Annals of Surgical Oncology (
2006)

DOI: 10.1245/s10434-006-9034-8

Surgical Resection of Gastrointestinal Stromal Tumors After

Treatment with Imatinib

Robert H. I. Andtbacka, MD, CM,1 Chaan S. Ng, MD,2 Courtney L. Scaife, MD,5
Janice N. Cormier, MD, MPH,1 Kelly K. Hunt, MD,1 Peter W. T. Pisters, MD,1
Raphael E. Pollock, MD, PhD,1 Robert S. Benjamin, MD,3 Michael A. Burgess, MD,3
Lei L. Chen, MD, PhD,3 Jonathan Trent, MD,3 Shreyaskumar R. Patel, MD,3
Kevin Raymond, MD,4 and Barry W. Feig, MD1

1
Department of Surgical Oncology, The University of Texas M. D. Anderson Cancer Center, Unit 444, 1515 Holcombe Blvd.,
Houston, TX 77030, USA
2
Department of Radiology, The University of Texas M. D. Anderson Cancer Center, Houston, TX 77030, USA
3
Department of Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, TX 77030, USA
4
Department of Pathology, The University of Texas M. D. Anderson Cancer Center, Houston, TX 77030, USA
5
Department of Surgery, University of Utah, Salt Lake City, UT 84112, USA

Background: Surgical resection of gastrointestinal stromal tumors (GISTs) has been the
most effective therapy for these rare tumors. Imatinib has been introduced as systemic therapy
for locally advanced and metastatic GIST. In this study, the surgical resection rates and long-
term outcomes of patients treated with preoperative imatinib for locally advanced primary,
recurrent, or metastatic GISTs were evaluated.
Methods: Patients were retrospectively assessed for completeness of surgical resection and
for disease-free and overall survival after resection.
Results: Forty-six patients underwent surgery after treatment with imatinib. Eleven were
treated for locally advanced primary GISTs for a median of 11.9 months, followed by com-
plete surgical resection. All eleven were alive at a median of 19.5 months, and ten were free of
disease. Thirty-five patients were treated for recurrent or metastatic GIST. Of these, eleven
underwent complete resection. Six of the eleven patients had recurrent disease at a median of
15.1 months. All eleven patients were alive at a median of 30.7 months. Patients with a partial
radiographic tumor response to imatinib had significantly higher complete resection rates than
patients with progressive disease (91% vs. 4%; P < .001). Of the 24 patients with incomplete
resection, 18 initially responded to imatinib but were unable to undergo complete resection
after they progressed before surgery.
Conclusions: Preoperative imatinib can decrease tumor volume and is associated with
complete surgical resection in locally advanced primary GISTs. Early surgical intervention
should be considered for imatinib-responsive recurrent or metastatic GIST, since complete
resection is rarely achieved once tumor progression occurs.
Key Words: Gastrointestinal stromal tumor—Imatinib—Surgery—Outcome.

INTRODUCTION
Received August 23, 2005; accepted March 6, 2006
Presented in part at the Annual Meeting of the Society of Sur- Gastrointestinal stromal tumors (GISTs) are the
gical Oncology, Atlanta, GA, March 2005. most common soft tissue tumors of the gastrointes-
Address correspondence and reprint requests to: Robert H. I.
Andtbacka, MD, CM; E-mail: [email protected]
tinal system. GISTs originate from mesenchymal
Published by Springer Science+Business Media, Inc.
2006 The Society of
stem cells that are believed to differentiate toward
Surgical Oncology, Inc. intestinal cells of Cajal, a network of pacemaker cells
 R. H. I. ANDTBACKA ET AL.

that coordinate peristalsis in the gastrointestinal trials are investigating the potential benefits of
system.1,2 Most GISTs express KIT (a tyrosine kinase preoperative imatinib in patients with primary
receptor) on their cell surface, which helps to distin- GISTs.17,18 The effect of preoperative imatinib on
guish them from other gastrointestinal mesenchymal surgical resection rates and postoperative outcome in
tumors.3 patients with locally advanced primary GIST is un-
Historically, treatment of recurrent or metastatic known.
GISTs with single or multi-agent chemotherapy The purpose of our study was to better define the
resulted in disappointingly low response rates of role of surgery and its long-term outcome in patients
<10%.4–6 The introduction of imatinib mesylate with locally advanced primary, recurrent, or meta-
(STI571, Gleevec or Glivec, Novartis Pharmaceuti- static GIST treated with imatinib preoperatively.
cals, East Hanover, NJ), a small-molecule receptor
tyrosine kinase inhibitor that targets KIT, has pro-
vided a much needed treatment regimen for patients METHODS
with unresectable and metastatic GISTs. Clinical
trials have demonstrated partial response rates of
Patients
40% to 69% and longer progression-free survival
times in patients treated with imatinib for unresec- Patients with locally advanced primary, recurrent,
table, recurrent, or metastatic GISTs.7–10 Although or metastatic GIST who were treated with imati-
complete responses are rare, many patients experi- nib were retrospectively identified from the
ence partial responses that often are maintained for M. D. Anderson Cancer Center prospective sarcoma
extended periods of time. database and the medical records database. Between
The current clinical practice at The University of January 2001 and October 2004, 46 patients with
Texas M. D. Anderson Cancer Center for patients KIT-positive GIST, as determined by CD117
with either locally advanced, recurrent, or metastatic (DakoCytomation California Inc., Carpinteria, CA)
GIST is to use imatinib as first-line treatment and immunohistochemical analysis underwent surgical
then consider surgical resection when the tumor has exploration at M. D. Anderson for tumor resection
demonstrated adequate response to allow for com- after imatinib treatment and were included in this
plete resection of any residual disease. However, the study. Permission to perform the study and a waiver
indications, timing, and role of surgery for these of informed consent were obtained from the insti-
patients are not well-established. Before the avail- tutional review board.
ability of imatinib, complete surgical resection of Patient data collected from the sarcoma database,
recurrent or metastatic GIST was achieved in only medical records database, and patient charts included
one third of patients, and the median postoperative age, sex, site of primary tumor, extent of disease at
disease-free survival and overall survival times for initial presentation, indication for preoperative
this subgroup was 12–19 months and 17.5–50 imatinib treatment, dose and duration of imatinib
months, respectively. The five-year survival was treatment, surgical outcome, and disease status at last
approximately 20%.11–15 We previously reported on follow-up. For the purpose of analysis, patients were
the feasibility of tumor resection in patients with divided into two main groups, those with locally
recurrent or metastatic GIST after imatinib treat- advanced primary GISTs and those with recurrent or
ment.16 In that report, 16 of 17 patients underwent metastatic GISTs.
complete resection of all macroscopic disease. At the Patients were assessed clinically and radiographi-
short median follow-up of six months, all 17 patients cally with CT imaging before and after surgery every
were alive, and 14 remained free of disease. It is not 1–9 months depending on their response to treat-
clear which patients may benefit the most from ment and the preference of the treating physician.
surgical resection after preoperative imatinib. In Patients were followed-up to February 2005 and
addition, the long-term effect of preoperative imati- complete follow-up data were available for all 46
nib on postresection disease-free and overall survival patients.
times in patients with recurrent or metastatic GIST
has not been established. Determining these factors
Definitions
would allow for improved treatment of patients with
recurrent or metastatic GIST. Locally advanced primary GIST was defined as
Surgery is currently the main treatment modality radiographic evidence of significant involvement of a
for patients with primary GIST. Two ongoing clinical single organ with tumor size ‡5 cm or extension of

Ann. Surg. Oncol. (
2006)

 SURGERY AFTER IMATINIB TREATMENT IN GIST

the tumor to adjacent organs. Recurrent disease was primary GIST before surgical resection; seven of
defined as the presence of histologically or these patients had synchronous metastasis and were
radiographically demonstrated recurrence of tumor included in the recurrent or metastatic GIST group.
after a previous surgical resection of the primary All 18 patients underwent surgical exploration at M.
GIST. Disease appearing in the region of the previous D. Anderson.
intraperitoneal tumor was called ‘‘recurrence,’’ and Patients with resectable primary GIST are
disease that had spread to noncontiguous distant sites not routinely treated with preoperative imatinib at
such as the liver or lung was called ‘‘metastasis.’’ M. D. Anderson. The patients in this study received
Complete resection was defined as removal of all preoperative imatinib for locally advanced primary
macroscopic disease, and incomplete resection was GIST that were deemed to be unresectable or bor-
defined as the presence of residual macroscopic dis- derline resectable with en bloc resection of adjacent
ease after resection. A complete pathologic response organs.
was defined as no evidence of residual tumor cells in The duration of preoperative imatinib treatment
the extirpated surgical specimen, whereas a partial was determined based on the radiographic response.
pathologic response was defined as evidence of Patients treated for locally advanced primary GIST
residual viable tumor cells within hyaline or myxoid underwent surgical resection after there were no
areas consistent with treatment effect. No pathologic further change in tumor size or enhancement between
response was defined as evidence of viable tumor cells two consecutive radiographic imaging studies 2–3
in the absence of treatment effect. Status of disease at months apart (n=10), or the tumor caused increased
last follow-up was determined using the most recent pain (n=1). Patients treated for recurrent or meta-
clinical and radiographic evaluation for that patient. static GIST received imatinib until no further change
If a patient had expired, the date of death and the was observed between consecutive CT imaging 2–3
disease status at death was recorded. months apart and the surgeon deemed that further
Disease-free survival time was defined as the time treatment would not change the surgical procedure.
from post-imatinib resection of locally advanced In patients who developed radiographic evidence of
primary, recurrent, or metastatic GIST to clinical or tumor progression on imatinib, the dose of imatinib
radiographic evidence of recurrent disease. Postop- was initially increased and if there was evidence of
erative survival time was defined as the time from continued progression the patient then underwent
post-imatinib resection of locally advanced primary, surgery.
recurrent, or metastatic GIST to the last documented
follow-up or patient death. Overall survival time was
defined as the time from initiation of imatinib for Evaluation of Tumor Response to Imatinib Treatment
locally advanced primary, recurrent, or metastatic
An experienced radiologist (C.S.N.) retrospectively
GIST to the time of last follow-up or patient death.
re-reviewed all radiographic data for the 46 patients
Postoperative follow-up time was the time from post-
in our study and categorized the longitudinal (i.e.,
imatinib surgery to the last follow-up or patient
over the entire preoperative imatinib treatment peri-
death. Total follow-up time was defined as the time
od) response to imatinib into four categories based on
from initiation of imatinib treatment to the last fol-
changes in tumor size, degree and extent of
low-up or patient death.
enhancement, and the presence or absence of solid
nodules within the tumor:
Surgery and Imatinib Treatment
1. Complete response: failure to identify a lesion
Twenty-eight (61%) of the 46 patients underwent previously identified on a radiographic image.
surgery as the initial treatment for their primary 2. Partial response:
GIST; 26 of these patients were operated on at
(a) Continuous regression: decreased tumor size or
other institutions and subsequently referred to
enhancement throughout treatment.
M. D. Anderson for follow-up, and two were oper-
(b) Initial regression then stabilization: initial de-
ated on at M. D. Anderson. All 28 of these patients
crease in tumor size or enhancement followed by
developed recurrent or metastatic GIST and were
a period of no further change.
treated with imatinib at M. D. Anderson and subse-
quently underwent surgical exploration at M. D. 3. Stable disease: no evidence of change in tumor size
Anderson. The remaining 18 (39%) patients received or enhancement throughout treatment.
preoperative imatinib at M. D. Anderson for intact 4. Progressive disease:

Ann. Surg. Oncol. (
2006)

 R. H. I. ANDTBACKA ET AL.

(a) Initial regression then progression: initial TABLE 1. Characteristics of primary GISTs (n = 46)
decrease in tumor size or enhancement followed Characteristic No. (%)
by development of new lesions within or outside
Site of primary GIST
the tumor and increase in tumor size or enhance- Small bowel 21 (46%)
ment over time. Stomach 13 (28%)
(b) Continuous progression: increase in tumor size or Colon 5 (11%)
Rectum 4 (9%)
enhancement throughout treatment. Esophagus 1 (2%)
Unknown 2 (4%)
The tumor volume in patients with locally ad- Extent of disease at diagnosis of primary GIST
vanced primary GIST was calculated from CT or Single intraperitoneal tumor 29 (63%)
magnetic resonance imaging (MRI) studies before Multiple intraperitoneal tumors 3 (7%)
Single perirectal tumor 4 (9%)
and after imatinib treatment and was approximated Single periesophageal tumor 1 (2%)
using the formula for a sphere, 4pr3/3, where ‘‘r’’ Single intraperitoneal tumor and liver metastasis 7 (15%)
represents the largest tumor radius. A change in tu- Multiple intraperitoneal tumors and liver metastasis 2 (4%)
mor volume with preoperative imatinib treatment GIST, gastrointestinal stromal tumor.
was expressed as an absolute change relative to the
pre-treatment volume.
Thirty-five (76%) of the 46 patients were treated for
recurrent or metastatic GIST; seven of these had an
Statistical Analysis intact primary GIST with synchronous liver metas-
tasis (one patient had an isolated liver metastasis,
Differences between proportions were tested with
whereas six patients had multiple liver metastases)
the chi-square test or FisherÕs exact test as appro-
(Table 2). The remaining eleven patients (24%) were
priate. The WilcoxonÕs rank-sum test was used to
treated for locally advanced primary GIST. The
analyze differences between medians. Survival times
median preoperative imatinib treatment duration for
were calculated by the Kaplan–Meier method, and
all 46 patients was 12.9 months (range, 2.8–38.1
differences in survival times were tested using the log-
months). For the 35 patients with recurrent or met-
rank test. A difference was accepted as significant
astatic GIST, the median time of preoperative
when P £ .05.
imatinib treatment was 15.2 months (range, 4.4–38.1
months), whereas the median time was significantly
RESULTS shorter for the eleven patients with locally advanced
primary GIST (11.9 months; range, 2.8–15.0 months;
P = .02). Thirty-two (70%) patients received 400 mg
Patient and Tumor Characteristics
daily, six (13%) received 600 mg daily, and eight
A total of 46 patients (22 men and 24 women) (17%) received 400 mg twice daily of imatinib at the
underwent surgery for GIST after preoperative time of surgery.
treatment with imatinib. The median age of the pa-
tients was 55.7 years (range, 23.0 to 75.9 years) at
Response to Preoperative Imatinib Treatment
diagnosis of primary GIST and 58.7 years (range,
28.4 to 80.9 years) at surgery after imatinib treat- Tumor response to imatinib was evaluated in most
ment. There was no significant difference in age dis- patients with serial CT imaging. One patient with
tribution between the genders at either time point. perirectal GIST was followed with serial MRI. Of the
Most patients were white (82%); 9% were black and 46 patients treated, one (2%) patient with a GIST at
9% were Hispanic. The small bowel was the most the gastroesophageal junction had a complete radio-
common site for primary GIST (Table 1). At initial graphic response (Table 2), although on surgical
presentation with GIST, 74% of patients had a single exploration, this patient was found to have a 2-cm
tumor focus without metastasis, 7% had multiple GIST with viable tumor cells. Nineteen (41%) pa-
intraperitoneal tumors without distant metastasis, tients had a partial response, one (2%) had stable
and 19% had synchronous liver metastasis. disease, and 25 (55%) had progressive disease (Ta-
ble 2). Of the 25 patients with progressive disease,
seven exhibited continuous progression on imatinib,
Preoperative Imatinib Treatment
whereas 18 had disease that initially regressed but
At the initiation of imatinib treatment, the extent subsequently progressed on imatinib. In these 18
of GISTs was assessed by CT or MRI imaging. patients, the median time from initiation of imatinib

Ann. Surg. Oncol. (
2006)

 SURGERY AFTER IMATINIB TREATMENT IN GIST

TABLE 2. Radiographic response to preoperative imatinib treatment for GISTs (n = 46)

Patients with recurrent or metastatic GIST
Patients with locally
All patients advanced primary GISTa Complete resection Incomplete resection
Response (n = 46) (n = 11) (n = 11) (n = 24)
Complete response (no., %) 1 (2%) 1 (9%) 0 (0%) 0 (0%)
Partial response (no., %) 19 (41%) 8 (73%) 10 (91%) 1 (4%)
Continuous regression (no.) 6 4 2 0
Initial regression then stable disease (no.) 13 4 8 1
Stable disease (no., %) 1 (2%) 1 (9%) 0 (0%) 0 (0%)
Progressive disease (no., %) 25 (55%) 1 (9%) 1 (9%) 23 (96%)
Initial regression then progression (no.) 18 0 0 18
Continuous progression (no.) 7 1 1 5

GIST, gastrointestinal stromal tumor.

a
All 11 patients underwent complete resection.

to evidence of disease progression was 18.0 months patients in our cohort, whereas 24 patients (52%) had
(range, 5.3–37.4 months). an incomplete resection (Tables 2 and 3). Half of the
Radiographic response to imatinib was based on complete resections were performed in patients with
changes in tumor size, degree and extent of locally advanced primary GIST and half in patients
enhancement, and the presence or absence of solid with recurrent or metastatic GIST, whereas all of the
nodules within the tumor over the entire preoperative incomplete resections occurred in patients with
treatment period. Of the 46 patients in our study, 38 recurrent or metastatic GIST; this distribution was
patients showed an initial radiographic response to significantly different (P < .001) (Table 4). The
imatinib. In these 38 patients, 28 (73%) exhibited a median total follow-up time was the same for patients
reduction in both tumor size and enhancement, nine who underwent complete resection and those who
(24%) had only a reduction in size and one (3%) had underwent incomplete resection (P = 0.8). In con-
only a reduction in enhancement. Of the 18 patients trast, the median postoperative follow-up time was
with GISTs that exhibited initial radiographic significantly longer for patients who underwent
regression to imatinib treatment but subsequently complete resection (P < .001).
progressed before surgery (Table 2), the progression
was evidenced by new intratumoral nodules without Resection of Locally Advanced Primary Tumors after
substantial alterations in the overall size of the Imatinib Treatment
dominant lesions in five patients. All eleven patients treated with imatinib for locally
Twenty-one (46%) of the 46 patients also under- advanced primary GIST underwent complete resec-
went [18F]-fluorodeoxyglucose positron emission tion. One patientÕs tumor did not respond to imatinib
tomography (FDG–PET) imaging before imatinib and showed radiographic evidence of tumor
treatment, and 18 of these underwent serial FDG– progression before resection; the remaining ten pa-
PET scanning during imatinib treatment. The CT and tients had radiographic evidence of complete re-
FDG-PET imaging results were concordant for 14 sponse (one patient), partial response (eight patients)
(78%) of these 18 patients. For the remaining four or stable disease (one patient) (Table 2). The nine
patients, the results were discordant in that the CT patients with a complete or partial response had an
imaging indicated residual disease or disease absolute median decrease in tumor volume of 85%
progression and the FDG–PET imaging demon- (range, 27–99%). Subjectively, the treating surgeons
strated no increased metabolic activity. Pathologic felt that all nine patients were able to undergo less
analysis of the surgical resection specimens indicated extensive surgery than was anticipated by the surgeon
residual viable disease in all four patients. Hence, before imatinib treatment.
FDG–PET imaging underestimated the tumor bur- Pathologic evaluation of the surgical specimens
den in these four patients. from all eleven patients indicated complete response
to imatinib in one patient and partial response in nine
patients. The patient with radiographic evidence of
Completeness of Surgical Resection and Outcome
tumor progression did not exhibit a pathologic
After Preoperative Imatinib Treatment
treatment response.
Complete surgical resection after treatment with The median total follow-up time for patients with
imatinib was accomplished in 22 (48%) of the 46 locally advanced primary GIST was 31.4 months

Ann. Surg. Oncol. (
2006)

 R. H. I. ANDTBACKA ET AL.

TABLE 3. Indications for preoperative imatinib treatment, follow-up times, and surgical outcomes, by completeness of resection
of GIST (n = 46)
Factor Complete resectiona (n = 22) Incomplete resection (n = 24) P
Indication for preoperative imatinib <.001*
Primary GIST (no., %) 11 (50%) 0 (0%)
Recurrence or metastasis (no., %) 11 (50%) 24 (100%)
Total follow-up time .8
Median (months) 36.2 36.2
Range (months) 15.5–48.1 11.6–48.0
Postoperative follow-up time <.001à
Median (months) 23.4 11.8
Range (months) 0.2–35.3 0.8–30.9
Disease status at last follow-up <.001
No evidence of disease (no., %) 16 (73%) 0 (0%)
Alive with disease (no., %) 6 (27%) 19 (79%)
Dead with disease (no., %) 0 (0%) 5 (21%)
Recurrences (no., %) 7b (32%) NA
Time to recurrence
Median (months) 12.5 NA
Range (months) 1.7–32.5 NA
Time to death
Median (months) NA 12.0
Range (months) NA 0.8–13.8

GIST, gastrointestinal stromal tumor; NA, not applicable.

a
Removal of all macroscopic disease.
b
One patient with recurrent disease underwent a second complete resection.
The following statistical analyses were used: * chi-square test, FisherÕs exact test, àWilcoxonÕs rank-sum test.

(range, 15.5–42.5 months), and the median postop- partial response to imatinib at the time of resection
erative follow-up time was 19.5 months (range, 7.2– were more likely to attain a complete resection (ten of
32.5 months). At last follow-up, all eleven patients eleven; 91%), whereas complete resection was rarely
were alive and ten (91%) were free of disease. The achieved in patients who had radiographic evidence
patient who exhibited radiographic evidence of tumor of progressive disease (one of 24; 4%). This difference
progression before surgery and no pathologic treat- in resectability based on radiographic response was
ment response at the time of surgery developed significant (P < .001). Patients who underwent
recurrent GIST with multiple intraabdominal im- complete resection were more likely to have multiple
plants at twelve months postresection. The median intraperitoneal recurrences, whereas patients who
disease-free survival time has not yet been reached for had an incomplete resection were more likely to have
these eleven patients. Postoperatively, eight patients multiple liver and extraperitoneal metastases.
continue on adjuvant imatinib at a median of 15.1 As indicated in Table 2, 23 (96%) of the 24 patients
months (range, 1.8–27.0 months). who had incomplete resection exhibited radiographic
evidence of progressive disease at the time of resec-
Resection of Recurrent or Metastatic Tumors After tion. However, 18 of these 23 patients initially
Imatinib Treatment responded to preoperative imatinib but then
Of the 35 patients with recurrent or metastatic progressed before surgical resection. The median time
GIST, eleven (31%) were able to have complete to progression in these 18 patients was 18.0 months
resection and 24 (69%) had incomplete resection. (range, 5.3–37.4 months), and the median time from
Tables 2 and 4 describe the differences between pa- the diagnosis of progression until surgery was
tients with recurrent and metastatic GIST who 5.7 months (range, 0.1–6.5 months). It is possible that
underwent complete and incomplete resections. some of the incompletely resected patients would
Completeness of resection was associated with have been able to undergo a complete resection had
response to and duration of imatinib treatment, they been operated on when the GIST was still
anatomic location and tumor burden. The median regressing on imatinib. To test this possibility, we
preoperative imatinib treatment time was signifi- compared the median preoperative imatinib treat-
cantly shorter for patients with complete resection ment duration in the ten completely resected patients
than for patients with incomplete resection (P = .03). who exhibited a partial response to imatinib at the
Patients who exhibited radiographic evidence of a time of resection (ten months) and the median time to

Ann. Surg. Oncol. (
2006)

 SURGERY AFTER IMATINIB TREATMENT IN GIST

TABLE 4. Indications for, duration of, and surgical outcomes after preoperative imatinib treatment among patients with
recurrent or metastatic disease, by completeness of resection of GIST (n = 35)
Factor Complete resection (n = 11) Incomplete resection (n = 24) P
Indication for preoperative imatinib
Recurrence
Single tumor (no., %) 0 (0%) 1 (4%)
Multiple tumors (no., %) 3 (27%) 3 (12%)
Isolated liver metastasis
Single tumor (no., %) 2 (18%) 0 (0%)
Multiple tumors (no., %) 0 (0%) 5 (21%)
Recurrence and liver metastasis
Single tumor (no., %) 3 (27%) 3 (12%)
Multiple tumors (no., %) 3 (27%) 9 (38%)
Recurrence and extraperitoneal metastasis (no., %) 0 (0%) 3 (12%)
Duration of preoperative imatinib .03à
Median (months) 10.0 22.9
Range (months) 6.9–37.5 4.4–38.1
Recurrences (no., %) 6 (55%) NA
Time to recurrence
Median (months) 15.1 NA
Range (months) 1.7–32.5 NA
Site of recurrence
Local (no.) 1 NA
Metastasis (no.) 3 NA
Local and metastasis (no.) 2 NA
Follow-up time since imatinib initiation .10à
Median (months) 39.1 36.2
Range (months) 28.7–48.1 11.6–48.0
Postoperative follow-up time <.001à
Median (months) 30.7 11.8
Range (months) 0.2–35.3 0.8–30.9
Status at last follow-up <.001 *
No evidence of disease (no., %) 6a (55%) 0 (0%)
Alive with disease (no., %) 5 (45%) 19 (79%)
Dead with disease (no., %) 0 (0%) 5 (21%)

GIST, gastrointestinal stromal tumor; NA, not applicable.

a
GIST recurred in one patient with a single liver metastasis who was rendered free of disease after a second liver resection.
The following statistical analyses were used: * chi-square test, à WilcoxonÕs rank-sum test.

progression in the 18 patients whose disease initially resection of multiple intraperitoneal tumors and
regressed on imatinib but then progressed before ethanol injection for liver metastasis, one for radio-
surgery (18.0 months). This difference was statisti- frequency ablation of multiple liver metastases, and
cally significant (P = .04). Of the 18 patients, only one for resection of a stable single liver metastasis.
two had disease that progressed before ten months of The patient with a single stable liver metastasis was
treatment. rendered free of disease by this second surgery and
Pathologic evaluation of the eleven completely remained free of disease at 25.6 months after the
resected recurrent or metastatic GISTs indicated a second operation. At the time of last follow-up, six of
complete response to imatinib in two patients and a the eleven patients, including the two patients with
partial response in nine patients. All eleven patients complete pathologic response, were free of disease at
were alive at a median postoperative follow-up time a median of 23.6 months (range, 0.2–31.6 months),
of 30.7 months (range, 0.2–35.3 months) (Table 4). whereas the other five patients have exhibited disease
All patients received imatinib postoperatively for a progression on imatinib.
median of 23.7 months (range, 1.7–33.8 months), and Among the 24 patients with incompletely resected
10 (91%) had continued on imatinib at last follow-up. GIST, the median overall survival time had not been
Six (55%) of the eleven patients had evidence of reached at the time of this analysis, and 19 (79%) of
recurrent disease at a median of 15.1 months (range, the 24 patients were alive at a median postoperative
1.7–32.5 months), resulting in an overall postopera- follow-up time of 11.8 months (range, 0.8–30.9
tive median disease-free survival time of 24.7 months. months) (Table 4). All 24 patients received imatinib
Three of the six patients with recurrent disease postoperatively for a median of 10.0 months (range,
underwent a second operation: one for palliative 0.8–31.9 months). Of the 19 patients who are alive,

Ann. Surg. Oncol. (
2006)

 R. H. I. ANDTBACKA ET AL.

twelve have developed progressive disease on imati- There was a reasonable concordance between a
nib despite dose escalation, and five of these twelve reduction in tumor size and tumor enhancement in
have discontinued imatinib and have been enrolled in tumors that responded to imatinib. In GIST which
either a phase I or phase II clinical trial. The five initially responded to imatinib but subsequently
patients who died with disease did so at a median of progressed, five out of 18 patients developed new
12.0 months (range, 0.8–13.8 months) after surgery. intratumoral lesions without any change in the
One of the patients died 24 days after surgery from a overall tumor size. Hence, at the present time, we
myocardial infarction, whereas the other four died as believe that using more expanded criteria in assessing
a result of disease progression. Hence, the 30-day radiographic tumor response to imatinib is appro-
postoperative mortality rate for all 46 patients was priate. We recognize that our findings are based on a
2.2%. relatively small number of patients and that pro-
spective validation of the response criteria and the
proposed response categories used in our study is
DISCUSSION necessary. Nonetheless, in our patient population,
radiographic evidence of disease regression was
Our study demonstrated that preoperative imatinib associated with complete surgical resection, indicat-
resulted in decreased tumor volume in patients with ing that it may be used as a guideline by which
locally advanced primary GIST, and this was asso- patients are assessed for the appropriateness of sur-
ciated with a high rate of complete surgical resection. gery in order to render them free of disease.
In patients with recurrent or metastatic GIST, those Tumor metabolism can also be used as a marker of
with imatinib-responsive GISTs were more likely to the response to imatinib. Several studies have indi-
undergo complete resection and had prolonged dis- cated the benefits of using the metabolic activity
ease-free survival. Patients with recurrent or meta- measurements of FDG–PET imaging in assessing
static GIST that initially regressed on imatinib but tumor response to imatinib,22,26,27 whereas one study
then progressed before resection rarely underwent found FDG–PET to be less predictive of the patho-
complete resection, indicating that early resection logic tumor response.16 In our study, four (22%) of 18
may be beneficial and should be considered for patients who underwent serial FDG–PET showed no
patients with imatinib-responsive recurrent or meta- evidence of active disease on PET scan just prior to
static GIST. Preoperative imatinib treatment resulted surgery but were found to have evidence of residual
in a reduction in both tumor size and tumor tumor on both CT imaging and pathologic analysis.
enhancement on radiographic evaluation, indicating Hence, FDG–PET may be beneficial in evaluating the
that size alone may not be the best measure of tumor initial response to imatinib treatment but may be
response. inaccurate as a predictor of residual disease at the
Radiographic response of solid tumors to chemo- time of resection.
therapeutic agents has traditionally been evaluated by Surgery remains the standard treatment for pri-
measurement of the two-dimensional19 or one- mary resectable GISTs. Ongoing trials by the Radi-
dimensional20 change in tumor size on serial imaging. ation Therapy Oncology Group (RTOG-S-0132)18
However, change in size alone, as defined by the and M. D. Anderson17 are evaluating the potential
Response Evaluation Criteria in Solid Tumors (RE- benefits of preoperative imatinib in treating resect-
CIST),20 has been shown to be a very inaccurate able localized or metastatic GIST. These trials are
predictor of tumor response in GISTs treated with still accruing patients and results are not expected for
imatinib.21–24 Evaluating other tumor changes such several years. Only a few anecdotal case reports have
as enhancement, density and development of new been published on the potential benefits of preoper-
intratumoral lesions have been shown to more accu- ative imatinib for primary GIST.28–30 In these case
rately predict the response of GISTs to imatinib than reports, preoperative imatinib treatment of unresec-
size alone.22 Currently, many investigators (including table GISTs resulted in tumor regression and enabled
a group at M.D. Anderson)25 are exploring the use of surgical resection. Our study is the first to report on
these radiographic variables to predict response to the surgical outcome after preoperative imatinib
imatinib. In our study, treatment responses were treatment for locally advanced primary GIST in a
assessed by changes in tumor size, enhancement and larger group of patients. We found that nine (82%) of
development of new intratumoral nodules. In addi- the eleven patients treated with preoperative imatinib
tion, these changes were determined longitudinally had a substantial reduction in tumor volume, which
over the entire preoperative treatment with imatinib. would potentially decrease the extent of surgical

Ann. Surg. Oncol. (
2006)

 SURGERY AFTER IMATINIB TREATMENT IN GIST

resection. Moreover, all eleven patients were able to response and had complete resection at the time of
have complete surgical resection, and one patient had surgery. Hence, one could postulate that complete
a pathologic complete response. resection might have been achieved in the 18 patients
Patients treated with preoperative imatinib for who underwent incomplete resection had they
locally advanced primary GIST had a longer median undergone surgery when their recurrent or metastatic
disease-free survival time compared with historical GIST was still demonstrating response to imatinib.
controls. After a median postoperative follow-up of This suggests that surgical intervention at an earlier
19.5 months, only one of the eleven patients with time in the preoperative therapy with imatinib may be
locally advanced primary GIST developed recurrent optimal in order to prevent tumor progression that
disease, and the median disease-free survival time has ultimately results from imatinib resistance. Resis-
not yet been reached. This outcome is much tance to imatinib in recurrent or metastatic GIST is
improved from that previously reported before now being reported,35–37 and the natural history of
imatinib was available. Ng et al.12 reported an imatinib-treated recurrent or metastatic GIST has
18-month median disease-free survival time after not yet been described. On the basis of the findings
complete resection of primary gastrointestinal leio- from our study, we believe that strong consideration
myosarcoma and a disease recurrence rate of 60% should be given to early surgical intervention in
within two years. Other investigators have reported patients with resectable recurrent or metastatic GIST
median disease-free survival times ranging from 7 to whose disease is imatinib-responsive or stable on
20 months.11,31 Whether the prolonged disease-free imatinib therapy. This may be especially true in those
survival time we observed in our study is generaliz- patients in whom further treatment with imatinib
able to a larger number of patients with primary would not alter the extent of surgical resection
GIST will need to be assessed in a larger cohort of required for complete resection. Although medical
patients. Eight of the eleven patients with locally treatment of GIST has improved, surgery remains the
advanced primary GIST in our study continued on only proven method to render these patients free of
imatinib postoperatively, and it is possible that the disease.
prolonged disease-free survival time is a result of The overall complete resection rate after preoper-
postoperative rather than preoperative imatinib ative imatinib for patients with recurrent or meta-
therapy. It is hoped that ongoing trials, including the static GISTs in our study was 31%. This is similar to
RTOG-S-0132,18 the M. D. Anderson trail,17 two the complete resection rates of 30–34% that were
adjuvant trials sponsored by the American College of reported before imatinib was available.13–15 However,
Surgeons Oncology Group (ACOSOG-Z9000 and - the postoperative median disease-free survival time in
Z9001),32,33 and an adjuvant trial conducted by the the eleven patients with recurrent or metastatic GIST
European Organization for Research and Treatment who underwent complete resection was 24.7 months,
of Cancer (EORTC-62024)34 will provide answers to which is longer than the 12–19 month disease-free
these clinical questions. survival reported in this group of patients before
Complete surgical resection can also be achieved in imatinib was available.11,13,14 The median postoper-
patients with recurrent or metastatic GISTs that ative survival time after complete resection of recur-
respond to preoperative imatinib therapy. However, rent or metastatic GIST before imatinib was available
the optimal time to proceed with surgical resection in ranged from 17.5 to 50 months.12,14,15 In the eleven
these patients has not been established. In our study, patients who underwent complete resection in our
ten (91%) of the eleven patients who underwent study, the median postoperative survival time has not
complete resection had radiographic evidence of a yet been reached, and all patients were alive at a
partial response to imatinib at the time of surgery. median follow-up time of 30.7 months. These results
In contrast, 23 (96%) of the 24 patients who had are encouraging, but longer follow-up is needed in
incomplete resection of their disease had radiographic this group of patients before we can make any
evidence of progressive disease at the time of surgery. definitive conclusions.
Interestingly, 18 of these 23 patients had initially The role for incomplete surgical resection (i.e.,
demonstrated response to imatinib but developed debulking) in recurrent or metastatic GIST has not
progression on imatinib before surgery. The median been established. The median postoperative survival
time to progression in these 18 patients was 18.0 time after incomplete resection ranged from 2 to 20
months, significantly longer than the median 10.0 months before imatinib was available.11,12,14,15 In our
months of preoperative imatinib treatment in the ten study, the median postoperative survival time has not
patients who had radiographic evidence of a partial yet been reached, and 19 (79%) of the 24 patients who

Ann. Surg. Oncol. (
2006)

 R. H. I. ANDTBACKA ET AL.

underwent incomplete resection were alive at a 2. Miettinen M, Sarlomo-Rikala M, Lasota J. Gastrointestinal

stromal tumors: recent advances in understanding of their
median postoperative follow-up time of 11.8 months. biology. Hum Pathol 1999; 30(10):1213–20.
Moreover, the median time since imatinib therapy 3. Fletcher CD, Berman JJ, Corless C, et al. Diagnosis of gas-
was started in these 24 patients was 36.2 months, and trointestinal stromal tumors: A consensus approach. Hum
Pathol 2002; 33(5):459–65.
the one-, two-, and three-year overall survival was 4. Le Cesne A, Judson I, Crowther D, et al. Randomized phase
95.8%, 91.0% and 79.8%, respectively. This two-year III study comparing conventional-dose doxorubicin plus
survival is better than the reported two-year overall ifosfamide versus high-dose doxorubicin plus ifosfamide plus
recombinant human granulocyte-macrophage colony-stimu-
survival rate of 69–79% in patients with recurrent, lating factor in advanced soft tissue sarcomas: A trial of the
metastatic, or unresectable GIST treated with imati- European Organization for Research and Treatment of Can-
nib alone.7,10 It is possible that some patients may cer/Soft Tissue and Bone Sarcoma Group. J Clin Oncol 2000;
18(14):2676–84.
benefit from incomplete resection, especially those 5. Nielsen OS, Judson I, van Hoesel Q, et al. Effect of high-dose
patients with recurrent or metastatic GIST that ifosfamide in advanced soft tissue sarcomas. A multicentre
exhibit a mixed response to imatinib (e.g., some phase II study of the EORTC Soft Tissue and Bone Sarcoma
Group. Eur J Cancer 2000; 36(1):61–7.
tumor nodules progress and others regress on imati- 6. Patel SR, Gandhi V, Jenkins J, et al. Phase II clinical investi-
nib). Resecting nodules that progress on imatinib gation of gemcitabine in advanced soft tissue sarcomas and
therapy may be beneficial if this proves to prolong the window evaluation of dose rate on gemcitabine triphosphate
accumulation. J Clin Oncol 2001; 19(15):3483–9.
overall survival time. Incomplete resection or ‘‘deb- 7. Verweij J, Casali PG, Zalcberg J, et al. Progression-free sur-
ulking surgery’’ can also be considered in select vival in gastrointestinal stromal tumours with high-dose
patients for alleviation of symptoms and palliation. imatinib: randomised trial. Lancet 2004; 364(9440):1127–34.
8. van Oosterom AT, Judson I, Verweij J, et al. Safety and effi-
Further studies are needed to determine which cacy of imatinib (STI571) in metastatic gastrointestinal stromal
patients benefit from partial resection. tumours: a phase I study. Lancet 2001; 358(9291):1421–3.
In summary, surgical resection after preoperative 9. Demetri GD, von Mehren M, Blanke CD, et al. Efficacy and
safety of imatinib mesylate in advanced gastrointestinal stro-
imatinib treatment is feasible and should be consid- mal tumors. N Engl J Med 2002; 347(7):472–80.
ered for patients with GISTs that are imatinib- 10. Rankin C, von Mehren M, Blanke C, et al. Dose effect of
responsive or stable on follow-up radiographic imatinib in patients with metastatic GIST—Phase III Sarcoma
Group Study S0033 [Abstract]. J Clin Oncol 2004;
studies without evidence of progression. Preoperative 22(14S):9005.
imatinib for locally advanced primary GISTs can 11. Mudan SS, Conlon KC, Woodruff JM, Lewis JJ, Brennan MF.
decrease tumor volume, is associated with a high rate Salvage surgery for patients with recurrent gastrointestinal
sarcoma: prognostic factors to guide patient selection. Cancer
of complete resection, and should be considered 2000; 88(1):66–74.
in these patients. In patients with recurrent or meta- 12. Ng EH, Pollock RE, Romsdahl MM. Prognostic implications
static GISTs, strong consideration should be given to of patterns of failure for gastrointestinal leiomyosarcomas.
Cancer 1992; 69(6):1334–41.
early resection of imatinib-responsive tumors, since 13. Ng EH, Pollock RE, Munsell MF, Atkinson EN,
disease-free survival can be prolonged in these Romsdahl MM. Prognostic factors influencing survival in
patients. When surgical resection is delayed and gastrointestinal leiomyosarcomas. Implications for surgical
management and staging. Ann Surg 1992; 215(1):68–77.
imatinib resistance develops, complete resection is 14. DeMatteo RP, Lewis JJ, Leung D, Mudan SS, Woodruff JM,
rarely achieved and the usefulness of surgical resec- Brennan MF. Two hundred gastrointestinal stromal tumors:
tion for these patients is questionable. recurrence patterns and prognostic factors for survival. Ann
Surg 2000; 231(1):51–8.
15. Clary BM, DeMatteo RP, Lewis JJ, Leung D, Brennan MF.
Gastrointestinal stromal tumors and leiomyosarcoma of the
ACKNOWLEDGMENTS abdomen and retroperitoneum: a clinical comparison. Ann
We thank Elizabeth L. Hess of the M. D. Ander- Surg Oncol 2001; 8(4):290–9.
16. Scaife CL, Hunt KK, Patel SR, et al. Is there a role for surgery
son Department of Scientific Publications for editing in patients with ‘‘unresectable’’ cKIT+ gastrointestinal
this manuscript and Dr. Joe Ensor of the M. D. stromal tumors treated with imatinib mesylate? Am J Surg
Anderson Department of Biostatistics and Applied 2003; 186(6):665–9.
17. Trent J, Hunt KK (2003) A prospective, randomized, phase II
Mathematics for assistance with statistical analysis. study of preoperative plus postoperative imatinib mesylate
(Gleevec, formerly STI-571) in patients with primary, recur-
rent, or metastatic resectable, kit-expressing, gastrointestinal
stromal tumor (GIST). Available: http://www.utm-
ext01a.mdacc.tmc.edu/dept/prot/clinicaltrialswp.nsf/index/
REFERENCES id03-0023 [accessed August 14, 2006].
18. Eisenberg B (2002) A phase II trail of neoadjuvant/adjuvant
1. Kindblom LG, Remotti HE, Aldenborg F, Meis-Kindblom JM. STI-571 (Gleevec NSC #716051) for primary and recurrent
Gastrointestinal pacemaker cell tumor (GIPACT): gastrointes- operable malignant GIST expressing the KIT receptor tyrosine
tinal stromal tumors show phenotypic characteristics of the kinase (CD117). Available: http://www.rtog.org/members/
interstitial cells of Cajal. Am J Pathol 1998; 152(5):1259–69. protocols/s0132/s0132.pdf [accessed August 14, 2006].

Ann. Surg. Oncol. (
2006)

 SURGERY AFTER IMATINIB TREATMENT IN GIST

19. Miller AB, Hoogstraten B, Staquet M, Winkler A. Reporting 28. Katz D, Segal A, Alberton Y, Jurim O, Reissman P, Catane R,
results of cancer treatment. Cancer 1981; 47(1):207–14. Cherny NI. Neoadjuvant imatinib for unresectable gastroin-
20. Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines testinal stromal tumor. Anticancer Drugs 2004; 15(6):599–602.
to evaluate the response to treatment in solid tumors. Euro- 29. Hiramatsu Y, Konno H, Kamiya K, et al. A case of gastric
pean Organization for Research and Treatment of Cancer, GIST treated preoperatively by imatinib mesylate. Gan To
National Cancer Institute of the United States, National Kagaku Ryoho 2004; 31(8):1219–23.
Cancer Institute of Canada. J Natl Cancer Inst 2000; 30. Bumming P, Andersson J, Meis-Kindblom JM, et al. Neoad-
92(3):205–16. juvant, adjuvant and palliative treatment of gastrointestinal
21. Casali PG, Fumagalli E, Messina A, et al. Tumor response to stromal tumours (GIST) with imatinib: a centre-based study of
imatinib mesylate in advanced GIST [Abstract]. J Clin Oncol 17 patients. Br J Cancer 2003; 89(3):460–4.
2004; 22(14S):9028. 31. Conlon KC, Casper ES, Brennan MF. Primary gastrointestinal
22. Choi H, Charnsangavej C, de Castro Faria S, et al. CT eval- sarcomas: analysis of prognostic variables. Ann Surg Oncol
uation of the response of gastrointestinal stromal tumors after 1995; 2(1):26–31.
imatinib mesylate treatment: a quantitative analysis correlated 32. DeMatteo RP (2001) A phase III randomized double-blind
with FDG PET findings. AJR Am J Roentgenol 2004; study of adjuvant STI571 (Gleevec) versus placebo in patients
183(6):1619–28. following the resection of primary gastrointestinal stromal tu-
23. Shankar S, van Sonnenberg E, Desai J, Dipiro PJ. mor (GIST). Available: http://www.acosog.org/studies/synop-
Van Den Abbeele A, Dimitri GD. Gastrointestinal stromal ses/Z9001_Synopsis.pdf [accessed April 20, 2005].
tumor: new nodule-within-a-mass pattern of recurrence after 33. DeMatteo RP (2001) A phase II study of adjuvant STI571
partial response to imatinib mesylate. Radiology 2005; (Gleevec) therapy in patients following completely resected
235(3):892–8. high-risk primary gastrointestinal stromal tumor (GIST).
24. Holdsworth CH, Manola J, Badawi RD, et al. Use of Available: http://www.acosog.org/studies/synopses/Z9000_
computerized tomography as an early prognostic indicator of Synopsis.pdf [accessed August 14, 2006].
response to imatinib mesylate in patients with gastrointesti- 34. Casali PG (2004) Phase III randomized study of adjuvant
nal stromal tumors [Abstract]. J Clin Oncol 2004; imatinib mesylate versus observation only in patients with
22(14S):3011. completely resected localized gastrointestinal stromal tumor at
25. Choi H, Charnsangavej C, Macapinilac HA, et al. Correlation intermediate- or high-risk of relapse. Available: http://
of computerized tomography and positron emission tomogra- www.eortc.be/protoc/Details.asp?Protocol=62024 [accessed
phy in patients with metastatic GIST treated at a single insti- August 14, 2006].
tion with imatinib mesylate [Abstract]. Proc Am Soc Clin Oncol 35. Chen LL, Trent JC, Wu EF, et al. A missense mutation in KIT
2003; 22:819. kinase domain 1 correlates with imatinib resistance in gastro-
26. Gayed I, Vu T, Iyer R, Johnson M, Macapinlac H, Swanston N, intestinal stromal tumors. Cancer Res 2004; 64(17):5913–9.
Podoloff D. The role of 18F-FDG PET in staging and early 36. Tamborini E, Bonadiman L, Greco A, et al. A new mutation in
prediction of response to therapy of recurrent gastrointestinal the KIT ATP pocket causes acquired resistance to imatinib in a
stromal tumors. J Nucl Med 2004; 45(1):17–21. gastrointestinal stromal tumor patient. Gastroenterology 2004;
27. Goerres GW, Stupp R, Barghouth G, et al. The value of 127(1):294–9.
PET, CT and in-line PET/CT in patients with gastrointestinal 37. Wakai T, Kanda T, Hirota S, et al. Late resistance to imatinib
stromal tumours: long-term outcome of treatment with therapy in a metastatic gastrointestinal stromal tumour is
imatinib mesylate. Eur J Nucl Med Mol Imaging 2005; associated with a second KIT mutation. Br J Cancer 2004;
32(2):153–62. 90(11):2059–61.

Ann. Surg. Oncol. (
2006)