A recent study by Kathy Albain, MD and colleagues is reported in Lancet and clearly describes the risks and benefits of treating early breast cancer with chemotherapy. As we all know, it is impossible to accurately predict (in spite of the various tests and scores and data which are used to make the decision) who needs it and who does not, which women would do fine with "just" surgery, radiation, and hormone therapy, when appropriate, and which women would relapse. The Oncotype DX has been a big help with this decision, but it is not, nothing is, 100% accurate in prediction. What, for example, is the right advice for all the women whose Oncotype DC-predicted recurrence risk is in the intermediate range? However the decision is made, we are sometimes, for every 100 treated women, left with 95 who didn't really need it and 5 whose lives likely were saved with chemotherapy.
The short term side effects of chemo are well known, and we can easily (?) convince ourselves that the temporary hair loss and general bad feelings are worth it. What about the rare but real long term side effects of cardiac damage or a secondary cancer caused by the treatment for breast cancer? I am not a neutral observer on this one, and, for myself, have opted for chemo in a rather gray situation. I have spent too many years with too many women with metastatic breast cancer, and I have enormous respect for the power and guile of the disease. My perspective, however, is my own, certainly not universally shared (nor should it be), and this decision is really tough.
Here is the (longish) summary of the the study:
Fighting overtreatment in adjuvant breast cancer therapy
In The Lancet today, Kathy Albain and colleagues
present 10-year results of the SWOG-8814 (INT-0100)
trial,1 which since its fi rst presentation has defi ned a
standard of care in breast cancer—ie, adjuvant cytotoxic
chemotherapy is benefi cial for node-positive patients.
However, use of nodal status as the main prognostic
factor leads to huge overtreatment for many patients.
Even in today's paper, the absolute 10-year overall survival
diff erence was just 5%, more than enough to justify a
desired standard of care nowadays but, ultimately, leaving
95% of patients with many side-eff ects.
Albain and colleagues have shown that addition of
tamoxifen to cyclophosphamide, doxorubicin, and
fl uorouracil (CAF) within the fi rst 6 months exerts a
longlasting eff ect. In terms of tumour biology, this
outcome is exciting and surprising from a theoretical
point of view. Whereas outcome curves for endocrineresponsive
breast cancer mimic those for chronic
diseases (ie, show an almost linear long-term event
pattern),2 addition of CAF seems to widen the survival
curve diff erence up to about year 5 after surgery.
What is the underlying biology? Are dormant micrometastases
eradicated successfully during those
4-6 months of chemotherapy, reducing the number
of later relapses permanently? Treatments that change
the microenvironment where these dormant cells
supposedly survive have gained attention in the
adjuvant setting.3 Early cytotoxic therapy could target
macrophages and aff ect the patient's immune response,
thereby contributing to a so-called carryover eff ect of
CAF in the fi rst 5 years.
A diff erent observation can be made for the comparison
of CAF and concurrent tamoxifen with CAF followed
by tamoxifen. Current standard practice is to avoid
concomitant tamoxifen and chemotherapy based on
previous fi ndings of SWOG-8814 and due to concerns
about preclinical antagonism, despite some controversial
reports.4 In today's report, the outcome diff erence
between concurrent and consecutive tamoxifen arose
late in the survival curves, provoking speculation about a
sustained early eff ect on the tumour micro environment.
These fi ndings lend support to a hypothesis about the
importance of indirect eff ects of adjuvant therapy, which
could act in a direct, toxic, tumour-cell-eliminating
manner to a lesser extent than we used to think.5
CAF, while statistically benefi cial for the cohort studied,
is not without substantial early and late toxic eff ects,
including early deaths, later congestive heart failure,
and secondary neoplasms. Thus long-term follow-up is
very important in such trials but is increasingly diffi cult
to fi nd funding for, particularly in industry-sponsored
studies. The ultimate risk-benefi t assessment will need
more than just achievement of the fi rst signifi cant
p value, which nowadays seems to be good enough for
major publication and even for regulatory approval of a
new therapeutic intervention.
There is a price to pay for a 5% overall survival benefi t
and, for that reason, identifi cation of subgroups of patients
who have an above-average outcome is import ant.
Unplanned subgroup analyses by Albain and col leagues
suggest patients at high risk of relapse (large tumours,
age <65 years, #4 aff ected nodes) mainly show a benefi t
of adjuvant therapy. Conversely, patients at low risk of
relapse (breast conservation, age #65 years, 1-3 aff ected
nodes) might show a marginal or no benefi t from CAF.
Since risk is something we cannot change, response
prediction is nowadays judged more important for
clinical decision making than risk itself,6 but calcu lation
of individual benefi t is still poor. Avoidance of
unnecessary or ineff ective treatment should be one
of the main goals in adjuvant breast oncology today.
Unfortunately, both patients and doctors hunt for tiny
statistical diff erences in survival curves. As outlined
elegantly by Dodwell and colleagues,7 this search could
not only lead to an oncological approach of unlimited
addition that we will not be able to aff ord, but would
also end inevitably in indeterminate polypharmacy with
substantial risks of unexpected toxic eff ects eating away
whatever progress we might make.
In SWOG-8814 and other adjuvant trials, overall
benefi t is mainly accounted for by the high-risk subgroup
of patients. This observation is substantiated by fi ndings
of another SWOG-8814 publication by Kathy Albain and
co-workers in The Lancet Oncology today,8 in which a
correlation between the 21-gene recurrence score and
CAF benefi t is described. On the basis of 367 patients
from SWOG-8814 for whom tumour RNA was available,
the overall CAF benefi t described in today's paper in
The Lancet was present exclusively in the highest risk
subgroup (recurrence score #31),8 which comprises not
even a third of the overall trial cohort.
Does the 21-gene assay truly provide a biological
approach to risk assessment and response prediction
compared with classic immunohistochemical measures?
This question is currently under investigation in large
prospective clinical trials, but similar data have been
reported for other multigenomic assays.9 Findings of
the molecular analysis presented in The Lancet Oncology
accord with those of the subgroup analysis in The Lancet—
that patients at low risk of relapse might not benefi t at
all from adjuvant chemotherapy. If this result is true,
overtreatment could be even more striking in patients
with node-negative breast cancer.
"First, do not harm" remains the main principle in
medicine. To be able to follow this rule, we need to better
understand the biology of breast cancer. The mistake of
"one treatment fi ts all" can only be ameliorated when we
critically review trial designs of adjuvant breast oncology.
Selection of precisely defi ned cohorts for phase 3 trials is
necessary, despite pressure to the contrary by scientifi c
ambition, pragmatism, and demands of industry.