Identifying variables with time-varying associations can help guide patient stratification and treatment strategies. Time-varying associations refer to the associations between variables of interest and the outcomes that change over time. Such associations of gene transcripts have never been examined in colorectal cancer (CRC). In the present study, we examined the transcripts of splicing factors and their associations with prognosis in CRC. We performed analyses using Cox PH models and Cox models with time-varying coefficients when appropriate. A transcript of pre-mRNA processing factor 38A (PRPF38A) was identified to be significantly associated with the progression-free survival after two years post-diagnosis, but not before that, making this transcript a candidate late-outcome marker in CRC. Further explorative analysis showed that this transcript correlated with the alternative splicing of genes involved in RNA binding and mitochondrial protein synthesis. Overall, our study, for the first time, revealed the association patterns of splicing factor transcripts over time in CRC, identified a transcript of PRPF38A as a candidate late-outcome marker, and provided mechanistic insights into outcome heterogeneity in CRC.AJCR Copyright © 2025.