© Springer International Publishing Switzerland 2017
Charles R. Scoggins, Chandrajit P. Raut and John T. Mullen (eds.)Gastrointestinal Stromal Tumors10.1007/978-3-319-42632-7_12Prognostic Factors for Advanced GIST
(1)
Medical Oncology, Sidney Kimmel Cancer Center Johns Hopkins, 1650 Orleans Street, CRB1 G89, Baltimore, MD, USA
Keywords
ImatinibTyrosine kinase inhibitor (TKI)Overall survival (OS)MutationKITPDGFRαExonCmin Circulating tumor DNA (ctDNA)1 Clinicopathologic Prognostic Factors
Prior to immunohistochemical identification of gastrointestinal stromal tumors (GISTs) utilizing CD117 and CD34 [1], many surgical reviews analyzing prognostic factors of gastrointestinal sarcomas included leiomyosarcomas in retrospective series that stretched over long periods of time [9–12]. In 2000, a retrospective analysis of GIST analyzed a 200 patient cohort at a single institution [8]. This cohort mixed primary, recurrent, and metastatic patients. Forty-seven percent of the patients presented with metastatic disease. Overall survival (OS) for those with metastatic disease was 19 months. One hundred fourteen of the 200 patients had surgery including 28 patients with metastatic disease. Multivariate analysis revealed that male sex, tumor size (>5 cm), and incomplete or unresectable tumors were poor prognostic signs. Although this analysis did not differentiate prognostic factors among the cohorts, it established a baseline representation of median overall survival in metastatic disease.
This baseline analysis of prognostic factors was further refined in the subsequent publication of several clinical trials of advanced and metastatic GIST patients. The B2222 trial, a phase II trial of advanced GIST patients treated with either 400 or 600 mg of imatinib daily, produced a clinical benefit (complete and partial response plus stable disease) in 80 % of the patients at a median follow-up of 24 weeks [13]. Given the initial robust response, a 4-year extension study and analysis was performed [7]. Of the initial 147 patients on the study, 56 of them were in the extended analysis with 46 of them taking imatinib for 5 years and 41 on treatment at the time of data analysis. With this longer follow-up, the clinical benefit rate was 83.7 %. Multivariate analysis of prognostic factors indicated worse outcomes for males, hypoalbuminemia (<3 g/dL), and patients with absolute neutrophil counts >4.5 × 109/L.
Two large phase III clinical trials in metastatic GIST and their subsequent meta-analysis also supported and extended these results [6, 14, 15]. The S0033 trial enrolled 746 patients who received imatinib either 400 mg daily or twice daily. Statistically significant worse prognostic factors identified by multivariate analysis included older age, poorer eastern cooperative oncology group (ECOG) performance status [2, 3], male sex, high absolute neutrophil counts (ANCs), and a low albumin (<3.5 g/dL) [15].
Early clinical trials of metastatic GIST noted that a small cohort of patients exhibited initial resistance to imatinib as defined as occurring in the first 3 months [13, 16]. The phase III trial run by the European Organization for Research and Treatment of Cancer, Italian Sarcoma Group and Australian Gastrointestinal Trials Group (EORTC-ISG-AGITG) 62005 identified prognostic factors in these initial and late resistance groups. Multivariate analysis revealed that lung but not liver metastases, a low hemoglobin level (<7 mmol/L), and a high granulocyte count characterized worse prognosis in the 116 patients with initial resistance to imatinib, while the 818 late resistance patients had a high baseline granulocyte count (>5.1 × 109/L), a larger tumor size (>12 cm), a 400-mg imatinib dose, and nongastric primary GIST [17]. The meta-GIST analysis combined both of these phase III trials identifying seven adverse prognostic factors for OS: hypoalbuminemia, male sex, larger tumor size, high ANC, older age, prior chemotherapy, and poor ECOG performance status [14]. A recent analysis of the EORTC-ISG-AGITG 62005 developed a nomogram for OS at 3 years on those patients who had KIT and platelet-derived growth factor receptor A (PDGFRA) genotyping available [18]. Five predictors for OS in this subset were identified: mitotic count per 50 high-powered fields (HPF) of the primary tumor, hemoglobin concentration, and ANC at the start of imatinib treatment, diameter of the largest metastasis, and tumor genotype. The nomogram was validated in a large data set of 236 patients spread over six international GIST referral centers and helped identify high-, intermediate-, and low-risk groups.
2 Treatment Prognostic Factors
2.1 Tyrosine Kinase Inhibitors
2.1.1 Imatinib Dose
The use of imatinib itself, though never formally compared to traditional chemotherapy in a randomized trial, clearly improves median overall survival. Previous traditional regimens performed poorly with one multiagent regimen having a median overall survival (OS) of 16.7 months [5]. In the EORTC-ISG-ATIGT 62005 study, a comparison between the two imatinib doses and historical controls from the EORTC database of GIST patients treated with doxorubicin revealed an approximate 10-month median OS with doxorubicin compared to a median OS that was not reached with both imatinib doses out to 30 months [6].
The dose of imatinib chosen can affect progression-free survival (PFS), but not OS in GIST patients. The initial phase I study of imatinib established a mean tolerated dose (MTD) of 400 mg twice daily (BID) with 82 % of the patients achieving a clinical benefit [16, 19]. Early phase clinical trials established a dose range of imatinib from 400 mg or 600 mg daily to 400 mg twice per day [13, 19]. The B2222 trial noted no difference in OS between the 400 and 600 mg daily dosing with a median overall survival of 57 months [7]. Further planning led to two separate phase III clinical trials testing 400 mg daily vs. 400 mg twice a day [15, 17] with a preplanned meta-analysis of these two trials, meta-GIST [14]. The S0033 trial did not find an OS difference between 400 mg daily and twice per day with a median survival of 55 or 51 months, respectively. The EORTC-ISG-ASG 62005 study tested progression-free survival (PFS) at those same doses and initially found a PFS benefit for the high-dose arm with the late resistance data favoring better outcomes with high-dose imatinib for small bowel GISTs. However, further analysis at a median follow-up of 40 months revealed no difference in PFS or OS. The meta-GIST analysis confirmed the findings showing no difference in OS between the high dose (800 mg) and standard dose (400 mg) arms.
Imatinib resistance in metastatic patients develops in approximately 20 months. Questions arose regarding an interrupted treatment strategy as a means of prolonging the duration of effectiveness and reducing resistance in advanced patients on imatinib. The French Sarcoma Group tested this hypothesis in the multicenter BRF14 trial interrupting treatment in a subset of patients after 1, 3, or 5 years of imatinib therapy. Fifty-eight patients were randomized between continuation (n = 26) and interruption (n = 32) groups after 1 year yielding progression in 26 of the 32 in the interruption group [20]. In the 3-year cohort, interruption led to a 2-year PFS of 16 % and in the 5-year cohort progression after 1 year of follow-up in 5 of the 11 patients [21, 22]. Analysis of the 1- and 3-year cohorts demonstrated that reintroduction of imatinib produced tumor control and similar mean times to secondary resistance without a difference in OS between continuation and interruption groups [20, 21]. Longer follow-up of the 71 patients with documented progressive disease (PD) who had interrupted and restarted therapy was carried out. Rechallenge with imatinib resulted in better PFS in those patients with longer imatinib-free intervals as well as those with a complete response (CR). However, this exploratory analysis was not powered to determine effects of OS [23]. Given the poor PFS with imatinib interruption, this is not recommended in metastatic patients stable on imatinib therapy.
2.1.2 Imatinib Trough Levels (Cmin)
Imatinib has excellent oral bioavailability and a 20-h half-life with 400 mg achieving expected pharmacodynamic effects [24, 25]. However, as indicated above in the EORTC-ISG-ASG 62005 study of metastatic GIST patients, dosing of imatinib influenced PFS though not OS. Furthermore, interpatient plasma imatinib levels fluctuate greatly [26]. Variability in plasma levels could influence response to therapy. An observational study of 38 GIST patients found that the imatinib-free drug levels (AUCu) significantly predicted response with an odd ratio (OR) of 2.6 (±1.1) [27]. A pharmacokinetic analysis of imatinib in metastatic patients done on 73 patients from the B2222 study evaluated imatinib Cmin levels at day 29. Those patients with concentrations below 1,100 ng/mL had reduced clinical benefit as measured by tumor response and time to progression (TTP) though it did not reach statistical significance [28]. Cmin levels were divided into quartiles (Q) for analysis and those with the lowest Cmin levels, Q1, had a TTP of 11.3 months compared to 30.6 months for Q2–3 and 33.1 months for Q4. Responders had Cmin levels of 1,446 ng/mL, while nonresponders were lower at 1,155 ng/mL. To apply Cmin levels prospectively in a clinical setting, 96 patients with advanced GIST treated were evaluated in an observational study. A Cmin level of 760 ng/mL predicted statistically significant differences in PFS whether stomach or small bowel in location [29]. Utilization of imatinib trough levels remains an area of active research without a defined role in routine clinical practice.
2.1.3 Sunitinib
After failure of imatinib in metastatic disease, sunitinib and regorafenib are indicated therapies in the second- and third-line setting, respectively [30, 31]. In the pivotal phase III GIST clinical trial of sunitinib, patients were randomized in a 2:1 design to sunitinib or placebo with sunitinib administered in a 4 weeks on, 2 weeks off, 50-mg dose regimen. Sunitinib produced a time to tumor progression (TTP) of 27 weeks compared with 6 weeks for placebo. Final analysis of the trial with conventional statistics revealed no difference between treatment and placebo arms in OS given the crossover design of the trial. However, an exploratory statistical analysis estimated a doubling of OS for sunitinib versus placebo of 73 versus 39 weeks. Multivariate analysis identified tumor size as a prognostic factor in this group of patients [32]. This overall survival endpoint is supported by further analysis of sunitinib in an international treatment-use trial of 1,124 patients, which revealed a similar OS time [33].
2.1.4 Regorafenib, Nilotinib, and Sorafenib
A number of other oral tyrosine kinase inhibitors have been utilized in GIST patients without prognostic factors identified to date. Regorafenib was recently approved in the third-line setting providing a PFS but not OS survival benefit. Further long-term follow-up is needed to determine if an OS benefit emerges [30]. Nilotinib in the third line did not show significant survival advantages in the intention-to-treat population though post hoc subgroup analysis suggested an OS benefit in a true population of patients who had received only two prior tyrosine kinase inhibitors [34]. Given prior earlier potential benefit, nilotinib was also tested as a first-line therapy against imatinib in the ENESTg1 trial. This trial was halted early for futility as it did not match imatinib efficacy [35]. Sorafenib was tested in the second and third line in an early phase trial demonstrating a progression-free survival of 5 months with most having stable disease [36]. Effectiveness was then studied in a larger community cohort of 124 patients. Sorafenib treatment in this third- or fourth-line line setting achieved 6.4 months [37]. Interestingly, a retrospective analysis of 223 GIST patients treated in the third-line setting revealed a PFS of 3.6 months and OS of 9.2 months. Factors associated with poor OS in this analysis were performance status ≥ ECOG 2 and albumin levels <35 g/L. Despite the advanced nature of GIST in these patients, further treatment with other tyrosine kinase inhibitors improved overall survival significantly compared to best supportive care in this study [38].
2.2 Circulating Factors
2.2.1 KIT/VEGF
Imatinib targets the KIT and platelet-derived growth factor receptors (PDGFRs), while sunitinib targets several receptors including KIT, fms-like tyrosine kinase-3 receptor (FLT-3), RET, PDGFRs, and the three vascular endothelial growth factor receptor (VEGFR) isoforms. Both therapies have been evaluated for mechanisms of resistance as well as molecular biomarkers for response to therapy. Soluble c-KIT (sKIT) and its ligand, stem cell factor (SCF), are present in the normal serum. sKIT results from proteolytic cleavage from the extracellular membrane and can bind circulating SCF, therefore possibly modulating its signaling [39, 40]. Preclinical data supported a role for sunitinib inhibiting multiple human and xenograft tumor models [41]. Further work supported its inhibition of angiogenesis, promotion of apoptosis in lung cancer and glioblastoma multiforme murine xenograft tumor models, and reduction of metastases in lung xenograft models. Analysis of its angiogenesis effects revealed the inhibition of neovascularization rather than the direct inhibition of existing tumor vasculature [42, 43]. Because sunitinib targets angiogenesis pathways, various VEGF and VEGFR proteins have been studied in clinical trials as potential biomarkers in a number of different cancers. VEGF levels tended to increase while soluble VEGFR-2 (sVEGFR-2), soluble VEGFR-3 (sVEGFR-3), and soluble cKIT (sKIT) levels decreased in renal cell carcinoma, hepatocellular carcinoma, and breast cancer [44–46].
As c-KIT plays a critical role in much of GIST pathogenesis and GISTs characteristically have increased vascularity, GIST trials evaluated these markers in metastatic patients. The B2222 imatinib trial measured VEGF, sKIT, and SCF levels in 66 of the 147 enrolled patients. While increases in SCF, VEGF, and the ratio between SCF and sKIT levels were observed, no prognostic information emerged between responders and nonresponders to imatinib. The analysis was hampered by imatinib’s success as there were only nine nonresponders in the population of sera analyzed [47].
These hypothesis-generating findings were subsequently analyzed in three sunitinib trials. In the phase I/II study of sunitinib, sKIT increases correlated with nonresponders while decreased sKIT correlated with responders [48, 49]. This was also supported by an open-label continuous daily dosing (CDD) sunitinib trial with an increasing statistical correlation between decreasing sKIT levels and OS [50]. The phase III trial testing sunitinib treatment at 50 mg/day for 4- of 6-week cycles originally showed a reduction of sKIT levels in the treatment arm during the first two cycles, which were a significant predictor of time to progression [51]. However, in the final analysis of this trial, sKIT levels did not correlate with OS. That correlation was found only with sVEGFR-2 baseline values and the sVEGFR-2 cycle 1, day 14 ratio to baseline.
2.2.2 Circulating Tumor Cells
Circulating Plasma DNA
Over the past two decades, novel advances in the direct detection of solid tumors through blood and plasma analysis reached clinical trial testing notably in the monitoring of circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA). Intense interest in these assays as biomarkers has been championed in multiple tumor types as possibly heralding earlier detection of residual disease after surgery, earlier identification of disease resistance, and effectiveness of neoadjuvant or adjuvant therapy long before radiographic detection [52–55]. Hematologic malignancies have shown both diagnostic and clinical values in ctDNA assays in various settings such as acute promyelocytic leukemia (APL), acute lymphoblastic leukemia (ALL), and chronic myelogenous leukemia (CML) [56–58]. Indeed, the detection of minimal residual disease (MRD) in ALL and early molecular response (EMR) in CML serve as strong prognostic markers in patients [59–61].
GISTs are ideal candidates for this analysis given the known mutations, effective targeted therapy, and onset of resistance in metastatic patients. However, the analyses are in the early stages of development. Maier et al. prospectively analyzed 291 plasma samples among 38 patients for ctDNA in a phase IIIb nonintervention trial [62]. Primary tumor samples were sequenced for the identification of candidate mutations and these were then evaluated in plasma samples by allele-specific ligation polymerase chain reaction (L-PCR). Eighteen patients had active disease and 20 had a postsurgical complete response (CR). Nine of the 18 patients with active disease and 6 of 17 postsurgical patients with an increased relapse risk had ctDNA carrying the mutation identified from the primary tumor. Furthermore, those patients with active disease had a high ratio of ctDNA to wild-type DNA. Lastly, they detected increases in ctDNA in patients whose active therapy was failing or in postsurgical CR patients prior to their relapse radiographically as well as decreases in ctDNA in some patients with responses to imatinib or sunitinib.
In the phase III GRID, ctDNA was analyzed by BEAMing technology in 163 baseline samples detecting mutations in approximately 60 % of samples [63]. Importantly, the ctDNA analysis led to detection of secondary kinase mutations in 47 % of the samples as compared to 12 % of the tumor samples. The group with secondary mutations had a shorter PFS than those in the study receiving placebo. This study demonstrated the feasibility and utility of ctDNA sample collection over repetitive tumor biopsies as a means of capturing the development of resistance in GIST.
This approach has been replicated in other studies. Two small case series evaluating three and four GIST patients, respectively, looked at ctDNA in primary and resistant settings in the context of imatinib and sunitinib treatment [64, 65]. They identified both primary and secondary mutations in the ctDNA. A serum biomarker analysis of samples collected from a phase II dovitinib study in TKI-refractory GIST, detected 5 primary and 11 secondary KIT mutations among 30 patients. The absence of secondary mutations in the serum of patients correlated significantly (P = 0.02) with a better median OS of 9.8 months [66, 67]. A similar study evaluating ponatinib in TKI-resistant GIST found ctDNA in 15 of 23 patients analyzed with associations found between decreased ctDNA and radiographic response [68].
2.3 Molecular Prognostic Factors
2.3.1 Mutational Landscape
KIT and PDGFRα are transmembrane receptor tyrosine kinase signaling molecules that play a fundamental role in the pathogenesis of most GISTs. Both c-KIT and PDGFRα signal through receptor homodimerization caused by binding of their ligands, SCF and PDGF, respectively [69, 70]. Gain of function mutations primarily in c-KIT and the PDGFRα drive tumorigenesis in GISTs and are mutually exclusive [2, 3, 71, 72]. The mutations occur in discrete regions of the receptors all of which result in receptor autophosphorylation and ligand-independent signaling. KIT mutations occur most often in exon 9 in the extracellular domain and exon 11 in the juxtamembrane (JM) domain, with decreased mutational frequencies in exons 13 and 17 of the kinase domains. In contrast, PDGFRα mutations occur infrequently in the juxtamembrane domain (exon 12), rarely in the first tyrosine kinase domain (exon 14) and most often in the activation loop of the second kinase domain (exon 18) [2, 71, 73].
The various KIT and PDGFRα mutations described above also correlate with morphology and anatomical location in several studies. KIT exon 11 mutations and PDGFRα exon 18 mutants occur with greatest frequency in gastric GISTs, while KIT exon 9, exon 17, and exon 13 mutations arise more often in GISTs of the small intestine. Exon 11 GISTs tend to have a spindled morphology while those with exon 9 have an epithelioid appearance [70, 71, 74–76]. The landscape of specific GIST mutations is quite broad. Exon 11 mutations tend to cluster at the 5′ end of the juxtamembrane (JM) region and include deletions between codons 550 and 561, missense mutations, point mutations, and internal tandem duplications (ITDs) [77–80]. Less frequently, ITDs represent three prime JM mutations [81]. Deletion mutations predicted worse survival outcomes than did missense mutations [82]. Exon 9 included ITDs and missense mutations and were associated with worse prognosis in early studies [2, 75, 83] while exons 13 and 17 exhibited substitution mutations [2, 80] that occur most frequently as resistance mechanisms to imatinib treatment [84]. PDGFRα harbors similar types of mutations. The exon 18 kinase domain and exon 12 JM domain contain missense mutations and deletions while exon 14 kinase domain has missense mutations [71–73, 85, 86]. The exon 18 mutations involve codons from 841 to 848 with a high percentage involving the D842V mutation [71, 85].
Further analysis of GIST mutational heterogeneity in the context of treatment led to correlations between specific mutations and response to therapy. In the B2222 study described above, both c-KIT and PDGFRα mutations were correlated for clinical outcome. Partial response rates to imatinib in exon 11 mutations (n = 85) were 83.5 % versus 48 % for those with exon 9, PDGFRα, or no mutations (n = 44). This translated into a longer OS in multivariate modeling at 29 months of follow-up. Although exon 9 was worse than exon 11 prognostically, it did show improved OS compared to GISTs with PDGFRα or no mutations. Furthermore, recapitulating in vitro modeling, patients with the PDGFRα D842V mutation were found to be unresponsive to imatinib [73]. Subsequent analysis at 63 months of follow-up maintained the association of exon 11 mutational status with overall survival [7].
These mutational data were confirmed in three phase III analyses. First, the S0033 study determined that the exon 11 genotype had improved responses to imatinib, longer TTP, and better OS when compared with exon 9 or wild-type GIST genotypes. Exon 9 genotype had improved responses to a higher imatinib dose but this did not correlate with OS [87]. Second, the EORTC-ISG-AGITG trial demonstrated that KIT exon 9 mutations were the worst prognostic factor when compared with exon 11 mutations decreasing both PFS and OS outcomes. Exon 9 mutations treated at the higher dose of imatinib had improved PFS [88]. Last, in the meta-GIST analysis of the two prior phase III studies, patients with exon 11 mutations had better OS than those with exon 9, wild-type, or other mutations. Furthermore, analysis of the patients on the high-dose imatinib arm revealed better PFS for exon 9 mutations though this did not translate to better overall survival [14]. Therefore, exon 11 mutations denote better prognostic significance in the context of imatinib treatment.
Consideration of genotype on sunitinib effects in the phase I/II study revealed improved benefit for specific mutational subsets as well. Fifty-eight percent of patients with primary exon 9 mutations had clinical benefit as opposed to 34 % with primary exon 11 mutations. This translated into PFS and OS benefit in the second-line setting. Furthermore, when analyzing sunitinib response based on secondary resistance genotype in imatinib-treated patients, secondary mutations in exon 13 or 14 of KIT, which correspond to the ATP binding pocket, correlated with a significantly longer PFS and OS than those secondary mutations in the kinase activation loop. Similar to imatinib treatment, the PDGFRα exon 18 mutation D842V was resistant to sunitinib therapy [89].
Related posts:
on Molecular Characterization and Targeted Therapies on GIST
Therapy for Metastatic Gastrointestinal Stromal Tumor
History and Prognosis of Localized Gastrointestinal Stromal Tumors in the Pre- and Post-imatinib Eras
Invasive Approaches to Gastrointestinal Stromal Tumors (GISTs)
Evaluation of Gastrointestinal Stromal Tumors
Management of Gastrointestinal Stromal Tumors
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
