No. One does not have to sum/ add the different TMSs.
Few things, as long as you only quantify and identify the same TMS version across all samples, it should be consistent and comparable. For instance, if you annotated Glucose-3TMS across all samples then that's enough. If you get Glucose-5TMSs across all runs then just areas under those curves/ peaks are good enough. Make sure that the software tool you are using does not switch between 5TMS and 3 TMS forms on its own or incorrectly annotates them among samples. To ensure that rely on a retntion time (RT) window to make sure that its Glucose-xTMS whatever 3 or 5 TMS, but its fixed. Remember due to 2 TMS difference their RT would be very different. Ratios as you mentin, would complicate it further, and the variable ratio is due to ion suppression/ enhancements between complex samples (with coeluting conditions with other compounds, 10s-100s may be!) and your pure standard just coming out on its own!: )
Also, summing is not a good idea, esp. as GC-MS or mass-sepctrometry techniques, the ionization efficiency is not 100% and does not correlate well with the absolute abundance of a given molecule (inept ESI/ EI ionization, plus ion suppression/ enhancements, and impurity in standards, and chromatographic losses etc.) even if you sum up, 3 TMS, 5TMS, 7TMS versions of molecule "M" you would not capture 100% of it. Also, the non-derivatized (non silylated version of M is rarely capture in GC-MS and inevitably that quantification is lost!) Its still Relative quantification" and given all the silylated forms come form the same molecule, measuring xTMS form of the metabolite M consitently across all samples is relative and good repeatable and reportable quantification!
" the variable ratio is due to ion suppression/ enhancements between complex samples, and your pure standard just coming out on its own "
In short I fear that, if I take a pure standard and a complex sample, with the same Glucose concentration, I might end up with different amounts of 3-TMS and 5-TMS.
Like the pure standard, lets say has more 3-TMS; the complex sample has more 5-TMS, even though they should have the same concentration.