Initial and Late Resistance to Imatinib in Advanced Gastrointestinal Stromal Tumors Are Predicted by Different Prognostic Factors: A Europea
http://www.100md.com
《临床肿瘤学》
the European Organisation for Research and Treatment of Cancer Data Center, Brussels
University Hospital Gasthuisberg, Leuven, Belgium
Erasmus University Medical Center, Rotterdam
Leiden University Medical Center, Leiden, the Netherlands
Istituto Tumori, Milano, Italy
Institut Gustave Roussy, Villejuif
Centre Leon Berard, Lyon, France
Charite Campus Buch, Robert Roessle Hospital, Berlin
Klinikum Grosshadern
Gesellschaft für Strahlenforschung–National Research Center for Environmental and Health, Munich, Germany
Prince of Wales Hospital, Randwick, Australia
Royal Marsden Hospital, London, United Kingdom
ABSTRACT
PURPOSE: The aim of this study was to identify factors predicting initial and late resistance of GI stromal tumor (GIST) patients to imatinib and to document the dose-response relationship in the prognostic subgroups. This study is based on the European Organisation for Research and Treatment of Cancer–Italian Sarcoma Group–Australasian Gastrointestinal Trials Group randomized trial comparing two doses of imatinib in advanced disease.
PATIENTS AND METHODS: Initial resistance was defined as progression within 3 months of randomization, and late resistance was defined as progression beyond 3 months. Investigated cofactors include imatinib dose, age, sex, performance status, original disease site, site and size of lesions at trial entry, and baseline hematologic and biologic parameters.
RESULTS: Initial resistance was recorded for 116 (12%) of 934 assessable patients and was independently predicted by the presence of lung and absence of liver metastases, low hemoglobin level, and high granulocyte count. Among 818 patients who were alive and progression free at 3 months, 347 subsequent progressions were recorded, and late resistance was independently predicted by high baseline granulocyte count, primary tumor outside of the stomach, large tumor size, and low initial imatinib dose. The impact of initial dose on late resistance was mainly significant in patients with a high baseline granulocyte count (> 5.109/L) and in patients with tumors of GI origin outside of the stomach and small intestine.
CONCLUSION: Our study identifies patients for whom initial and/or long-term treatment needs to be improved and patients who require a high initial dose. Correlation of these results with immunohistochemistry and molecular parameters may further help to understand the biologic mechanisms of resistance.
INTRODUCTION
Soft tissue sarcomas represent 1% of adult malignancies and are a heterogeneous group of neoplasms whose only common denominator is their derivation from mesenchymal tissue. GI stromal tumors (GIST) are a subset of soft tissue sarcomas that were classified relatively recently. Their local treatment essentially consists of surgery. After the stage of resection, these tumors have proven to be insensitive to chemotherapy and radiotherapy.1
Imatinib is a small-molecule tyrosine kinase inhibitor that is active against BCR-ABL, KIT, and PDGFR. KIT is expressed in the vast majority of GISTs and is frequently mutated, leading to constitutive activation in these tumors. A European Organisation for Research and Treatment of Cancer (EORTC) phase I study2 identified the highest feasible dose of imatinib to be 400 mg bid and indicated extensive activity of imatinib in GIST. Phase II studies showed activity at all doses tested (400 to 800 mg).3,4 Two large, randomized, phase III studies comparing doses of 400 mg once a day to 400 mg bid have confirmed the activity of imatinib in terms of progression-free survival and overall survival.5,6 One of these studies has also documented a small but significant benefit with the high-dose regimen (400 mg bid) in terms of progression-free survival.5 Finally, a randomized trial from the French Sarcoma Group has demonstrated that imatinib therapy should be continued indefinitely, even after complete response.7
Response of GIST to imatinib does not always result in an immediate decrease of the size of the lesions but, rather, in an initial inhibition of growth. Objective response (according to Response Evaluation Criteria in Solid Tumors [RECIST] criteria8) has been reported in approximately half of the patients, but time to onset varies largely among patients, and some responses have been first documented more than 1 year after start of therapy.5 Therefore, response to imatinib is frequently defined as absence of progression at the time of the first formal disease evaluation (2 to 3 months after starting therapy), whereas progression at this time point is considered as initial or primary resistance. In patients who have experienced an initial stabilization, further progressions (or relapses) are considered late or secondary resistance. These two distinct mechanisms of drug resistance are reflected in progression-free survival curves by a rapid drop off at the time of first evaluation (initial resistance), followed by a slower continued decrease with a small hazard rate (late resistance). Analysis of genomic and biologic profiles has suggested that heterogeneous biologic mechanisms may be responsible for drug resistance; some of the mechanisms may already be present and active at baseline, and others may be activated later or result from acquisition of new mutations.9,10
The current article reports on an analysis of the clinical and biologic factors affecting initial and late resistance to imatinib, based on the data of the randomized trial jointly conducted by the EORTC, the Italian Sarcoma Group, and the Australasian Gastrointestinal Trials Group. This article also explores whether the recently reported advantage of high-dose imatinib5 is homogeneous among prognostic subgroups.
PATIENTS AND METHODS
Eligibility Criteria
Patients with histologically proven advanced and/or metastatic unresectable GIST characterized by c-KIT expression as assessed by polyclonal CD117 antibodies (Dako Cytomation, Glostrup, Denmark) were eligible for this trial. Any prior chemotherapy was accepted if discontinued for more than 4 weeks. Patients with measurable or nonmeasurable disease that was documented by conventional scan imaging or physical examination were eligible. Other eligibility criteria are described elsewhere.5 Each participating institution obtained the approval of the competent ethical review board, and all patients gave written informed consent.
Prestudy and Follow-Up Investigations
Within 14 days before treatment, a physical examination was performed, CBC count and serum chemistry were assessed, and relevant computed tomography scans were performed for tumor assessment. Computed tomography scans were repeated after 2, 4, and 6 months and every 3 months thereafter until progression of disease. The RECIST8 method was used for evaluation of response and for documentation of progression.
After completion of recruitment, paraffin-embedded tumor blocks were collected for a central pathology review and obtained for approximately half of the patients. Results of this review will be analyzed in the subgroup of patients for whom material is available and published separately.
Treatment and Dose Modifications
Patients were randomly assigned to receive either 400 mg administered orally once daily or 400 mg bid. All patients were scheduled to continue treatment until disease progression or unacceptable toxicity. Dose modifications requested in case of toxicity are described in detail elsewhere.5 In case of disease progression in a patient randomly assigned to the 400 mg once daily dose, a cross over to the 400 mg bid dose was allowed, regardless of the dose the patient was taking at the moment of progression.
Statistical Analysis
Initial resistance was defined as objective disease progression (according to RECIST8) within 3 months of randomization. The cutoff point was selected to include progressions documented at the first disease evaluation (after 2 months) but exclude progressions documented at the second disease evaluation (after 4 months). This end point was analyzed as a binary variable. Patients who either died in the absence of progression or who were lost to follow-up within 3 months of randomization were excluded from the analysis.
Late resistance was analyzed as a time to event variable (time to objective progression) with a 3-month landmark period. Patients who experienced progression, died, or were lost to follow-up within 3 months of random assignment were excluded from this analysis. Patients who died in the absence of progression after 3 months were censored at the date of death.
Cofactors investigated in the analysis included the initial daily dose of imatinib (randomized), age, sex, performance status at trial inclusion, primary site of disease (abdominal, stomach, small bowel, other GI, or other site), time since first GIST diagnosis, prior treatments for GIST (surgery, radiotherapy, and chemotherapy), site (primary tumor, liver metastases, or lung metastases) and size of lesions (diameter of the largest lesion) at the time of trial inclusion, and baseline hematologic and biologic parameters (WBCs, granulocytes, platelets, hemoglobin, creatinine, bilirubin, and albumin). Continuous variables were not recoded for building the prognostic models, but prognostic variables had to be recoded for drawing time to progression curves; in such cases, values close to quartiles were chosen as category cutoff points.
Both univariate and multivariate analyses used logistic regression (initial resistance) and Cox regression (late resistance) models. Factors found to be significant in the univariate analysis at the P = .05 level were subsequently included in a step-down multivariate model. Correlation between cofactors was measured by the Spearman rank correlation coefficient.
Integrating results of the pathology review in this analysis would not have been possible without losing substantial power, but the analyses have been repeated on the subgroup of patients for whom the GIST histology had been externally confirmed as sensitivity analyses. The impact of significant prognostic factors is detailed in overall time to progression curves, which were estimated by the Kaplan-Meier method. All randomly assigned patients are included in those curves. The prognostic value of the randomly allocated initial imatinib dose has been subsequently explored in prognostic subgroups using the Wald test adjusted for repetitive testing.
RESULTS
A total of 946 patients were included in the trial. At the time of this analysis, the median follow-up was 25 months (1- and 2-year follow-up rates, 98% and 58%, respectively). Comparisons of efficacy and toxicity parameters between therapeutic arms have been published elsewhere.5
Demographic Data
Cofactors and their distribution are listed in Table 1. There was no major imbalance between the randomized arms.
Initial Resistance
Among the 946 randomly assigned patients, 11 died within 3 months without evidence of progression (six patients from the 400 mg once daily arm and five from the 400 mg bid arm) because of toxicity (n = 2), infection (n = 3), hemorrhage (n = 3), severe diarrhea and vomiting (n = 1), and cardiac disease (n = 2). One ineligible patient (non-GIST, 400 mg once daily arm) was lost to follow-up. These 12 patients were excluded from the analysis. Among the 934 remaining patients, 116 (12%) experienced progression within 3 months (initially resistant).
The following prognostic factors of initial resistance were identified by univariate analysis (in order of significance): presence of lung metastases, low baseline hemoglobin level, high baseline granulocyte and platelet count, poor performance status, low baseline albumin level, absence of liver metastases, and short interval since the initial diagnosis of the disease (Table 2). None of the other cofactors showed any significant correlation with initial resistance.
In the multivariate model, presence of lung metastases, low baseline hemoglobin level, and absence of liver metastases were highly significant adverse prognostic factors, and high baseline granulocyte count showed borderline significance (Table 2). Highly significant (P < .005) correlation coefficients (Spearman) were observed between baseline hemoglobin level and albumin level (r = 0.51), performance status (r = –0.32), platelet count (r = –0.26), granulocyte count (r = –0.092), and time since GIST diagnosis (r = 0.097), and between time since GIST diagnosis and liver metastases (r = 0.24). The logistic regression model was applied to 456 assessable patients with an independently confirmed GIST diagnosis. In this sensitivity analysis, only baseline hemoglobin level and granulocyte count retained a significant prognostic value.
Late Resistance
The late resistance analysis is based on the 818 patients who where progression free and alive at 3 months (404 patients from the 400 mg once daily arm and 414 patients from the 400 mg bid arm). A total of 347 progressions were subsequently recorded. Patients who died without evidence of progression because of toxicity (n = 3), drug- and disease-unrelated events (n = 14), and unknown causes (n = 7) were censored at the date of death (10 patients from the 400 mg once daily arm and 14 patients from the 400 mg bid arm).
The following prognostic factors of late resistance were identified in the univariate analysis (in order of significance): high baseline granulocyte count, tumor size (largest diameter of the largest lesion), high baseline WBC count, poor performance status, nongastric primary tumor, small bowel primary tumor, low baseline albumin, prior chemotherapy, and random assignment to 400 mg once daily (Table 3). None of the other cofactors showed any significant correlation with late resistance.
In the multivariate analysis, only four factors remained as significant independent factors of adverse prognosis: tumor size, high baseline granulocyte count, nongastric primary tumor, and random assignment to imatinib 400 mg once daily (Table 3). Highly significant (P < .0001) correlation coefficients (Spearman) were observed between baseline granulocyte count and WBC count (r = 0.90), tumor size (r = 0.33), performance status (r = 0.26), and albumin level (r = –0.24), and between tumor size and albumin level (r = –0.34) and performance status (r = 0.32). The final Cox regression model was also applied to the subgroup of 421 patients assessable for late resistance and with independent confirmation of the GIST diagnosis. The results of this sensitivity analysis were similar to the results observed for the whole cohort (with lower significance levels).
Impact of the Most Significant Cofactors
Liver and lung lesions. Initial resistance was documented in 96 (11%) of 857 patients without lung lesions, 10 (20%) of 50 patients with both lung and liver lesions, and 11 (41%) of 27 patients with lung but no liver lesions; in patients with externally confirmed GIST diagnosis, these progression rates were 8% (34 of 426 patients), 7% (two of 27 patients), and 0% (zero of 11 patients), respectively.
Baseline hemoglobin level. Figure 1 illustrates the increased initial resistance to imatinib in patients with a low baseline hemoglobin level (< 7 mmol/L or 11.27 mg/100 mL). After the first 3 months, the curves remained parallel, which is reflected in the lack of prognostic significance of this factor for late resistance.
Baseline granulocyte count. As shown in Figure 2, baseline granulocyte count slightly affected the initial drug resistance but largely affected the late resistance, which is substantially increased in patients with a baseline count greater than 5 x 109/L.
Tumor size. Figure 3 illustrates the impact of tumor size on late resistance, which was mainly observed after 1 year of imatinib therapy and with an increasing failure rate in patients with large lesions (> 12 cm).
Site of primary disease. Figure 4 shows the time to progression for tumors according to the site of primary disease. Patients with a disease of gastric origin have a better prognosis than patients with disease originating in the small bowel. In other subgroups, the limited sample size does not allow any formal comparison.
Competing risk. The aim of this study was to identify factors that could predict resistance to imatinib, and therefore, progression and time to progression have been chosen as primary end points. The competing risk of death in the absence of progression was ignored in the principal analyses (those patients were censored). However, in the same data set, a cumulative incidence analysis demonstrated a limited contribution (< 10%) of intercurrent deaths on progression-free survival.5 We also performed a sensitivity analysis, considering all deaths as events, and obtained similar results.
Subgroup analysis. The impact of the randomly allocated initial dose on time to progression was evaluated in the following subgroups: patients with a baseline granulocyte count greater or less than 5 x 109/L; patients with tumors smaller or larger than 12 cm; and patients with tumors of stomach origin, small bowel origin, or other GI origin (the number of events was too small in other subgroups). Table 4 lists the estimates of the hazard ratios, their 95% CIs, and the results of the Wald test for all subgroups.
The advantage of the high initial dose of imatinib (in terms of time to progression) was statistically significant in the following two subgroups (overall P < .05, nominal P < .007): patients with a high (> 5 x 109/L) baseline granulocyte count (Fig 5) and patients with tumors of GI origin outside of stomach or small bowel (Fig 6). However, time to progression was not affected by the initial dose in patients with tumors of small bowel origin.
DISCUSSION
This prognostic factor analysis is based on the largest available series of patients with advanced or metastatic GIST who were consistently treated with imatinib, observed, and documented. The large sample size provides the appropriate power to identify with high confidence those factors that have an independent prognostic value.
Information on all cofactors investigated in this study is usually available for individual patients in any clinical practice. Therefore, our results do provide immediate prognostic information for any patient diagnosed with advanced GIST and treated with imatinib. The results may guide decisions on individual treatment with imatinib and help to identify patients who require an initial high dose or who may not benefit from imatinib and for whom a different treatment approach may be considered.
The obtained prognostic models have not been validated. We could have built the prognostic models on a randomly selected subset of the data (a training sample) and validated the results on the remaining data (validation sample), but this would have reduced the power of the analyses. Because other large similar data series will become available, we elected to build the model on the whole study cohort, assuming that external validation will be carried out by independent groups and will provide more reliable results than internal validation.
We have demonstrated that initial and late resistance to imatinib are predicted by different clinical and biologic factors. This is analogous to the existence of different competing mechanisms of resistance as identified on the basis of the analysis of genomic and molecular profiles.9,10 Analysis of KIT and/or PDGF mutations will probably provide additional prognostic information, as already suggested.11 However, nonbiomolecular mechanisms may also play a different role in the initial and late resistance setting.
In the whole cohort of patients, initial resistance was predicted by the following four independent factors: baseline hemoglobin level, baseline granulocyte count, presence of lung metastases, and absence of liver metastases. The two last factors probably characterize a small proportion of misdiagnosed non-GIST patients. Sarcomas other than GIST have been proven to be unresponsive to imatinib.4 Inclusion of a small proportion of non-GIST patients can largely affect the prognostic model for initial resistance because it is based on a 12% progression rate. This hypothesis is reinforced by the fact that those factors lose their significance when the analysis is restricted to the subgroup of patients with an external confirmation of the GIST diagnosis. Disease presentation with lung and/or without liver metastases should be an indication for external review of the pathologic diagnosis, but these factors probably do not affect resistance to imatinib in true GIST patients.
Hemoglobin level has also been reported to be a prognostic factor in patients treated with imatinib for chronic myeloid leukemia.12,13 In a previous EORTC study, low hemoglobin has been found to be correlated with pharmacokinetic parameters including small distribution volume, short half-life, low clearance (in L/h), and high area under the curve,14,15 and a first hypothesis is that hemoglobin level could affect the drug transport and delivery, resulting in insufficient intratumoral drug levels to inhibit disease proliferation in some patients with low hemoglobin. Indeed, small distribution volume (ie, high concentration) associated with short half-life may suggest that the drug remains in the blood instead of being distributed to organs (and to the tumor), which is in contrast to a high concentration associated with prolonged half-life that results from drug accumulation in the whole body (blood and tumor).
The role of hemoglobin in drug transport is further suggested by the fact that significant amounts of imatinib could be quantified in the erythrocyte sediments of patients treated with the drug.16 Erythrocyte loading was dose dependent in both volunteer and patient blood, and partition ratios of erythrocytes versus plasma ranged between 0.01 and 0.5. However, when patients have been treated with imatinib over prolonged periods of time, partition ratios can increase beyond 3 (Prenen et al, personal communication, April 2004). Various anticancer agents are capable of inducing changes in their own partition ratios17; the implications for imatinib are subject of further investigations.
Another hypothesis is that low hemoglobin reflects a more aggressive type of disease or a more advanced stage that is associated with mucosal ulceration and tumor bleeding and is less responsive to imatinib. Any other hypotheses suggesting that one of the molecular mechanisms of drug resistance either affects or is affected by hemoglobin levels could also explain this finding.
The same kind of hypotheses can be made for granulocyte counts, which apparently affect both initial and late drug resistance. A high granulocyte count may reflect an inflammatory reaction induced by an aggressive type of tumor that tends to progress earlier. Molecular classification of patients with low hemoglobin levels and high granulocyte counts may help in understanding those findings.
Late resistance to imatinib is independently predicted by the size of the lesions present at treatment start. Tumor size has also been reported as a prognostic factor for primary disease by several groups and, therefore, has been included in the definition of risk groups for GIST established by a consensus meeting.18 In our study, tumor size did not affect initial resistance to the drug (as demonstrated both by the univariate and the multivariate analyses), suggesting that patients with an advanced stage of disease are initially responsive to imatinib (as opposed to what is observed with conventional chemotherapy in solid tumors). However, tumor size has a significant impact on late resistance, which seems to increase with time, suggesting correlation with a delayed mechanism of drug resistance. Patients with large lesions are probably more exposed to the risk of secondary mutations because of a higher likelihood of emergence of new clones, which explains the increasing risk of late resistance.
According to our data, tumors of gastric origin tend to progress later than tumors of small bowel origin. This has also been reported for primary disease, and it has been suggested that the proportion of benign (as opposed to malignant) GISTs is higher in the stomach than in the small bowel19 and that the mitotic index is highly site dependent.20 Our model demonstrates that, in advanced disease, site of origin adds independent prognostic information to baseline granulocyte count and tumor size, but it still needs to be investigated whether this prognostic information is also independent of mitotic index.
Despite the inclusion of 20% of patients older than the age of 70 years, age did not show any prognostic value for either initial or late resistance to imatinib, even in the univariate analysis. The same observation has been reported for chronic myeloid leukemia, with the conclusion that the adverse prognosis of elderly patients is generally related to the treatment rather than the intrinsic biology of the disease.21 This statement is probably also true for GIST.
Finally, the subgroup analysis suggests that delayed resistance to imatinib in patients allocated to the high-dose regimen is mainly observed in patients with a high baseline granulocyte count (> 5 x 109/L) and in patients with a GI tumor origin outside of the stomach and the small bowel; our data do not suggest such an advantage in patients with tumors of small bowel origin. However, cross over from 400 mg once daily to 400 mg bid after progression has been shown to induce further disease stabilization in some patients,22 and only survival data will be able to demonstrate whether increasing the initial dose of imatinib results in a clinical benefit for all patients or subgroups of patients with advanced GIST.
Our results may be used to identify patients for whom initial and long-term treatment should be improved. In particular, imatinib resistance can be delayed by increasing the initial dose in patients with high granulocyte counts and in patients with tumors of GI origin outside of the stomach and small bowel. These results may also be helpful in the investigation or confirmation of the different biologic mechanisms of drug resistance. Correlation of identified prognostic factors with immunohistochemical and molecular parameters could improve the knowledge of this disease.
Appendix
The following investigators also contributed to this study: Dr J. Zalcberg, Peter Maccalum Cancer Institute, Melbourne, Australia; Dr S. Leyvraz, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland; Dr B. Bui, Institut Bergonie, Bordeaux, France; Dr P. Sch?ffksi, Medizinische Hochschule, Hannover, Germany; Dr A. Lopez Pousa, Hospital De La Santa Creu I Sant Pau, Barcelona, Spain; Dr D. Kotasek, Ashford Cancer Centre, Ashford, Australia; Dr T. De Pas, Istituto Europeo di Oncologia, Milan, Italy; Dr S. Rodenhuis, the Netherlands Cancer Institute, Amsterdam, the Netherlands; Dr W. Ruka, Maria Sklodowska-Curie Memorial Cancer Centre, Warsaw, Poland; Dr G. Grignani, Institute For Cancer Research and Treatment, Torino, Italy; Dr F. Duffaud, Centre Hospitalier Universitaire de la Timone, Marseilles, France; Dr J. Radford, Christie Hospital, Manchester, United Kingdom; Dr M. Findlay, Wellington Hospital, Wellington, New Zealand; Dr. C. Chevreau, Centre Claudius Regaud, Toulouse, France; Dr J. Whelan, Middlesex Hospital, London (Middlesex), United Kingdom; Dr L. Paz Arez, Hospital Universitario 12 De Octubre, Madrid, Spain; Dr M. Leahy, St James's University Hospital, Leeds, United Kingdom; Dr D. Hossfeld, Universitaets-Krankenhaus Eppendorf, Hamburg, Germany; Dr S. Frustaci, Centro Di Riferimento Oncologico, Aviano, Italy; Dr N. Deligny, Centre Oscar Lambret, Lille, France; Dr A. Krarup-Hansen, Herlev Hospital, Herlev, Denmark; Dr G. Apice, Istituto Nazionale Per Lo Studio E La Cura Dei Tumori, Napoli, Italy; Dr F. Cowie, Western Infirmary, Glasgow, United Kingdom; Dr K. Siong, National Cancer Center, Singapore; Dr G. Van Hazel, Sir Charles Gairdner Hospital, Perth, Australia; Dr W. van der Graaf, Academisch Ziekenhuis Groningen, Groningen, the Netherlands; Dr P. Lorigan, Westin Park Hospital, Sheffield, United Kingdom; Dr D. Grimes, Wesley Clinic for Hematology and Oncology, Brisbane, Australia; Dr M. Links, St George Hospital, Sydney, Australia; Dr A. Comandone, Ospedale Gradenigo, Torino, Italy; Dr H. Gelderblom, Leiden University Medical Center, Leiden, the Netherlands; Dr S. Clarke, Royal Prince Alfred Hospital, Sydney, Australia; Dr D. Wyld, Royal Brisbane Hospital, Brisbane, Australia; Dr J. Vermorken, Universitair Ziekenhuis Antwerp, Antwerp, Belgium; Dr O. Nielsen, Aarhus University Hospital, Aarhus, Denmark; Dr F. Kirsten, Bankstown-Lidcombe Hospital, Sydney, Australia; Dr J. Buesa, Hospital General de Asturias, Oviedo, Spain; Dr A. Poveda, Instituto Valenciano de Oncologia, Valencia, Spain; Dr N. Wilcken, Westmead Hospital, Sydney, Australia; Dr M. Green, Royal Melbourne Hospital, Melbourne, Australia; Dr R. McLennan, The Geelong Hospital, Geelong, Australia; Dr D. Ransom, Royal Perth Hospital, Perth, Australia; Dr C. Karapetis, Flinders Medical Centre, Adelaide, Australia; Dr I. Byard, Launceston General Hospital, Launceston, Australia; Dr P. Woll, Nottingham City Hospital, Nottingham, United Kingdom; Dr D. Bell, Royal North Shore Hospital, Sydney, Australia; and Dr E. Walpole, Princess Alexandra Hospital, Brisbane, Australia.
Authors' Disclosures of Potential Conflicts of Interest
Acknowledgment
We thank H. Dhillon (National Health and Medical Research Council Clinical Trials Centre, Sydney, Australia) and L. Mariani (Milan, Italy) for data management.
NOTES
Supported by an unrestricted grant from Novartis Oncology and by grant Nos. 2U10 CA11488-29 through 5U10 CA11488-34 from the National Cancer Institute (Bethesda, MD).
Presented in part at the 10th Annual Meeting of the Connective Tissue Oncology Society, Montreal, Quebec, Canada, November 11-13, 2004.
The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Joensuu H, Fletcher C, Dimitrijevic S, et al: Management of malignant gastrointestinal stromal tumours. Lancet Oncol 3:655-664, 2002
van Oosterom AT, Judson I, Verweij J, et al: Safety and efficacy of imatinib (STI571) in metastatic gastrointestinal stromal tumours: A phase I study. Lancet 358:1421-1423, 2001
Demetri GD, von Mehren M, Blanke CD, et al: Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N Engl J Med 347:472-480, 2002
Verweij J, van Oosterom A, Blay JY, et al: Imatinib mesylate (STI-571 Glivec, Gleevec) is an active agent for gastrointestinal stromal tumours, but does not yield responses in other soft-tissue sarcomas that are unselected for a molecular target: Results from an EORTC Soft Tissue and Bone Sarcoma Group phase II study. Eur J Cancer 39:2006-2011, 2003
Verweij J, Casali PG, Zalcberg J, et al: Progression-free survival in gastrointestinal stromal tumours with high-dose imatinib: Randomized trial. Lancet 364:1127-1134, 2004
Rankin C, Von Mehren M, Blanke C, et al: Dose effect of imatinib in patients with metastatic GIST: Phase III sarcoma group study S0033. J Clin Oncol 22:819s, 2004 (abstr 9005, suppl 14)
Blay J-Y, Berthaud P, Perol D, et al: Continuous vs intermittent imatinib treatment in advanced GIST after one year: A prospective randomized phase III trial of the French Sarcoma Group. J Clin Oncol 22:819s, 2004 (abstr 9006, suppl 14)
Therasse P, Arbuck SG, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumors: European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92:205-216, 2000
Fletcher JA, Corless CL, Dimitrijevic S, et al: Mechanisms of resistance to imatinib mesylate (IM) in advanced gastrointestinal stromal tumor (GIST). Proc Am Soc Clin Oncol 22:815, 2003 (abstr 3275)
Debiec-Rychter M, Cools J, Dumez H, et al: Mechanisms of resistance to imatinib mesylate in gastrointestinal stromal tumors and activity of the PKC412 inhibitor against imatinib-resistant mutants. Gastroenterology 128:270-279, 2005
Corless CL, Fletcher AJ, Heinrich MC: Biology of gastrointestinal stromal tumors. J Clin Oncol 22:3813-3825, 2004
Kantarjian H, O'Brien S, Cortes J, et al: Survival advantage with imatinib mesylate therapy in chronic-phase chronic myelogenous leukemia (CML-CP) after IFN-alpha failure and in late CML-CP, comparison with historical controls. Clin Cancer Res 10:68-75, 2004
O'Dwyer ME, Mauro MJ, Blasdel C, et al: Clonal evolution and lack of cytogenetic response are adverse prognostic factors for hematologic relapse of chronic phase CML patients treated with imatinib mesylate. Blood 103:451-455, 2004
Judson I, Donato di Paola E, Verweij J, et al: Population pharmacokinetic (PK) analysis and PK-pharmacodynamic (PD) correlations in phase I/II trial of imatinib in gastrointestinal stromal tumours (GIST) conducted by the European Organisation for Research and Treatment of Cancer Soft Tissue and Bone Sarcoma Group. Proc Am Soc Clin Oncol 22:818, 2003 (abstr 3287)
Judson I, Ma P, Peng B, et al: Imatinib pharmacokinetics in patients with gastrointestinal stromal tumour, a retrospective population pharmacokinetic study over time: EORTC Soft Tissue and Bone Sarcoma Group. Cancer Chemother Pharmacol 55:379-386, 2005
Guetens G, De Boeck G, Highley M, et al: Quantification of the anticancer agent STI-571 in erythrocytes and plasma by measurement of sediment technology and liquid chromatography-tandem mass spectrometry. J Chromatogr A 1020:27-34, 2003
Dumez H, Reinhart WH, Guetens G, et al: Human red blood cells: Rheological aspects, uptake, and release of cytotoxic drugs. Crit Rev Clin Lab Sci 41:159-188, 2004
Fletcher C, Berman JJ, Corless C, et al: Diagnosis of gastrointestinal stromal tumors: A consensus approach. Hum Pathol 33:459-465, 2002
Miettinen M, El-Rifai W, Sobin L, et al: Evaluation of malignancy and prognosis of gastrointestinal stromal tumors: A review. Hum Pathol 33:478-483, 2002
Emory TS, Sobin LH, Lukes L, et al: Prognosis of gastrointestinal smooth-muscle (stromal) tumors: Dependence on anatomic site. Am J Surg Pathol 23:82-87, 1999
Cortes J, Talpaz M, O'Brien S, et al: Effects of age on prognosis with imatinib mesylate therapy for patients with Philadelphia chromosome-positive chronic myelogenous leukemia. Cancer 98:1105-1113, 2003
Zalcberg JR, Verweij J, Casali PG, et al: Outcome of patients with advanced gastro-intestinal stromal tumors (GIST) crossing over to a daily imatinib dose of 800mg after progression on 400mg. Eur J Cancer (in press)(Martine Van Glabbeke, Jaa)
University Hospital Gasthuisberg, Leuven, Belgium
Erasmus University Medical Center, Rotterdam
Leiden University Medical Center, Leiden, the Netherlands
Istituto Tumori, Milano, Italy
Institut Gustave Roussy, Villejuif
Centre Leon Berard, Lyon, France
Charite Campus Buch, Robert Roessle Hospital, Berlin
Klinikum Grosshadern
Gesellschaft für Strahlenforschung–National Research Center for Environmental and Health, Munich, Germany
Prince of Wales Hospital, Randwick, Australia
Royal Marsden Hospital, London, United Kingdom
ABSTRACT
PURPOSE: The aim of this study was to identify factors predicting initial and late resistance of GI stromal tumor (GIST) patients to imatinib and to document the dose-response relationship in the prognostic subgroups. This study is based on the European Organisation for Research and Treatment of Cancer–Italian Sarcoma Group–Australasian Gastrointestinal Trials Group randomized trial comparing two doses of imatinib in advanced disease.
PATIENTS AND METHODS: Initial resistance was defined as progression within 3 months of randomization, and late resistance was defined as progression beyond 3 months. Investigated cofactors include imatinib dose, age, sex, performance status, original disease site, site and size of lesions at trial entry, and baseline hematologic and biologic parameters.
RESULTS: Initial resistance was recorded for 116 (12%) of 934 assessable patients and was independently predicted by the presence of lung and absence of liver metastases, low hemoglobin level, and high granulocyte count. Among 818 patients who were alive and progression free at 3 months, 347 subsequent progressions were recorded, and late resistance was independently predicted by high baseline granulocyte count, primary tumor outside of the stomach, large tumor size, and low initial imatinib dose. The impact of initial dose on late resistance was mainly significant in patients with a high baseline granulocyte count (> 5.109/L) and in patients with tumors of GI origin outside of the stomach and small intestine.
CONCLUSION: Our study identifies patients for whom initial and/or long-term treatment needs to be improved and patients who require a high initial dose. Correlation of these results with immunohistochemistry and molecular parameters may further help to understand the biologic mechanisms of resistance.
INTRODUCTION
Soft tissue sarcomas represent 1% of adult malignancies and are a heterogeneous group of neoplasms whose only common denominator is their derivation from mesenchymal tissue. GI stromal tumors (GIST) are a subset of soft tissue sarcomas that were classified relatively recently. Their local treatment essentially consists of surgery. After the stage of resection, these tumors have proven to be insensitive to chemotherapy and radiotherapy.1
Imatinib is a small-molecule tyrosine kinase inhibitor that is active against BCR-ABL, KIT, and PDGFR. KIT is expressed in the vast majority of GISTs and is frequently mutated, leading to constitutive activation in these tumors. A European Organisation for Research and Treatment of Cancer (EORTC) phase I study2 identified the highest feasible dose of imatinib to be 400 mg bid and indicated extensive activity of imatinib in GIST. Phase II studies showed activity at all doses tested (400 to 800 mg).3,4 Two large, randomized, phase III studies comparing doses of 400 mg once a day to 400 mg bid have confirmed the activity of imatinib in terms of progression-free survival and overall survival.5,6 One of these studies has also documented a small but significant benefit with the high-dose regimen (400 mg bid) in terms of progression-free survival.5 Finally, a randomized trial from the French Sarcoma Group has demonstrated that imatinib therapy should be continued indefinitely, even after complete response.7
Response of GIST to imatinib does not always result in an immediate decrease of the size of the lesions but, rather, in an initial inhibition of growth. Objective response (according to Response Evaluation Criteria in Solid Tumors [RECIST] criteria8) has been reported in approximately half of the patients, but time to onset varies largely among patients, and some responses have been first documented more than 1 year after start of therapy.5 Therefore, response to imatinib is frequently defined as absence of progression at the time of the first formal disease evaluation (2 to 3 months after starting therapy), whereas progression at this time point is considered as initial or primary resistance. In patients who have experienced an initial stabilization, further progressions (or relapses) are considered late or secondary resistance. These two distinct mechanisms of drug resistance are reflected in progression-free survival curves by a rapid drop off at the time of first evaluation (initial resistance), followed by a slower continued decrease with a small hazard rate (late resistance). Analysis of genomic and biologic profiles has suggested that heterogeneous biologic mechanisms may be responsible for drug resistance; some of the mechanisms may already be present and active at baseline, and others may be activated later or result from acquisition of new mutations.9,10
The current article reports on an analysis of the clinical and biologic factors affecting initial and late resistance to imatinib, based on the data of the randomized trial jointly conducted by the EORTC, the Italian Sarcoma Group, and the Australasian Gastrointestinal Trials Group. This article also explores whether the recently reported advantage of high-dose imatinib5 is homogeneous among prognostic subgroups.
PATIENTS AND METHODS
Eligibility Criteria
Patients with histologically proven advanced and/or metastatic unresectable GIST characterized by c-KIT expression as assessed by polyclonal CD117 antibodies (Dako Cytomation, Glostrup, Denmark) were eligible for this trial. Any prior chemotherapy was accepted if discontinued for more than 4 weeks. Patients with measurable or nonmeasurable disease that was documented by conventional scan imaging or physical examination were eligible. Other eligibility criteria are described elsewhere.5 Each participating institution obtained the approval of the competent ethical review board, and all patients gave written informed consent.
Prestudy and Follow-Up Investigations
Within 14 days before treatment, a physical examination was performed, CBC count and serum chemistry were assessed, and relevant computed tomography scans were performed for tumor assessment. Computed tomography scans were repeated after 2, 4, and 6 months and every 3 months thereafter until progression of disease. The RECIST8 method was used for evaluation of response and for documentation of progression.
After completion of recruitment, paraffin-embedded tumor blocks were collected for a central pathology review and obtained for approximately half of the patients. Results of this review will be analyzed in the subgroup of patients for whom material is available and published separately.
Treatment and Dose Modifications
Patients were randomly assigned to receive either 400 mg administered orally once daily or 400 mg bid. All patients were scheduled to continue treatment until disease progression or unacceptable toxicity. Dose modifications requested in case of toxicity are described in detail elsewhere.5 In case of disease progression in a patient randomly assigned to the 400 mg once daily dose, a cross over to the 400 mg bid dose was allowed, regardless of the dose the patient was taking at the moment of progression.
Statistical Analysis
Initial resistance was defined as objective disease progression (according to RECIST8) within 3 months of randomization. The cutoff point was selected to include progressions documented at the first disease evaluation (after 2 months) but exclude progressions documented at the second disease evaluation (after 4 months). This end point was analyzed as a binary variable. Patients who either died in the absence of progression or who were lost to follow-up within 3 months of randomization were excluded from the analysis.
Late resistance was analyzed as a time to event variable (time to objective progression) with a 3-month landmark period. Patients who experienced progression, died, or were lost to follow-up within 3 months of random assignment were excluded from this analysis. Patients who died in the absence of progression after 3 months were censored at the date of death.
Cofactors investigated in the analysis included the initial daily dose of imatinib (randomized), age, sex, performance status at trial inclusion, primary site of disease (abdominal, stomach, small bowel, other GI, or other site), time since first GIST diagnosis, prior treatments for GIST (surgery, radiotherapy, and chemotherapy), site (primary tumor, liver metastases, or lung metastases) and size of lesions (diameter of the largest lesion) at the time of trial inclusion, and baseline hematologic and biologic parameters (WBCs, granulocytes, platelets, hemoglobin, creatinine, bilirubin, and albumin). Continuous variables were not recoded for building the prognostic models, but prognostic variables had to be recoded for drawing time to progression curves; in such cases, values close to quartiles were chosen as category cutoff points.
Both univariate and multivariate analyses used logistic regression (initial resistance) and Cox regression (late resistance) models. Factors found to be significant in the univariate analysis at the P = .05 level were subsequently included in a step-down multivariate model. Correlation between cofactors was measured by the Spearman rank correlation coefficient.
Integrating results of the pathology review in this analysis would not have been possible without losing substantial power, but the analyses have been repeated on the subgroup of patients for whom the GIST histology had been externally confirmed as sensitivity analyses. The impact of significant prognostic factors is detailed in overall time to progression curves, which were estimated by the Kaplan-Meier method. All randomly assigned patients are included in those curves. The prognostic value of the randomly allocated initial imatinib dose has been subsequently explored in prognostic subgroups using the Wald test adjusted for repetitive testing.
RESULTS
A total of 946 patients were included in the trial. At the time of this analysis, the median follow-up was 25 months (1- and 2-year follow-up rates, 98% and 58%, respectively). Comparisons of efficacy and toxicity parameters between therapeutic arms have been published elsewhere.5
Demographic Data
Cofactors and their distribution are listed in Table 1. There was no major imbalance between the randomized arms.
Initial Resistance
Among the 946 randomly assigned patients, 11 died within 3 months without evidence of progression (six patients from the 400 mg once daily arm and five from the 400 mg bid arm) because of toxicity (n = 2), infection (n = 3), hemorrhage (n = 3), severe diarrhea and vomiting (n = 1), and cardiac disease (n = 2). One ineligible patient (non-GIST, 400 mg once daily arm) was lost to follow-up. These 12 patients were excluded from the analysis. Among the 934 remaining patients, 116 (12%) experienced progression within 3 months (initially resistant).
The following prognostic factors of initial resistance were identified by univariate analysis (in order of significance): presence of lung metastases, low baseline hemoglobin level, high baseline granulocyte and platelet count, poor performance status, low baseline albumin level, absence of liver metastases, and short interval since the initial diagnosis of the disease (Table 2). None of the other cofactors showed any significant correlation with initial resistance.
In the multivariate model, presence of lung metastases, low baseline hemoglobin level, and absence of liver metastases were highly significant adverse prognostic factors, and high baseline granulocyte count showed borderline significance (Table 2). Highly significant (P < .005) correlation coefficients (Spearman) were observed between baseline hemoglobin level and albumin level (r = 0.51), performance status (r = –0.32), platelet count (r = –0.26), granulocyte count (r = –0.092), and time since GIST diagnosis (r = 0.097), and between time since GIST diagnosis and liver metastases (r = 0.24). The logistic regression model was applied to 456 assessable patients with an independently confirmed GIST diagnosis. In this sensitivity analysis, only baseline hemoglobin level and granulocyte count retained a significant prognostic value.
Late Resistance
The late resistance analysis is based on the 818 patients who where progression free and alive at 3 months (404 patients from the 400 mg once daily arm and 414 patients from the 400 mg bid arm). A total of 347 progressions were subsequently recorded. Patients who died without evidence of progression because of toxicity (n = 3), drug- and disease-unrelated events (n = 14), and unknown causes (n = 7) were censored at the date of death (10 patients from the 400 mg once daily arm and 14 patients from the 400 mg bid arm).
The following prognostic factors of late resistance were identified in the univariate analysis (in order of significance): high baseline granulocyte count, tumor size (largest diameter of the largest lesion), high baseline WBC count, poor performance status, nongastric primary tumor, small bowel primary tumor, low baseline albumin, prior chemotherapy, and random assignment to 400 mg once daily (Table 3). None of the other cofactors showed any significant correlation with late resistance.
In the multivariate analysis, only four factors remained as significant independent factors of adverse prognosis: tumor size, high baseline granulocyte count, nongastric primary tumor, and random assignment to imatinib 400 mg once daily (Table 3). Highly significant (P < .0001) correlation coefficients (Spearman) were observed between baseline granulocyte count and WBC count (r = 0.90), tumor size (r = 0.33), performance status (r = 0.26), and albumin level (r = –0.24), and between tumor size and albumin level (r = –0.34) and performance status (r = 0.32). The final Cox regression model was also applied to the subgroup of 421 patients assessable for late resistance and with independent confirmation of the GIST diagnosis. The results of this sensitivity analysis were similar to the results observed for the whole cohort (with lower significance levels).
Impact of the Most Significant Cofactors
Liver and lung lesions. Initial resistance was documented in 96 (11%) of 857 patients without lung lesions, 10 (20%) of 50 patients with both lung and liver lesions, and 11 (41%) of 27 patients with lung but no liver lesions; in patients with externally confirmed GIST diagnosis, these progression rates were 8% (34 of 426 patients), 7% (two of 27 patients), and 0% (zero of 11 patients), respectively.
Baseline hemoglobin level. Figure 1 illustrates the increased initial resistance to imatinib in patients with a low baseline hemoglobin level (< 7 mmol/L or 11.27 mg/100 mL). After the first 3 months, the curves remained parallel, which is reflected in the lack of prognostic significance of this factor for late resistance.
Baseline granulocyte count. As shown in Figure 2, baseline granulocyte count slightly affected the initial drug resistance but largely affected the late resistance, which is substantially increased in patients with a baseline count greater than 5 x 109/L.
Tumor size. Figure 3 illustrates the impact of tumor size on late resistance, which was mainly observed after 1 year of imatinib therapy and with an increasing failure rate in patients with large lesions (> 12 cm).
Site of primary disease. Figure 4 shows the time to progression for tumors according to the site of primary disease. Patients with a disease of gastric origin have a better prognosis than patients with disease originating in the small bowel. In other subgroups, the limited sample size does not allow any formal comparison.
Competing risk. The aim of this study was to identify factors that could predict resistance to imatinib, and therefore, progression and time to progression have been chosen as primary end points. The competing risk of death in the absence of progression was ignored in the principal analyses (those patients were censored). However, in the same data set, a cumulative incidence analysis demonstrated a limited contribution (< 10%) of intercurrent deaths on progression-free survival.5 We also performed a sensitivity analysis, considering all deaths as events, and obtained similar results.
Subgroup analysis. The impact of the randomly allocated initial dose on time to progression was evaluated in the following subgroups: patients with a baseline granulocyte count greater or less than 5 x 109/L; patients with tumors smaller or larger than 12 cm; and patients with tumors of stomach origin, small bowel origin, or other GI origin (the number of events was too small in other subgroups). Table 4 lists the estimates of the hazard ratios, their 95% CIs, and the results of the Wald test for all subgroups.
The advantage of the high initial dose of imatinib (in terms of time to progression) was statistically significant in the following two subgroups (overall P < .05, nominal P < .007): patients with a high (> 5 x 109/L) baseline granulocyte count (Fig 5) and patients with tumors of GI origin outside of stomach or small bowel (Fig 6). However, time to progression was not affected by the initial dose in patients with tumors of small bowel origin.
DISCUSSION
This prognostic factor analysis is based on the largest available series of patients with advanced or metastatic GIST who were consistently treated with imatinib, observed, and documented. The large sample size provides the appropriate power to identify with high confidence those factors that have an independent prognostic value.
Information on all cofactors investigated in this study is usually available for individual patients in any clinical practice. Therefore, our results do provide immediate prognostic information for any patient diagnosed with advanced GIST and treated with imatinib. The results may guide decisions on individual treatment with imatinib and help to identify patients who require an initial high dose or who may not benefit from imatinib and for whom a different treatment approach may be considered.
The obtained prognostic models have not been validated. We could have built the prognostic models on a randomly selected subset of the data (a training sample) and validated the results on the remaining data (validation sample), but this would have reduced the power of the analyses. Because other large similar data series will become available, we elected to build the model on the whole study cohort, assuming that external validation will be carried out by independent groups and will provide more reliable results than internal validation.
We have demonstrated that initial and late resistance to imatinib are predicted by different clinical and biologic factors. This is analogous to the existence of different competing mechanisms of resistance as identified on the basis of the analysis of genomic and molecular profiles.9,10 Analysis of KIT and/or PDGF mutations will probably provide additional prognostic information, as already suggested.11 However, nonbiomolecular mechanisms may also play a different role in the initial and late resistance setting.
In the whole cohort of patients, initial resistance was predicted by the following four independent factors: baseline hemoglobin level, baseline granulocyte count, presence of lung metastases, and absence of liver metastases. The two last factors probably characterize a small proportion of misdiagnosed non-GIST patients. Sarcomas other than GIST have been proven to be unresponsive to imatinib.4 Inclusion of a small proportion of non-GIST patients can largely affect the prognostic model for initial resistance because it is based on a 12% progression rate. This hypothesis is reinforced by the fact that those factors lose their significance when the analysis is restricted to the subgroup of patients with an external confirmation of the GIST diagnosis. Disease presentation with lung and/or without liver metastases should be an indication for external review of the pathologic diagnosis, but these factors probably do not affect resistance to imatinib in true GIST patients.
Hemoglobin level has also been reported to be a prognostic factor in patients treated with imatinib for chronic myeloid leukemia.12,13 In a previous EORTC study, low hemoglobin has been found to be correlated with pharmacokinetic parameters including small distribution volume, short half-life, low clearance (in L/h), and high area under the curve,14,15 and a first hypothesis is that hemoglobin level could affect the drug transport and delivery, resulting in insufficient intratumoral drug levels to inhibit disease proliferation in some patients with low hemoglobin. Indeed, small distribution volume (ie, high concentration) associated with short half-life may suggest that the drug remains in the blood instead of being distributed to organs (and to the tumor), which is in contrast to a high concentration associated with prolonged half-life that results from drug accumulation in the whole body (blood and tumor).
The role of hemoglobin in drug transport is further suggested by the fact that significant amounts of imatinib could be quantified in the erythrocyte sediments of patients treated with the drug.16 Erythrocyte loading was dose dependent in both volunteer and patient blood, and partition ratios of erythrocytes versus plasma ranged between 0.01 and 0.5. However, when patients have been treated with imatinib over prolonged periods of time, partition ratios can increase beyond 3 (Prenen et al, personal communication, April 2004). Various anticancer agents are capable of inducing changes in their own partition ratios17; the implications for imatinib are subject of further investigations.
Another hypothesis is that low hemoglobin reflects a more aggressive type of disease or a more advanced stage that is associated with mucosal ulceration and tumor bleeding and is less responsive to imatinib. Any other hypotheses suggesting that one of the molecular mechanisms of drug resistance either affects or is affected by hemoglobin levels could also explain this finding.
The same kind of hypotheses can be made for granulocyte counts, which apparently affect both initial and late drug resistance. A high granulocyte count may reflect an inflammatory reaction induced by an aggressive type of tumor that tends to progress earlier. Molecular classification of patients with low hemoglobin levels and high granulocyte counts may help in understanding those findings.
Late resistance to imatinib is independently predicted by the size of the lesions present at treatment start. Tumor size has also been reported as a prognostic factor for primary disease by several groups and, therefore, has been included in the definition of risk groups for GIST established by a consensus meeting.18 In our study, tumor size did not affect initial resistance to the drug (as demonstrated both by the univariate and the multivariate analyses), suggesting that patients with an advanced stage of disease are initially responsive to imatinib (as opposed to what is observed with conventional chemotherapy in solid tumors). However, tumor size has a significant impact on late resistance, which seems to increase with time, suggesting correlation with a delayed mechanism of drug resistance. Patients with large lesions are probably more exposed to the risk of secondary mutations because of a higher likelihood of emergence of new clones, which explains the increasing risk of late resistance.
According to our data, tumors of gastric origin tend to progress later than tumors of small bowel origin. This has also been reported for primary disease, and it has been suggested that the proportion of benign (as opposed to malignant) GISTs is higher in the stomach than in the small bowel19 and that the mitotic index is highly site dependent.20 Our model demonstrates that, in advanced disease, site of origin adds independent prognostic information to baseline granulocyte count and tumor size, but it still needs to be investigated whether this prognostic information is also independent of mitotic index.
Despite the inclusion of 20% of patients older than the age of 70 years, age did not show any prognostic value for either initial or late resistance to imatinib, even in the univariate analysis. The same observation has been reported for chronic myeloid leukemia, with the conclusion that the adverse prognosis of elderly patients is generally related to the treatment rather than the intrinsic biology of the disease.21 This statement is probably also true for GIST.
Finally, the subgroup analysis suggests that delayed resistance to imatinib in patients allocated to the high-dose regimen is mainly observed in patients with a high baseline granulocyte count (> 5 x 109/L) and in patients with a GI tumor origin outside of the stomach and the small bowel; our data do not suggest such an advantage in patients with tumors of small bowel origin. However, cross over from 400 mg once daily to 400 mg bid after progression has been shown to induce further disease stabilization in some patients,22 and only survival data will be able to demonstrate whether increasing the initial dose of imatinib results in a clinical benefit for all patients or subgroups of patients with advanced GIST.
Our results may be used to identify patients for whom initial and long-term treatment should be improved. In particular, imatinib resistance can be delayed by increasing the initial dose in patients with high granulocyte counts and in patients with tumors of GI origin outside of the stomach and small bowel. These results may also be helpful in the investigation or confirmation of the different biologic mechanisms of drug resistance. Correlation of identified prognostic factors with immunohistochemical and molecular parameters could improve the knowledge of this disease.
Appendix
The following investigators also contributed to this study: Dr J. Zalcberg, Peter Maccalum Cancer Institute, Melbourne, Australia; Dr S. Leyvraz, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland; Dr B. Bui, Institut Bergonie, Bordeaux, France; Dr P. Sch?ffksi, Medizinische Hochschule, Hannover, Germany; Dr A. Lopez Pousa, Hospital De La Santa Creu I Sant Pau, Barcelona, Spain; Dr D. Kotasek, Ashford Cancer Centre, Ashford, Australia; Dr T. De Pas, Istituto Europeo di Oncologia, Milan, Italy; Dr S. Rodenhuis, the Netherlands Cancer Institute, Amsterdam, the Netherlands; Dr W. Ruka, Maria Sklodowska-Curie Memorial Cancer Centre, Warsaw, Poland; Dr G. Grignani, Institute For Cancer Research and Treatment, Torino, Italy; Dr F. Duffaud, Centre Hospitalier Universitaire de la Timone, Marseilles, France; Dr J. Radford, Christie Hospital, Manchester, United Kingdom; Dr M. Findlay, Wellington Hospital, Wellington, New Zealand; Dr. C. Chevreau, Centre Claudius Regaud, Toulouse, France; Dr J. Whelan, Middlesex Hospital, London (Middlesex), United Kingdom; Dr L. Paz Arez, Hospital Universitario 12 De Octubre, Madrid, Spain; Dr M. Leahy, St James's University Hospital, Leeds, United Kingdom; Dr D. Hossfeld, Universitaets-Krankenhaus Eppendorf, Hamburg, Germany; Dr S. Frustaci, Centro Di Riferimento Oncologico, Aviano, Italy; Dr N. Deligny, Centre Oscar Lambret, Lille, France; Dr A. Krarup-Hansen, Herlev Hospital, Herlev, Denmark; Dr G. Apice, Istituto Nazionale Per Lo Studio E La Cura Dei Tumori, Napoli, Italy; Dr F. Cowie, Western Infirmary, Glasgow, United Kingdom; Dr K. Siong, National Cancer Center, Singapore; Dr G. Van Hazel, Sir Charles Gairdner Hospital, Perth, Australia; Dr W. van der Graaf, Academisch Ziekenhuis Groningen, Groningen, the Netherlands; Dr P. Lorigan, Westin Park Hospital, Sheffield, United Kingdom; Dr D. Grimes, Wesley Clinic for Hematology and Oncology, Brisbane, Australia; Dr M. Links, St George Hospital, Sydney, Australia; Dr A. Comandone, Ospedale Gradenigo, Torino, Italy; Dr H. Gelderblom, Leiden University Medical Center, Leiden, the Netherlands; Dr S. Clarke, Royal Prince Alfred Hospital, Sydney, Australia; Dr D. Wyld, Royal Brisbane Hospital, Brisbane, Australia; Dr J. Vermorken, Universitair Ziekenhuis Antwerp, Antwerp, Belgium; Dr O. Nielsen, Aarhus University Hospital, Aarhus, Denmark; Dr F. Kirsten, Bankstown-Lidcombe Hospital, Sydney, Australia; Dr J. Buesa, Hospital General de Asturias, Oviedo, Spain; Dr A. Poveda, Instituto Valenciano de Oncologia, Valencia, Spain; Dr N. Wilcken, Westmead Hospital, Sydney, Australia; Dr M. Green, Royal Melbourne Hospital, Melbourne, Australia; Dr R. McLennan, The Geelong Hospital, Geelong, Australia; Dr D. Ransom, Royal Perth Hospital, Perth, Australia; Dr C. Karapetis, Flinders Medical Centre, Adelaide, Australia; Dr I. Byard, Launceston General Hospital, Launceston, Australia; Dr P. Woll, Nottingham City Hospital, Nottingham, United Kingdom; Dr D. Bell, Royal North Shore Hospital, Sydney, Australia; and Dr E. Walpole, Princess Alexandra Hospital, Brisbane, Australia.
Authors' Disclosures of Potential Conflicts of Interest
Acknowledgment
We thank H. Dhillon (National Health and Medical Research Council Clinical Trials Centre, Sydney, Australia) and L. Mariani (Milan, Italy) for data management.
NOTES
Supported by an unrestricted grant from Novartis Oncology and by grant Nos. 2U10 CA11488-29 through 5U10 CA11488-34 from the National Cancer Institute (Bethesda, MD).
Presented in part at the 10th Annual Meeting of the Connective Tissue Oncology Society, Montreal, Quebec, Canada, November 11-13, 2004.
The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Joensuu H, Fletcher C, Dimitrijevic S, et al: Management of malignant gastrointestinal stromal tumours. Lancet Oncol 3:655-664, 2002
van Oosterom AT, Judson I, Verweij J, et al: Safety and efficacy of imatinib (STI571) in metastatic gastrointestinal stromal tumours: A phase I study. Lancet 358:1421-1423, 2001
Demetri GD, von Mehren M, Blanke CD, et al: Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N Engl J Med 347:472-480, 2002
Verweij J, van Oosterom A, Blay JY, et al: Imatinib mesylate (STI-571 Glivec, Gleevec) is an active agent for gastrointestinal stromal tumours, but does not yield responses in other soft-tissue sarcomas that are unselected for a molecular target: Results from an EORTC Soft Tissue and Bone Sarcoma Group phase II study. Eur J Cancer 39:2006-2011, 2003
Verweij J, Casali PG, Zalcberg J, et al: Progression-free survival in gastrointestinal stromal tumours with high-dose imatinib: Randomized trial. Lancet 364:1127-1134, 2004
Rankin C, Von Mehren M, Blanke C, et al: Dose effect of imatinib in patients with metastatic GIST: Phase III sarcoma group study S0033. J Clin Oncol 22:819s, 2004 (abstr 9005, suppl 14)
Blay J-Y, Berthaud P, Perol D, et al: Continuous vs intermittent imatinib treatment in advanced GIST after one year: A prospective randomized phase III trial of the French Sarcoma Group. J Clin Oncol 22:819s, 2004 (abstr 9006, suppl 14)
Therasse P, Arbuck SG, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumors: European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92:205-216, 2000
Fletcher JA, Corless CL, Dimitrijevic S, et al: Mechanisms of resistance to imatinib mesylate (IM) in advanced gastrointestinal stromal tumor (GIST). Proc Am Soc Clin Oncol 22:815, 2003 (abstr 3275)
Debiec-Rychter M, Cools J, Dumez H, et al: Mechanisms of resistance to imatinib mesylate in gastrointestinal stromal tumors and activity of the PKC412 inhibitor against imatinib-resistant mutants. Gastroenterology 128:270-279, 2005
Corless CL, Fletcher AJ, Heinrich MC: Biology of gastrointestinal stromal tumors. J Clin Oncol 22:3813-3825, 2004
Kantarjian H, O'Brien S, Cortes J, et al: Survival advantage with imatinib mesylate therapy in chronic-phase chronic myelogenous leukemia (CML-CP) after IFN-alpha failure and in late CML-CP, comparison with historical controls. Clin Cancer Res 10:68-75, 2004
O'Dwyer ME, Mauro MJ, Blasdel C, et al: Clonal evolution and lack of cytogenetic response are adverse prognostic factors for hematologic relapse of chronic phase CML patients treated with imatinib mesylate. Blood 103:451-455, 2004
Judson I, Donato di Paola E, Verweij J, et al: Population pharmacokinetic (PK) analysis and PK-pharmacodynamic (PD) correlations in phase I/II trial of imatinib in gastrointestinal stromal tumours (GIST) conducted by the European Organisation for Research and Treatment of Cancer Soft Tissue and Bone Sarcoma Group. Proc Am Soc Clin Oncol 22:818, 2003 (abstr 3287)
Judson I, Ma P, Peng B, et al: Imatinib pharmacokinetics in patients with gastrointestinal stromal tumour, a retrospective population pharmacokinetic study over time: EORTC Soft Tissue and Bone Sarcoma Group. Cancer Chemother Pharmacol 55:379-386, 2005
Guetens G, De Boeck G, Highley M, et al: Quantification of the anticancer agent STI-571 in erythrocytes and plasma by measurement of sediment technology and liquid chromatography-tandem mass spectrometry. J Chromatogr A 1020:27-34, 2003
Dumez H, Reinhart WH, Guetens G, et al: Human red blood cells: Rheological aspects, uptake, and release of cytotoxic drugs. Crit Rev Clin Lab Sci 41:159-188, 2004
Fletcher C, Berman JJ, Corless C, et al: Diagnosis of gastrointestinal stromal tumors: A consensus approach. Hum Pathol 33:459-465, 2002
Miettinen M, El-Rifai W, Sobin L, et al: Evaluation of malignancy and prognosis of gastrointestinal stromal tumors: A review. Hum Pathol 33:478-483, 2002
Emory TS, Sobin LH, Lukes L, et al: Prognosis of gastrointestinal smooth-muscle (stromal) tumors: Dependence on anatomic site. Am J Surg Pathol 23:82-87, 1999
Cortes J, Talpaz M, O'Brien S, et al: Effects of age on prognosis with imatinib mesylate therapy for patients with Philadelphia chromosome-positive chronic myelogenous leukemia. Cancer 98:1105-1113, 2003
Zalcberg JR, Verweij J, Casali PG, et al: Outcome of patients with advanced gastro-intestinal stromal tumors (GIST) crossing over to a daily imatinib dose of 800mg after progression on 400mg. Eur J Cancer (in press)(Martine Van Glabbeke, Jaa)