Search

Knowledge of the molecular background of functional magnetic resonance (MR) images is required to fully exploit their potential in cancer management. We explored the prognostic impact of dynamic contrast-enhanced MR imaging (DCE-MRI) parameters in cervical cancer combined with global gene expression data to reveal their underlying molecular phenotype and construct a representative gene signature for the relevant parameter. On the basis of 78 patients with cervical cancer subjected to curative chemoradiotherapy, we identified the prognostic DCE-MRI parameter ABrix by pharmacokinetic analysis of pretreatment images based on the Brix model, in which tumors with low ABrix appeared to be most aggressive. Gene set analysis of 46 tumors with pairwise DCE-MRI and gene expression data showed a significant correlation between ABrix and the hypoxia gene sets, whereas gene sets related to other tumor phenotypes were not significant. Hypoxia gene sets specific for cervical cancer created in cell culture experiments, including both targets of the hypoxia inducible factor (HIF1α) and the unfolded protein response, were the most significant. In the remaining 32 tumors, low ABrix was associated with upregulation of HIF1α protein expression, as assessed by immunohistochemistry, consistent with increased hypoxia. On the basis of the hypoxia gene sets, a signature of 31 genes that were upregulated in tumors with low ABrix was constructed. This DCE-MRI hypoxia gene signature showed prognostic impact in an independent validation cohort of 109 patients. Our findings reveal the molecular basis of an aggressive hypoxic phenotype and suggest the use of DCE-MRI to noninvasively identify patients with hypoxia-related chemoradioresistance. Cancer Res; 72(20); 5285–95. ©2012 AACR.

Purpose We have applied the sensitive and specific in situ proximity ligation assay (PLA) to characterize Tyr1068 phosphorylation of the epidermal growth factor receptor (EGFR) in cervical cancer in relation to the protein level and gene dosage. Materials and methods Pretreatment tumor biopsies from 178 patients were analyzed. EGFR protein level was determined by immunohistochemistry, and Tyr1068 phosphorylation was detected with PLA in 97 EGFR positive tumors. EGFR gene dosage was derived from array comparative genomic hybridization of 86 cases. Results EGFR was expressed in most tumors, whereas phosphorylation was seen in about half of the EGFR positive ones. A correlation was found between the expression of EGFR and phosphorylated EGFR ( p =0.016, membrane; p =0.012, cytoplasm). However, tumor regions with high protein level without phosphorylation were occasionally seen and the percentage of EGFR positive cells was higher than the phosphorylated percentage ( p p =0.014) and cytoplasm ( p =0.002) was seen in 11 tumors with gain of EGFR. The protein level was not correlated with gene dosage. Conclusion In contrast to gain of the EGFR chromosomal region, high EGFR protein level may not necessarily indicate Tyr1068 phosphorylation and thereby receptor activation in cervical cancer.

Purpose: We compared the prognostic significance of ectodomain isoforms of the epidermal growth factor receptor (EGFR), which lack the tyrosine kinase (TK) domain, with that of the full-length receptor and its autophosphorylation status in cervical cancers treated with conventional chemoradiotherapy. Experimental Design: Expression of EGFR isoforms was assessed by immunohistochemistry in a prospectively collected cohort of 178 patients with squamous cell cervical carcinoma, and their detection was confirmed with Western blotting and reverse transcriptase PCR. A proximity ligation immunohistochemistry assay was used to assess EGFR-specific autophosphorylation. Pathways associated with the expression of ectodomain isoforms were studied by gene expression analysis with Illumina beadarrays in 110 patients and validated in an independent cohort of 41 patients. Results: Membranous expression of ectodomain isoforms alone, without the coexpression of the full-length receptor, showed correlations to poor clinical outcome that were highly significant for lymph node–negative patients (locoregional control, P = 0.0002; progression-free survival, P < 0.0001; disease-specific survival, P = 0.005 in the log-rank test) and independent of clinical variables. The ectodomain isoforms were primarily 60-kD products of alternative EGFR transcripts. Their membranous expression correlated with transcriptional regulation of oncogenic pathways including activation of MYC and MAX, which was significantly associated with poor outcome. This aggressive phenotype of ectodomain EGFR expressing tumors was confirmed in the independent cohort. Neither total nor full-length EGFR protein level, or autophosphorylation status, showed prognostic significance. Conclusion: Membranous expression of ectodomain EGFR isoforms, and not TK activation, predicts poor outcome after chemoradiotherapy for patients with lymph node–negative cervical cancer. Clin Cancer Res; 17(16); 5501–12. ©2011 AACR.

Tumor hypoxia is an adverse factor in cervical cancer that would be valuable to implement in the clinical decision-making. In a previous study, we identified a hypoxia gene signature associated with aggressiveness in locally advanced cervical cancer. It consisted of 31 genes, and was predictive of progression free survival in an independent cohort treated with chemoradiotherapy. The signature could be useful in the development of a prognostic test to select patients at risk of failure. The aim of the current study was to obtain a deeper understanding of the pathways reflected in the signature. This could provide an increased insight into the hypoxia-related aggressiveness of cervical cancer and be useful to fully exploit the clinical potential of the signature. The gene signature included a total of 12 genes which are known direct targets of HIF1, and 4 genes involved in the unfolded protein response (UPR), including 3 of the 12 HIF1-targets. To obtain a metric for comparison with other relevant factors, a score was calculated from the expression level of the 31 genes in 154 patients. This hypoxia score correlated significantly with HIF1α protein expression (p Cox survival analysis of the hypoxia score was performed on the 154 patients. When stratifying the score on lymph node status, it was predictive of poor outcome only for the patients without metastases to the lymph nodes (89 patients) (p=0.001). When assessing the prognostic value of HIF1α protein expression in the whole patient group, it was not correlated with survival. However, in patients with no lymph node metastases, high levels of HIF1α were found to be a prognostic indicator of poor outcome (p=0.043). Multivariate cox survival analysis was performed on the lymph node negative patients, including the hypoxia score, HIF1α, and clinical parameters. Only the hypoxia score and FIGO stage were found to be independently related to patient survival. In conclusion, it appears that HIF1α is an important mediator of the aggressive phenotype associated with our hypoxia gene signature. However, since the hypoxia score emerged as a significantly better prognostic factor than HIF1α, other signaling pathways may also contribute. Furthermore, this hypoxic phenotype seems to be associated with aggressiveness particularly in patients with no metastases to the lymph nodes. Citation Format: Cathinka Halle, Eva-Katrine Aarnes, Ruth Holm, Gunnar B. Kristensen, Heidi Lyng. HIF1α signaling contributes to an aggressive hypoxic phenotype in cervical cancer. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 4037. doi:10.1158/1538-7445.AM2013-4037