How likely do cDNAs with start-loss variants appear in mRNA-seq?

Question

[John Ma]

I know that many transcripts having frameshift mutations will be degraded by nonsense- or readthrough-mediated degradation, subject to rules that are well-known.

How how about start-loss? Most likely they won't be translated if they don't have an alternate start codon, but will they be degraded quick enough that they won't show up in RNAseq data?