Abstract
Abstract
Disruptive research is typically recognized after publication through citation patterns, leaving limited evidence about how it is evaluated before entering the scholarly record. Peer review is the main setting in which manuscripts are judged, revised, and legitimized, yet little is known about how reviewers' concerns develop across rounds or how authors respond. Using peer review histories from PLOS ONE, we link reviewer reports, author responses, review rounds, editorial timing, and citation-based disruption scores for 18,776 accepted articles. Reviewer comments are classified into seven evaluative aspects and three sentiment categories, while matched author replies are coded as revisions, explanations, or rebuttals. High-disruption papers follow slightly longer submission-to-acceptance trajectories, mainly because of longer author-response intervals rather than reviewer or editor processing. They do not receive more substantive comments overall, but attract a higher proportion of negative comments, especially on clarity, substance, and originality. Across rounds, clarity concerns show only suggestive evidence of slower attenuation, whereas replicability concerns are less persistent. For matched concern–response pairs, authors use revision more often and explanation less often, without significantly increasing rebuttal. These findings suggest that disruptive research faces a selective problem of evaluative legibility before publication, and illuminate how scholarly information is reframed and stabilized.