BACKGROUNDThe modified docetaxel, cisplatin, and fluorouracil (mDCF) regimen has shown efficacy and safety as first-line treatment for advanced squamous cell carcinoma of the anus, making it a standard regimen. Inhibitors of programmed cell death protein 1 and its ligand, such as pembrolizumab, nivolumab, retifanlimab, avelumab, and atezolizumab, have shown some antitumour activity as monotherapy in advanced squamous cell carcinoma of the anus that is refractory to chemotherapy. This phase 2 study evaluated the combination of mDCF and atezolizumab as first-line treatment in advanced squamous cell carcinoma of the anus.METHODSIn this randomised, open-label, non-comparative, phase 2 study, participants from 21 centres (academic, private, and community hospitals and cancer research centres) across France with chemo-naive, metastatic, or unresectable locally advanced recurrent squamous cell carcinoma of the anus, aged 18 years or older, and with an Eastern Cooperative Oncology Group performance status of 0 or 1, were randomly allocated (2:1) to receive either atezolizumab (800 mg intravenously every 2 weeks up to 1 year) plus mDCF (eight cycles of 40 mg per m2 docetaxel and 40 mg per m2 cisplatin on day 1 and 1200 mg per m2 per day of fluorouracil for 2 days, every 2 weeks intravenously; group A) or mDCF alone (group B). Randomisation was done centrally using a minimisation technique and was stratified by age (<65 years vs ≥65 years) and disease status. The primary endpoint was investigator-assessed 12-month progression-free survival in the modified intention-to-treat population in group A (35% for the null hypothesis and 50% for the alternative hypothesis). This trial is registered with ClinicalTrials.gov, NCT03519295, and is closed to new participants.FINDINGS97 evaluable participants (64 in group A and 33 in group B) were enrolled between July 3, 2018, and Aug 19, 2020. The median follow-up was 26·5 months (95% CI 24·8-28·4). The median age of participants was 64·1 years (IQR 56·2-71·6), and 71 (73%) were female. 12-month progression-free survival was 45% (90% CI 35-55) in group A and 43% (29-58) in group B. In participants with a PD-L1 combined positive score of 5 or greater, 12-month progression-free survival was 70% (95% CI 47-100) in group A and 40% (19-85) in group B (interaction p=0·051) Both groups showed high compliance. Adverse events of grade 3 or higher were observed in 39 (61%) participants in group A and 14 (42%) in group B. The most common grade 3-4 adverse events were neutropenia (nine [14%] participants in group A vs five [15%] in group B), anaemia (nine [14%] vs one [3%]), fatigue (three [5%] vs four [12%]), and diarrhoea (seven [11%] vs one [3%]). Serious adverse events occurred in 16 (25%) participants in group A and four (12%) in group B, and these were mDCF-related in seven (11%) participants in group A and four (12%) in group B. Atezolizumab-related serious adverse events occurred in nine (14%) participants in group A, including grade 2 infusion-related reaction in three (5%), grade 3 infection in two (3%), and grade 2 colitis, grade 3 acute kidney injury, grade 3 sarcoidosis, and a grade 4 platelet count decrease each in one participant (2%). There were no treatment-related deaths.INTERPRETATIONDespite a higher incidence of adverse events, combining atezolizumab with mDCF is feasible, with similar dose intensity in both groups, although the primary efficacy endpoint was not met. The predictive value of a PD-L1 combined positive score of 5 or greater now needs to be confirmed in future studies.FUNDINGGERCOR, Roche.