The issue is that a hypothetical diagnostic attempting to sequence this region would likely be pre-processed with a PCR to facilitate sequencing of only the region of interest... this necessarily involves a cDNA step.
You could sequence the whole genome at higher depth, but this would be more expensive.
Edit: My mistake, the patent covers the reverse transcription step exclusively, not the act of transcription in general - which means PCR from genomic DNA is fine, but rtPCR or cDNA library construction is not... no scientific consistency there, but looks like cheaper BCRA tests in the near future.