A standardised dose-reduction strategy has not been established for the widely used gemcitabine plus nab-paclitaxel regimen in patients with metastatic pancreatic ductal adenocarcinoma. We aimed to investigate the efficacy and tolerability of alternating treatment cycles of nab-paclitaxel-gemcitabine combination therapy and gemcitabine alone versus continuous treatment with the nab-paclitaxel-gemcitabine combination. ALPACA was a randomised, open-label, phase 2 trial conducted at 29 study centres across Germany. Patients aged 18 years or older with a histologically or cytologically confirmed diagnosis of metastatic pancreatic ductal adenocarcinoma who had not been previously treated for advanced disease were enrolled. After an induction phase with three cycles of nab-paclitaxel-gemcitabine combination therapy (nab-paclitaxel 125 mg/m2 and gemcitabine 1000 mg/m2 administered intravenously on days 1, 8, and 15 of each 28-day cycle), patients were randomly assigned (1:1) by stratified permuted block randomisation either to continue treatment with standard nab-paclitaxel-gemcitabine or to receive alternating cycles of nab-paclitaxel-gemcitabine and gemcitabine alone. Patients and investigators were not masked to treatment allocation. Randomisation was done centrally by the study statistician using a computer-generated randomisation list, and was stratified by Karnofsky Performance Status and presence of liver metastases. The primary endpoint was the derivation of an unbiased point estimate and an associated confidence interval with a confidence coefficient of 80% for the hazard ratio (HR) for overall survival after randomisation, without testing a specific hypothesis, analysed by intention to treat in all patients who started randomised treatment. Safety was analysed according to treatment received. This trial is registered with ClinicalTrials.gov, NCT02564146, and is completed. Between May 27, 2016, and May 27, 2021, 325 patients were enrolled. Following three cycles of induction treatment, 174 patients were randomly assigned: 85 to continue receiving standard nab-paclitaxel-gemcitabine, of whom 79 started treatment, and 89 to the alternating treatment schedule, of whom 88 started treatment. Of the 167 patients who started randomised treatment, 88 (53%) were female and 79 (47%) were male. Median overall survival after randomisation was 10·4 months (80% CI 9·2-12·0) in the group that received standard treatment and 10·5 months (10·2-11·1) in the group that received alternating treatment (HR 0·90, 80% CI 0·72-1·13; p=0·56). The most common adverse events of any grade were peripheral neuropathy (59 [74%] of 80 patients in the continuous treatment group vs 53 [62%] of 85 patients in the alternating treatment group) and fatigue (43 [54%] vs 44 [52%]). Treatment-emergent serious adverse events after randomisation occurred in 40 (50%) patients in the continuous treatment group and in 28 (33%) in the alternating treatment group. Fewer treatment-emergent adverse events of grade 3 or higher occurred in patients treated with alternating cycles compared with those receiving standard therapy, especially for peripheral neuropathy (17 [21%] patients in the continuous treatment group vs 12 [14%] in the alternating treatment group) and infections (16 [20%] vs nine [11%]). There were two treatment-related deaths after randomisation, both in the continuous treatment group (one multiple organ dysfunction syndrome, not treated after randomisation, and one interstitial lung disease). Our findings suggest that a dose-reduced regimen with alternating cycles of nab-paclitaxel-gemcitabine and gemcitabine alone after three induction cycles is associated with similar overall survival to that for standard treatment with nab-paclitaxel-gemcitabine, but with improved tolerability. We therefore propose that a switch to the alternating schedule could be considered in a clinical setting for patients with metastatic pancreatic cancer who have at least stable disease after three cycles of nab-paclitaxel-gemcitabine treatment. Celgene/Bristol Myers Squibb.