Understanding cancer biology is fundamental for specific target selection in developing drugs for modern oncology. Our study focused on the EGFR-PI3K cascade in OC, which demonstrated a high frequency of dysregulated factors. In addition to the frequently observed EGFRwt and GCN variations, EGFRvIII and the absence of PTEN were also notable. However, PIK3CA gene aberrations were rare in OC. Together, the pathway anomalies led to activated pAKT, which impacted patient survival. Moreover, our study reinforced the indispensable role of EGFR in OC, with EGFRvIII in particular being the major contributory element that influenced patient survival and outcome. To our knowledge, we are the first to report survival differences by EGFRvIII classification in OC.
The truncated variant EGFRvIII draws attention to the constitutive signaling activity, which is independent of ligand binding . The actual mechanism of the production of this mutant receptor remains unknown. In a HNC study conducted by Sok, et al., EGFRvIII was exclusively expressed in coexistant with EGFRwt . The different observation in our study may therefore suggest a distinguished feature specified to OC comparing to other HNC diseases, or rather a unique feature in betel-nut prevalent area. In fact, a breast cancer study has indicated that EGFRvIII expression is not a consequence of EGFR locus rearrangement or amplification but is rather due to alterative splicing events . This notion was supported by our study and other brain tumor studies, as the appearance of EGFRvIII was not necessarily co-existed or co-related with EGFRwt protein expression or gene amplification [30, 31]. In these cancer diseases, the probable indispensable roles of the mutant protein in tumorigenesis are therefore to be expected. Unlike in brain tumors, in which the impact of EGFRvIII on survival is known [32–34], its function in OC and other HNC are ambiguous, though frequently detected. In addition to our study, previous studies have failed to prove the determinant role of the truncated protein for survival outcome [27, 35]. This failure may be due to the preceding reason and due to the different methods utilized in the investigations. In fact, accurately detecting the mutant protein is challenging because of interference from EGFRwt and the nearby non-tumor tissue. Different from the two studies utilizing RT-PCR, we used IHC staining because of its direct indication of the signal distribution. In addition, IHC possesses a specificity that is comparable to PCR over laser capture microdissected tissue . The application of this method led us to demonstrate the crucial role of EGFRvIII. This result was not unexpected because of the association with disease stage, which has also been shown by Keller and colleagues .
In our studies, we found that the major subcellular distribution area of the variant protein was in the cytoplasm, whereas EGFRwt expression was detected at both the membrane and in the cytoplasm. This result was consistent with those observed in the human glioblastoma samples and in the breast cancer samples [36, 38]. The definite causes leading to the distinct phenomenon are not yet clear. It has been suggested that trivial differences in signaling and functioning distinguish EGFRvIII and its wild-type counterpart. One possible example is the defective downregulation of the truncated receptor due to the hypophosphorylation of residue Y1045, leading to escape from ubiquitination by c-Cbl . Another example was the demonstration of a large intracellular pool of EGFRvIII functioning with Src to reduce the glucose dependency when relocalizing to the mitochondria . Taken together, these studies along with our data support the role of cytoplasmic EGFRvIII in tumorigenesis. Therefore, further research to elucidate the mechanism of intracellular redistribution is imperative.
Though positive correlation was shown between the EGFR expression and the GCN amplification, we found that they were not completely overlapped. In recent reports, discrepancies have been noted between these two factors in OC [25, 41]. The exact mechanisms in the samples that had controversies within the two factors remain unclear. Regarding the protein producing process, the accurate regulatory mechanism of transcription and translation from the EGFR coding gene is not fully understood. Complicated modulation against EGFR gene transcription was noted, for example, that the regulation mechanism depends on the length of a CA repeat in intron 1 . In addition, Wheeler and colleagues found despite of correlation with EGFR gene amplification status and the protein levels, the level of EGFR mRNA was not associated to both factors . This indicated post-translational regulation was at least in part crucial to the protein expression. Therefore, it was not surprising that some samples have positive EGFR protein expression without GCN amplification, and vice versa.
Recent investigations have emphasized the role of PI3K in HNC . Our study of the PIK3CA gene, however, indicated that GCN amplification or hotspot point mutations were rare. This finding was in contrast to several reports that highlighted the frequent aberrations of the gene itself. Examples of such reports include a study that found that 34.8% of OC samples exhibit high GCN amplifications  and another showing that 11% of HNC samples carry hotspot point mutations . Nevertheless, contradictory results were also noted in the OC study conducted by Kozaki and colleagues . In the 108 OC primary tumor samples analyzed, 16.7% of those showed altered copy number for a 1.3 ~ 3.4-fold increase, and four samples were detected of bearing hotspot point mutations. Taken together with our study, these data suggest that GCN variations of PIK3CA encoding areas were not as significant as EGFR alterations. Finally, the possibility that the discrepancy in these observations is due to population differences should be further examined.
The pAKT expression levels in our samples were high and correlated with patient survival. As the preferred signaling protein, it was unexpected that its associations with EGFRvIII or EGFRwt (data not shown) expression were not significant . One reasonable potential explanation is the interruption of the other regulatory factors as compensation. For instance, PTEN is known to be a negative modulator of this pathway. Because AKT, and likely STAT3 , are oppositely regulated by EGFRvIII and PTEN, its activation was thus speculated as a result of feedback reaction. In addition, further studies have uncovered alternative routes that may be responsible for tumorigenesis by the truncated protein, such as through CXCR4 and cyclooxygenase-2 [48, 49]. These proteins bypass the cascade, attenuating the dependency of the AKT pathway without dispelling EGFRvIII tumorigenic impact.