13283890 (P.T.A.B. Sep. 5, 2018)

Ex Parte SCHWARTZ et al

Patent Trial and Appeal BoardSep 5, 2018

13283890 (P.T.A.B. Sep. 5, 2018)

UNITED STA TES p A TENT AND TRADEMARK OFFICE APPLICATION NO. FILING DATE 13/283,890 10/28/2011 27752 7590 09/07/2018 THE PROCTER & GAMBLE COMPANY Global IP Services Central Building, C9 One Procter and Gamble Plaza CINCINNATI, OH 45202 FIRST NAMED INVENTOR James Robert SCHWARTZ UNITED STATES DEPARTMENT OF COMMERCE United States Patent and Trademark Office Address: COMMISSIONER FOR PATENTS P.O. Box 1450 Alexandria, Virginia 22313-1450 www .uspto.gov ATTORNEY DOCKET NO. CONFIRMATION NO. 11927M 1101 EXAMINER GEMBEH, SHIRLEY V ART UNIT PAPER NUMBER 1628 NOTIFICATION DATE DELIVERY MODE 09/07/2018 ELECTRONIC Please find below and/or attached an Office communication concerning this application or proceeding. The time period for reply, if any, is set in the attached communication. Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the following e-mail address(es): centraldocket.im @pg.com pair_pg@firsttofile.com mayer.jk@pg.com PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE BEFORE THE PATENT TRIAL AND APPEAL BOARD Ex parte JAMES ROBERT SCHWARTZ, CHARLES WINSTON SAUNDERS, ROBERT SCOTT YOUNGQUIST, CASEY PATRICK KELLY, JEFFREY KENNETH DOMSIC, ROBIE LYNN LUCAS, and JUN XU Appeal2017-007109 Application 13/283,890 Technology Center 1600 Before RICHARD M. LEBOVITZ, JEFFREY N. FRED MAN, and ULRIKE W. JENKS, Administrative Patent Judges. FREDMAN, Administrative Patent Judge. DECISION ON APPEAL This is an appeal 1,2 under 35 U.S.C. Â§ 134 involving claims to a personal care composition. The Examiner rejected the claims as obvious under 35 U.S.C. Â§ 103(a). We have jurisdiction under 35 U.S.C. Â§ 6(b ). We affirm, but designate our affirmance as a New Ground .. 1 Appellants identify the Real Party in Interest as The Procter & Gamble Company (see App. Br. 1 ). 2 We have considered and herein refer to the Specification of Oct. 28, 2011 ("Spec."); Final Office Action of Feb. 1, 2016 ("Final Action"); Appeal Brief of Nov. 1, 2016 ("App. Br."); and Examiner's Answer of Jan. 12, 2017 ("Answer"). Appeal2017-007109 Application 13/283,890 Statement of the Case Background "Zinc Pyrithione (ZPT) is a common anti-dandruff active used in shampoos and other treatments" (Spec. 1 ). The Specification teaches "the use of appropriate iron chelators in combination with ZPT is a key insight to developing significantly more effective anti-dandruff formulas" (id.). The Claims Claims 1, 2, and 4--14 are on appeal. Claim 1 is representative and reads as follows: 1. A personal care composition comprising: c) an effective amount of a pyrithione or a polyvalent metal salt of a pyrithione; d) an effective amount of an iron chelator or a material which chelates iron; wherein the combination of the iron chelator and the pyrithione or a polyvalent metal salt of a pyrithione has a fractional inhibitor concentration of less than or equal to 1 and wherein the iron chelator has an AlogP value of greater than or equal to 0.4. The Rejections A. The Examiner has rejected claims 1, 2, 4--8, and 12-14 under 35 U.S.C. Â§ I03(a) as obvious over Polson3 and Bohn4 (Final Act. 2--4). B. The Examiner has rejected claims 9-11 under 35 U.S.C. Â§ I03(a) as obvious over Polson, Bohn, and Pippard5 (Final Act. 4--5). 3 Polson et al., US 2009/0298805 Al, published Dec. 3, 2009. 4 Bohn et al., Dermatopharmacology of ciclopirox nail lacquer topical solution 8% in the treatment of onychomycosis, 43 J. AM. AcAD. DERMATOLOGY S57---69 (2000). 2 Appeal2017-007109 Application 13/283,890 A. 35 US.C. Â§ 103(a) over Polson and Bohn The Examiner finds that Polson teaches "the use of zinc pyrithione (ZPT) to treat nail fungus" and Bohn teaches "topical administration of ciclopirox in nail lacquer is effective to treat onychomycosis and that its antifungal activity is due to the hydroxypyridone moiety" (Final Act. 2). The Examiner finds because "both ZPT and ciclopirox are both taught to be effective to treat nail fungus, it would have been prima facie obvious to combine them in a single composition" (Final Act. 2, citing In re Kerkhoven, 626 F.2d 846 (CCPA 1980) "It isprimafacie obvious to combine two compositions each of which is taught by the prior art to be useful for the same purpose, in order to form a third composition to be used for the very same purpose."). The issue with respect to this rejection is: Does a preponderance of the evidence of record support the Examiner's conclusion that Polson and Bohn render the claims obvious? Findings of Fact 1. Polson teaches: "Zinc pyrithione is a very effective antifungal additive that is useful in treating nail fungus organisms when compared with other topical antifungal drugs" (Polson ,r 11 ). 2. Polson teaches "the composition containing the metal pyrithione complex is topically applied to the infected nail preferably as a nail polish or lacquer" (Polson ,r 10). 5 Pippard et al., Iron chelation using subcutaneous infusions of diethylene triamine penta-acetic acid (DTPA), 36 SCAND. J. HAEMOTOLOGY 466-72 (1986). 3 Appeal2017-007109 Application 13/283,890 3. Bohn teaches: "Ciclopirox exhibits fungal inhibitory activity (minimum inhibitory concentration< 4 Âµg/ml for dermatophytes) as well as fungicidal activity; to date resistance to the drug has not been identified. Ciclopirox has been formulated in a nail lacquer delivery system" (Bohn, abstract). Principles of Law "The combination of familiar elements according to known methods is likely to be obvious when it does no more than yield predictable results." KSR Int'! Co. v. Teleflex Inc., 550 U.S. 398,416 (2007). Analysis We adopt the Examiner's findings of fact and conclusions of law (see Final Act. 2--4; FF 1-3) and agree that Polson and Bohn render the claims obvious. We address Appellants' arguments below. Appellants contend "that combining anti-fungals (or anti-microbials in general) does not necessarily result in synergistic or additive activities. In fact, that case is the exception, not the rule. The present invention compositions are limited to additive/synergistic combinations via an FIC value" (App. Br. 4--5). Appellants contend "Polson in view of Bohm and Kraemer does not disclose combinations of materials or synergy or additive activities resulting from those combinations" (Id. at 5; emphasis omitted). We find this argument unpersuasive because, while many factors impact synergy, the evidence of the Specification does not appear to compare dose response curves but rather compares single values of chemicals as "additive" or "synergistic" (see Spec. 9). In particular, the Specification calculates the combinatorial effect by measuring the lowest 4 Appeal2017-007109 Application 13/283,890 inhibitory dose of each chemical divided by that dose plus a dose with the other chemical, and then adding the result (see Spec. 9). However, the analysis of the Specification is incomplete because the data does not demonstrate that the dose responses of the chemicals are linear, so the evidence resulting from combining two drugs at a single dose doesn't necessarily mean that even an improved result is additive or synergistic, because the improved result might have been obtained by simply doubling the dose of either chemical alone. More simply put, as discussed by Berenbaum at 2, 6'7 suppose that two chemicals, A and B, each suppress a response by 10% when given at a dosage of 1 mg/kg but that, when both are given together, each at this dose, the suppression is not 20% but 90%. This might appear as a case of marked synergy. But the dose-effect curves of these chemicals may demonstrate that a 2 mg dose of either chemical alone also produces a 90% suppression. As 2 mg of the combination produces exactly the same effect as 2 mg of either drug by itself, the drugs are clearly not more effective when used in combination than when used singly. This is the situation in the instant case, since there is no evidence that the chemical response is linear (see Spec. 11 ). Nor does the data in Table 2 demonstrate that the dose response curves are linear (id.). Finally, the evidence does not clearly "represent a 'difference in kind' that is required to 6 Berenbaum, M., Synergy, additivism and antagonism in immunosuppression, 28 Clin. Exp. Immunol. 1-18 (1977). 7 We note that Berenbaum was not previously cited by the Examiner, and due to our citation, we designate our rejection as a New Ground. 5 Appeal2017-007109 Application 13/283,890 show unexpected results." In re Harris, 409 F.3d 1339, 1344 (Fed. Cir. 2005). Therefore, there is no persuasive demonstration that the combinations are additive or synergistic solely based on the fractional inhibitor concentration presented in the Specification. Appellants contend the intrinsic anti-fungal potency is being used as a predictor of the potential to have either synergistic or additive anti-microbial benefits when combined with another anti-fungal agent. The data of Table 1 clearly show both hydrophobi[ ci]ty and iron affinity parameters must meet certain minimal criteria to result in appreciable independent anti-fungal activity. (App. Br. 6). We find this argument unpersuasive because, to the extent that there is evidence in Table 2 of synergy for nine specific compounds, claim 1 encompasses any iron chelators whatsoever. Therefore, the scope of claim 1 far exceeds the supporting evidence. Unexpected results must be "commensurate in scope with the degree of protection sought by the claimed subject matter." Harris, 409 F.3d at 1344. Indeed, Appellants themselves acknowledge that "common iron chelators such as EDT A are observed not to be effective" (App. Br. 6) and the Kelly Deel. II8 states the "measured FIC values for 1,10-Phenanthroline is 3.0 ... Hence this material is considered indifferent in the presence of ZPT" (Kelly Deel. II at 2). Therefore, even if we agreed that the results for the particular tested compounds was "synergistic", these results are not commensurate in scope with the recitation 8 Declaration of Dr. Casey Kelly, dated June 2, 2015 ("Kelly Deel. II"). 6 Appeal2017-007109 Application 13/283,890 in claim 1. Appellants have not explained how the stated "synergy" for several of the tested compounds would predict synergy for the entire scope of the claim, Appellants contend they have provided evidence that the long list of iron chelators in previous cited art cannot be presumed to be combined with Polson and be: 1) hydrophobic with the claimed AlogP and further 2) cannot be presumed to have the claimed FIC if one combines the iron chelant of cited art with ZPT. This is a hindsight reconstruction of the prior art, impermissibly based on the applicants' disclosure. (App. Br. 6). We find this argument unpersuasive because the Examiner's obviousness rejection relies upon teaching by Polson that zinc pyrithione is a "very effective antifungal additive that is useful in treating nail fungus" (FF 1) and the teaching by Bohn that ciclopirox "exhibits fungal inhibitory activity" and "has been formulated in a nail lacquer delivery system" (FF 3). Thus, both references suggest antifungal compounds for treatment of fingernails and therefore claim 1 "'simply arranges old elements with each performing the same function it had been known to perform' and yields no more than one would expect from such an arrangement, the combination is obvious." KSR, 550 U.S. at 417, citing Sakraida v. Ag Pro, Inc., 425 U.S. 273, 282 (1976). We note the reason "in the prior art to combine the references does not have to be identical to that of the applicant to establish obviousness." In re Kemps, 97 F.3d 1427, 1430 (Fed. Cir. 1996). We also find the hindsight argument unpersuasive. While we are fully aware that hindsight bias may plague determinations of obviousness, Graham v. John Deere Co., 383 U.S. 1, 36 (1966), we are also mindful that 7 Appeal2017-007109 Application 13/283,890 the Supreme Court has clearly stated that the "combination of familiar elements according to known methods is likely to be obvious when it does no more than yield predictable results." KSR, 550 U.S. at 416. Polson and Bohn establish that zinc pyrithione and ciclopirox are both known anti- fungal agents used for treatment of fingernails (FF 1, 3). As already discussed, we do not find Appellants have established evidence supporting unpredictable results for the broad generic claim, much less the narrower species of zinc pyrithione and ciclopirox suggested by Polson and Bohn. We have also considered the Kelly Deel. I9, which contends that "many of the chelators listed in the Lawyer and Lawyer and the Guthrie reference do not meet the AlogP requirement required in the present invention" (Kelly Deel. I ,r 5). However, this evidence does not address the current rejection, which relies on neither Lawyer nor Guthrie, but rather upon Bohn's teaching of ciclopirox, a chelator that was recognized as satisfying the requirements of the claim in the Specification itself (see Spec. 3). Appellant presents no evidence that ciclopirox would not satisfy the AlogP requirement of claim 1. Conclusion of Law A preponderance of the evidence of record supports the Examiner's conclusion that Polson and Bohn render the claims obvious. B. 35 USCÂ§ 103(a) over Polson, Bohn, and Pippard Appellants contend "combining anti-fungals (or anti-microbials in general) does not necessarily result in synergistic or additive activities. In 9 Declaration of Dr. Casey Kelly, dated May 1, 2014 ("Kelly Deel. I"). 8 Appeal2017-007109 Application 13/283,890 fact, that case is the exception, not the rule. The present invention compositions are limited to additive/synergistic combinations via an FIC value" (App. Br. 7). We remain unpersuaded by this argument for the reasons given above. Briefly, the claimed fractional inhibitor concentration does not necessarily require synergism because it analyzes single doses, not dose response curves, and therefore does not necessarily result in synergistic results as discussed above. See Berenbaum (fn. 7). We also remain unpersuaded for the reasons given above by Appellants' argument that: "one would not be motivated to arrange elements in the same way as Claim 1" (App. Br. 8); that the Examiner's rejection is "a hindsight reconstruction of the prior art, impermissibly based on the applicants' disclosure" (id.); and remain unpersuaded by either of the Kelly Declarations discussed above. We therefore affirm the rejection of claims 9-11. SUMMARY In summary, we affirm the rejection of claim 1 under 35 U.S.C. Â§ 103(a) as obvious over Polson and Bohn. Claims 2, 4--8, and 12-14 fall with claim 1. We affirm the rejection of claims 9-11 under 35 U.S.C. Â§ 103(a) as obvious over Polson, Bohn, and Pippard. Because we cite new evidentiary prior art, we designate our affirmance as a new ground pursuant to 37 C.F.R. Â§ 4I.50(b). Section 4I.50(b) provides "[a] new ground of rejection pursuant to this paragraph shall not be considered final for judicial review." Section 41.50(b) also provides: 9 Appeal2017-007109 Application 13/283,890 When the Board enters such a non-final decision, the appellant, within two months from the date of the decision, must exercise one of the following two options with respect to the new ground of rejection to avoid termination of the appeal as to the rejected claims: ( 1) Reopen prosecution. Submit an appropriate amendment of the claims so rejected or new Evidence relating to the claims so rejected, or both, and have the matter reconsidered by the examiner, in which event the prosecution will be remanded to the examiner. The new ground of rejection is binding upon the examiner unless an amendment or new Evidence not previously of Record is made which, in the opinion of the examiner, overcomes the new ground of rejection designated in the decision. Should the examiner reject the claims, appellant may again appeal to the Board pursuant to this subpart. (2) Request rehearing. Request that the proceeding be reheard underÂ§ 41.52 by the Board upon the same Record. The request for rehearing must address any new ground of rejection and state with particularity the points believed to have been misapprehended or overlooked in entering the new ground of rejection and also state all other grounds upon which rehearing is sought. Further guidance on responding to a new ground of rejection can be found in the Manual of Patent Examining ProcedureÂ§ 1214.01. AFFIRMED; 37 C.F.R. Â§ 4I.50(b) 10 Application/Control Applicant(s)/Patent Under Reexamination No. 13/283,890 Notice of References Cited Examiner Art Unit Gembeh 1628 Page 1 of 1 U.S. PATENT DOCUMENTS * Document Number Date CPC Classification US Classification Country Code-Number-Kind Code MM-YYYY Name A US- B US- C US- D US- E US- F US- G US- H US- I US- J US- K US- L US- M US- FOREIGN PATENT DOCUMENTS * Document Number Date Country Code-Number-Kind Code MM-YYYY Country Name CPC Classification N 0 p Q R s T NON-PATENT DOCUMENTS * Include as applicable: Author, Title Date, Publisher, Edition or Volume, Pertinent Pages) u Berenbaum, M., Synergy, additivism and antagonism in immunosuppression, 28 Clin. Exp. lmmunol. 1-18 (1977) .. V w X *A copy of this reference 1s not being furnished with this Office action. (See MPEP Â§ 707.05(a).) Dates in MM-YYYY format are publication dates. Classifications may be US or foreign. U.S. Patent and Trademark Office PT0-892 (Rev. 01-2001) Notice of References Cited Part of Paper No. C/in. exp. Immunol. (1977) 28, 1-18. Synergy, additivism and antagonism in immunosuppression A CRITICAL REVIEW M. C. BERENBAUM Wellcome Laboratories of Experimental Pathology, Variety Club Research Wing, St. Mary's Hospital Medical School, London (Received 11 June 1976) INTRODUCTION In spite of the successes of immunosuppressive therapy, the agents in use are not as effective as desired, partly because their dosage is to a large extent limited by their toxicity. Methods to increase their effective- ness are therefore of practical importance. One of the approaches to this end is the search for synergistic combinations of agents and, over the past 25 years, there has been a steady stream of publica- tions in this field. Synergy, however, is a topic on which confusion reigns. The relevant pharmacological literature is often obscure (some papers, indeed, are models of incomprehensibility) and is profusely littered with technical terms that are not always clearly defined. Several different terms are used to describe the same phenomenon and the same term means different things to different authors. Goldin & Mantel (1957) listed seven different definitions of synergy and implied that, in this chaotic state of affairs, the choice of definition was a matter of personal preference. So far as most published work is concerned, there appears to have been little progress since that time. Reference to pharmacological text-books does not help; they generally deal with the topic cursorily and/or incorrectly, or avoid it altogether. It is not surprising, therefore, that this confusion is reflected in investigations on synergy in immuno- suppression. Experiments are commonly designed in such a way that they could not detect synergy if it were present, results are interpreted as showing synergy when there is no evidence for it, as showing additivism where there is clear antagonism, and so on. The basic difficulty is that most investigators use fallacious criteria for determining the nature of drug interactions-they compare the effect of the agents used in combination with the sum of their effects when used alone. This comparison is experimentally straight-forward but, as it is based on assumptions that are wrong, it leads to endless confusion, and conclusions based on it are generally valueless. The correct method for analysing drug interactions is, in most cases, more laborious and involved, but conclusions based on it may be relied on. It is worth pointing out that procedures for analysing interactions between various agents have a wide application in immunology and are not restricted to the study of immunosuppressive agents. Topics eminently suitable for the application of such methods are the study of cell interactions and the treat- ment of neoplasms by combined immunization and chemotherapy. These fields, especially the latter, have suffered from the effects of confused ideas about synergy and they would benefit from the more rigorous approach to be described here. THE FALLACY IN ADDING DRUG EFFECTS A combination of agents that is more effective than is expected from the effectiveness of its constituents is said to show synergy ( other terms loosely used in the immunological literature are augmentation, potentiation, super-additivism and sometimes, quite wrongly, additivism). One less effective than expected is said to show antagonism (or depotentiation, negative interaction, negative synergy, etc.), and one no more and no less effective than expected is said to show additivism. Synergy and antagonism imply that the different constituents affect each other's actions, i.e., they interact pharmacologically; Correspondence: Dr M. C. Berenbaum, Department of Experimental Pathology, St. Mary's Hospital Medical School, London W2 IPG. 2 M. C. Berenbaum additivism implies that they do not. Of course, the key question is what is to be 'expected' from a combination of agents. Most workers simply add the effects of the constituents used separately, and synergy, antagonism and additivism are deemed to be present if the effect of the combination is respectively more than, less than or equal to this sum. For example, Bareham, Griswold & Calabresi ( 1974), in a study of the effect of methotrexate (MTX) and 5-fluorouracil (FU) on the antibody response of mice to sheep red cells, found that 1 mg/kg of MTX suppressed the response by 35Â·9% and 50 mg/kg FU suppressed it by 35Â·4%. They reasoned that both drugs together should suppress the response by 35Â·9+35Â·4%, or 71Â·3%. When, however, both were given (the MTX 30 min before FU), the response was suppressed by 93Â·3%, and they con- cluded that this was a case of synergy. Similar reasoning has been followed by the great majority of workers in this field. The fallacy in this approach may be illustrated as follows. Suppose that two drugs, A and B, each suppress a response by 10% when given at a dosage of 1 mg/kg but that, when both are given together at this dose, the suppression is not 20 but 90%. One might conclude that this is a case of marked synergy but, as the dose-effect curves of these drugs show (Fig. 1 ), a 2 mg dose of either drug alone also produces 0Â·9 0Â·8 0-7 ., 0Â·6 .,, C 8. 0Â·5 ,,, a:: 0Â·4 0Â·3 0Â·2 0Â·1 0 I 0 I A (mg) B (mg) Fm. I. Dose-effect curves of agents A and B, to show that the nature of a drug interaction cannot be deter- mined by adding drug effects. 1 mg of A or of B each suppress the response by 10% and it is known that both together (i.e., 2 mg of a mixture of A and B) suppress it by 90%. This does not indicate synergy, for the dose-effect curves show that 2 mg of either drug alone also suppresses the response by 90%. a 90% suppression. As 2 mg of the combination produces exactly the same effect as 2 mg of either drug by itself, the drugs are clearly not more effective when used in combination than when used singly. It is equally wrong to conclude that a drug combination is synergistic merely because the doses chosen produce no effect on their own but are effective when given together, for dose-effect curves may have thresholds. In Fig. 1, for example, doses up to 0Â·5 mg have no evident effect and, if the effect of a combination were the sum of effects of its constituents, we would have to conclude that any number of doses of 0Â·5 mg given together should also have no effect, which would be absurd. This approach would be correct only if the effects of drugs were simply proportional to dose, when the effect of a dose of one drug would be the sum of the effects of its constituent quanta. If two or more such drugs given together did not interact pharmacologica11y, the effect of the combination should similarly be the sum of effects of its constituent quanta. However, because of the nature of drug-receptor interac- tions, dose-effect curves for biologically active agents are rarely if ever linear (Ariens, 1964). The dose- effect curves for immunosuppressive agents so far investigated have all been found to be markedly non- linear (Berenbaum, 1969) and, considering the complexities of the responses they affect, the likelihood of any not yet investigated being linear appears to be remote, so that this condition is unlikely to be met. There is only one common and relevant situation in which the effect of an agent is proportional to Synergy in immunosuppression 3 dose, and that is when ionizing radiation is used and the duration, not the intensity, of the effect is measured. Proportionality to dose is seen, for instance, in the time taken by irradiated animals to recover immunity to tetanus (Silverman & Chin, 1956), to form serum haemolysins after injection of sheep red cells (Taliaferro & Taliaferro, 1964) and to reject skin allografts (Brent & Medawar, 1966). Ionizing radiation impairs cell reproductive integrity exponentially with dose, and the proportionality could be explained by assuming that (1) depletion of cells by radiation followed a simple exponential curve with no shoulder, so that log surviving fraction of cells was inversely proportional to dose and (2) recovery of the response was due to these survivors proliferating exponentially without a lag and at a rate inde- pendent of the degree of depletion. Thus, if two or more agents behaving like radiation were used together and did not interact pharmacologically, the duration of the effect of the combination could reasonably be expected to be the sum of the durations of the effects of its constituents. However, no other agent has so far been found to resemble radiation in showing simple proportionality between dose and duration of immunosuppression. It appears then that the ideas underlying a good deal of work on drug interactions, and certainly most work on drug interactions in immunology, are wrong. Except in special circumstances, where the effect measured is known to be simply proportional to dose, the effect of a combination of drugs can rarely if ever be expected to be the sum of the effects ofits constituents and, except in these circumstances, the assumption that it should be leads to absurd conclusions. However, it is easy to avoid the snare represented by this approach. The proper way to compare different agents having the same effect and non-linear dose-effect curves is to find what amount or concentration of each produces the same quantitative effect, i.e., to titrate them, a procedure long familiar to immunologists. Titration of different agents to the same, easily identifiable end-point is performed as readily with combinations of agents as with single ones and it can therefore be used to compare the effectiveness of combinations of drugs with the effectiveness of their constituents. This approach avoids the pitfall of non-linearity of dose-effect relations and, as will now be shown, it enables the formulation of unequivocal definitions of synergy, additiveness and antagonism which can be described in simple mathematical terms. METHODS FOR DETERMINING THE NATURE OF DRUG INTERACTIONS Algebraic method. It is self-evident that combinations of various doses of one and the same drug cannot produce effects greater or less than those expected from its dose-effect curve, i.e., such combina- tions can only show additive effects; they cannot be synergistic or antagonistic. Consider, for example, a drug that doubles the survival time of skin allografts when given in a dose of 10 mg. We put this drug into two containers, labelled A and B, and conduct an experiment to find out what combinations of A and B will produce a given quantitative effect (in this case, doubling of graft-survival time). Of course, we find that this specified effect is produced by 10 mg of sample A, or 10 mg of sample B, or 5 mg A+ 5 mg B, or 1 mg A+ 9 mg B, and so on. Expressing this algebraically, let the doses of A and B producing the same quantitative effect (the equi-effective doses) be Ae and Be respectively (in the example we are considering, Ae=Be= 10 mg). Then, if the relation between A and B is additive, our example shows that this specified effect will also be produced by any combination of A and B such that: dose of A+ dose of B __ l. Ae Be (Equation 1) Suppose now that sample B is mixed with equal parts of inert material which has no effect itself and does not influence the effect of the drug. Again, there can be neither synergy nor antagonism between samples A and B, but only additive effects. Now, Ae is 10 mg and Be 20 mg, and a doubling of graft survival time is now produced by 10 mg of A, 20 mg ofB, or by such combinations as 5 mg A+ 10 mg B, 7 mg A+6 mg B, 1 mg A+ 18 mg B, and so on. Equation (1) evidently still holds, so that the fact that the two samples now have different equi-effective doses does not alter the algebraic description of their 4 M. C. Berenbaum additivism. Additive relations between pairs of drugs can therefore be expressed by equation (1), whether their equi-effective doses are the same or not. Now, suppose that A and Bare not samples of one and the same drug but of two different drugs, which may show additivism, synergy or antagonism, as the case may be. If the relation between A and B is additive, it will satisfy equation ( 1 ). If it is synergistic, the drugs will be more effective in combination than separately; therefore, smaller than expected fractions of Ae and Be given together will produce the same effect as either of them given alone, so that the sum of the fractions in equation ( 1) is less than 1. If the relation is antagonistic, greater than expected fractions of Ae and Be are needed to produce this effect when they are given together than when either is given separately, so that the sum of the fractions exceeds 1. The method may be illustrated by the data in Table 1 for suppression of antibody production in mice, taken from Bieber et al. (1962). We first find some effect that is produced by a combination and by each drug alone in the dose-range covered. For thioguanine and thiodeoxyuridine (Table la) we shall use depression of the 'antibody index' to OÂ· 37 of control levels, an effect produced by a combination of OÂ· 3 mg/kg thioguanine and 50 mg/kg thiodeoxyuridine. The equi-effective doses for this effect, found by interpolation in the respective dose-effect curves, are 0Â·9 mg/kg thioguanine and 275 mg/kg thiodeoxyuridine. Substituting in equation (I), we have, for this specified effect: As the sum of these fractions is < 1, this dose-combination is synergistic. Similarly, by interpolation, Table lb suggests that the doses of azathioprine and bromodeoxyuridine that depress the response to 0Â·58 of controls are about 37 mg/kg and 9 mg/kg respectively. In combina- tion, this same effect is produced by 20 mg/kg azathioprine with 10 mg/kg bromodeoxyurine. The respective fractions of the equi-effective doses here are OÂ· 54 and 1 Â· 11. This combination therefore shows antagonism, for the sum of these fractions is greater than 1. It will be obvious that the sum of fractions calculated in this way allows one to quantitate the degree TABLE I. (a) Synergy of 6-thioguanine and 4-thiodeoxyuridine and (b) an- tagonism of Â·azathioprine and 5-bromodeoxyuridine in inhibiting haemag- glutinin response of mice to sheep red cells. Values are 'antibody indices' (Bieber et al., 1962) (a) Dose of thioguanine Dose of (mg/kg) thiodeoxyuridine (mg/kg) 0 0Â·3 lÂ·O 3Â·0 0 1Â·00 0Â·77 0Â·33 0Â·19 50 0Â·72 0Â·37 0Â·16 150 0Â·52 0Â·31 0Â·13 450 0Â·31 (b) Dose of azathioprine Dose of (mg/kg) bromodeoxyuridine (mg/kg) 0 7 20 60 0 lÂ·OO 0Â·92 0Â·72 0Â·48 3 0Â·93 0Â·72 0Â·79 0Â·48 IO 0Â·56 0Â·69 0Â·58 0Â·42 30 0Â·31 0Â·56 0Â·46 0Â·30 Synergy in immunosuppression 5 of synergy or antagonism shown by a drug combination and thus provides a useful yardstick for comparing different combinations. These examples illustrate some of the problems in determining the nature of a drug interaction from examination of tabulated results. Generally, the equi-effective doses must be found by interpolation, and it is almost impossible from a table to envisage the nature of the drug interaction over the whole dose range investigated, especially when many dose-combinations are studied. Geometric method. lsoboles. The algebraic method described above has an elegant geometric counter- part. If combinations of drugs A and B are represented by points in a Cartesian plane, the two axes of which represent doses of A and B, then equation { 1 ), describing the additive relationship, is the equa- tion of the straight line joining Ae and Be (for example, the straight line in Fig. 2). Therefore, whether any particular dose-combination is additive, synergistic or antagonistic is shown immediately by whether CD 0 "' "' 0 a s. Dose of A FIG. 2. Isoboles of additivism, synergy and antagonism. A. and B. are doses of drugs A and B that produce an equal effect, and the three isoboles shown are lines joining points representing dose-combinations of equal effect. For all combinations of A and B represented by points on the additivism isobole, which is the straight line joining A. and B., the sum of the fractions of A. and B. constituting the combination equals I (e.g., combination X consists of0Â·6 A.+0Â·4 B., and so on). For all combinations on the synergy isobole, which is concave, the sum of these fractions is < I (e.g., combination Y consists of0Â·2 A.+0Â·5 B.). For combinations on the antagonism isobole, which is convex, the sum of fractions is> I (e.g. combination Z consists of0Â·5 A0 ---J- 0Â·8 B.). the point representing that combination lies on, below or above the straight line joining the doses Ae and Be of the two drugs that, when given alone, produce the same effect as that combination. When a dose combination is synergistic the sum of the fractions in equation (1) is < 1, and therefore the point representing the combination lies below the line (e.g., combination Yin Fig. 2); when it is antagonistic, the sum of these fractions is > 1 and the point representing it lies above the line { e.g., combination Z in Fig. 2). The line joining all doses and dose combinations producing any given quantitative effect is termed the isobole for that effect. When all the dose combinations producing that effect are synergistic, the isobole is concave; when they are antagonistic, the isobole is convex (Fig. 2). The term 'isobole' was coined by Loewe & Muischnek (1926), who described the characteristic isoboles for synergy, additivism and antagonism, and their application has been discussed in detail by Loewe (1928; 1953; 1957) and de Jongh (1961). However, the method is a good deal older than this, for Fraser (1870-1: 1872) was the first to realize the convenience and expository power of representing drug interactions in this way, and who first illustrated an isobole showing drug antagonism. The data in Table 1 may be used again to illustrate the graphic method. With combinations of thio- deoxyuridine and thioguanine, it is evident that the isoboles for antibody responses 0Â·2, 0Â·3 and 0Â·5 of normal are concave, showing that these two drugs act synergistically. With combinations of azathioprine and bromodeoxyuridine, the isoboles for responses 0Â·5 and 0Â·6 of controls are markedly convex, showing antagonism (Fig. 3). 6 Sf ' CJ' E 300 0 0Â·3 M. C. Berenbaum I 2 Thioguonine (mg/kg) ~ ' O' E 30 -; 20 C ~ :; >- ( bl 5 ~ !0Â·50! J io 1~1:1 ~o~ 3 0-93 0Â·72 0Â·7.: ~ 0Â·19 0 1Â·00 0Â·92 0Â·72 3 0 7 20 40 Azothioprine (mg /kg) 0Â·30 0Â·42 60 FIG. 3. Isoboles showing (a) synergy of thioguanine and thiodeoxyuridine and (b) antagonism of azathioprine and bromodeoxyuridine in inhibiting antibody production in mice. Data of Bieber et al., (1962) as in Table I. Isoboles have been fitted by eye. Boxed figures at the ends of isoboles indicate the effects produced by dose- combinations lying on the respective isoboles. The advantage of the graphic over the algebraic method is that it makes it possible to depict in readily comprehensible form the interactions of two agents in several dose-combinations. In fact, the more combinations used, the more confidence can be attached to the isoboles and the more informative they are. Plotting isoboles makes it easy to identify combinations of maximum synergistic or antagonistic effect, to show which parts of the dose-combination map are worth more detailed exploration and to locate combinations with anomalous effects. Its disadvantage is that it is readily used only with combina- tions of two agents: application to combinations of three agents is cumbersome, for the graph is in three dimensions (see below), and it is not applicable to combinations of four or more agents. The algebraic method has no such limitations. Combinations of three or mfJre agents. The argument used earlier for two samples of the same drug, one of which was mixed with inert material, can be extended to any number of samples mixed with different amounts of inert material. So, equation (1) can be generalized for use with any number of agents: dose of A dose of B dose of C dose of X 1 r dd. Â· Â· (E Â· 2) Â· { < 1 for synergy ----+ + + ... + = ior a 1ttv1sm quat1on Ac Be Ce Xe > 1 for antagonism. For combinations of three agents, equation (2) be.:omes the equation of the plane passing thmugh Ae, Bc and Ce when doses of A, B and C respectively are represented by three coordinate axes. It follows that, if the interaction between drugs A, B and C is additive, and their doses are represented by points on three axes, then the points representing all doses and dose combinations with the same effect as Ae, Be or Ce will lie on the plane connecting these three points. (The two-drug combinations A-B, B-C and C-A are represented by the three lines bounding the plane) (Fig. 4). Synergistic combinations producing the same effect as Ae, Be or Ce are represented by points below the plane; when all dose combinations producing this effect show synergy, they make up a concave isobolar surface. Antagonistic combinations producing this effect are represented by points above the plane and, when they all show antagonism, the isobolar plane is convex. Logarithmic scale isoboles. A problem arises in plotting isoboles when the drugs studied have very shallow dose-effect curves, for the doses used may extend over a wide range, sometimes several orders of magnitude. It is then impracticable to use linear scales for drug doses and logarithmic scales are Synergy in immunosuppression c. Additivism Synergy Antagonism Fm. 4. Isobolar surfaces for additivism, synergy and antagonism with combinations of three agents, A, Band C. A., B. and c. are doses producing an equal effect. The isobolar surface is flat when all combinations of A, B and C producing that effect are additive, and the sum of fractions of A., B. and C. constituting any combination on this surface equals I (e.g., combination X consists of0Â·33 A.+0Â·33 B 0 +0Â·33 C.). The isobolar surface for synergy is concave and the sum of these fractions is < I (e.g., combination Y consists of0Â·25 A.+0Â·25 B.+ 0Â·25 C.). The isobolar surface for antagonism is convex and the sum of fractions is > I (e.g., combination Z consists of0Â·6 A.+0Â·4 B.+0Â·5 C.). 7 preferred. When isoboles are plotted on such scales, they are greatly distorted. Figure 5 illustrates this effect, and shows that the additive isobole becomes markedly convex on a log-log plot and that any less marked convexity, or a straight line, indicates synergy. Examples of such isoboles are given elsewhere (Berenbaum, Cope & Bundick, 1977). It is in any case possible to demonstrate the nature of the drug interaction involved by transferring such isoboles back to the less convenient linear scales, or by measur- ing the sums of the fractions of the equi-effective doses which together form equi-effective combinations as described above. (al 100 10 Syn2 0 10 ( b) Fm. 5. Isoboles of additivism (Add), antagonism (Ant) and synergy, moderate (Syni) or marked (Syn 2 ) plotted on (a) linear and (b) logarithmic scales. When the logarithmic scale covers two or more decades, isoboles with anything less than a marked convexity indicate synergy. HETERERGIC COMBINATIONS The interactions considered above are between drugs each of which can independently produce the effect being measured. Effects that all the constituents of a combination have in common are termed homergic (Loewe, 1957). It is clear that such agents may act either synergistically (Fig. 3a), additively or antagonistically (Fig. 3b) with respect to the effect they have in common. We have now to consider interactions in which one agent increases or decreases the effect of another without itself being able to produce that effect. Effects produced by some but not all the constituents of a drug combination are termed heterergic (Loewe, 1957). One agent may, for example, delay or accelerate the activation, degradation or excretion of another agent, or sensitize cells to its action, without itself having the effect specific to that agent, or it may have an effect itself on the target cells which is the 8 M. C. Berenbaum opposite of the effect under consideration. For instance, the immunosuppressive effects of 6-mercapto- purine are increased by administration of the xanthine-oxidase inhibitor HPP (Elion et al., 1963) or of duazomycin A (Rosenberg & Calabresi, 1963; Vogel & Calabresi, 1969), and the effect of 6-azauracil is said to be increased by chloramphenicol (Fischer, Cassidy & Welch, 1966), but it has been claimed that HPP, duazomycin A and chloramphenicol are not in themselves immunosuppressive. On the other hand, folinic acid antagonizes the immunosuppressive effects of methotrexate (Berenbaum & Brown, 1965), the microsomal enzyme inhibitor SKF-525A reduces the immunosuppressive effects of cyclo- phosphamide (in rats) (Berenbaum, Cope & Double, 1973) and cysteine reduces the immunosuppressive effects of nitrogen mustard (Addison & Berenbaum, 1971 ). Claims that an effect is heterergic rather than homergic require to be made with caution for they depend on showing that the agent lacking the specific effect under consideration does not do so merely because of inadequate dosage. No claim so far is entirely satisfactory in this respect. In the work of Elion et al. (1963) and Fischer et al. (1966), both HPP and chloramphenicol showed weak immuno- suppressive activity in the single experiments in which they were tested alone and, in the work of Rosenberg & Calabresi (1963) and Vogel & Calabresi (1969), duazomycin A was tested only in low and non-toxic dosage. Folinic acid, SKF-525A and cysteine have not so far been adequately investigated for possible immunosuppressive effects but this is perhaps unimportant in the present context, for the mechanisms by which they antagonize the effects of methotrexate, cyclophosphamide and nitrogen mustard are well understood. In the pharmacological literature, heterergic effects have acquired their own special terminology (Loewe, 1957; de Jongh, 1961). Ignoring this, as it complicates matters unnecessarily, we may say that heterergic combinations, like homergic, may show synergy (when the immunologically ineffective agent increases the effect of the immunosuppressive agent), antagonism (when it reduces its effect) or addi- tivism (when it has no modifying effect), and there is no difficulty in using equation (2) to analyse such cases. The fraction corresponding to the immunologically ineffective agent in equation (2) is zero as its equi-effective dose is theoretically infinite, and isoboles for such combinations do not intersect the dose-axis of the ineffective agent. EXPERIMENT AL DESIGN Minimum requirements for demonstrating synergy or additivism Equation (2) shows that dose combinations claimed to show synergy must be made up of fractions of equi-effective doses of the agents used alone such that the sum of these fractions is less than 1. For dose combinations claimed to be additive, the sum of the fractions must be L It follows that, in experi- ments designed to investigate possible synergy or additivism, the dose of each agent in the combination(s) in question must be less than at least one of its doses used alone, otherwise the sum of the fractions can never be shown to be < L Therefore, each agent must be tested in at least the following three dose levels: (a) one level of the agent used alone; (b) one level, less than (a), used in a combination; (c) a nil-dose level (as a corollary of (a), if each agent is tested alone, each of its partners must be tested at a nil-dose level). Similarly, to show that an isobole runs a concave or straight course between its two ends, we require at least three points, i.e., its ends (each representing a dose of each agent used alone with a nil dose of the other) and at least one intermediate point (representing a combination in which each agent is given in a dosage less than its dose used alone). Experiments in which any of the agents is tested at less than these three levels can therefore never demonstrate synergy or additivism. This simple consideration alone is enough to invalidate the great majority of published claims for synergy in immunosuppression, for most of these are based on experi- ments in which each agent was tested in only two doses, including a nil dose, forming a 2 x 2 array when plotted graphically ( or a 2 x 2 x 2 array when three agents were used). Consider, for example, the data of Bareham et al. (1974) referred to at the outset of this paper, in which 1 mg/kg MTX and 50 mg/kg FU together (the former given 30 min before the latter) sup- Synergy in immunosuppression 9 pressed the response by 92Â·3%, and each given alone caused 35Â·9 and 35Â·4% suppression respectively. Substituting in equation (2) for 92Â·3% suppression we have: 1 50 -1+~0<2-> >) This result does not permit us to draw any conclusions as to the nature of the interaction involved- it is consistent equally with synergy, additiveness and antagonism and, when the data are plotted graphically, it is evident that the shape of the isobole for the effect of the combination is quite inde- terminate (Fig. 6). 50 35Â·4 0 0 0 MTX (mg/kg) ? -~ 92Â·3 ""'? 35Â·9 FIG. 6. Impossibility of demonstrating synergy with a 2 x 2 array of dose-levels. Data ofBareham et al., (1974) for suppression of plaque-forming cell response of mice by methotrexate and fluorouracil, the former being given 30 min before the latter. The shape of the isobole for 92Â·3% reduction in the response cannot be deter- mined from the data. The same consideration applies to all investigations in which one of the agents is tested at two levels only (including the nil-dose level), irrespective of the number of dose levels of the other(s). The work of Hansen et al. (1964) will illustrate this point. These authors claimed that azathioprine and azaserine acted synergistically in suppressing antibody production in the rat. Azaserine was tested at five dose levels but azathioprine at two only, 0 and 10 mg/kg. The impossibility of demonstrating synergy or additivism with this sort of experimental design is illustrated by Fig. 7. The ten groups form a 2 x 5 rectangular array and therefore cannot provide the information required for drawing a concave or straight isobole, whatever the results. Requirements for demonstrating antagonism Antagonism is obviously present when a drug combination is less effective than one or more of its constituents (for instance, in Fig. 3b, the combinations of 7 mg/kg of azathioprine with IO or 30 mg/kg bromodeoxyuridine have less effect than these doses of bromodeoxyuridine on their own). Positive claims to have demonstrated antagonism in immunosuppression usually refer to antagonism of this marked degree, which needs no elaborate procedure for its detection. In such cases, the isobole is so convex that it forms an angle of more than 90Â° with one or more of the dose-axes (Fig. 3b, Fig. 10). The situation is not so simple when antagonism is less marked and the combination is more effective than any of its constituents used alone. Antagonism is then detectable only by making measurements sufficient to show that the sum of fractions in equation (2) is > I or that the isobole for the effect of the combination is convex. This sort of situation causes great confusion when incorrect criteria are used for 10 "' .Â£ 1i 0 :.c 0 :;_ 10 210 104 0 360 140 0 0Â·25 M. C. Berenbaum 0 \ 0Â·5 lÂ·O 2Â·0 Azosc,,ne (r.-.g/kg/doy) Fm. 7. Impossibility of demonstrating synergy with a 2x n array of dose-levels. Data of Hansen et al. (1964) for suppression of production of antibody to a mouse tumour in the rat by combinations of azathioprine and azaserine. Values are titresx 10- 6 â€¢ The approximate directions of the lower parts of the isoboles can be guessed (they are drawn here as straight lines although there is no evidence from the data that they must be so), but the ends of these isoboles on the azathioprine dose-axis, and therefore their overall shapes, are quite in- determinate. Although as many as ten groups of animals were used, the experimental design (a 2 x 5 array) has made it impossible to demonstrate synergy. analysing drug interactions. For example, Roth, Friedman & Syverton (1957) found that Candida- infected mice given cortisone and radiation together had a shorter survival time and higher mortality than mice given either agent alone, and they concluded from this that the combination was synergistic. Their results are shown in Fig. 8a and h, and it is clear from the convexity of the isoboles that the com- bination was, in fact, antagonistic, a conclusion that may readily he confirmed by appropriate use of equation (2). It should he emphasized further that a combination may be antagonistic even when its effect exceeds the sum of the effects of its constituents, as shown in Fig. 8c. This shows again, if any further demonstration were needed, that drug interactions cannot usefully he examined by comparing the effect of a combination with the sum of the effects of its constituents. (a) ( b) 400 8 6 4 71 79 83 3 16 ~ ., "' ---........ 3~6~58 0 -0 0 -~ 200 18 '\ 12 "' 0 .31 '5 0 a:: 0 >30 18 14 17 33 50 0 O 2 4 0 2 4 0 2 3 Corlisone acetate {mg/kg) Oc,e of X Fm. 8. Antagonism may be present when the effect of a combination exceeds that of any of its constituents or even the sum of their effects. (a) and (b) show the effect of cortisone acetate and radiation on mean survival time in days (a) and per cent mortality (b) in mice infected with Candida albicans (Roth et al., 1957). The authors concluded that the combinations were synergistic because their effects exceeded those of their constituents, but the convex isoboles show that they were, in fact, antagonistic. (c) The four hypothetical combinations represented by the ringed points are all antagonistic, as shown by convex isoboles, but the effect of each exceeds the sum of the effects of its constituents. Synergy in immunosuppression Determination from incomplete data 11 Ideally, experiments on drug interactions should be designed to allow determination of equi-effective doses for all the agents used, which means constructing a sufficiently detailed dose-effect curve for each agent. However, it is often possible to demonstrate synergy or antagonism without going to these lengths, as all we require to know is whether the sum of the fractions in equation (2) is less or more than 1. We may be sure that this is the case if the minimum or maximum limits for doses of the agents used alone that are equi-effective with the combination examined are sufficiently high or low, even if we do not know their values more precisely. A case in point is provided by the data of Friedman, Gelfand & Bernheimer (1971), who measured Â·;:; ::, .0 E 0 2 .s:: L) 2 8Â·3 I 12Â·5 66Â·7 0 0 0 27Â·8 0 10 100 Methylprednisolonc ( mg /ky) FIG. 9. Demonstration of synergy with incomplete data. Effect of chlorambucil and methylprednisolone on tetanus antitoxin production in mice. Selected data of Friedman et al. (1971). Values are percentages of animals in which drugs prevented the protective effect of immunization. The isobole for 66Â·7% prevention must be concave, indicating synergy, although the doses of the two drugs that produce this effect on their own were not determined. the effect of chlorambucil and methylprednisolone on tetanus antitoxin production in mice (assayed by protection against toxin). The data are illustrated in Fig. 9. Abolition of protection in 66Â·7% of animals is obtained with a combination of 1 mg/kg chlorambucil and 10 mg/kg methylprednisolone but, if these drugs are used alone, it would require considerably more than 2 mg/kg chlorambucil or 100 mg/kg methylprednisolone to produce the same effect. Substituting in equation (2) we have: 1 10 >2 + > 100 = < 0Â·6Â· As this sum is less than 1 (and the isobole for 66Â·7% protection is correspondingly concave), this com- bination is undoubtedly synergistic, although these results do not enable us to say what doses of the two agents alone would produce this effect nor, indeed, do they enable us to determine the equi-effective doses for any effect. An even simpler experimental design may suffice to prove the existence of antagonism. Consider, for example, the data of Bareham et al. ( 1974) for the effect of 5-fluorouracil and methotrexate on the antibody response of mice to sheep red cells, illustrated in Fig. 10. A 12Â·4% suppression of the response is produced by 1 mg/kg of methotrexate with 50 mg/kg fluorouracil provided the former is given 30 min after the latter. We do not know what doses of these drugs separately have this effect, only that they are respectively, less than 1 mg/kg methotrexate and less than 50 mg/kg fluorouracil. Substituting in equation (2) we have: 12 M. C. Berenbaum 50 I <50+ 2Â· AB this sum exceeds 1, the combination is antagonistic, as shown also by the marked convexity of the isobole for this effect. 50 35Â·4 0 0 35.9 0 MTX !mg/kg) FIG. 10. Demonstration of antagonism with a 2 x 2 array of dose-levels. Data of Bareham et al. (1974) for suppression of plaque-forming cell response of mice by methotrexate and fluorouracil, the former being given 30 min after the latter. The isobole for 12Â·4% reduction in the response must be markedly convex, showing antagonism, although the doses of the two agents that produce this effect on their own were not determined. However, in most published work on drug interactions in immunosuppression, the results do not allow us to obtain a value for the sum of the fractions that is unequivocally less or greater than 1. Efficiency and economy of experimental design. Diagonal arrays Inspection of Fig. 2 shows that, for combinations of two agents, departure from linearity of the isobole for any specified effect would most efficiently be detected by testing the combination at the mid-point of the straight line joining the equi-effective doses, i.e., that made up of half the equi-effective dose of each agent. For combinations of three agents, Fig. 4 shows that concavity or convexity of the isobolar surface is detected most easily by testing the combination at the mid-point of the flat plane joining equi-effective doses i.e., that made up of one-third of the equi-effective dose of each agent. In either case, the degree of synergy or antagonism is measured by titrating along the line joining the point in question to the origin, the arrangement thus constituting what will be termed a diagonal array. In general, therefore, the most efficient and economical procedure for n agents is as follows: (a) Test each agent alone at several dose levels. Construct dose-effect curves and so find doses of the different agents that have equal specified effects (e.g., reduction of antibody levels to 0Â·5, 0Â· 1 and 0Â·01 of controls, prolongation of graft survival by 5, 10 and 20 days, and so on). Usually these equi-effective doses must be found by interpolation in the dose-effect curve, and the confidence to be attached to them depends on the closeness of the dose levels tested, the errors of the measurements and the regularity of the curve. (b) Test each agent again (1) at the above-determined equi-effective doses for one (or more) of the specified effects and (2) in combinations made up of 1/n of the equi-effective dose of each agent, and of serially decreasing or increasing doses in the ratio of this combination ( represented by the line passing through this combination and the origin) so as to find a combination that produces the specified effect. Determine the nature of the drug interaction by use of equation (2) or by plotting isoboles. (c) Explore in more detail doses and dose-combinations found to be of interest in step (b). Synergy in immunosuppression 13 VALID CLAIMS FOR SYNERGY IN IMMUNOSUPPRESSION Some fifty-six papers dealing with synergy or additiveness in immunosuppression have been examined, appearing principally in the main immunological journals over the past 25 years. These may be taken as representative of published work in this field and probably constitute the bulk of it. These papers incorporate 104 claims that particular combinations of immunosuppressive agents are synergistic and seventeen claims to have shown additivism. The great majority of these claims can be dismissed without further consideration because the experimental design used (a 2 x 2 array or other 2 x n array) could not have demonstrated the existence of synergy or additivism even if they had been present. The most consistent offenders are workers in transplantation; these account for seventy-seven of the claims, only one of which is valid (Floersheim, 1969), and even this was based on the incorrect grounds that the effect of the combination exceeded that 'which would have been obtained by simple addition of the individual prolongations'. In fifteen papers, the use of 3 x 3 or higher order assays is reported, and the validity of the claims has been tested by using the results to calculate in each case the sum of fractions in equation (2). In three of the papers (Roth et al., 1957; McKneally et al., 1964; Fischer el al., 1966), none of the combinations tested was synergistic according to the criteria used here. The remaining papers are summarized in Table 2. Fourteen of the claims for synergy are valid. In another seven cases, equi-effective doses for use in equation (2) had to be found by interpolation between widely spaced dose levels. They are there- fore approximate only and the sum of fractions is accordingly uncertain but, where it is probably < l, the claim for synergy has been deemed to be probably valid. Perhaps the most important conclusion to be drawn from Table 2 (and the papers excluded from it) is that the paucity of valid claims for synergy in immunosuppression may be due mainly to the wide- spread use of inappropriate experimental designs. When appropriate designs are employed, commonly used immunosuppressive agents quite often show synergy. It should be emphasized, however, that many of the claims deemed to be valid or probably valid were based on single experiments and/or single dose-combinations in experimental systems of notorious inherent variability, and they therefore require confirmation. There is little point in detailing here the invalid claims that constitute the vast majority. However, it is worth noting that, in the papers reviewed, only one of the groups of workers investigating the potentially promising manoeuvre of treating transplant recipients with combinations of donor antigen and immunosuppressive agents, especially ALG, adopted an experimental design capable of demonstrat- ing synergy. Yet their claims to have shown that such combinations are synergistic underlie the belief that 'this is probably the next major application of antilymphoid globulin' (Monaco & Codish, 1976). (The sole exception was that of McKneaJly et al. (1964), and their claim that azathioprine and radiation acted synergistically with donor antigen was invalid.) It should also he pointed out that at least one current hypothesis about the mechanism of immunological enhancement (Davies, 1976) is said to be supported by the finding that enhancing antisera and ALG act synergistically (Russell, 1971; Batchelor, Fabre & Morris, 1972). However, these claims were based on experiments with 2 x 2 arrays of dose levels which, as shown above, cannot detect synergy. Of course, these criticisms of the bulk of published work do not imply that combinations claimed to be synergistic were not so, but that the experimental designs used were incapable of demonstrating the fact. As explained earlier, the detection of marked antagonism presents little difficulty; claims to have demonstrated this are generally correct and will not be examined further here. It should he noted, however, that the confusion in this field is so pervasive that it leads some authors to overlook (Cerilli & Hattan, 1972) or positively to deny (Koene et al., 1975, Table 3) the existence of antagonism shown in their experiments, and even (Roth et al., 1957) to claim that clearly antagonistic combinations are synergistic (Fig. 8). TABLE 2. Valid claims for synergy in immunosuppression Experi- mental Reference Experimental system Agents designâ€¢ Antibody production Bieber et al. (1962) Table l CD mouse, SRC agglutinins 6MP, BUdR 4x4 Table 2 CD mouse, SRC agglutinins 6TG, ThioUdR 4x4 Elion et al. (1963) Table 2 C57BL mouse, SRC agglutinins 6CP, HPP 4x3 Friedman et al. (1971) Tables 1, 2 Swiss mouse, antitetanus immunity MePr, Chlor 6x6 Tables 1, 2 Swiss mouse, antitetanus immunity MePr, AzaT 6x6 Tables 1, 3 Swiss mouse, antitetanus immunity MePr, Rad 6x4 Tables 1, 3 Swiss mouse, antitetanus immunity AzaT, Rad 6x4 Tables 1, 2 Swiss mouse, antitetanus immunity MePr, AzaT, Chlor 6x6x6 Tables 1, 3, 4 Swiss mouse, antitetanus immunity MePr, AzaT, Rad 6x6x4 Tables 1, 3, 4 Swiss mouse, antitetanus immunity AzaT, Chlor, Rad 6x6x4 Tables 1, 3, 4 Swiss mouse, antitetanus immunity MePr, AzaT, Chlor, Rad 6x6x6x4 Hyman et al. (1975) Figs 2, 3, 5, 6, 7 NZB/W mouse, antibody to poly I. Cyclo, MTX 7x 7 poly C. Combina- Relevant tions combina- tested tionst 9 4 4 4 6 2 l 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I 5 5 Sum of fractionsi 0Â·75, <0Â·67, 0Â·75, <0Â·43 <0Â·60, <0Â·67, 0Â·52, <0Â·44 < 0Â·46, < 0Â·68 <0Â·25 <0Â·22 <0Â·46 <0Â·58 <0Â·34 <0Â·51 <0Â·58 <0Â·60 2Â·9, 0Â·89, 0Â·46, 0Â·81, 2Â·13 -..... ~ ~ . ~ ~ ~ ~ I;:! ~ Graft rejection Floersheim (1969) Table 3 BALB/c skin to CBA Graft-vs-host disease Glucksberg & Fefer (1973) Tables I, 2 C57BL cells to BALB/c Autoimmune disease Gelfand & Steinberg (1972) Tables 1, 2} Gelfand et al. (1972) Table I NZB/W mouse nephritis Resistance to infection Syverton et al. (1952) Fig. 2 Swiss mouse, Coxsackie virus, 30-day survival In vitro systems Mendelsohn et al. (1973) Table 1 Human lymphocytes, PHA response Table I Human lymphocytes, unstimulated, thymidine uptake Schrek & Stefani (1976) Table 3 Human lymphocytes, unstimulated, 5-day survival Berenbaum et al. (1977) Fig. 1 Human lymphocytes, PHA response ALS, Ro-4/6824 4x 3 Cyclo, Procarb. Sx 5 AzaT, MePr, Cyclo 3x3x3 Rad, Cort Ac 6x6 Cortisol, DBcAMP 3x3 Cortisol, DBcAMP 3x 3 CArab, Rad 3x 3 Cortisol, PGE2 4x6 6 4 4 4 4 15 2 4 8 <0Â·79, <1Â·07 <0Â·58 < lÂ·O 0Â·37, 0Â·42, 0Â·52, CÂ·65 0Â·32 0Â·19 <0Â·66 90% of values < 1 ; 50% <0Â·25 Where more than one relevant combination was tested, fractions are given in the order of the combinations in an isobologram, from top left to bottom right, the dose-scale of the first named agent along the abscissa. Fractions in bold type show valid claims for synergy, i.e., doses of the individual agents equi-effective with the combinations tested were found by interpolation between reasonably close dose-levels or their lower limits could be determined. Fractions < 1 not in bold type show claims deemed probably valid, i.e., lower limits for the equi-effective doses could not be determined or dose levels were too widely spaced for confident interpolation. Aza T = azathioprine; BU dR = 5 bromodeoxyuridine; CArab = cytosine arabinoside; Chlor = chlorambucil; 6CP = 6-chloropurine; Cyclo = cyclophosphamide; Cort Ac= cortisone acetate; DBcAMP = dibutyryl cyclic AMP; HPP = 4-hydroxypyrazolo (3,4-d) pyrimidine; MePr = methylprednisolone; 6MP = 6-mercaptopurine; MTX = methotrexate; Procarb = procarbazine; Rad= radiation; ThioUdR = 4-thiodeoxyuridine; 6TG = 6-thioguanine; SRC = sheep red cells. â€¢ Number of dose-levels of each agent, including nil dose, in same order as in preceding column. t Combinations that include the maximum dose level of any agent are irrelevant for determination of synergy (see text). ; Calculated by equation (2). ~ ~ "" 6 ~Â· Â§Â· ~ ~ ;-:::: <:::) .... ~ l cÂ· ~ Vl 16 M. C. Berenbaum CLINICAL IMPLICATIONS The demonstration that a particular combination of drugs shows synergy in the effect being measured does not guarantee that it will show synergy with regard to other effects. Indeed, if all the biological effects of the combination showed equal degrees of synergy or antagonism, there could be no therapeutic gain or loss. Synergy in the pharmacological sense would be disadvantageous and antagonism ad- vantageous if they were more marked for the toxic than the therapeutic effects of the drugs used. It might seem therefore that, in practice, we need not be concerned unduly with the detailed analysis of drug interactions and that what is important is simply the relation between the therapeutic and toxic effects of various dose-combinations, which may be determined without elaborately measuring equi- effective doses, plotting isoboles, and so on. Certainly, information of great practical use may be obtained in this way, but this is not an argument for restricting investigations to this simple level, but for extending them. If a dose combination of two or more agents, chosen more or less at random from the innumerable possible combinations of those agents, turns out to be therapeutically better than some other combination, is it not highly probable that other combinations, not tested, will be better still? Indeed, the laws of chance, and the fact that isoboles for qualitatively different effects such as immunosuppression and 0Â·60 10Â·5e1 1Â·0 0Â·79/33 0Â·50/27 oi Aâ€¢ .JC -~ -- -.!:; - - - 0 â€¢s ..... [fill -~ .... / C: ...... ~ 0Â·5 0Â·81/8 -- 0-5~1~ -Q. E "' -~ ... ..2 .c - -(.) 0 1â€¢0/0 0Â·58/0 0Â·56/4 0 90 180 270 6-Azauracil (mg/kg) Fm. 11. Immunosuppressive and toxic elfects of chloramphenicol and 6-azauracil in the mouse. (Data of Fischer et al., 1966.) In each pair of values, the first indicates the 'antibody index' (Bieber et al., 1962) for the response to sheep red cells, the second the percent mortality. Isoboles for antibody response (--), mortality) (- - -). Boxed figures indicate the elfect produced by dose-combinations on the isobole. toxicity are not parallel, make this a virtual certainty. Consider, for instance, the experiments of Fischer et al. (1966), who measured both the immunosuppressive and toxic effects of combinations of chloram- phenicol and 6-azauracil in mice. Their results are illustrated in Fig. 11, which shows isoboles for the degree of immunosuppression and for mortality. It is immediately apparent, from the positions and directions of these isoboles, that combinations A, B, C and D for example, in that order, show not only increasing immunosuppressive effects but decreasing toxicity and therefore that, in searching for com- binations of these two drugs of optimal therapeutic effectiveness, the region most profitable to explore is the bottom right-hand corner of the graph and the area to the right of this. It is unlikely that this conclusion would easily have been reached as a result simply of comparing one or two randomly selected combinations. Optimum therapy therefore requires the exploration of a range of dose combinations sufficient to provide the information for constructing a map of isoboles for both therapeutic and toxic effects. Further, without constructing such maps, it may be difficult to detect, or even suspect, the existence of markedly anomalous effects produced in particular ranges of dose-combinations. This consideration is especially important when it is proposed to explore in man therapeutic regimens sug- gested by work in laboratory animals. ,. Synergy in immunosuppression 17 Many substances found in vivo have immunosuppressive effects, for instance, corticosteroids, prosta- glandins, some tumour products (such as Q'.-foeto-protein), certain a-globulins, fibrin-degradation products, interferon and probably other, as yet, unidentified materials. In view of the frequency of immunological impairment found in malignant disease, in some chronic non-malignant diseases, after trauma, during pregnancy and so on, the possibility that some of these substances act synergistically to cause clinically significant immunosuppression requires examination (Berenbaum, Cope & Bundick, 1977). The relation between immunosuppressive and toxic effects is irrelevant here and, in any case, most of these substances are not significantly toxic at levels occurring in vivo. Investigation of the possibility that depressed immunity in vivo is due to synergistic interactions of a number of naturally occurring substances requires exploration of the whole range of concentrations found in vivo. Lastly, it is not only for scientific reasons that the study of synergy is important. Cost and shortage of supply are potent factors (although not the only ones) in limiting the dosage in man of at least one agent, ALG. Indeed, this limitation may be partly responsible for the doubtful therapeutic status of this agent in man, for there is no doubt of its effectiveness in laboratory animals which are given pro- portionately much higher doses. ALG of greater selectivity and less toxicity than at present available, and therefore potentially capable of being given in larger doses, will doubtless be even more expensive and limited in supply, and the same may be true of sera with special selectivities, e.g., for particular sub-populations of T cells or B cells, that may eventually be required clinically. There is thus a direct economic motive in searching for synergistic combinations of ALG and other antisera with other agents, so as to permit the use of smaller amounts of serum, even where toxicity is not a limiting factor. I am indebted to the Medical Research Council and Cancer Research Campaign for financial support. REFERENCES ADDISON, I.E. & BERENBAUM, M.C. (1971) The effect of cysteine on the immunosuppressive activity of busul- phan, cyclophosphamide and nitrogen mustard. Brit. :J. Cancer 25, 172. ARIENS, E.J. (ed.) (1964) Molecular Pharmacology. The Mode of Action of Biologically Active Compounds, volume 1 Academic Press, New York. BAREHAM, C.R., GRISWOLD, D.E. & ULABRESI, P. (1974) Synergism of methotrexate with lmuran and with 5- fluorouracil and their effects on hemolysin plaque-forming cell production in the mouse. Cancer Res. 34, 571. BATCHELOR, J.R., FABRE,}. & MORRIS P.J. (1972) Passive enhancement of kidney allografts. Potentiation with anti- thymocyte serum. Transplantation, 13, 610. BERENBAUM, M.C. (1969) Dose-response curves for agents that impair cell reproductive integrity. A fundamental difference between dose-response curves for antimetabo- lites and those for radiation and alkylating agents. Brit. :J. Cancer, 23, 426. BERENBAUM, M.C. & BROWN, I.N. (1965) The effect of delayed administration of folinic acid on immunological inhibition by methotrexate. Immunology, 8, 251. BERENBAUM, M.C., CoPE, W.A. & BUNDICK, R.V. (1977) Synergistic effect of cortisol and prostaglandin E 2 on the PHA response. Relation to immunosuppression in- duced by trauma. Clin. exp. Immunol. (In Press.) BERENBAUM, M.C., COPE, W.A. & DOUBLE, J.A. (1973) The effect of microsomal enzyme inhibition on the immunosuppressive and toxic effects of cyclophospha- mide. Clin. exp. Immunol. 14, 257. BIEBER, s., ELION G.B., HITCHINGS, G.H., HOOPER, D.C. & NATHAN, H.C. (1962) Suppression of the immune re- sponse by drugs in combination. Proc. Soc. exp. Biol. (N. Y.), Ill, 334. BR.;NT, L. & MEDAWAR, P. (1966) Quantitative studies on B tissue transplantation immunity. VIII. The effects of irradiation. Proc. roy. Soc. B. 165, 4I3. CERILLI, J. & HATTAN, D. {1972) The effect of combined immunosuppressive agents and cell-free antigen on xenograft survival. Transplantation, 14, 207. DAVIES, D.A.L. (1976) The true mechanism of immuno- logic enhancement, deduced from studies of heart transplants in rats. Transplant. Proc. 8, 37. ELION, G.B., CALLAHAN, S., NATHAN, M., BIEBER, s., RUNDLES, R.W. & HITCHINGS, G.H. (1963) Potentiation by inhibition of drug degradation: 6-substituted purines and xanthine oxidase. Biochem. Pharmacol. 12, 85. FISCHER, D.S., CASSIDY, E.P. & WELCH, A.D. (1966) lmmunosuppression by pyrimidine nucleoside analogs. Biochem. Pharmacol. 15, IOI3. FLOERSHEIM, G.L. (1969) A study of combined treatment with chemical immunosuppressants and antilymphocytic serum to prolong skin allograft survival. Transplantation, 8, 392. FRASER, T.R. (1870-1) An experimental research on the antagonism between the actions of physostigma and atropia. Proc. roy. Soc. Edinb. 7, 506. FRASER, T.R. (1872) The antagonism between the actions of active substances. Brit. med. :f. ii, 485. FRIEDMAN, E.A., GELFAND, M. & BERNHEIMER, H.P. {1971) Synergism in immunosuppression. l. Effect of methyl- prednisolone, azathioprine, chlorambucil, and radiation on tetanus antitoxin production. Transplantation, ll, 479. GELFAND, M.C. & STEINBERG, A.D. (1972) Therapeutic studies in NZB/W mice. II. Relative efficacy of azathio- prine, cyclophosphamide and methylprednisolone. Arthr. and Rheum. 15, 247. GELFAND, ~1.C., STEINBERG, A.D., NAGLE, R. & KNEP- SHIELD, J.H. (1972) Therapeutic studies in NZB/W mice. 1. Synergy of azathioprine, cyclophosphamide and methyl- 18 M. C. Berenbaum prednisolone in combination. Arthr. and Rheum. 15, 239. GLUCKSBERG, H. & FEFER, A. (1973) Combination chemo- therapy for clinically established graft-versus-host disease in mice. Cancer Res. 33, 859. GOLDIN, A. & MANTEL, N. (1957) The employment of combinations of drugs in the chemotherapy of neo- plasia: a review. Cancer Res. 17, 635. HANsEN, H.J., VANDEVOORDE J.P., BENNETT K.J., GILES, W.G. & NADLER, S.B. (1964) Azaserine and thiopurines. I. Inhibition of S-180 mouse tumor and antibody syn- thesis . .1. Lab. clin. Med. 63, 801. HYMAN, L.R., KOVACS, K. & STEINBERG, A.D. (1975) Drug-induced tolerance: selective induction with im- munosuppressive drugs and their synergistic interaction. Int. Arch. Allergy, 48, 248. DE }ONGH, S.E. (1961) lsoboles. Quantitative Methods in Pharmacology (ed. by H. de Jonge), p. 318. North- Holland Publishing Co., Amsterdam. KoENE, R.A.P., GERLAG, P.G.G., McKENz1E, I.F.C., HAGEMANN, J.F.H.M. & WIJDEVELD, P.G.A.B. (1975) Enhancement of skin grafts in the mouse. The combined use of specific and non-specific immunosuppression. Transplantation 20, 241. LOEWE, S. (1928) Die quantitativen Probleme der Pharma- kologie. Ergebn. Physiol. 27, 47. LoEWE, S. (1953) The problem of synergism and antagonism of combined drugs. Arzneimitteiforsch. 3, 285. LOEWE S. (1957) Antagonisms and antagonists. Pharmacol. Rev. 9,237. LOEWE, S. & Mu1sCHNEK H. (1926) Ober Kombinations- wirkungen. I. Mitteilung: Hilfsmittel der Fragestellung. Arch.f. exper. Path. u. Pharm. 114, 313. McKNEALLY, M.F. KELLY, W.D., RESNICK, J. & MARTINEZ, C. (1964) Induction of tolerance with disrupted antigen, Imuran, and sublethal irradiation. Surg. Forum 15, 131. MENDELSOHN, J., MULTER, M.M. & BOONE R.F. (1973) Enhanced effects of prostaglandin E 1 and dibutyryl cyclic AMP upon human lymphocytes in the presence of cortisol. .1. clin. Invest. 52, 2129. MONACO, A.P. & ComsH, S.D. (1976) Survey of the current status of the clinical uses of antilymphocyte serum. Surg. Gyn. Obstet. 142, 417. ROSENBERG, S. & CAl.ABRESI, P. (1963) Enhanced sup- pression of the secondary immune response by combina- tion of 6-mercaptopurine 'Duazomycin A'. Nature (Lond.), 199, 1101. RoTH, F.]., FRIEDMAN, J. & SYVERTON, J.T. (1957) Effects of roentgen radiation and cortisone on susceptibility of mice to Candida albicans . .1. Immunol. 78, 122. RussELL, P.S. (1971) Immunological enhancement. Trans- plant. Proc. 3, 960. SCHREK, R. & STEFANI, S.S. (1976) Cytarabine: cytocidal effect on normal and leukemic lymphocytes. Synergism with X-rays and comparison with mechlorethamine. Exper. mo/. Path. 24, 84. SILVERMAN, M.S. & CHIN, P.H. (1956) The effect of whole body X-irradiation of mice on immunity to tetanus toxoid. II. The delayed immune response to injections of tetanus toxoid . .1- Immunol. 77, 266. SYVERTON, J.T., WERDER, A.A., FRIEDMAN, J., ROTH, F.J. JR., GRAHAM, A.B. & M1RA, O.J. (1952) Cortisone and roentgen radiation as synergistic agents for production of lethal infections. Proc. Soc. exp. Biol. (N. Y.), 80, 123. TALIAFERRO, W.H. & TALIAFERRO, L.G. (1964) The relation of radiation dosage to enhancement, depression, and recovery of the initial Forssman hemolysin response in rabbits . .1. infect. Dis. 114, 285. VOGEL, C.L. & CALABRES1, P. (1969) Enhanced suppression of experimental allergic encephalomyelitis by combination therapy with duazomycin-A and 6-mercaptopurine. Proc. Soc. exp. Biol. (N. Y.), 131, 251.