OBJECTIVES: A propensity score matching analysis (PSA) was conducted to compare the efficacy of polihexanide (PHMB) 0.08% vs currently used off-label therapies in patients with acanthamoeba keratitis (AK).

METHODS: The PSA was based on the individual patient data (IPD) for PHMB 0.08% from the phase 3 trial 043/SI (NCT03274895) and IPD data from Papa et al. 2020, a retrospective study in AK, for current therapies. For the retrospective study, two populations were analysed: 1) the whole study population, referred to as ‘any initial pharmacological treatment’; and 2) patients treated with PHMB 0.02% plus a diamidine 0.1%. The endpoint of interest was the difference in the clinical resolution rate (CRR) with no surgery within 12 months of treatment. Discontinuations from baseline therapy were considered as ‘failure’. A PSA with overlap weighting was used. IPD from the two studies were reweighted to balance study populations for key prognostic factors and/or treatment effect modifiers: age; gender; AK disease stage; prior use of corticosteroids; prior use of antivirals; delay in starting treatment from diagnosis. The overlap weights were normalised to account for the study sample size. The calculated estimate was based on the average treatment effect. CRR was assessed using logistic regression methods to estimate the absolute difference between treatments and the corresponding 95% confidence interval (CI).

RESULTS: The absolute difference in CRR rate was 36.4% (95%CI=24.9%-47.9%; p<0.001) for PHMB 0.08% (n=66) compared with any initial treatment (n=227) and 24.2% (11.3%, 37.1%; p<0.001) for PHMB 0.08% compared with PHMB 0.02% plus a diamidine (n=111), after weighting. The baseline characteristics and effective sample size in the weighted populations suggested that the indirect treatment comparison had successfully aligned the analysis populations.

CONCLUSIONS: The PSA suggested an improved efficacy with PHMB 0.08% compared with currently used therapies in achieving clinical resolution with no surgery in AK.