Ductal Carcinoma
In Situ
399
An attempt to improve the prognostic value of VNPI 2003
through the replacement of nuclear grade by genomic grade
index (GGI, a 97-gene measure of histologic tumor grade) to
generate the VNPI–GGI index has been described.
408
This
unusual (and currently rather impractical) attempt to merge
morphologic and molecular information is unlikely to be
widely applied. The novel use of USC/VNPI to assess post-
mastectomy risk of recurrence has also been proposed.
409
A nomogram has been established for the risk of relapse
after breast-conserving therapy on the basis of 10 predictive
factors derived from a Cox multivariate analysis of retro-
spective data from 1,681 patients who underwent breast-
conserving surgery at Memorial Sloan-Kettering Cancer
Center in New York.
410
Factors with the greatest influence
on the risk of ipsilateral recurrence included age, family
history of breast carcinoma, margin status, number of ex-
cisions, adjuvant radiation or endocrine therapy, and treat-
ment time period. Notably, neither tumor size nor any of
the commonly used biomarkers (ER, PR, and HER2) were
included as variables in this nomogram. This is the first no-
mogram to offer such a decision tool.
410
A recent attempt to
validate this nomogram in an independent data set found
that it overestimated the risk of recurrence in some subsets
of patients.
411
Breast Conservation for Mammographically
Detected DCIS
Few studies have analyzed data based exclusively on mam-
mographically detected DCIS treated by breast conservation
with radiotherapy.
404,405
The 10-year breast recurrence rates
ranged from 4% to 7% in patients with negative final exci-
sion margins to as high as 30% for women with positive or
close margins. Time to recurrence appeared to be shorter
for patients with positive margins (median 3.6 years) than
for those with negative (median 4.3 years) or indetermi-
nate (median 5.2 years) margins.
404
In patients with mam-
mographically detected DCIS, pathologic features such as
nuclear grade, necrosis, and architecture (comedo vs. non-
comedo) were not significantly related to the risk of local
recurrence. The lack of association with pathologic charac-
teristics indicates the importance of stratifying patients by
detection modality in the analysis of risk factors for local
breast recurrence after conservation therapy.
Age at diagnosis (less than 45 vs. greater than or equal
to 45 years) was found to be a significant predictor of local
recurrence after breast-conserving surgery with radiation
in patients with mammographically detected DCIS.
269
In
this study, the 10-year actuarial rate of local failure in the
breast was 23.4% for women younger than 45 years when
treated and 7.1% among those 45 years or older. The au-
thors were unable to apply the VNPI to their analyses be-
cause tumor size could not be determined in 58% of the
cases. Pathologic study of the DCIS revealed several factors
that might have predisposed the younger women to local
recurrence.
412
These included smaller diagnostic biopsy
specimens and more frequent lesions with high nuclear
grade and necrosis.
or the number of slides with DCIS do not provide measure-
ments suitable for the VNPI.
404,405
There are likely to be many
patients for whom a VNPI cannot be determined or for whom
the calculated VNPI is of questionable accuracy.
An updated report from the Van Nuys Center published in
1998 did not classify patients according to the VNPI.
406
The
series of 707 of nonrandomized patients included 208 women
treated by lumpectomy and radiotherapy and 240 treated by
excision alone. Breast recurrences were detected in 36 women
in each group, representing 17% and 15%, respectively, and
approximately half of the recurrences were invasive in each
group. Distant metastases were diagnosed in six patients, five
of whom had been treated originally by lumpectomy and ra-
diotherapy. Five of the patients (0.7%) in the entire series died
of breast carcinoma with four in the radiated group. The me-
dian follow-up for the 35 patients who had invasive recurrent
carcinoma was 127 months (58 months from initial diagnosis
to invasive recurrence and 69 additional months after recur-
rence). The distant recurrence rate in the subset of 35 patients
with invasive recurrence in the breast was 27.1%, and the
mortality rate due to breast carcinoma was 14.4% at 8 years.
The VNPI has evolved over the years. In 1995, the Van
Nuys classification using a combination of nuclear grade
and necrosis was proposed as a tool for the prediction of
local recurrence. The 1996 version of VNPI (VNPI-1996)
was based upon the
size
of DCIS, its
pathologic grade
and
margins
. The modified University of Southern California
(USC)/VNPI-2003 included
age
as the fourth factor to the
estimation, although the introduction of age did not appear
to cause any significant shift in treatment modalities.
In 2010, the USC-VNPI was fine-tuned further, on the
basis of the observations in three times as many patients as
were included 15 years previously.
407
The five quantifiable
prognostic factors (size, margin width, nuclear grade, come-
donecrosis, and age) were retained; however, the recommen-
dations were revised. On the basis of cumulative data that
included patients treated as early as l979, it was concluded
that to achieve a local recurrence rate of less than 20% at 12
years, excision alone for patients scoring 4, 5, or 6, and for
those with a score of 7 and margin widths of 3 mm or more,
would be appropriate. Excision plus radiation therapy would
achieve the same goal for patients with a score of 7 with mar-
gins less than 3 mm, patients with a score of 8 and margins of
3 mm or more, and patients whose score is 9 with margins of
5 mm or more. Mastectomy was recommended for patients
who score 8 and less than 3-mmmargins, those with a score 9
and margins less than 5 mm, and for all patients with a score
of 10, 11, or 12. As noted above, the VNPI is not evidence
based, and its formulation and reformulation have been based
on a relatively small retrospective series of cases from a single
institution where treatment was not randomized. In fact, it
is evident that the treatment program was revised with each
reanalysis of the data and lacked consistency throughout the
early three decades during which the information was assem-
bled. These factors, changes in the data points used to create
the VNPI, and the unreliability of some of the measurements
are reasons to be very cautious in accepting the VNPI cat-
egories as an absolute basis for making therapeutic decisions.
