Direct, indirect and total effect of HIV coinfection on the risk of non‐liver‐related cancer in hepatitis C virus‐infected patients treated by direct‐acting antivirals: a mediation analysis

Abstract

Objectives HIV‐coinfected patients experience higher incidences of non‐liver‐related cancers than HCV‐monoinfected patients. Chronic inflammation, immunosuppression, but also higher tobacco or alcohol consumption and metabolic dysregulation could explain this higher risk. We aimed to estimate the direct, indirect and total effects of HIV coinfection on the risk of non‐liver‐related cancers in HCV participants treated with direct‐acting antivirals (DAAs). Methods Up to four HCV‐monoinfected participants from the ANRS CO22 HEPATHER cohort were matched by age and sex to HIV/HCV‐coinfected participants from the ANRS CO13 HEPAVIH cohort. Participants were followed from DAA initiation until the occurrence of a non‐liver‐related cancer. Counterfactual mediation analysis was carried out to estimate the direct (chronic inflammation and immunosuppression), indirect (tobacco and alcohol consumption and metabolic syndrome) and total effect of HIV coinfection on the risk of non‐liver‐related cancers. Results 548 HIV/HCV‐coinfected and 2016 monoinfected participants were included. Overall, HIV coinfection was associated with a 3.7‐fold [95% confidence interval (CI): 1.7–7.0] higher risk of non‐liver‐related cancers in HCV participants. This increased risk was explained by significant direct effect [hazard ratio (HR) = 3.4, 95% CI: 1.7–6.6] but not indirect effect (HR = 1.1, 95% CI: 0.8–1.5) of HIV coinfection. Conclusions In HCV participants treated with DAAs, the direct effect of HIV coinfection, reflecting chronic inflammation and immunosuppression, was associated with a 3.7‐fold higher risk of non‐liver‐related cancer. By contrast, the indirect effect of HIV coinfection, reflecting higher tobacco and alcohol consumption and metabolic dysregulation, was not significantly associated with the risk of non‐liver‐related cancers.