Do saliva concentrations predict plasma unbound theophylline concentrations? A problem re-examined.

On the assumption that plasma unbound drug concentrations are therapeutically active, the value of saliva concentrations in predicting plasma unbound theophylline concentrations was investigated in 25 ambulatory adults (aged 27 to 84 years) receiving theophylline (225-1350 mg aminophylline daily) for asthma or chronic bronchitis. Plasma samples from all patients were ultrafiltered, and the plasma unbound theophylline (F) concentrations were compared with the corresponding total plasma (P), citric acid stimulated saliva (S) and non-stimulated saliva (Ns) theophylline concentrations. Plasma unbound theophylline concentrations correlated significantly with P (r = 0.97) and S (r = 0.973), but less well with Ns (r = 0.883), emphasising the benefit of saliva stimulation. The ability of S to predict F theophylline concentrations was assessed using the mean ratio of 0.7297. In 92% of the patients, predicted F concentrations were within +/- 1 microgram/ml of the measured concentrations. Similarly, using the mean F/P ratio of 0.418, predicted P were within +/- 1 microgram/ml of obtained P in 84% patients, and using the mean S/P ratio of 0.568, predicted P were within +/- 1 microgram/ml of obtained P in 81%. An accuracy of +/- 1 microgram/ml in estimating F from S concentrations would be sufficient to indicate appropriate dose adjustments, and we therefore advocate the use of stimulated saliva samples for routine monitoring of theophylline therapy.