In the case of MAOA R297R it's largely down to basic genetics. A synonymous variant which produces exactly the same amino acid ("R" = Arginine) will almost never have an impact. The research from
http://sci-hub.cc/10.1097/jcp.0b013e3181ac4aaf is sometimes used to claim it does have an impact, but that research demonstrates that it's not statistically significant. The only statistical significance from that paper is for a haplotype, which is often a nice way of saying that the researchers pushed the data around until it gave a false positive result. So in this case, with no corrections made for multiple comparisons, and a small sample size, even the haplotype is probably insignificant.
There are no problematic up-regulations of the CBS gene, unless you have three copies of it (Downs Syndrome) or half of it has been lopped off in an experiment (mutant lab yeast). CBS C699T "+/+" has been found to be mildly beneficial in the published research, with no evidence of even the possibility of harm ever shown:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2366099/
VDR Taq is again a synonymous variant. It's in the coding section, and produces exactly the same amino acid with either version. Any impact seen in research is probably due to it being in tight linkage disequilibrium with VDR Bsm, which is in a non-coding section of the gene and might be capable of having a very small impact. Yasko reports VDR Taq and VDR Bsm opposite of each other, meaning everyone using her interpretation will be flagged as being +/+ for one or +/- for both. VDR Taq is the one to ignore.
There's no research into the Yasko BHMT SNPs. Hence, she's making stuff up, and your doctor is just repeating it. Unfortunately I can't prove the lack of something's existence, so if you do find proof of such research, please do share it with us.
Some of the MTRR variants have very little or no impact. Others do, but on that gene they have to be homozygous (+/+) to have an impact. I'm not sure if that's explicitly stated in something published, but all of the MTRR research I've looked at shows that the +/+ mutations which have a substantial impact have almost no effect when +/-. The most common significant mutation is A66G +/+, which is discussed somewhat at
http://www.ncbi.nlm.nih.gov/pubmed/12416982?dopt=Abstract
I don't deny that COMT V158M can have an impact on enzyme function either way (though the human body seems very well-equipped to handle the variations in function), but you've got the most common and "balanced" genotype for it. 50% of humanity has it.
MTHFR +/+, which doesn't seem to be discussed for treatment, indicates a 70% reduction in enzyme activity. This isn't very uncommon, however, and the primary risk associated with it is birth defects in the fetus of an affected mother. But research also shows that risk disappears if the mother is supplementing a normal dose of folic acid (active folate might be a better idea) or eating a diet with a good amount of vegetables.
http://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=1123&context=lawfacpub