I have a degree in chemistry and am familiar with gas chromatography, and I have a question that has been burning in my mind for some time.  Maybe someone can put my ideas in perspective so that I can let this thing go.  I don't have much knowledge of the things I'm about to suggest, and only a little knowledge of gas chromatography, and so I would like very much to know where my error lies in this.

We have heard of dogs being trained to identify the urine of a person with bladder cancer with a high rate of accuracy.  So why could researchers not use this in combination with gas chromatography, as I am about to explain, to develop a method of identifying cancer patients simply and accurately.

A known sample of urine from a cancer patient that has been experimentally identified by dogs could be subjected to fractional distillation to separate the components to the degree possible, and dogs could be used to identify the fraction that contains the cancer-related compound they are trained to detect.  If necessary, that identified fraction could then be further separated into components by whatever means suitable, and the detection process could be repeated with dogs.   Finally, the fraction(s) identified by the dogs could be introduced to a gas chromatograph with suitable substrate to produce the characteristic peaks on a chart. 

This may permit an easier identification of the peak or peaks produced by cancer-related compouinds that are detected by the dogs, and could lead to a simple test by gas chromatography of patient urine samples, and more effective, non-invasive detection of prostate and bladder cancer, -couldn't it?  In other words, it may be possible to substitute gas chromatography for the dogs' noses.

What am I missing in this over-simplified idea?  Has this idea been considered?