From Professor J. C. Dearden, MRPharmS (Hon)
SIR,-"Onlooker" (PJ, August 12, p216) drew attention to the problem of drugs in human breast milk, citing a review by Ito.1 It was pointed out that the ratio of drug concentrations in milk and plasma (M/P) is a function of protein-binding, hydrophobicity, molecular size and extent of ionisation, although Ito gave no indication of how these physicochemical factors might be used quantitatively to model and predict M/P.
In fact, a quantitative structure-activity (QSAR) study by Meskin and Lien2 has shown that M/P can be correlated quantitatively with the above factors. They found that separate correlations were required for acidic and basic drugs, and reported the following QSARs:
Acidic drugs
Log M/P = 2.068 – 0.162 (MW) – 0.185 log P
n = 20 r = 0.807 s = 0.457
Basic drugs
Log M/P = 0.265 – 0.153 log P – 0.128 log U/D
n = 15 r = 0.801 s = 0.297
where MW = molecular weight, log P = common logarithm of the octanol-water partition coefficient, U/D = ratio of undissociated to dissociated drug concentrations in plasma, n = number of drugs used in derivation of equation, r = correlation coefficient, and s = standard error of the estimate.
Log P reflects not only hydrophobicity but also protein-binding,3 which explains why high log P values cause a reduction in M/P, since protein-binding lowers membrane permeation. It is not surprising that M/P is reduced for large molecules, since these are well known to have greater difficulty in penetrating membranes, but it is unexpected that MW is not a significant factor in controlling the M/P ratio of basic drugs. The negative coefficient on log U/D appears to go against the pH-partition hypothesis, but since the pH of milk (~7.0) is lower than that of plasma (~7.4), a basic drug moving into milk is trapped by the increase in ionisation at the lower pH (the ion-trap concept).
The above equations can be used to predict the M/P ratios of most common drugs with reasonable accuracy. Log U/D for basic drugs is calculated simply as pH (7.4) – pKa. Log P values are available for most drugs,4 and can also be calculated by many currently available software packages (eg, ClogP for Windows, KOWWIN, TSAR, SciLogP Ultra); there is also a web-site (www.logp.com) that calculates log P values from the structure of a molecule inputted as a SMILES string.5
As "Onlooker" pointed out, the M/P ratio is not the only determinant of risk to infants from drug intake via breast milk; rate of clearance is also of great importance. QSAR correlations for drug clearance are well established,6 and again depend largely on log P and pKa.
Thus QSAR methodology allows pharmacists, clinicians, midwives and others a rapid and reasonably accurate prediction of the risk to infants of toxicity from drugs ingested through breast milk.
John Dearden
Liverpool
| 1. Ito S. Drug therapy for breast-feeding women. N Engl J Med 2000;343:118-26. |
| 2. Meskin MS, Lien EJ. QSAR analysis of drug excretion into human breast milk. J Clin Hosp Pharm 1985;10: 269-78. |
| 3. Dearden JC, Tomlinson E. Physico-chemical studies of analgesics. The protein-binding of some p-substituted acetanilides. J Pharm Pharmacol 1970;22:Suppl, 53S-59S. |
| 4. Hansch C (Editor). Comprehensive medicinal chemistry, Vol 6. Oxford: Pergamon Press; 1990. |
| 5. Weininger D, Weininger JL. Chemical structures and computers. In Comprehensive medicinal chemistry, Vol 4 (Editor, Ramsden CA). Oxford: Pergamon Press; 1990, pp59-82. |
| 6. Dearden JC. Molecular structure and drug transport. In Comprehensive medicinal chemistry, Vol 4 (Editor, Ramsden CA). Oxford: Pergamon Press; 1990, pp375-411. |