Annals of the Rheumatic Diseases 1993; 52: 241-243 241

HYPOTHESIS

Why are non-steroidal anti-inflammatory drugs so variable in their efficacy? A description of ion trapping

G A Ellis, D R Blake

Hypothesis Non-steroidal anti-inflammatory drugs vary in Our hypothesis is that acidic non-steroidal their aqueous solubility, but most are highly anti-inflammatory drugs (NSAIDs) undergo albumin bound and will be co-distributed with ion trapping in acidotic tissue and that this this protein. Diffusion of an NSAID into the affects not only their gastrotoxicity, which is cell follows Fick's law, as it is dependent on the well known, but also their selectivity for, and concentration gradient across the plasma their efficacy in, inflammatory joint disease. membrane and the lipid/water partition Tissue acidosis may not be a feature of all coefficient of the drug molecule. chronic inflammatory disease states, but it does For weak acids, such as most NSAIDs, the occur in inflammatory joint disease and is effective concentration gradient is that of the highly variable depending on the extent of the non-ionised form, as ionised molecules are not mismatch between synovial perfusion and lipid soluble metabolic demand. This mismatch is influ- HD D- + H+ enced not only by disease activity, but also by Non-ionised Ionised joint movement, which causes transient hypoxic episodes. Interventions, such as rest The position ofthe equilibrium depends on the and drug treatment, would serve to reduce ion intrinsic ionisation constant of the molecule trapping and, therefore, NSAID efficacy (pKa) and on the pH value of the aqueous through their effects on disease activity. The phase. The influence of pH is most profound influence of these factors on pH dependent ion around the pKa value of the molecule (fig 1). trapping may contribute to the variable efficacy For acidic NSAIDs, decreasing the pH of NSAIDs between disease states, between increases the proportion of the non-ionised patients within a disease population, and form, raises the overall lipophilicity, and within a single patient over time. enhances transport across the lipid membrane. have been found in increased The pKa values for several NSAIDs at 25°C amounts in many chronic inflammatory states, are given in the table. Within tissue, the actual including rheumatoid arthritis' and osteo- values will vary slightly with temperature and arthritis,2 where they contribute to the severity the ionisation potential of the environment. and persistence of inflammation. They are Where pH gradients exist across lipid derived from by the action of membranes ionisable molecules are subject to cyclo-oxygenase and it is principally through ion trapping. The non-ionised form will tend the inhibition of this enzyme that NSAIDs to equilibrate across the membrane and will exert therapeutic effects.3 Despite their move down its concentration gradient into the mechanism of action NSAIDs are not high pH compartment where it will be ionised. universally effective in all chronic inflamma- tory disease states and this may be due, in part, to the of the drugs at the cellular level. 100 [ To reach their active site within inflam- matory cells NSAIDs must be distributed to the extravascular compartment. Passage across E the vascular endothelium can be either via gaps 0 between the endothelial cells or by simple 'a Inflammation them. There ._n0 50 Research Group, diffusion through are, however, . Bone and Joint restrictions and both routes can be altered by Research Unit, disease. 0 25-29 Ashfield Street, z London El 2AD, Normally the gaps between endothelial cells United Kingdom are small, allowing only limited egress of small G A Ellis solutes, but restricting loss of protein. The D R Blake presence of inflammation weakens the Correspondence to: allow Ms Ellis. endothelial barrier, and the wider gaps water and Increasing pH Accepted for publication the extravasation of hydrated 3 November 1992 molecules and proteins, including albumin.4 Figure 1 Ionisation of a weak acid. 242 Ellis, Blake

pKa Values ofvarious non-steroidal anti-inflammatory reduction in prostaglandins.'0 Further down drugs (NSAIDs) the intestinal tract pH gradients may occur but, NSAID pKa Value if correctly orientated, are much smaller as the 3 0 lumenal pH in these regions only varies 3-3 between 7 0 and 8&0. As a class the acidic Acetylsalicylic acid 3-5 3-9 NSAIDs have considerable gastrotoxicity and 4-2 their reduction of endogenous prostaglandins 4-4 4-5 is a major, but not the only, factor in this Indomethacin 4-5 effect." It is interesting that the pattern of ion 4.7 6-Methoxy-2-naphthylacetic acid* 5-0 trapping parallels that of the gastrotoxicity 5-1 profile. The incidence of endoscopically *Active metabolite ofnabumetone. proved gastric ulcers in patients treated with NSAIDs is estimated to be about 20%,/12 whereas in the intestine the adverse effects of More non-ionised drug is then pulled across NSAIDs are generally limited to subclinical the membrane and this dynamic process inflammation, and radiologically demonstrable continues until equilibrium is achieved (fig 2). ulceration and strictures occur only Ion trapping of NSAIDs within cells occasionally. '3 therefore occurs in situations where the pH of McCormack and Brune" considered the the extracellular fluid (pHe) is lower than that physicochemical characteristics of a few acidic of the intracellular fluid (pHi), such as in the NSAIDs with a lower than average gastrotoxic gastric mucosa and in acidotic diseased tissue. potential. They noted that these drugs did not Normally the pHi of a cell is about 7 1 whereas reach high concentrations in the gastric the pHe is 7 5. In vitro studies have shown, mucosa. For , , and however, that the orientation of this pH diflunisal this was suggested to be owing to gradient is reversed when the pHe decreases their enhanced lipophilicity, whereas for below approximately 7X 1 in cells that have the it was thought to be because it mechanisms to cope with acidification.56 is a cation at low pH values. In most situations the magnitude of the Ion trapping also occurs in tissues made measurable pHe-pHi gradient will be small, acidotic by pathophysiological mechanisms but nevertheless it can have a considerable such as in tumours and in inflammation. effect on the drug concentration at the active Research with ionisable 5-fluorouracil in rat site. In human erythrocytes in vitro, decreasing fibrosarcomas in vivo, using nuclear magnetic the pHe ofthe phosphate buffer from 7X4 to 6 8 resonance spectroscopy techniques, has shown approximately doubled the concentration of that a localised pHe reduction from 7-3 to 6-9 phenylbutazone in the intracellular compart- enhanced the cellular uptake of the drug, ment, without affecting that of the basic causing a 2-5-fold increase in the elimination compound, antipyrine.7 Physical effects can half life.'5 Such direct studies have not been enhance the pH gradient. The electrostatic conducted in inflammation. Acidic NSAIDs interaction between negatively charged groups have been shown to accumulate in inflamed in the plasma membrane reduces the pHe at tissue in several animal models, however.'6 1' the external surface of the cell. For example, In patients with joint inflammation, total with a bulk pHe of 7 0 the surface pHe is NSAID concentrations increase in the synovial calculated to be only about 6.2.8 In addition, fluid and tissue as inflammation becomes more good buffering in the unstirred layer at the severe and, therefore, more acidotic.'8 '9 surface amplifies the ion trapping effect. Although increased extravasation of albumin Replacing sodium chloride in a buffer solution bound NSAIDs will contribute to the higher with sodium citrate produced a 43-fold NSAID concentrations, ion trapping could increase in the ion trapping of salicylate.9 also have a role. The gastric mucosa has the largest pH If ion trapping does occur with NSAIDs, it gradients in the body and is the prime site for would explain why acidic NSAIDs are more NSAID trapping. These drugs accumulate in effective in vivo than NSAIDs such as the cells of this tissue and produce a prolonged and indoxole, which, although fairly potent cyclo-oxygenase inhibitors in vitro, are not acidic.20 It might also explain why more than 90% of patients with gout, with its severe synovitis, respond to NSAID treatment Low pHe Hig pH within one day, with complete resolution of / (relative) \ symptoms in five to eight days2' and why the patients with rheumatoid arthritis most likely HA HA to respond well to NSAID treatment are those with more aggressive disease.22 The acidosis of osteoarthritis is less severe and probably less persistent than in rheumatoid arthritis.23 Given H+ +±A- H+ +A- this, NSAIDs would be expected to be most effective during inflammatory and acidotic episodes and less so in the intervening periods. Differences in NSAID ion trapping may Figure 2 Ion trapping ofweak acids. pHe=pH ofthe account for some ofthe variation in therapeutic extracellularfluid; pHi=pH ofthe intracellularfluid. response within a disease group,24 as acidosis NSAIDs anid ion trapping 243

3 Vane J R. Inhibition of synthesis as a is a dynamic variable and is under the influence mechanism of action for -like drugs. Natirc 197 1; of diverse factors. In rheumatoid arthritis the 231: 232-5. extent of acidosis differs considerably between 4 Simkin P A, Pizzorno J E. Svnovial permeabilitv and rheumatoid arthritis. Arthritis Rlicioiii 1979; 22: 689-96. patients and seems to be compounded by 5 Nuccitelli R, Heiple J M. Summary of the evidence and movement.'9 2 2' Flexion of an inflamed joint discussion concerning the involvement of pHi in the control of cellular functions. In: Nuccitelli R, Deamer produces a small but significant decrease in D \X, eds. Ioitis-icelllla7i pH: its MiLs'llieMewt, regc'gidatloi auci pH. This correlates with the flexion induced ittili/zatioii ini .bJllrlc1 /it7ictiMiis. New York: Liss, 1982: 567-86. increase in intra-articular pressure, which, in 6 Seisjo B K, Messeter K. Factors determining intracellular turn, is accentuated by the presence of an pH. In: Seisjo B K, Sorensen S C, eds. Ioui hoto,eostasi's of the br)iii. Copenhagen: Munksgaard, 1971: 244-62. effusion.27 Joint movement would, therefore, 7 Brune K, Graf P. Nonsteroidal anti-inflammatory drugs: be expected to have a positive influence on the influence of extracellular pH on biodistribution and pharmacological effects. 13i0iL1hui P/ia,iicacol 1978; 27: distribution of NSAIDs and may have 525-30. contributed to the beneficial effects of physical 8 Ohshima H, Kondo T. pH dependence of electrostatic interaction betseen ion-penetrable membranes. Bi0plhYs training observed in patients with moderate Cheiii 1988; 32: 161 -6. rheumatoid arthritis.28 29 By the same token, 9 Gutknetch T, Tosteson D C. Diffusion of weak acids across lipid bilaser membranes: effects of chemical reactions in any intervention, be it bed rest or drug the unstirred layers. Scie7ic 1973; 182: 1258-61. treatment, that reduces disease activity would 10 Blowver P. TIhe scientific-equixvalent efficacy and diminished decrease NSAID ion trapping and therefore risk. Jtiur. R/lciuiiiatol Iuifla,iuiii 1991; 11: 29-37. 11 'Whittle B J R. Temporal relationship between cvclo- NSAID efficacy. In fact, NSAIDs could alter oxvgenase inhibition, as measured bv prostacvclin the synovial microenvironment in such a way biosynthesis, and gastrointestinal damage induced bs indomethacin in the rat. Gastroewocer'n0og 1981; 80: 94-8. as to attenuate their own efficacy over a period 12 Gabriele S E, Bombardier C. NSAID induced ulcers. An of time. The constantly changing profile of the emerging epidemic. Y Rhe/iciiatol 1990; 17: 1-4. 13 Bjarnason I, Zanelli G, Smith T, et al. Nonsteroidal influences on tissue pH could account for not antiinflammatorv drug-induced intestinal inflammation in only some of the variability between patients, humans. Gastrociteruology 1987; 93: 480-9. 14 McCormack K, Brune K. Classical absorption theory and but also that within a single patient over time. the development of gastric mucosal damage associated The recently developed NSAID nabu- wvith non-steroidal anti-inflammatory drugs. Arcli Toxicol 1987;60:261-7. metone uses the concept of ion trapping to 15 Guerquin-Kern J, Leteutre F, CroisN A, Lhoste J. pH good effect. itself is not acidic dependence of 5-fluorouracil uptake observed by in-vivo 'iP and "F nuclear magnetic resonance spectroscopy. and is without significant pharmacological CauicerRcs 1991; 51: 5770-3. effects. It does not undergo ion trapping in the 16 Brune K, Glatt M, Graf P. Mechanisms of action of anti- stomach and has low gastrotoxicity in animal inflammatory drugs. Gmuu Phariiiacol 1976; 7: 27-33. 17 Brune K, Rainsford K D, Schweitzer A. Biodistribution of models.' It is metabolised in the liver to mild . Bu7 C(Ii/u Pharnuuacol 1980; 10: 279S-84S. 6-methoxy-2-naphthylacetic acid, which is 18 Gaucher A, Netter P, Faure G, Schoeller J P, Gerardin A. Diffusion of oxyphenbutazone into synovial fluid, synovial acidic, possesses similar anti-inflammatory tissue, joint cartilage and cerebrospinal fluid. Eluo 7 Cl/i activity to other acidic NSAIDs, and can Pharnuuaco/ 1983; 25: 107-12. 19 Falchuk K H, Goetzl E J, Kulka J P. Respiratorv gases of potentially be ion trapped in inflammatory synovial fluids. An approach to svnovial tissue circulatory- cells. 3" metabolic imbalance in rheumatoid arthritis. Au)i Y7 Med 1970; 49: 223-31. In conclusion, we suggest that the variable 20 Brune K, Otterness I. In-vivo and in-vitro assessment of efficacy of NSAIDs in different diseases, non-steroidal anti-inflammatory drugs. Bail/e/res Cli/ Rheiu)ouatol 1988; 2: 295-307. between patients for a selected disease, and 21 Fam A G. Strategies and controversies in the treatment of within the same patient over time may be gout and hvperuricaemia. Ba/lli/ers Cl/i/ Rheuooiatol 1990; 4:177-92. accounted for by the acidic nature of NSAIDs 22 Cush J J, Jasin H E, Johnson R, Lipsky P E. Relationship and their propensity for ion trapping. Detailed between clinical efficacy and laboratorv correlates of inflammators and immunologic activitv in rheumatoid metabolic studies using nuclear magnetic arthritis patients treated with non-steroidal anti- resonance spectroscopy techniques that allow inflammatory drugs. Ar-thtitis Rheiuoui 1990: 33: 623-33. accurate 23 Naughton D P, Whelan M A, Grootveld M, Blake D R. Nessv for quantification of tissue pH, and metabolic markers for arthritic disease etiology. In: intracellular concentrations of suitably labelled Proceediugs (of the 55th Auienicaui College of Rheuiuiatologv NSAIDs, should be able to assess this Meetnig; 17-21 Nosvember 1991; Boston. S145. 24 Huskisson E C, Dieppe P A, Scott J, Jones H. Diclofenac hypothesis. In the meantime, should our sodium, diflunisal and : patient preferences for hypothesis be true, clinical rheumatologists anti-inflammatory drugs in rheumatoid arthritis. Rhetimzato/ Rehabil 1982; 21: 238-42. may like to consider some of the more obvious 25 Farr M, Garrey K, Bold A M, Kendall M J, Bacon P A. consequences. Movement of the joint may Significance of hvdrogen ion concentration in svnovial fluid in rheumatoid arthritis. Clini Exp Rheiniiatol 1985; 3: produce beneficial effects on drug distribution 99-104. there be a 26 James M J, Cleland L G, Rofe A M, Leslie A L. Intra- and, therefore, efficacy. Also, may articular pressure and the relationship between synovial rationale for switching between NSAIDs, with perfusion and metabolic demand. 7 Rheiipiiatol 1990; 17: their different to 521 7. physicochemical properties, 27 Merrs P, Williams R, Cox N, King J B, Blake D R. A give continuing anti-inflammatory activity in a comparative studv of intra-articular pressure dynamics in microenvironment. acute traumatic and chronic inflammatorv joint effusions: constantly changing potential implications for hypoxic-reperfusion injury. Aoiz Rheuiiii Dis 1991; 50: 917-20. We thank the Arthritis and Rheumatism Council for their 28 Harkcom T M, Lampman R M, Banwell B F, Castor C W. financial support to the Bone and Joint Research Unit and Dr Therapeutic value of graded aerobic exercise training in P Blower for his help with the manuscript. rheumatoid arthritis. A-thritis Rheum?i 1985; 28: 32-9. 29 Lyngberg K, Danneskiold-Samsoe B, Haskov 0. The effect of phvsical training on patients with rheumatoid arthritis: 1 Trang L E, Granstrom E, Lovgren 0. Levels of changes in disease activity, muscle strength and aerobic prostaglandin F2-alpha and E2 and B2 in capacity. A clinically controlled minimized cross-over joint fluid in rheumatoid arthritis. Scand J R/iezioatrol study. C/i7i Exp Rheuuuiatol 1988; 6: 253-60. 1977; 6: 151-4. 30 Haddock R E, Jeffery D J, Mangan F R, Pardon I S. 2 Sahap-Atik 0. Leukotriene B4 and -like Penetration of nabumetone into the inflammatory activity in synovial fluid in osteoarthritis. Prostaglanidinls exudates in the rat. Y Phianui Phanuuacol 1983; 35: Leuikot Esse,i Fattv Acids 1990; 39: 253-4. 358-62.