Treatment results of locally advanced cervical cancer are suboptimal despite the recent success of first line therapy with concurrent chemoradiation [31, 32]; or with and cisplatin plus topotecan for recurrent and metastatic . Nevertheless, it is necessary to evaluate newer forms of therapy. Preclinical and clinical development of newer agents targeting molecular defects present in tumors  open the way for the identification of molecular targets in cervical cancer.
The availability of the tyrosin kinase inhibitor imatinib which not only targets bcr/abl but also c-kit and PDGFR has led to investigate the participation of these two receptors in several solid malignancies, particularly in small cell lung cancer and prostate carcinoma which over-express the c-kit receptor. However, early results using imatinib as a single agent have not been encouraging [35–37] which suggests that the sole presence of the putative target may not be sufficient to achieve tumor responses, besides, it is necessary to demonstrate that the pathway is functional and of relevance for one or more characteristics of the malignant phenotype in the tumor to be targeted.
So far, the role of the PDGFR pathway for the development of cervical cancer has been only studied in in vitro models, derived from the observation of the interaction between the E5 papilloma virus protein with the PDGFR [19, 20]. There is a single study showing that the four primary cervical cancers analyzed by immunohistochemistry expressed the PDGFRβ . In several tumors, where the participation of PDGF family has been demonstrated, a simultaneous expression of the receptors and the ligands seems to be important, suggesting an autocrine stimulation mechanism . In the present study we evaluated a larger sample of primary tumors for the expression of the ligands and receptors of PDGF at both the malignant and stroma components of the biopsy. The expression of these growth factor family members in stroma is important as they are regulators of mesenchymal cell proliferation and migration during development, and therefore could play important roles in stromal fibroblast recruitment and tumor progression [39, 40].
Our results indicate that around half of the tumors express the PDGFRα, PDGFRβ and PDGFA, in both tumor and stroma as shown in Figure 1. We found no expression of the B ligand in tumor samples, which was surprising as 5 out of 8 cervical cancer cell lines expressed this ligand. In addition, both ligands have been found expressed in a number of other tumor types . Whether this feature is unique to primary cervical cancer tumors merits further study.
Interestingly 78% of the primary tumors express either the alpha or beta receptor. This elevated frequency of expression and co-expression highly suggest autocrine and paracrine functioning in cervical cancer tumors. This high frequency of expression and co-expression also occurred in the cervical cancer cell lines studied. It has been shown that the sole expression of PDGF receptors correlates with known adverse prognostic factors such as axillary lymph node metastases in breast cancer . However, in this study we found no correlation between any combination of PDGF members expression with neither clinical characteristics of patients nor survival (data not shown). This observation does not exclude the possible prognostic significance of these receptors in cervical cancer. We also studied the mutational status of the PDGFRα, based on the identification of activating gene mutations within a subset of GISTs that lacked KIT gene mutation, and where constitutive phosphorylation of PDGFRα was observed, and the corresponding PDGFRα isoforms demonstrated ligand-independent kinase activity . These observations have profound implications for the treatment of solid tumors with imatinib, as some of these mutations confer either sensitivity or resistance to this tyrosine kinase inhibitor . This can be explained on the lack of differences in the activation of downstream signaling intermediates between KIT-mutant and PDGFRα-mutant tumors, suggesting that mutant PDGFRα provides oncogenic signals that parallel those of mutant KIT .
Our sequence results indicate that the cervical cancer cell lines and primary tumors analyzed showed a number of intronic and exonic variations, most of them previously unreported, (Tables 2 and 3). The mutation G>A in codon 571 leading to a Glu>Lys substitution found in three of the cell lines has not been reported; however, the deletion/substitution SPDGHE566-571R (already reported) involves codon 571 which corresponds to the juxtamembrane domain of the protein. The significance of this change in regard to imatinib sensitivity is unknown  but this deletion increases PDGFRα activation in the presence of PDGFA  suggesting that the mutation that we found, could affect the PDGFRα functioning. In an attempt to study the possible effect of this mutation on the protein, we used the simulation program PolyPhen  to predict whether this mutation, found in the cell lines, is likely to have biological meaning. The results indicate that the G>A at codon 571 of exon 12 is "bening". It is clear however, that this is only a theoretical approximation hence; the meaning of changes found must be investigated in experimental systems . Besides, most of the intronic changes that we found are located in the proximity of the exon boundaries, and their possible meaning has to be studied. Regarding the samples from patients, we found the silent mutation at codon 824 already reported, in 10 out of 17 patients (59%). In the normal cervical samples, we found that 6 out of 11 presented the polymorphism at codon 824 (55%), indicating that this polymorphism is very common in our population. In exon 12, two samples presented the intronic change C>A IVS12+22 already reported [rs2307051] with no known consequence.
As stated above, the sole expression of the PDGFR or ligand is not an indication that the pathway is functional and that regulates the growth of the expressing tumor. Hence, we wanted to evaluate whether CaLo cell line which express the receptor alpha and both ligands is responsive to the growth factor and its inhibitor. Our results clearly demonstrate that while CaLo cells exhibit growth stimulation by PDGFBB, imatinib has on their own a small but significant growth inhibitory effect in serum-deprived cells, and that imatinib also partly blocks the PDGFB-induced growth stimulation. These effects on growth are likely due to activation of the PDGFR pathway as the results of western blots employing a primary antibody specific for the activated form of the receptor alpha indicate phosphorylation after the stimulation with PDGFBB, and that is back to unstimulated levels upon exposure to imatinib. It is noteworthy however, that in untreated cells despite being inhibited in their growth by imatinib, the basal level of phosphorylation of the receptor was unchanged after imatinib treatment, which may suggest that under the experimental conditions tested, the method has not enough sensitivity to detect a certain degree of phosphorylation inhibition. Even that we did not evaluate the activation of known downstream phosphorylation targets of the PDGFR pathway, the results demonstrate that as observed in other tumors such as osteosarcoma  and ovarian carcinoma , the blocking of this pathway could have therapeutical implications. Nevertheless, further functional characterization of the pathways either Src-PI3K-ERK, JNK1-p21 or other , known to be activated by PDGFR stimulation in other systems, is needed in order to optimize the potential use of PDGFR inhibitors in this malignancy. Likewise, is desirable to increase the number of primary tumors tested for activating mutations of the PDGFR to rule out the existence of activating mutations at this receptor that could be specifically blocked by PDGFR inhibitors. In addition, it remains to be explored whether the in in vivo blocking the receptor affects the interstitial pressure and would facilitate drug penetration into tumors.