Prognostic and predictive significance of plasma HGF and IL-8 in a phase III trial of chemoradiation with or without tirapazamine in locoregionally advanced head and neck cancer

Abstract

Purpose: Hepatocyte growth factor (HGF) is a hypoxia-induced secreted protein that binds to cMet and regulates interleukin (IL)-8 expression. We evaluated the role of circulating HGF and IL-8 as prognostic and predictive factors for efficacy of tirapazamine (TPZ), a hypoxic cell cytotoxin.

Experimental design: Patients with stages III to IV head and neck cancer were randomized to receive radiotherapy with cisplatin (CIS) or CIS plus TPZ (TPZ/CIS). Eligibility for the substudy included plasma sample availability for HGF and IL-8 assay by ELISA and no major radiation deviations (N = 498). Analyses included adjustment for major prognostic factors. p16(INK4A) staining (human papillomavirus surrogate) was carried out on available tumors. Thirty-nine patients had hypoxia imaging with (18)F-fluoroazomycin arabinoside ((18)FAZA)-positron emission tomography.

Results: Elevated IL-8 level was associated with worse overall survival (OS) irrespective of treatment. There was an interaction between HGF and treatment arm (P = 0.053); elevated HGF was associated with worse OS in the control but not in the TPZ/CIS arm. Similar trends were observed in analyses restricted to p16(INK4A)-negative patients. Four subgroups defined by high and low HGF/IL-8 levels were examined for TPZ effect; the test for interaction with arm was P = 0.099. TPZ/CIS seemed to be beneficial for patients with high HGF and IL-8 but adverse for low HGF and high IL-8. Only HGF correlated with (18)FAZA tumor standard uptake value.

Conclusions: IL-8 is an independent prognostic factor irrespective of treatment. There is an interaction between HGF and treatment arm. Certain subgroups based on IL-8/HGF levels seemed to do better with TPZ/CIS while others did worse, highlighting the complexity of hypoxia targeting in unselected patients.

Figure 1

Overall survival for 498 patients by pre-treatment HGF levels and treatment arm

Figure 2

Overall survival for 498 patients by pre-treatment IL8 levels and treatment arm

Figure 3

A. Interaction plot of relative risk for death (relative hazard rates, RHR, log scale) by treatment arm and IL8/HGF groups, adjusting for prognostic factors. The graph shows the RHRs rates for each of the eight combinations of treatment arm and IL8 and HGF levels, where a line joins the RHRs of CIS and TPZ/CIS arms for each IL8/HGF combination. No interaction would be indicated by all lines being parallel. The slope of each line is the log of the hazard ratio for arm (TPZ/CIS: CIS) for the given IL8/HGF subgroup. The figure suggests that OS is worse with TPZ/CIS for high IL8/low HGF patients and better for high IL8/high HGF patients, compared with CIS. B. Overall survival by treatment arm and HGF levels within the high IL8 group

Figure 3

A. Interaction plot of relative risk for death (relative hazard rates, RHR, log scale) by treatment arm and IL8/HGF groups, adjusting for prognostic factors. The graph shows the RHRs rates for each of the eight combinations of treatment arm and IL8 and HGF levels, where a line joins the RHRs of CIS and TPZ/CIS arms for each IL8/HGF combination. No interaction would be indicated by all lines being parallel. The slope of each line is the log of the hazard ratio for arm (TPZ/CIS: CIS) for the given IL8/HGF subgroup. The figure suggests that OS is worse with TPZ/CIS for high IL8/low HGF patients and better for high IL8/high HGF patients, compared with CIS. B. Overall survival by treatment arm and HGF levels within the high IL8 group