Imatinib treatment of poor prognosis mesenchymal-type primary colon cancer: a proof-of-concept study in the preoperative window period (ImPACCT)

Mortality from colon cancer is almost invariably due to the development of distant metastases. In current practice, pathological (TNM stage) and clinical characteristics (age, co-morbidity) mainly determine the choice of adjuvant chemotherapy. However, these features have limited value in predicting which patients are at risk of developing metastases. In clinical trials, the five-year recurrence rate in stage III colon cancer patients is approximately 50% without adjuvant chemotherapy. With adjuvant treatment this is reduced to ~35%, implying that such treatment is only effective in a subgroup of patients [1]. Consequently, the majority of patients are currently being under- or over-treated. It is therefore important to be able to identify patients who are at high risk of recurrence and to develop more effective therapies to prevent relapse. Relapse-prevention trials in the adjuvant setting are challenging however, due to the long follow-up periods and the large numbers of patients that are required for sufficient statistical power. Prior evidence of drug activity and the availability of a companion diagnostic tool for patient selection could greatly facilitate the design and increase the quality of such studies.

Novel adjuvant therapies should be based on an understanding of the pathways that drive metastasis formation. Recent studies on molecular classification of colon cancer have provided insight into these pathways. Several research groups have independently developed classification systems for primary colon cancer based on gene expression profiles [2‐8]. Cross-cohort analysis of the results has led to the identification of four consensus molecular subtypes (CMS1–4). Of these, CMS4 (~25% of colon cancers) is associated with a significantly worse disease-free and overall survival [9]. Novel treatment strategies for this subtype are thus particularly needed.

The pro-metastatic pathways that are upregulated in CMS4 provide opportunities for subtype-targeted therapy. CMS4 tumours are characterised by high expression of stem cell and mesenchymal genes, and a high stromal content [9]. We have previously shown that Platelet-Derived Growth Factor Receptors (PDGFRs) and KIT are highly expressed in mesenchymal-type colon tumours and that their expression strongly correlates with disease-free survival. Moreover, in vitro and in vivo inhibition of PDGFR and KIT tyrosine kinase signalling reduced invasiveness, metastatic potential and stem-like characteristics of mesenchymal-type colon cancer [10‐12]. Based on these findings we hypothesise that patients with poor-prognosis mesenchymal-type colon cancers could benefit from treatment with imatinib, a tyrosine kinase inhibitor with high selectivity for PDGFR and KIT.

To test this hypothesis in a proof-of-concept study, we designed the ImPACCT trial (Imatinib as Pre-operative Anti-Colon Cancer Targeted therapy). In ImPACCT, patients with CMS4 colon cancer are identified with a recently developed 4-gene RT-qPCR test that measures PDGFRA, PDGFRB, PDGFC and KIT expression levels in diagnostic tumour biopsies [13]. Pre-selected patients with CMS4 tumours are then treated with imatinib during the pre-operative window period (the time between initial cancer diagnosis and surgery). This allows comparison of pre-treatment diagnostic tumour biopsies with biopsies obtained from the resection specimen after treatment. The primary objective is to assess whether imatinib treatment reduces the aggressive phenotype of CMS4 tumours in colon cancer patients. ImPACCT may not only form the basis for future adjuvant studies with imatinib, but could also serve as a blueprint for other proof-of-concept studies with subtype-targeted therapies.

Targeted anti-cancer therapies that seem promising in pre-clinical and early-phase clinical trials often fail to show benefit in phase III randomized controlled trials. Up to 60 % of phase III trials have negative outcomes [14, 15]. This high failure rate underscores the need for optimization of trial design, including better patient selection. Since the successful addition of oxaliplatin to the adjuvant chemotherapy regimen in 2004 [16], no further advances have been made in the outcome of patients with stage II/III colon cancer. The lack of recent positive adjuvant chemotherapy trials is partly due to the fact that hardly any new drugs are actually tested in the adjuvant setting. Novel treatments are generally first tested in patients with metastatic disease who have no regular treatment options left, and who have often received multiple lines of systemic treatment. Therapies aimed at preventing the development or outgrowth of metastases are probably most effective at early disease stages, but their clinical development may be terminated due to lack of efficacy in late-stage disease [17]. Moreover, new (combination) treatments that are effective in metastatic disease do not necessarily have value in the adjuvant setting, as exemplified by trials PETACC-03 (5-FU/LV plus irinotecan) [18], NSABP C-08 (FOLFOX6 plus bevacizumab) [19] and N0147 (FOLFIRI plus cetuximab) [20]. To address this problem, design and approval of clinical trials in which promising drugs are tested in treatment-naïve patients with early-stage disease, rather than in late-stage patients who progressed under standard treatment, is needed.

Prospective stratification and/or inclusion based on predictive molecular biomarkers will presumably improve trial results by enrichment for responsive patients [21]. However, microsatellite instability is currently the only molecular marker that is being used in the clinical decision process for adjuvant chemotherapy in colon cancer [22]. The four recently identified CMSs in colon cancer show marked differences in the activity of various biological pathways, which could provide the basis for subtype-specific targeted therapy [9]. CMS4 (the mesenchymal/stem-like/stroma-rich subtype) has the poorest prognosis and, importantly, this subtype seems associated with a lack of benefit from oxaliplatin treatment [23]. These findings call for clinical trials with novel therapies specifically targeting CMS4 tumours. In ImPACCT we aim to deliver proof-of-concept that PDGFR/KIT inhibition with imatinib reduces the aggressive phenotype of newly diagnosed primary CMS4 colon tumours. Delivering proof-of-concept for drug activity in early stages of drug development is pivotal to prevent unfounded trial phase transition [14, 15].

ImPACCT is conducted with treatment-naïve patients during the pre-operative window period, which allows us to obtain high quality tissue material before and after imatinib therapy without additional interventions: pre-treatment samples will be obtained during diagnostic colonoscopy and post-treatment biopsies are collected from the resection specimen; both procedures are part of standard of care. By comparison of pre- and post-treatment tissue samples, the effects of imatinib on CMS4 tumours can be evaluated at a cellular level. The treatment period of 14 days is within the normal time frame from diagnosis to surgery (the pre-operative window). We expect that two weeks of treatment will be sufficient to induce changes in gene expression: given that steady state plasma concentrations will be attained within four days (the plasma half-life of imatinib is 18 h [24]), patients will be exposed to the full dose for ten days. The chosen daily dose of 400 mg is the standard dose for chronic myeloid leukaemia, gastrointestinal stromal tumours (GIST) and myelodysplastic syndrome.

Experimental therapy during the pre-operative window period could potentially lead to an increase in postoperative complications. However, based on two phase II trials that evaluated the use of neoadjuvant imatinib therapy for otherwise irresectable or metastatic GIST, we believe it is safe to administer the last dose of imatinib within 12 h before surgery. In these trials imatinib was administered for several months and stopped one day prior to surgery. Both studies concluded that this approach was feasible, and the reported postoperative complications were acceptable and not out of the ordinary considering the extensive abdominal surgery [25, 26].

Instead of assessing a clinical endpoint (e.g. disease-free survival), this proof-of-concept trial investigates a surrogate endpoint (gene expression changes) to demonstrate a ‘biological’ treatment effect. This endpoint requires a relatively limited sample size and can be determined after only two weeks of therapy, which limits the burden placed on participants. If treatment effects are indeed demonstrated, knowledge of the size of the effect on tumour biology can be used to drive the design of a larger randomised phase II trial.

The design of the ImPACCT trial presented here allows rapid evaluation of the mechanism of action of a targeted therapy in a subtype-stratified patient population by analysis of paired pre- and post-treatment biopsies, without interfering with or delaying standard-of-care, by making use of the pre-operative window period. This design can serve as a blueprint for subtype-directed proof-of-concept trials in colon cancer, with the ultimate goal of designing effective adjuvant therapy that eradicates occult metastases.