Although tumor DNA sequencing uncovers frequent mutations in non-coding areas, their role in cancer development remains poorly understood. To address this gap, we investigated the impact of these mutations in regulatory regions of genes linked to the aggressive behavior of triple-negative breast cancer cells. During our exploration, COL1A1 stood up as a strong candidate, with a putative mutation in an unannotated intronic alternative promoter. COL1A1 gene codes for type I collagen, a constituent of the extracellular matrix and is a canonical target of the TGFb pathway, with a predominantly main protein-coding isoform consisting of 51 exons and only one annotated promoter. However, CAGE data from breast cancer (BC) cell lines showed the predominant usage of an alternative transcription start site, located close to the 3’ end of the first intron, coinciding with a reported enhancer. This potential transcript was confirmed by RT-PCR from extracted RNA of multiple BC cell lines. We could further validate these findings by looking at RNA splicing patterns across various cell lines with RNA-seq. To account for potential artifacts arising from culturing cells in a flat dish (2D), which may not fully represent real tumors, we compared these findings to data from MDA-MB-231 cells grown in a 3D culture system and observed the same results. However, when we analyzed RNA-seq data from eight BC patients, all of them categorically used the canonical promoter. We conclude that the misregulation of COL1A1 promoter usage is probably inherent to a cell line artifact and not inherent to BC tumor biology. Since this could be the case for several “cancer-specific isoforms”, we are undertaking a systematic comparison between RNA isoform usage in BC cell lines and patients’ samples. BC cell lines are great models to explore cancer biology, although they might present many anomalies with uncertain causes hence the extrapolation of results to clinical patients must be made with caution.