2012 EDUCATIONAL BOOK - American Society of Clinical Oncology

ADJUVANT THERAPY IN HIGH-RISK GASTROINTESTINAL STROMAL TUMORS
and has thereafter been reviewed by the EORTC IDMC on a
yearly basis. Based on IDMC recommendations and the
recent awareness that late imatinib given at progression of
disease could in theory make up for early adjuvant imatinib
without the theoretical disadvantage of inducing early resistance
and that in the overall assessment, retreatment with
imatinib also has to be taken into account, it was decided to
change the primary endpoint of this study to time to imatinib
failure: the start of any new systemic therapy other
than imatinib or death due to any cause. The first analysis
based on this new endpoint is currently expected in the first
half of 2012.
What Do These Studies Tell Us?
Importantly, the data details of the ACOSOG and SSG
studies have not been publically presented in the same way.
This renders it difficult to assess potential effects of differences
in populations in prognostic parameters. Yet the
majority of patients in the studies had tumors with high risk
of relapse so the observations likely hold for this population.
Both studies indicate that adjuvant imatinib given for either
1 or 3 years is able to improve RFS. But as indicated this is
not the primary aim of adjuvant therapy, and in an editorial
Hohenberger has even described the use of RFS in this
setting as a self-fulfilling prophecy. 18 Bearing in mind that
there may be differences in the study populations, it is
interesting to analyze the RFS curves. It is remarkable that
the RFS in the patients given 1 year of imatinib in the SSG
study is similar to the RFS of the control group in the
ACOSOG study, while the curves for the respective 3-year
and 1-year groups also seem to overlap. Of concern is the fact
that at the time of reporting both studies showed bananashaped
curves, suggesting that we can yet not exclude that
the beneficial effects, if any, may be temporary.
Importantly, if we also project in the curves (both for RFS
and OS) of the nonrandomized comparative study 14 it shows
that the control group in that study performed worst of all
control groups, while the treated group outperformed all
treatment groups. This in itself supports the notion that
nonrandomized studies can be heavily biased by patient
selection and should be ignored in deciding on practice
recommendations.
Plotting the data of the two published studies from the
time that patients went off drug, there again are remarkable
differences but also some similarities. The SSG study arms
seem to perform worse than those of the ACOSOG study, but
all arms suggest a steady drop-off rate, and the curves run in
parallel. This seems to suggest that stopping imatinib in
some patients leads to rapid relapse. This would be in line
with the observations in the FSG study in metastatic disease
19 that has shown that patients in whom imatinib was
withheld while disease was not progressing rapidly experienced
relapse. Yet, while in metastatic disease these data
support the continuation of imatinib as long as patients
seem to benefit, the absence of indisputable overall survival
benefit in the adjuvant setting should lead to a conservative
approach to the duration of adjuvant imatinib treatment.
For OS a limitation of all studies is that they could not
exclude cross-over to or retreatment with imatinib at recurrence,
which may affect OS results. This is why the EORTC
decided to change its primary endpoint after the study had
fully accrued. The ACOSOG study 15 has only presented OS
up to 4 years, while the SSG study 16 has presented longer
follow-up. Again, remarkably, at 4 years the 1-year treated
group in the SSG study seems to have worse outcome than
any of the ACOSOG study arms. This may point to differences
in the study populations. The ACOSOG study has not
yet shown any benefit in overall survival. The SSG study, as
indicated, shows a statistically significant difference in favor
of the 3-year treatment. If the ACOSOG study does not show
a difference between no treatment and 1-year treatment, in
theory this 1-year treatment is an adequate control in the
SSG study. However, if the above-mentioned differences
between the 1-year groups in the two studies cannot be
explained on population differences, the observed difference
in the SSG study might solely be based on a worse performance
of the 1-year group. Yet, since the statistics support
this, for the time being we can consider the outcome of the
SSG study as evidence supporting the use of 3 years of
adjuvant imatinib in the high-risk population. Still, caution
needs to be taken in interpreting the outcome of the study
since it is relatively small and differences are based on small
numbers. For the same reason of small patient numbers in
individual subgroups, further investigations are still warranted.
One could wonder why the IDMC of the EORTC
study has not decided to unblind the data of that study now
that the SSG study shows a survival benefit. In theory, this
could mean the study does not show a survival benefit and
data need to further mature. It will be very interesting to see
the report of that study later this year or early next year.
Conclusion and Recommendation
With all of the caveats mentioned, for the time being
adjuvant imatinib can be considered the standard of care
in patients with a high risk of GIST relapse. The most
important support for this statement comes from the overall
survival benefit suggested in the SSG trial. The current
consensus seems to be to consider patients with a greater
than 50% chance of relapse within 5 years according to the
Gold nomogram as candidates for such treatment. Although
not fully based on evidence, because of their rare occurrence,
patients with neurofibromatosis type 1–associated GIST,
which strongly expresses wild-type KIT and has a very
indolent behavior, and those with PDGFR-alpha D842V
mutations, which are known to be insensitive to imatinib,
17,20 may be excluded from this recommendation. Patients
with tumor rupture may best be considered as having
metastatic disease and be treated accordingly.
Longer-term follow-up data as well as publication of the
results of the EORTC study are eagerly awaited and should
lead to a critical reconsideration of the above-mentioned
recommendation.
Based on the studies, adjuvant imatinib should be initiated
within 3 months after definitive surgery and preferably
be handled by an expert team after multidisciplinary discussion.
Since all studies have applied a dose of 400 mg daily
regardless of the mutation status of the primary tumor, this
is also the dose recommended for standard application of
adjuvant imatinib. Based on the data of the SSG study, this
dose should be given for 3 years. In GIST cases where
preoperative adjuvant imatinib therapy is used for localized
disease for the purpose of organ-sparing tumor shrinkage or
patient refusal of surgery, the duration of neoadjuvant
therapy should be considered as part of the whole adjuvant
treatment duration of 3 years.
Based on the performed studies, during adjuvant treat-
661