2012 EDUCATIONAL BOOK - American Society of Clinical Oncology

een reached by other modeling efforts designed to examine
trends in the incidence of breast cancer. With their model,
Mandelblatt and colleagues 25 consistently underestimated
rates of DCIS as compared with rates drawn from SEER
data, suggesting that not all DCIS will become clinically
relevant. In a similar model, Fryback and colleagues 26
identified a substantial proportion of tumors having limited
potential to become malignant in order to account for the
observed DCIS rate.
Although there are few data, evidence indicates that
low-grade DCIS treated by biopsy alone does not progress to
invasive breast cancer. 27,28 Additionally, autopsy data suggest
that untreated DCIS may not develop into invasive
breast cancer in a time period that will affect a woman’s
life. 29 It appears feasible that there exists a reservoir of
DCIS in the population that is never diagnosed and never
attains clinical relevance, compelling us to ask if we have
overestimated the potential of DCIS to progress to invasive
breast cancer. If so, are we overdiagnosing and therefore
overtreating DCIS?
Researchers have tried to identify which in situ cancers
are the likely precursors of invasive disease and, of those,
which are of such low-risk that they might require minimal
local treatment (i.e., surgery alone or even active surveillance).
Until recently, researchers have tried to identify
these low-risk groups by standard clinical-pathologic features.
30 For example, high-grade palpable lesions have been
shown to carry a higher risk for recurrence in the first 5
years after diagnosis. 31 A number of molecular markers
have been proposed to stratify risk but have not yet been
validated in large studies. 32 The Van Nuys score, based on
combinations of patient age, tumor size, tumor grade, and
surgical margins, has been used to determine the value of
radiation therapy. A particular problem however, is the
current use of nuclear grade as overall grade, causing as
much as a 50% increase in the classification of high-grade
lesions.
Interestingly, a recent analysis of grade and molecular
profiling of in situ lesions suggested that low-grade lesions
were not the precursors of low-grade tumors, but high-grade
DCIS lesions were consistently predictive of high-grade
invasive tumors. However, when molecular subtyping was
performed, the vast majority (85%) of the lobular carcinoma
in situ (LCIS) and DCIS lesions were luminal A subtype.
The molecular subtype of the subsequent invasive tumor
was largely maintained: 85% for LCIS, and 69% for DCIS. 33
Although the numbers in the study were small, the findings
suggest the possibility that grade is not a reliable indicator
of risk and that many DCIS lesions are in fact precursors of
more indolent cancers.
The unmet need for better risk stratification is now being
served with molecular profiling in the same way it has for
invasive cancers. Researchers at Genomic Health have now
applied their expertise in molecular profiling to DCIS. Recent
data from their studies showed that they are able to
identify a low-risk category for which surgery alone would be
adequate local therapy. In fact, not only were they able to
identify risk for local recurrence but they were also able to
distinguish between total recurrences (in situ plus invasive)
and invasive cancer recurrence only. These data suggest
that a substantial number of low-risk lesions have, on
average, a 5% risk of an invasive local recurrence at 10 years
after lumpectomy alone. This risk is the equivalent of a Gail
e42
Risk Score of 2.5, indicating a 2.5% risk of invasive cancer at
5 years. This risk is also similar to the average risk of
invasive cancer for a woman in her mid-60s. Again, these
findings mean that DCIS, with excision alone, is appropriately
categorized as a high-risk lesion such as atypia. These
new data are extremely important because they may finally
provide a more precise way of identifying women who are
being overtreated for a diagnosis that may have little bearing
on their life for the next 20–30 years.
It is important to recognize that when screening was first
introduced, the goal was to identify invasive cancers. Over
the years, the focus has broadened to include calcifications
and the identification of DCIS. In the United States, the
biopsy rates are higher than they are in Europe, 34 and the
cancer-to-biopsy ratios are in the range of 25% and 50%,
respectively. The proportion of DCIS detected is higher, and
more cases of detected DCIS are lower grade. These findings
are important to keep in mind when evaluating results of
different trials. For example, in the [United Kingdom New
Zealand trial], the majority of DCIS detected was high
grade, which may explain why tamoxifen was not found to be
effective in reducing the risk of ipsilateral invasive cancer.
The introduction of more standard molecular tools may
prove important in comparing DCIS types and in allowing a
different approach to DCIS in the future.
Despite the recognition that some breast cancers we
detect are indolent in nature, there have been almost no
trials of active surveillance, as there have been in prostate
cancer. Even DCIS, which is thought to be a precursor of
invasive cancer, has rarely been approached with active
surveillance. It is hoped that the availability of tools to
profile DCIS lesions in a standard manner will usher in a
more progressive approach to DCIS, not unlike what has
been adopted in the treatment of prostate cancer, especially
given the availability of preventive interventions.
What Is the Solution?
ALVARADO, OZANNE, AND ESSERMAN
Change the Approach to DCIS from Cancer Treatment to
Prevention of Invasive Cancer
The evidence for overtreatment of DCIS is extremely
compelling, given the associated risk. Classic treatment
strategies have almost always included either breast conservation
with whole breast radiation or mastectomy. It is now
becoming apparent that this type of “precursor lesion” is
being overtreated. With continued evidence that we are not
only overdetecting but also overtreating breast cancer in
this country and abroad, researchers need to think seriously
about developing protocols for women at the lowest risk.
With regard specifically to DCIS, it would not be unreasonable
to develop algorithms to identify women who could have
active surveillance after carcinoma in situ is found on
evaluation of a biopsy specimen. 35 Many groups have discussed
ways in which to stratify risk, and with the availability
of molecular markers, it should be an immediate
action item to finally address the overtreatment of earlystage
breast cancer. Previous research attempted to identify
risk groups that were typically based on classic, clinicalpathologic
features such as patient age, tumor size, and
tumor grade. Numerous data with molecular profiling have
indicated that standard clinical-pathologic features are a
poor way of truly risk stratifying newly diagnosed breast
cancer. In fact, grade appears to be nonreproducible across