In Reply We thank Golmard et al and Rousseau et al for their comments on our Research Letter.1 The major concern was the inclusion of all the nonsynonymous variants in the analysis. We agree that not all POLE/POLD1 variants can lead to an ultramutated phenotype. A previous study2 has shown that some POLE/POLD1 variants outside of the exonuclease domain are also capable of damaging their proofreading function and leading to an ultramutated phenotype. Thus, defining pathogenic variants according to the variant location is not accurate enough. In our further analysis, the results show that patients with variants outside of the exonuclease domain also had a better prognosis than patients with wild-type variants in the immune checkpoint inhibitor (ICI) treatment cohort. Variants of POLE/POLD1 can serve as a biomarker in immunotherapy regardless of whether they are located in the exonuclease domain.