You need of course design primers that would amplify DNA (in exon, exon-intron boundary). RNA present in the sample, if untranscribed, it wouldn't amplify of course, but can inhibit reaction, so pretreatment with RNase is better. What are you trying to asess is a copy number variation of your gene (CNV).
But I think more problem lies in the fact, that transcription levels can be many times up or downregulated, but if you're looking for loss of heterozygosity of one allele on DNA level, the observed difference between one allele and two alleles is very small. That's a 2 -fold difference only.
You need a to choose a gene acting as reference and have several heatlhy controls. You need to establish a ratio between the reference gene and gene of interest in the controls to get what is supposed to be a normal ratio (because each quatification has a bit different amplification efficincy). Then measure the ratio of your sample to see if there is a decrease in your gene of interest.
Because of the very small differences, it's important you have only very small variations between replicates, and Ct values within the "ideal" range (20-25).
Also, with DNA it's very easy to overload the reaction, if possible, go for low concentration, lower than 50 ng in reaction. You mentioned SYBR, that is specifically important in this case, as SYBR binds any dsDNA, so the less DNA template in reaction, the better (but still remember the Cts shouldn't be too high, the higher they get after 25, the more variable they are, after 30 they generaly stop being quantifiable). The measured samples should of course have the same concentration. With SYBR you can check the baseline fluorescence, just after the reaction start, it should be low and there shouldn't be differences between samples.
Taken care of all those, you could be able to measure as little as 2-fold difference.
But generaly.. unless it's a tumor tissue, that can have vastly increased CNV of certain oncogenes, the transcriptional regulation is most likely the only cause of your decrease. For example, if you have 10-fold decreased expression of a gene, then 2-fold reduction in the DNA template is unlikely causing it. Sometimes the loss of one allele on DNA level can be even compensated for by increased transcription from the remaining allele, that is possible in certain settings and certain pathways.