Therapeutic Drug Monitoring in Saliva

Summary

The observation that drug concentrations in saliva are often proportional to the concentrations in plasma has led to the suggestion that in therapeutic drug monitoring, or in pharmacokinetic studies in general, saliva might be substituted for plasma. Saliva can be collected by non-invasive techniques, generally following the stimulation of salivary flow by chewing on some semi-solid material or by spraying citric acid on the tongue.

There is evidence that many organic compounds enter saliva by a passive diffusion process, where lipid solubility and the degree of ionisation — dependent on plasma and salivary pH —are important factors. In addition, it has been shown for some drugs that their concentration in saliva equals the free or protein-unbound concentration in plasma, which is an advantage because drug plasma concentrations generally represent both bound and unbound drug. Lithium is the most well-known example of a drug that is actively secreted in saliva, but active transport mechanisms have also been proposed for organic drugs since discrepancies and time dependencies in the saliva/plasma concentration ratios have been reported, especially in single dose studies (e.g. theophylline).

If saliva is to be used in therapeutic monitoring then the saliva to plasma concentration ratio should be constant over a wide plasma concentration range. For the anticonvulsant drugs Phenytoin, primidone, ethosuximide and carbamazepine, the existence of a consistent correlation between their concentrations in saliva and plasma has been established under steady state conditions. Saliva concentrations appear to be the same as the protein unbound concentrations in plasma, and the S/P ratio is not influenced by concurrent therapy with other anticonvulsants. For phenobarbitone, a larger variability in the S/P ratio has been observed, probably due to the fact that its salivary concentration is dependent on salivary pH.

For digoxin, a substantial interindividual variation in the S/P ratio has been found in patients on long-term therapy, which makes the use of saliva for digoxin monitoring doubtful. Due to the substantial inter- and intraindividual variation in the S/P ratio of procainamide in patients on long-term therapy it is impossible to predict its plasma concentration on the basis of saliva concentrations. However, clinically, the saliva concentration of procainamide may be relevant, since this concentration and the pharmacological effect are correlated. The value of salivary drug concentration measurements in monitoring quinidine therapy is still unclear, due to the lack of information concerning the salivary excretion of quinidine in patients on long-term therapy.

Salivary concentration measurements are of limited value in monitoring theophylline therapy since this drug shows a strong time-dependency in the S/P ratio, resulting in large inter- and intraindividual variations especially during the absorption phase in single oral and rectal dose studies. A substantial interindividual variability also occurs in the S/P ratio with lithium. However, within an individual subject this value remains constant over a long period. Therefore once the S/P ratio has been established, measurement of saliva concentrations provides all information necessary for rational dosage adjustment in lithium therapy. Due to the lack of information concerning the salivary excretion of salicylate under steady state conditions and in the therapeutic concentration range, the value of saliva salicylate concentrations in therapeutic monitoring remains unclear.

For antipyrine, S/P values in the range of about I are usually observed. Therefore, saliva data permit the calculation of pharmacokinetic parameters like plasma half-life, apparent volume of distribution and total body clearance in antipyrine studies. That salivary drug concentration measurements may be a useful approach to acetylation phenotyping has been illustrated with drugs like sulphasalazine and isoniazid. The S/P ratio for a number of other drugs has been summarised but these are probably not of great value for therapeutic drug monitoring purposes.