Targeting c-KIT, PDGFR in cancer of unknown primary: a screening study for molecular markers of benefit

Introduction

Cancer of unknown primary (CUP) makes up to 3–5% of metastatic malignancies. It is characterised by failure to identify the primary site with appropriate clinical, imaging and laboratory work-up. The majority of CUP patients present with widespread metastases after a short symptomatic period, do not benefit from systemic therapy and succumb to malignancy within a few months. Although some progress has been made recently with the identification of favourable prognosis subgroups of CUP patients who occasionally enjoy long-term survival with chemotherapy, the molecular pathophysiology of the disease is still poorly understood (Pentheroudakis et al. 2007). No molecular markers of established predictive or prognostic significance have been found to date (Pentheroudakis and Pavlidis 2006).

The proto-oncogene c-kit encodes a type III transmembrane receptor (CD117) with cytoplasmic tyrosine kinase activity involved in inhibition of apoptosis, regulation of cell adhesion and induction of cellular proliferation. Constitutive ligand-independent KIT activation can occur by gain-of-function mutations mainly observed in gastrointestinal stromal tumours (GIST), an early oncogenic event that contributes to malignant transformation of the interstitial Cajal cells of the myenteric plexus. The most common such mutations are seen in exon 11 of the gene, the mutated exon encoding a juxtamembrane domain that facilitates spontaneous KIT receptor dimerisation and activation of tyrosine kinase (Hirota et al. 1998).

Two structurally related transmembrane protein tyrosine kinase receptor subunits, α- and β-, represent the PDGF receptor with activity involved in regulation of cell growth, actin reorganisation and chemotaxis, among others (Heldin and Westermark 1999). Constitutive ligand-independent PDGFRa activation can occur by gain-of-function mutations mainly observed in GISTs and most commonly found in exons 12 and 18 of the gene, the mutated exon encoding a juxtamembrane and a tyrosine kinase II domain, respectively (Hirota et al. 2003; Heinrich et al. 2003a).

The advent of receptor tyrosine kinase inhibitors provided impetus for the identification of target molecules driving neoplastic growth. Imatinib (STI 571, Gleevec^®) inhibits KIT, BCR-ABL and PDGFR tyrosine kinases, producing remarkable clinical benefit in patients with GIST and chronic myelogenous leukemia (Buchdunger et al. 2000; Demetri et al. 2002). Sunitinib inhibits VEGFR, PDGFR, KIT and Fms-like tyrosine kinase 3 (Marx 2005), producing benefit in renal-cell carcinoma (Motzer et al. 2006a, b) and GIST patients, among others. The lack of insight into the molecular biology of CUP, lack of data on KIT and PDGFR status along with the prospect for beneficial tyrosine kinase inhibition with targeted drugs have prompted us to screen for exon 11 c-kit and exon 12 and 18 PDGFR gene mutations and respective protein expression in CUP.

Materials and methods

Tumour tissue from formalin-fixed paraffin embedded biopsy specimens was collected from 50 patients with CUP managed at the Department of Medical Oncology at the Ioannina University Hospital. All patients were treated with combination chemotherapy according to Hellenic Cooperative Oncology Group (HeCOG) protocols. Histological sections were used for histopathologic diagnosis with hematoxylin-eosin staining, for evaluation of CD117, PDGFRa protein expression and for mutational gene analysis. For all tumour samples included in the analysis, the number of malignant cells represented at least 75% of all nucleated cells, as judged by hematoxylin-eosin staining. The protocol was approved by the local research ethics committee of the Ioannina University Hospital and all patients provided written informed consent for research use of their biologic material.

Results

Discussion

The c-kit and PDGFRa genes, located both on the long arm of chromosome 4, encode for the KIT and PDGFRa receptor tyrosine kinases consisting of an extracellular domain, the transmembrane segment and the juxtamembrane and cytoplasmic domains (Yarden et al. 1987; Claesson-Welsh et al. 1989). The juxtamembrane domain is responsible for the dimerisation of two KIT receptors upon binding to the extracellular segment of its ligand, Stem Cell Factor (SCF). The extracellular Ig domain 4 is responsible for the dimerisation of two PDGF receptors, in addition to the bridging effect of its ligand, PDGF. After ligand-induced activation and dimerisation, the cytoplasmic segment of both receptors undergoes autophosphorylation of tyrosine residues and activation of signal transduction pathways favouring cell survival and proliferation (Kitamura and Hirotab 2004; Kazlauskas and Cooper 1989).

The pivotal oncogenic effect of KIT activation has been demonstrated primarily in malignant gastrointestinal stromal tumours (GISTs). Gain-of-function mutations of the c-kit gene that result in ligand-independent constitutive KIT receptor protein dimerisation and tyrosine kinase activation has been found in 85–90% of these tumours (Rubin et al. 2001). These small in-frame deletions, insertions or point mutations were most frequently found (80% of mutated cases) at exon 11 encoding for the juxtamembrane domain (mostly in codons 550–562) and occasionally at exons 9, 13, 14 and 17, encoding for the extracellular and tyrosine kinase domains. Preclinical experiments along with clinical observations proved that C-KIT constitutive activation is an early, necessary and sufficient oncogenic stimulus for malignant transformation and that GISTs remain dependent on it for continuing growth (Hirota et al. 1998; Kitamura and Hirotab 2004; de Silva and Reid 2003).

Expression of PDGF receptor has been linked to the development of several tumours, including gliomas and GISTs (Shih and Holland 2006; Corless et al. 2004). Activating mutations were found in exons 12 and 18 of the PDGF receptor, which encode parts of the juxtamembrane and the tyrosine kinase regions. The signal transduction profiles of the PDGFR-mutant GISTs were indistinguishable from KIT-mutant tumours, suggesting that PDGFRa can substitute for KIT in GIST oncogenesis (Heinrich et al. 2003).

The advent of imatinib mesylate (Gleevec^®, Novartis) and sunitinib (Sutent^®, Pfizer) has inaugurated new therapeutic avenues with drugs targeting tumour-specific molecular abnormalities. Dramatic responses are seen in metastatic GIST patients treated with imatinib (Buchdunger et al. 2000; Demetri et al. 2002) and sunitinib (Demetri et al. 2006). Interestingly, GIST patients with c-kit mutations in exon 11 respond better to imatinib (tumour regression rate 72%) and live longer than patients with mutations in other exons or wild type c-kit (tumour regression rate 32 and 12%, respectively) (Heinrich et al. 2003b; Debiec-Rychter et al. 2004). Implications have also been made concerning a possible correlation between PDGF receptor mutations and response to imatinib treatment (Corless et al. 2005).

The results of our study show that in patients with unknown primary carcinoma neither KIT oncoprotein expression, present in 81% of our cases, nor PDGFRa protein expression, present in 50% of our cases, are associated with exon 11 of c-kit or exons 12 or 18 of PDGFR gene activating mutations. PCR-based SSCP is a fast, economic and highly sensitive molecular technique that allows reliable detection of nucleotide changes in samples containing as little as 10% of mutated DNA. Immunohistochemical KIT protein expression has been reported by other investigators in as many as 70% of small cell lung carcinomas, 84% of seminomas, 65% of adenoid cystic carcinomas, 35% of melanomas and occasionally in 47 additional tumour types (Went et al. 2004). Immunohistochemical PDGFRa expression has been reported in head-and-neck (Ongkeko et al. 2005), melanoma (Ugurel et al. 2005), renal cell carcinoma (Sulzbacher et al. 2003a), osteosarcoma (Sulzbacher et al. 2003b), mesothelioma (Langerak et al. 1996) and ovarian carcinoma (Wilczynski et al. 2005), among others. Despite the relatively frequent oncoprotein expression, activating c-kit gene mutations were only observed in GISTs and never reported in other tumour types (Went et al. 2004; Burger et al. 2003; Sihto et al. 2005). Similarly, none or few exon 12 or 18 PDGFR gene mutations have been so far reported in the various series analysed (Liegl et al. 2006; Gomes et al. 2007; Peterson et al. 2006; Carvalho et al. 2005). In contrast to our findings, Rashid et al. recently reported KIT immunohistochemical expression in only 9 out of 76 patients with CUP (Rashid et al. 2005). This highlights the inherent heterogeneity that probably characterises every CUP patient cohort. KIT protein expression did not have prognostic significance for survival in our cohort, in keeping with previous experience published (Debiec-Rychter et al. 2004). We failed to find prognostic significance for PDGFRa protein expression either, at a time when no other published studies, to our current knowledge, have examined immunohistochemical expression of PDGF receptor in CUP patient cohorts. The moderate sample size, a restriction common in all CUP studies published, makes it impossible to rule out a weaker prognostic utility of these molecular markers.

The absence of activating mutations of c-kit as well as the lack of firm evidence that the oncoprotein activity is important for cell transformation and survival in tumour types other than GIST may explain the failure of imatinib therapy in patients with soft tissue sarcomas, small cell lung cancer and germ cell tumours (Kitamura and Hirotab 2004; Heinrich et al. 2003b; Johnson et al. 2003). The prognostic implications of PDGFR gene mutations, if any, is not clarified from the evidence collected so far. Our data highlight absence of established molecular predictors of benefit from tyrosine kinase inhibitors in CUP, adding to accumulating evidence that its molecular pathogenesis is complex and heterogeneous. We recently failed to find epidermal growth factor receptor (EGFR) activating mutations or amplification in 50 CUP tumours (Dova et al. 2007). Although mechanisms other than gain-of-function mutations such as wild-type gene amplification, SCF overproduction, cross-activation by other kinases, gene promoter demethylation may result in KIT/PDGFR activation in solid tumours, our data pinpoint lack of known molecular markers of benefit from imatinib or sunitinib (Sihto et al. 2005). Consequently, further preclinical studies exploring alternative routes of activation of C-KIT and PDGFR should be carried out before embarking on CUP clinical trials that involve administration of TKI compounds.