• No se han encontrado resultados

CAPÍTULO III. EFECTO DE LA TEMPERATURA, HUMEDAD RELATIVA Y LUZ

3.8 DISCUSIÓN

To understand the clinical implications of PKN kinase expression in breast cancer patients, the expression levels of PKN kinases were first investigated. Analysis of PKN kinase expression levels across all available breast cancer datasets showed no significant differences between the three PKN kinase isoforms in normal individuals and patients with different subtypes of breast cancer (Fig. 2). This comparison includes between patients with luminal A, luminal B, Her 2, basal and TNBC. The data was also divided to compare patients with TNBC and non-TNBC. Overall, there were no differences in PKN kinase expression across different breast cancer types nor in comparison to normal samples.

67

Figure 2. PKN kinase expression profile across different breast cancer subtypes

These plots show the expression values (log2 expression) of A) PKN1, B) PKN2 and C) PKN3 in different molecular subtypes of breast cancer and normal samples. The number of samples in each subtype is shown in the box below the plots. Each sample was assigned to a PAM50 molecular subtype based on the expression of the intrinsic gene list (Parker et al. 2009) - LumA, LumB, Basal- like, HER2-enriched, normal-like. The TN and non-TN definitions are also based on molecular classifications. The ER, PR and HER2 receptor status of each sample were defined by implementing functions within the MCLUST R library. The MCLUST algorithm was set to calculate the Bayesian Information Criterion (BIC) for a 2-component Gaussian distribution model. TN samples were isolated and the levels of gene expression compared to samples allocated to the non-TN group.

3.1.1.2 Analysing the effect of PKN kinase expression on breast cancer survival

Despite the lack of association between PKN kinase expression levels across different subtypes of breast cancer, the correlation between PKN kinase expression and survival was investigated using an online Kaplan-Meyer analysis tool (Györffy et al. 2010). The correlation between PKN kinase expression and relapse-free survival was analysed, as this had the best number of patients for breast cancer. This analysis was automated for selecting the best performing threshold for defining low and high PKN kinase expression and involved only the best probe sets selected by JetSet algorithms. As multiple probes can detect a gene, Li et al developed this JetSet scoring system to assign one probe for a given

68 gene that scores best in three criteria: high specificity, ability to detect splice variants and unaffected by transcript degradation (Li et al. 2011).

The PAM50 single sample predictor (Parker et al. 2009) as applied to the breast cancer data to assign each sample to one of the five molecular subtypes (luminal A, luminal B, Her2+, normal breast-like or basal). Similarly, triple negative samples were also determined, as described by Lehmann et al, on the empirical expression distributions of ER, PR and Her2 (Lehmann, Bauer, Chen, Sanders, Chakravarthy, and JA 2011).

The expression levels of PKN kinase expression was plotted for each group (Fraley et al. 2012). Systematic comparisons were made between different combinations of PKN kinase isoforms and breast cancer classifications. The most intriguing comparisons are those involving comparisons between TNBC, non-TNBC, basal and mesenchymal-like breast cancer, due to the effect of PKN2 knockout in our inducible MEFs and the destructive effect on the mesodermal compartment of mice embryo upon systemic PKN2 deletion (Quétier et al. 2016).

Amongst all breast cancer datasets, high PKN1 expression has a significant (P<0.001) but modestly better RFS than patients with low PKN1 levels (Fig. 3A). This survival advantage is more apparent in patients with basal and ER-negative breast cancer (Fig. 3B, C), with no significant differences in ER-positive patients (Fig. 3D). Similar patterns were observed for the PKN3 isoform. Elevated PKN3 expression is associated with better RFS across all breast cancer data sets (Fig.5A). In basal, ER-negative and ER-positive breast cancer low PKN3 expression has a moderate but non-significant positive effect on RFS (Fig. 5B, C, D).

In contrast, PKN2 expression had no impact on RFS across all breast cancer datasets (Fig. 4A) and in basal breast cancer (Fig. 4B). However, this global analysis masks significant differences in behaviour between disease subtypes. There is significantly improved RFS in basal (P=0.0067) and ER-negative (P=0.0028) breast cancer patients with low PKN2

69 expression (Fig. 4C), whereas the converse is true for ER-positive patients (Fig.4D). As dependence on PKN2 in TNBC for cell survival has recently been demonstrated in several studies, the RFS for patients with low and high PKN2 expression was further investigated. Interestingly, analysis of overall TNBC data showed no difference in RFS between low and high PKN2-expressing patients (Fig.4E). We next analysed distinct subsets of TNBC using the Pietenpol classifications (Chen et al. 2012).

Figure 3. PKN1 expression analysis on breast cancer survival

Figure showing probability of relapse free survival (RFS) in patients with low and high PKN1 expression, generated using an online tool (Györffy et al. 2010). PKN high and low expression threshold was selected at best cut-off for best performing threshold. PKN1 (affymetrix ID:

202161_at) was analysed for its prognostic relevance in A. all (n=3951), B. intrinsic basal (n=618), C. ER-negative (n=801) and D. ER-positive (n=2061) breast cancer.

70

Figure 4. PKN2 expression analysis of breast cancer survival

Figure showing probability of relapse free survival (RFS) in patients with low and high PKN2 expression, generated using an online tool (Györffy et al. 2010). High and low expression threshold was selected at best cut-off for best performing threshold. Out of three PKN2 probes, the one classed as the best probe by Jetset algorithm was chosen (affymetrix ID: 212628_at). A. All types of breast cancers, B. Intrinsic basal subtype C. ER-negative, D. ER-positive, E. TNBC – ER-negative, PR- negative and HER2-negative, and the Pietenpol subtypes: F. Basal-like 1 (BL1), G. Basal-like 2 (BL2) and H. Mesenchymal-like (ML). Number of patients at risk shown below the plots.

71 Those patients with TNBC classified as basal-like (BL1, BL2) or mesenchymal-like (ML), and with low PKN2 expression, have significantly improved RFS survival (Fig.4F, G, H).

However, the low number of patients involved in this analysis must be considered (detailed in figure 4). Additionally, as mentioned above, intrinsic basal breast cancer classification shows the same pattern of behaviour (Fig. 4C). Together, these data suggest that high PKN2 expression is associated with poor survival in a subset of basal-like and mesenchymal-like breast cancer patients, in contrast to the PKN1 and PKN3 isoforms. This is particularly interesting as we have demonstrated that systemic deletion of PKN2 in mice embryo resulted in abnormal development of the mesodermal compartment (Quétier et al. 2016). Furthermore, MEFs from these embryos show dependence on PKN2 for proliferation in vitro.

Figure 5. PKN3 breast cancer survival analysis.

Figure showing probability of relapse free survival (RFS) in patients with low and high PKN3 expression considering A. all breast cancer types, B. basal, C. ER-negative and D. ER-positive breast cancers. High and low expression threshold was selected at best cut-off for best performing threshold. Amongst three probes for PKN3, the one classed as the best probe set by Jetset algorithms was chosen (affymetrix ID: 226299_at). Plot generated using an online tool (Györffy et al. 2010). Number of patients at risk for each time point is indicated in numbers under the graphs.

72 In the following section, the effect of genetic and pharmacological inhibition of PKN kinases on viability of breast cancer cell lines in vitro will be discussed.

Documento similar