (TBCI) of North America
(the so-called Program for the Assessment of
Cancer Clinical Tests [PACCT] trial) for patients
with node-negative, ER-positive breast cancer. In
the PACCT trial, patients with low recurrence
scores will all receive endocrine therapy and will
be asked to participate in a registry study for future
prognostic assay validation. Those with high recur-
rence scores will be treated with endocrine therapy
and chemotherapy, preferably within other pro-
spective trials. Those with intermediate recurrence
scores, for whom we have equipoise regarding the
benefits of adjuvant chemotherapy, will be ran-
domly assigned to endocrine therapy only or
endocrine and chemotherapy. This trial takes on
even greater importance in the light of ongoing
progress in the field. It is likely that the overall
prognosis of low and intermediate score patients
will be even better in the modern era with
application of aromatase inhibitors instead of or
in sequence with tamoxifen. In contrast, it is also
 ARTICLE IN PRESS
498
likely that modern chemotherapy, incorporating
anthracyclines and taxanes in a dose-dense fashion,
will be more effective than the standard CMF used
in B20. Thus, especially in the intermediate group,
chemotherapy may be needed less (because hor-
mone therapy is better), but may be more
effective, further increasing the clinician’s equi-
poise for this population.
Other factors
Promising new predictive factors for endocrine
agents include HER2 and ER co-activators and co-
repressor expression.28–30 In vitro chemo-resistance
assays for selected chemotherapeutic agents are
appealing, but evidence supporting their clinical
utility in breast cancer is lacking.31 Multi-gene
expression assays have been applied in preliminary
neo-adjuvant studies with exciting results, but
these require substantial confirmation to be clini-
cally helpful.32,33
Conclusion
In conclusion, although new technologies offer
exciting and promising new approaches to deter-
mining a patient’s prognosis and whether she will or
will not benefit from specific therapies, few have
been validated in well-designed, Level of Evidence
I studies.34,35 In particular, available data are often
confounded by patient selection and the effects of
systemic therapy, which are often not determined
prospectively, not included in analyses, and not
reported adequately.34–37 It is gratifying to see
newer prognostic and predictive factor studies
follow well-designed strategies so that validation
of clinical utility will occur more quickly and with
greater reliability.
Acknowledgments
This work was supported in part by NIH CA092461
and by a grant from the Fashion Footwear Founda-
tion/QVC Presents Shoes on Sales.
References
1. Levi F, Lucchini F, Negri E, La Vecchia C. Trends in mortality
from major cancers in the European Union, including
acceding countries, in 2004. Cancer 2004;101:2843–50.
2. Jatoi I, Miller AB. Why is breast-cancer mortality declining?
Lancet Oncol 2003;4:251–4.
D.F. Hayes
3. Peto R, Boreham J, Clarke M, Davies C, Beral V. UK and USA
breast cancer deaths down 25% in year 2000 at ages 20–69
years. Lancet 2000;355:1822.
4. Isaacs C, Stearns V, Hayes DF. New prognostic factors for
breast cancer recurrence. Semin Oncol 2001;28:53–67.
5. Singletary SE, Allred C, Ashley P, et al. Staging system for
breast cancer: revisions for the 6th edition of the AJCC
Cancer Staging Manual. Surg Clin North Am 2003;83:803–19.
6. Bast Jr. RC, Ravdin P, Hayes DF, et al. 2000 Update of
recommendations for the use of tumor markers in breast and
colorectal cancer: clinical practice guidelines of the Amer-
ican Society of Clinical Oncology. J Clin Oncol 2001;19:
1865–78.
7. Braun S, Naume B. Circulating and disseminated tumor cells.
J Clin Oncol 2005;23:1623–6.
8. Braun S, Vogl FD, Naume B, Janni W, Osborne MP, Coombes
RC, et al. A pooled analysis of bone marrow micrometastasis
in breast cancer. N Engl J Med 2005;353(8):793–802.
9. Wiedswang G, Borgen E, Karesen R, et al. Isolated tumor
cells in bone marrow three years after diagnosis in disease-
free breast cancer patients predict unfavorable clinical
outcome. Clin Cancer Res 2004;10:5342–8.
10. Wiedswang G, Borgen E, Karesen R, et al. Detection of
isolated tumor cells in bone marrow is an independent
prognostic factor in breast cancer. J Clin Oncol 2003;21:
3469–78.
11. Bast Jr. RC, Ravdin P, Hayes DF, et al. Errata: 2000 update of
recommendations for the use of tumor markers in breast and
colorectal cancer: clinical practice guidelines of the Amer-
ican Society of Clinical Oncology. J Clin Oncol 2001;19:
4185–8.
12. Look M, van Putten W, Duffy M, et al. Pooled analysis of
prognostic impact of uPA and PAI-1 in breast cancer patients.
Thromb Haemost 2003;90:538–48.
13. Janicke F, Prechtl A, Thomssen C, et al. Randomized
adjuvant chemotherapy trial in high-risk, lymph node-
negative breast cancer patients identified by urokinase-type
plasminogen activator and plasminogen activator inhibitor
type 1. J Natl Cancer Inst 2001;93:913–20.
14. Zemzoum I, Kates RE, Ross JS, et al. Invasion factors uPA/
PAI-1 and HER2 status provide independent and complemen-
tary information on patient outcome in node-negative
breast cancer. J Clin Oncol 2003;21:1022–8.
15. Cady B. New era in breast cancer. impact of screening on
disease presentation. Surg Oncol Clin N Am 1997;6:195–202.
16. Perou CM, Jeffrey SS, van de Rijn M, et al. Distinctive
gene expression patterns in human mammary epithelial
cells and breast cancers. Proc Natl Acad Sci USA 1999;96:
9212–7.
17. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of
human breast tumours. Nature 2000;406:747–52.
18. Sorlie T, Perou CM, Tibshirani R, et al. Gene expression
patterns of breast carcinomas distinguish tumor subclasses
with clinical implications. Proc Natl Acad Sci USA 2001;98:
10869–74.
19. Van ‘t Veer LJ, Dai H, van de Vijver MJ, et al. Gene
expression profiling predicts clinical outcome of breast
cancer. Nature 2002;415:530–6.
20. Van de Vijver MJ, He YD, van’t Veer LJ, et al. A gene-
expression signature as a predictor of survival in breast
cancer. N Engl J Med 2002;347:1999–2009.
21. Paik S, Shak S, Tang G, et al. A multi-gene RT-PCR assay using
fixed, paraffin-embedded tumor tissue to predict the like-
lihood of breast cancer recurrence in node negative,
estrogen receptor positive, tamoxifen-treated patients. N
Engl J Med 2004;351:2817–26.
 ARTICLE IN PRESS
Prognostic and predictive factors
22. Piccart MJ, Loi S, van ‘t Veer LJ, et al. Multi-center external
validation study of the Amsterdam 70-gene prognostic
signature in node negative untreated breast cancer: are
the results still outperforming the clinical-pathological
criteria? Breast Cancer Res Treat 2004;88:S17 (abstract 38).
23. Goldhirsch A, Wood WC, Gelber RD, Coates AS, Thurlimann
B, Senn HJ. Meeting highlights: updated international expert
consensus on the primary therapy of early breast cancer.
J Clin Oncol 2003;21:3357–65.
24. Cady B, Stone MD, Schuler JG, Thakur R, Wanner MA, Lavin
PT. The new era in breast cancer. Invasion, size, and nodal
involvement dramatically decreasing as a result of mammo-
graphic screening. Arch Surg 1996;131:301–8.
25. Habel L, Quesenberry CP, Jacobs MK, et al. A large
case–control study of gene expression and breast cancer
death in the Northern California Kaiser Permanente popula-
tion. Breast Cancer Res Treat 2004;88:S118 (abstract 3019).
26. Paik S, Shak S, Tang G, et al. Expression of 21 genes in
the recurrence score assay and prediction of clinical
benefit from tamoxifen in NSABP study B-14 and chemo-
therapy in NSABP B-20. Breast Cancer Res Treat 2004;88:S15
(abstract 24).
27. Gianni L, Zambetti M, Clark K, et al. Gene expression of
paraffin-embedded core biopsy tissue predicts response to
chemotherapy in patients with locally advanced breast
cancer. Proc Am Soc Clin Oncol 2004;23:3 (abstract 501).
28. Yamauchi H, Stearns V, Hayes DF. When is a tumor marker
ready for prime time? A case study of c-erbB-2 as a
predictive factor in breast cancer. J Clin Oncol 2001;19:
2334–56.
29. Ellis MJ, Coop A, Singh B, et al. Letrozole is more effective
neoadjuvant endocrine therapy than tamoxifen for ErbB-1-
and/or ErbB-2-positive, estrogen receptor-positive primary
499
breast cancer: evidence from a phase III randomized trial.
J Clin Oncol 2001;19:3808–16.
30. Osborne CK, Bardou V, Hopp TA, et al. Role of the estrogen
receptor coactivator AIB1 (SRC-3) and HER-2/neu in tamox-
ifen resistance in breast cancer. J Natl Cancer Inst 2003;95:
353–61.
31. Schrag D, Garewal HS, Burstein HJ, Samson DJ, Von Hoff DD,
Somerfield MR. American Society of Clinical Oncology
Technology Assessment: chemotherapy sensitivity and re-
sistance assays. J Clin Oncol 2004;22:3631–8.
32. Chang JC, Wooten EC, Tsimelzon A, et al. Gene expression
profiling for the prediction of therapeutic response to
docetaxel in patients with breast cancer. Lancet 2003;362:
362–9.
33. Ayers M, Symmans WF, Stec J, et al. Gene expression profiles
predict complete pathologic response to neoadjuvant
paclitaxel and fluorouracil, doxorubicin, and cyclophospha-
mide chemotherapy in breast cancer. J Clin Oncol 2004;22:
2284–93.
34. Hayes DF, Bast R, Desch CE, et al. A tumor marker utility
grading system (TMUGS): a framework to evaluate clinical
utility of tumor markers. J Natl Cancer Inst 1996;88:
1456–66.
35. Hayes DF. Clinical importance of prognostic factors: moving
from scientifically interesting to clinically useful. In:
Bronchud MH, Foote M, Giaccone G, Olapade O, Workman
P, editors. Principles of molecular oncology. 2nd ed. Totowa,
NJ: Humana; 2004. p. 51–72.
36. Hayes DF, Trock B, Harris A. Assessing the clinical impact of
prognostic factors: when is ‘‘statistically significant’’ clini-
cally useful? Breast Cancer Res Treat 1998;52:305–19.
37. Ransohoff DF. Cancer. Developing molecular biomarkers for
cancer. Science 2003;299:1679–80.