Lampalizumab Phase 2 (MAHALO).

Study characteristics
Methods	Study design: parallel‐group Randomisation: web‐based, blocked and stratified by baseline GA lesion size (dichotomised at 10 mm2) Masking: participants, outcomes assessors Intention‐to‐treat: yes Statistical methods:  For the primary analyses of GA and BCVA change, marginal means were derived from ANOVA stratified by baseline GA lesion size or BCVA, respectively For the secondary analysis of GA change, marginal means were derived from a mixed model for repeated measures (MMRM) adjusted for baseline GA area as a continuous variable, time, treatment, time‐by‐treatment interaction, and treatment‐by‐CFI status, and baseline GA category (≥ 10 mm2 versus < 10 mm2) Missing data: missing values of BCVA and GA lesion size change were handled by last observation carried forward (LOCF) for primary analyses. Missing values of GA lesion size change were additionally explored through multiple random imputation using Markov Chain Monte Carlo (MCMC) sampling in sensitivity analyses.
Participants	Total number analysed: 129 Setting: multicentre Diagnostic tool: FAF, near‐infrared imaging, and digital colour fundus photograph  Age: 60 to 89 years (mean 79 years) Key ocular eligibility criteria (study eye):  BCVA of 20/50 to 20/400 (Snellen equivalent) using ETDRS charts GA lesion size of ≥ 2.5 and ≤ 17.5 mm2 GA lesion residing completely within imaging field Presence of perilesional hyperautofluorescence of either banded or diffuse patterns No previous retinal surgery, or other therapeutic procedures for AMD No subfoveal focal laser photocoagulation No previous IVT therapy Key ocular eligibility criteria (both eyes): GA secondary to AMD in both eyes No evidence of prior or active CNV in either eye No select concurrent ocular and systemic conditions
Interventions	Agent: lampalizumab 10 mg/100 μL Route of delivery: IVT  Frequency of delivery: every 4 or 8 weeks Duration of treatment: up to 18 months Controls: sham injection
Outcomes	Primary outcome measures: Growth rate of GA lesion area, as measured by FAF (time frame: 18 months) Secondary outcome measures: Mean change in BCVA as assessed by ETDRS chart (time frame: 18 months)
Notes	Ocular history of exudative AMD in the fellow eye is exclusionary. Patients who developed exudative AMD during the study were not withdrawn.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Web‐based randomisation. However, if both eyes met the inclusion criteria, it is unclear which eye was designated as the study eye.
Allocation concealment (selection bias)	Low risk	Blocked randomisation based only on GA lesion size but across multiple sites makes it unlikely that the investigator/participant would know or influence the intervention group before an eligible participant entered the study.
Masking of outcome assessment (detection bias) BCVA	Low risk	Masking of outcome assessment ensured, and unlikely that the masking could have been broken.
Masking of outcome assessment (detection bias) GA lesion size	Low risk	Masking of outcome assessment ensured, and unlikely that the masking could have been broken.
Masking of outcome assessment (detection bias) Adverse event reporting	Low risk	Masking of outcome assessment ensured, and unlikely that the masking could have been broken.
Masking of participants and personnel (performance bias) BCVA	Low risk	Masking of participants and key study personnel ensured, and unlikely that the masking could have been broken.
Masking of participants and personnel (performance bias) GA lesion size	Low risk	Masking of participants and key study personnel ensured, and unlikely that the masking could have been broken.
Masking of participants and personnel (performance bias) Adverse event reporting	Low risk	Masking of participants and key study personnel ensured, and unlikely that the masking could have been broken.
Selective reporting (reporting bias)	Low risk	The protocol for the extension study is available, and all the study’s pre‐specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre‐specified way.
Incomplete outcome data (attrition bias) BCVA	Unclear risk	Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups. As AMD outcomes tend to deteriorate with time, the use of last observation carried forward (LOCF) procedures for imputation is inappropriate as this could have biased the effect estimate in favour of the drug. However, it is unclear whether the time interval is sufficient for this to affect BCVA.
Incomplete outcome data (attrition bias) GA lesion size	Unclear risk	Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups. Various sensitivity analyses and imputation methods to assess the impact of missing data showed that these did not significantly influence the study outcomes. Although derived using inappropriate last observation carried forward (LOCF) procedures, the GA lesion size outcomes are supported by linear mixed‐effect models.
Incomplete outcome data (attrition bias) Adverse event reporting	Low risk	Missing outcome data were balanced in numbers across intervention and sham groups, with similar reasons for missing data across groups. All adverse events were reported as participants were withdrawn from the study.
Other bias	Low risk	We did not identify other potential threats to validity.