Patient Characteristics Including Tumor Mutation Status
Nineteen patients were accrued from August 8, 2003, to March 26, 2007, at MDACC, TX, USA. The trial schematic is illustrated in Supplemental Data 1. The baseline patient characteristics are provided in Table 1.
Baseline core needle biopsies of tumor were collected prior to imatinib treatment from 14 (74%) of 19 patients. We did not acquire tumor tissue by core-needle biopsy at baseline from five patients owing to lack of consent or untoward risk of biopsy.
Mutation analyses revealed 18 (95%) patients had tumor harboring mutations in either kit or PDGFR-α (Table 1). Interestingly, one patient had both a kit exon 11 duplication and a PDGFR-α exon 12 point mutation. The most common type of mutation was deletion (47% of patients, Supplemental Data 2).
Surgical Intervention and Perioperative Morbidity
Seventeen of 19 patients underwent a timely resection of their GIST after receiving imatinib. In one patient, the surgical resection was delayed by 3 weeks owing to imatinib-associated toxicity. One patient withdrew consent prior to surgical resection for personal reasons not related to the trial.
The surgical procedures are summarized in Table 2A. All patients had exploratory laparotomy and en bloc resection of tumor and pseudo-capsule, with resection of any gross metastatic deposits.
There were no cases of intraoperative tumor rupture or hemorrhage. Additionally, we observed no episodes of wound dehiscence or delayed healing. One patient had a pelvic abscess that was drained percutaneously and resolved on antibiotic therapy. Three (18%) of 17 patients required brief (<3 days) care in the intensive care unit: need for serum glucose monitoring secondary to type I diabetes mellitus, a non-Q-wave myocardial ischemic event on postoperative day 1, and hyperglycemia secondary to type II diabetes mellitus. Patients began postoperative imatinib a median of 22 days (range 7–59 days) after tumor resection.
As compared with a retrospective analysis of 27 imatinib-naïve patients undergoing surgical resection, similar mean estimated blood loss (605.8 mL, p = 0.58), transfusion rates (p = 1.0), and perioperative complication rates (p = 0.1) were demonstrated (Table 2B).
Medical Intervention and Adverse Events
The toxicity profile of imatinib in this clinical trial was similar to that in patients with metastatic GIST (Table 3).6,19 Most patients tolerated therapy well, and there were no deaths. Five grade 4 events occurred in four (22%) patients. Perioperative vascular events were infrequent (12%, 2 of 17 patients), but one patient had non-Q-wave myocardial ischemia and one patient had a transient ischemic event. Both patients had known, pre-existing vascular disease and both fully recovered to baseline function prior to discharge from the hospital. One patient had grade 4 anemia postoperatively secondary to endometriosis and uterine bleeding.
Radiographic Efficacy of Preoperative Imatinib
Sixteen (84%) of 19 patients were assessable for response by 18FDG PET (Fig. 1). One patient did not complete the imaging studies owing to scheduling errors. Two patients had tumors that did not demonstrate glucose uptake on baseline 18FDG PET. Eleven (69%) of 16 patients had tumors that demonstrated ≥ 40% decrease in SUVmax. Moreover, 10 (63%) of 16 patients had tumors with residual SUVmax ≤ 3.9.
Seventeen (89%) of 19 patients had GISTs that were assessable by dCT (Fig. 1). Two patients did not complete the imaging studies because of scheduling errors. Of these 17 patients, 12 (71%) had a >10% (range 11.53–75.14%) decrease in blood flow in the viable regions of their tumors.
Collectively, all patients that were assessed for radiographic response responded to preoperative imatinib by one or more of our criteria. Interestingly, 62% of patients responded exclusively by either a decrease in tumor cell glucose metabolism (PET) or by a decrease in tumor blood flow (dCT) but not both, whereas only 38% responded by both modalities. Thus, we next sought to determine the early cellular effects of imatinib.
Antitumor Efficacy of Preoperative Imatinib
To better understand the tumoral events leading to the radiographic responses evidenced by 18FDG PET and dCT, we comparatively analyzed matched baseline core-needle biopsies and surgical specimens using light microscopy (Fig. 2).
Histologically, equivalently cellular tumor tissue was seen at both baseline and surgery without evidence of myxoid degeneration, a feature characteristic of more prolonged treatment with imatinib. Thus, pathologic response and cytoreduction after this short interval of preoperative imatinib was not appreciated by histological examination in any of the surgical tissues.
We hypothesized that a potential mechanism of initiation of myxoid degeneration was tumor apoptosis. Therefore, we assessed matched pre and post-imatinib tumor tissue for TUNEL. Ten of 14 pretreatment biopsies (4 biopsy specimens were inadequate) and 17 of 17 surgical tumor specimens were analyzed for apoptosis (Fig. 2). Rare TUNEL-positive cells were seen in the pre-imatinib tissue. However, numerous TUNEL-positive cells were observed in the post-imatinib surgical specimens with an absolute increase of 12% for matched tissues (Table 4A).
Moreover, the rate of tumor cell apoptosis was found to be dependent on the duration of preoperative imatinib therapy (Table 4B). Tumor cell apoptosis increased incrementally with duration of imatinib, where patients treated for 7 days with preoperative imatinib had the greatest rate of tumor cell apoptosis (15%, p = 0.04).
A high rate of TUNEL-positive cells in the resected tumor specimens tended to be associated with dCT response; however, this trend did not reach statistical significance (p = 0.13). Interestingly, PET response by either response criteria had no association with GIST cell apoptosis.
Patients with tumors harboring kit exon 11 mutations had a 14% increase in tumor cell apoptosis after preoperative imatinib (p = 0.03) (Table 4C). Conversely, the one patient with tumor harboring a kit exon 9 mutation had no increase in apoptosis with imatinib therapy of 600 mg daily.
At the time of this analysis, eight (42%) patients have successfully completed 2 years of postoperative imatinib. Three (16%) patients continue to take postoperative imatinib on the study. Eight (42%) patients discontinued imatinib prior to completing 2 years of postoperative imatinib: four owing to toxicity, two due to withdrawal of consent, and two being lost to follow-up.
With median follow-up of 32 months, median duration of DFS was 46 months (Fig. 3). Actuarial DFS rates were 94% and 87% at 1 and 2 years, respectively. Six (32%) of the 19 patients have had a recurrence, but no patients have died. Randomization to 3, 5, or 7 days of preoperative imatinib did not affect DFS (p = 0.71).
Interestingly, all patients who had a recurrence on the study had disease that originated in the small bowel. Moreover, these patients either had a large tumor burden (>10 cm) or presented with metastatic and/or recurrent disease. Not surprisingly, larger tumor size predicted shorter DFS (p = 0.02). There was a trend between longer DFS and response measured by dCT and higher levels of tumor apoptosis, but neither reached statistical significance. Moreover, PET response was not predictive of DFS duration.
All six patients who had tumor recurrence had stopped taking imatinib due to completion of the study, noncompliance, or toxicity at the time of the recurrence. Five of these six patients harbored kit exon 11 mutation and were treated for 24, 24, 24, 8, and 15 months postoperatively. The other patient harbored a kit exon 9 mutation and was treated for 24 months postoperatively. Median time to disease progression for these patients after discontinuation of imatinib was 4 months. However, the likelihood of recurrence was not associated with whether a patient was receiving imatinib (p = 0.99), accounting for the fact that the actual schedule of receiving or not receiving imatinib over time varied between patients.
No patient recurred at the cutaneous site of tumor biopsy but rather in the peritoneum (three) and at hepatic sites (three). The five patients that restarted imatinib therapy (one of these patients was lost to follow-up) had a subsequent response as assessed by contrast-enhanced CT.