1983 review paper: The Diagnostic Potential of Breath Analysis

Approximately 200 compounds have been detected in human breath, some of which have been correlated to various diseases. With the advent of new technology that may permit the rapid analysis of breath, further progress can be anticipated in the use of breath metabolites for the diagnosis of disease, including neonatal screening, toxicology, and metabolic disease.
This 1983 paper is of interest mainly because of it's lists (which may or may not be outdated), demonstrating the potential of detecting compounds through breath. Particularly from alveolar breath, which is breath from the lungs that is being exhaled from the system. Such as the way breathalyzers detect alcohol. The lists demonstrate the authors thoughts on potential compounds that could be detected at the time. Trimethylamine gets a brief mention.

Breath analyzers certainly seem a potentially useful tool in diagnosing systemic body odor and both types of halitosis. Perhaps someday there may even be portable breathalyzers for the detection of compounds such as trimethylamine, so that sufferers can monitor their trimethylamine levels. Or trimethylamine test papers so that urine can be easily tested. Perhaps this sort of technology is already out there but they don't realise there is a market.

The metabolites excreted in the breath may be divided into five groups:

1. Lipid degradation products: Numerous diseases will affect the concentration of total serum fatty acid or the fatty acid chromatographic pattern in the breath. Breath acetone has already been shown to be useful in monitoring diabetes(13).
2. Aromatic compounds: Toluene and other alkylbenzenes, furan, naphthalene, and p-tolualdehyde have been detected in the breath (9-11). The origin of these compounds in the breath is generally not known.
3. Thio compounds: Methanethiol, ethanethiol, dimethylsulfides, and, in smaller concentrations, higher alkanethiol and alkylsulfides are present in human breath (9-11). Increased concentrations of specific thio-compounds have already been shown to have diagnostic significance in cirrhosis(14, 15) and ovulation (16).
4. Ammonia and amines: Ammonia would be expected to be increased in hepatic disease (although serum ammonia does not correlate well with hepatic coma) (17). Dimethylamine and triethylamine are increased in uremia (18).
5. Halogenated compounds: These are probably derived from inhaled, injected, or absorbed environmental pollutants and are of interest in industrial toxicology (8).

http://www.clinchem.org/cgi/reprint/29/1/5