Targeting EGFR and PI3K pathways in ovarian cancer

Abstract

Background: The epidermal growth factor receptor (EGFR) is expressed in ovarian cancer, but agents targeting this pathway have shown little effect as single agents. This may be due to the presence of alternative pathways, particularly activation of the PI3K/Akt/MTOR pathway.

Methods: We have therefore examined the effect of inhibitors of this pathway (ZSTK474 and sirolimus) in combination with the EGFR inhibitors erlotinib and gefitinib in ovarian cancer primary cell cultures.

Results: The single-agent EGFR inhibitors showed little activity, although some activity was seen with the single-agent PI3K inhibitor, ZSTK474. Combinations of ZSTK474 with EGFR inhibitors showed enhanced activity with some evidence of synergy, whereas sirolimus combinations were less active. The results were not explicable on the basis of PIK3CA mutation or amplification, or PTEN loss, although one tumour with a KRAS mutation showed resistance to EGFR inhibitors. However, there was correlation of the EGFR expression with sensitivity to EGFR and resistance to PI3K active agents, and inverse correlation in the sensitivity of individual tumours to agents active against these pathways, suggesting a mechanism of action for the combination.

Conclusion: Phase I/II clinical trials with these agents should include further pharmacodynamic endpoints and molecular characterisation to identify patients most likely to benefit from this strategy.

Figure 1

Example ovarian tumour primary cell culture results, showing the effect on cell viability as assessed by an ATP-based chemosensitivity assay. There is relatively little activity of the single agents tested. However, the combinations are more active. (A) Tumour 9 shows synergy to erlotinib+ZSTK474 and (B) tumour 1 to gefitinib+sirolimus. In (A) the degree of inhibition in the ATP-TCA reaches 90% at clinically relevant concentrations, whereas in (B) the degree of inhibition is somewhat less impressive. The % inhibition (y-axis) is calculated from the ATP counts from medium only (MO) wells as described in the methods section, whereas the final concentration of each drug is given on the x-axis. Each point represents the mean ATP count derived from triplicate wells, normalised to the mean MO well ATP count as described.

Figure 2

Waterfall plots of sensitivity index (IndexSUM) in ovarian cancer samples for (A) single agents and (B) combinations. A sensitivity index of 0=complete inhibition and 600=no inhibition. The tables below each plot show the gene expression ratio (GER) data for the tumours identified by number below each bar. The darker shading indicates a low gene expression, whereas the lighter shading shows higher GER. Higher expression of EGFR, HER2 or HER4 appears to be associated with increased sensitivity to EGFR inhibitors, whereas the reverse is true for PI3K pathway inhibitors. This relationship is not observed with combinations. Tumour 8 had a KRAS mutation and was resistant to EGFR inhibition, although it remained relatively sensitive to PI3K inhibition.

Figure 3

Comparison of sensitivity by IndexSUM for all nine ovarian tumours for (A) erlotinib and sirolimus; (B) gefitinib and sirolimus; (C) gefitinib and ZSTK474; (D) erlotinib and ZSTK474. There is an inverse correlation for each comparison (r²=0.52, 0.66, 0.51 and 0.57, respectively; all P <0.05). Each symbol represents an individual tumour. IndexSUM calculated as 600−Sum[Inhibition 6.25…200] as previously described (Andreotti et al., 1995), such that complete inhibition=0 and complete resistance=600.

Figure 4

Comparison of single-agent sensitivity (IndexSUM) with gene expression ratio (GER) showing (A) correlation of increasing gefitinib activity with increased EGFR expression (Pearson's correlation, r²=0.46, P <0.05); (B) decreasing ZSTK474 activity with increased EGFR expression (r²=0.48, P <0.04); (C) increasing gefitinib activity with increased HER2 expression (r²=0.45, P <0.05); (D) decreasing ZSTK474 activity with increased HER2 expression (r²=0.43, NS); (E) decreasing gefinitib activity with increased IGF1 expression (r²=0.52, P <0.03); (F) increasing ZSTK474 activity with increased IGF1 expression (r²=0.23, NS); (G) decreasing gefitinib activity with increased IGF2R expression (r²=0.50, P <0.04); (H) increasing ZSTK474 activity with increased IGF2R expression (r²=0.36, NS). While the numbers are small and the results are of borderline statistical significance, there is clearly an inverse relationship between the sensitivity to EGFR and PI3K inhibitors with expression of key genes in these pathways.

Figure 5

Human epidermal growth factor receptor and PI3K pathways in ovarian cancer.