I would like to look for functional polymorphisms upstream of the transcription start site of my gene of interest. Has anyone got any experience in doing this in populations of cancer patients?
If you already have the sequences you could always align and look for SNPs, indels etc? The sample size would have to be quite large to be statistically significant, and you'd want to look at a control group (i.e. non-cancer patients) as well. There's probably quite a lot of existing sequence data out there already for a variety of cell types.
To test what functional effect these may have you could do as Jana suggests and do some Luciferase assays (or EMSA I guess), there are probably some relevant cell lines available already.
If you want to do functional study, you can use luciferase assay. If you want to figure out the relative transcription factors binding to different SNP sites, probably you want to do DNA affinity chromatography, EMSA, ChIP assay, etc. You may find the following paper useful. http://www.ncbi.nlm.nih.gov/pubmed/12176321