The CMS measurement of dijet pseudorapidity distributions in pPb versus pp collisions at 5.02 TeV provides a probe for nuclear gluon PDFs. We show that while the predicted pPb pseudorapidity distributions suffer from sizable free-proton PDF uncertainties, the ratios of the pPb and pp distributions are practically insensitive to scale and free-proton PDF choices. We find the CMS data on pPb to pp ratios to be in good agreement with the EPPS16 nuclear modifications. Using a non-quadratic extension of the Hessian PDF reweighting method, we study the impact of these data on the EPPS16 nuclear PDFs. Relative to EPPS16, we find stronger evidence for mid-x gluon antishadowing as well as indication for larger gluon shadowing at small x. The data are also able to further constrain the gluon PDF in the EMC region.