Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
DEMONSTRATION OF MACHINE BIOCHEMORPHOLOGY CORRELATION OF ANTIHISTAMINIC ACTIVITY AND CHEMICAL STRUCTURE BY MEANS OF THE NATIONAL RESEARCH COUNCIL CHEMICAL-BIOLOGICAL COORDINATION CENTER IBM PUNCH CARDS David Fielding Marsh Professor of Pharmacology West Virginia University Morgantown, West Virginia
142 INTRODUCTORY NOTE The following material has been prepared to illustrate the type of machine derivation of simple generalizations between biological activity and chemical constitution, and correlative data, that the Center can supply. There are certain limitations inherent in any coding and machine sorting procedure; these will become apparent. With certain types of questions, the machines can save months of effort by employing the punch cards already prepared from the extensive data compiled by the Center. The present machines are not intended to replace the various abstracting journals for answering simple search questions. For coordination work, any problem that involves much less than a thousand compounds or entries can usually be handled by visual inspection. Simple card systems, such as the Keysort, are usually adequate for five hundred to twenty-five hundred entries if a limited number of types of data are recorded. The machines become useful when a large number of compounds with a large number of different or qualified types of data are available. With such problems the machines can rapidly sort the compounds into groups so that only a few cards are left for final analysis. The present problem, that of determining any possible generalizations concerning the relationships between chemical constitution and antihistaminic activity, has been chosen since the f;eld is a large one, most biological and chemical workers are generally aware of the results, and a large number of modern entries are available. Since the antihistaminic structure-activity relationships have been surveyed in several excellent review articles and are widely known, the problem is an excellent one to demonstrate the usefulness of the procedure and to stimulate the formulation of future problems. Since the general solution to the problem is known to most workers in this field, the development may seem naive. As far as humanly possible, no recourse to prior knowledge has been used to reach any preconceived or expected results, and the machines have used only such cards as would be routinely available at the time of any sorting. Obviously, errors committed by the original worker, publisher of journal article, abstracter, coder, punch operator, or machine operator will not be eliminated by the machines. In a general problem, a few mistakes are not important. However, any compound that is apparently missing in a series, or any compound or experimental result that appears out of place can be easily checked by referring to the chemical file, the data sheets, or the original publication. It is necessary to have some definition of terms. Antihistaminic activity may have several meanings. Considerable confusion exists in some of our minds concerning "activity" and "use". For example, some of you know that clinical use of a few of the agents in this discussion is not predicated on their relative activity at all, nor on available published data, and some of the very active compounds are not clinically available for patent or other economic reasons. Obviously, the machines will not have this information. In general, a chemical is said to have antihistaminic activity if it has little or no direct action on a biological structure but will prevent any typical action of histamine on this structure, or if the histamine is already present, if the chemical will overcome this effect or return the condition of the structure toward the normal. Since the first demonstration of this subject was presented in May, 1950, numerous additional publications concerning antihistaminics have appeared. Consequently, the entire procedure has been rerun, and is complete, as far as the Center is concerned, through May, 1951. Of course, some published articles are still in the possession of abstracters, some data sheets remain to be coded, and some cards to be punched. This, coupled with the delays involved in publication, means that the demonstration is several months out of date; for purposes of illustration of the general method, this is of little practical importance. The present demonstration has followed the general line of attack of the earlier one, although some items have been followed more thoroughly and some have been discarded as simple duplications of included material. It should not be overlooked that the Center has several types of information with which to work. The abstracter prepares a work sheet to contain pertinent data about each compound listed in an original article. This data sheet is the primary reference to the original publication. The
143 chemical agent is next listed in the chemical file and given a reference number, or if it is already listed in the chemical file, the reference number is supplied. All the data sheets of a given chemical are kept together according to this CBC number, with secondary numbers to indicate individual data sheets in case there are more than one. The information on the sheet is coded, and the coded material is punched on a standard IBM card. This punch card has the compound number which is also the data sheet number, a secondary number to indicate which data sheet is involved if there is more than one, and the coded material printed on the top of the card, as well as having it punched into the body of the card. It is possible, then, to go from punch card to chemical file to find just the formula of the compound, or to the name file to find its names or synonyms, or to the sheet to check the coded data or to find the reference so that the original article may be found and examined. For rapid determination of generalizations between structure and activity, the investigation should be limited to the punch cards with the use of the chemical or formula file as necessary. The data sheets should be avoided as far as possible as they are very bulky and relatively clumsy to handle if any number of items is involved. It is necessary that the questions asked be in a form that the machines can handle. Careful examination of the general biological code and the chemical code will help to formulate reasonable questions, but final consultation with the machine operator is advantageous; his experience in handling the cards and the system will often enable him to use shortened or simplified procedures or to point out impractical questions. This is necessary, since the CBCC is accumulating punch cards so rapidly that it will soon require a matter of days or even weeks to make a single sort of all the available cards. Replacing the sorted cards in numerical order also takes much machine time. I am grateful to Dr. W. R. Kirner and Dr. C. P. Huttrer of the Center for suggesting the problem and much assistance, to Mr. Ballard for carefully explaining the machine operations and carrying out numerous sorting procedures, and to Dr. Filippi and Dr. Geer for assistance with the Biological and Chemical codes. MACHINE QUESTIONS One primary question has been asked and nine subsidiary groups of questions. Obviouslyt the answer to the first question is the summarized data of the answers of the other nine. The questions, which are the same sort one might ask for any problem relating biological activity and chemical constitution, regardless of primary subject, are listed below. 1. Is there any relationship between antihistaminic activity and chemical constitution or chemical structure ? 2. Since I am relatively unacquainted with the biological code, are there appropriate headings that will allow for the isolation of punch cards that refer to compounds that have antihistaminic activity; i.e. , agents that prevent, overcome, or inhibit the effects of histamine ? 3. In a sort or search under these headings, how many compounds or cards are found? 4. Have all of these compounds been tested by the same pharmacological method? 5. If not, are there any outstanding or common methods that have been used in sufficient numbers that a correlation might be possible?
144 6. If two or more methods have been used to determine antihistaminic activity, is it possible to rank the compounds according to increasing activity by these methods, and are the more active compounds and less active compounds always found in the same place regardless of the method used? If they are not, then the test methods are not comparable and any correlation will be limited. 7. By visual examination of the most active compounds of any classification, are there any common chemical features that are found? Do the very active compounds occur in only one chemical family or are they not chemically related at all or do they occur in a few chemical types or groups ? If they occur only in one family or a few families, are any of the compounds in the inactive or relatively inactive groups that also belong to this family ? 8. Have the compounds that have been tested for antihistaminic activity been tested for any other type of biological activity? If so, is there any group or groups of data that might be correlated with antihistaminic activity? Are the most active compounds in any group also the most antihistaminic or is there no apparent correlation? 9. Have any of these compounds been tested in man? If so, is there any relationship between the relative activity of compounds tested in man and any tests carried out in animals ? 10. Of the compounds that have been found with high antihistaminic activity, are there any common chemical features? Do they all have a common chemical feature, such as a methoxy group, a nitrogen atom, a benzene ring, an ethylene chain, or a halogen atom? If any chemical grouping or pair of chemical groupings seems common in the high or moderately active compounds, what is found when the entire groups of compounds in the files is searched for this? Are compounds found with this grouping that have been inactive on test? Are compounds found that have not been tested as antihistaminics ? RESULTS Z^ With any general problem, it should be possible to approach from either the chemical or biological side to demonstrate any possible correlation. In this particular instance we are assuming that we know nothing of the chemistry of any antihistaminic and can only approach the problem from the biological side for the first approximation. For an understanding of the application of the general biological code, reference to the division of the punch card is helpful. The general biological card consists of 80 vertical columns, each 10 number spaces high with space for alphabetical overpunches. The columns 1-6 are used for the CBCC serial number, columns 7 and 8 if a particular salt of the compound is mentioned, columns 12 - 14 to specify which of several data sheets with this particular CBC number is involved. If an additional, or secondary chemical is involved in the test, such as a mutual antagonism problem, it is listed by CBC number in columns 44 - 49 (field H of the biological code). Or a comparison standard chemical may be listed in 44 - 49. Column 19 is used for the route of administration of the secondary chemical (if any is used) and column 22 is used for the route of administration of the primary chemical listed in columns 1-6. The various routes are listed in field K of the biological code. Columns 33 - 35 (field T) are for the response criterion, 53 - 60 (field A-l) to describe the animal species used, 62 - 64 (field C-l) to describe the part of the animal involved, 68 (field F-l) indicates the direction of response, 69 - 72 (field F-2) lists
145 the specific action of the primary chemical on the organism, while 73 - 75 (field F-3) may give the general action of the primary chemical. If it is possible to express the potency in a quantitative summary according to field G of the biological code; this is listed in column 76. Insufficient data may be given to use all of the columns and fields listed and it may not be necessary to use all of the information in these columns for a particular problem. It is also possible to have data that mean essentially the same thing listed in more than one way in the biological code, depending on whether the activity has been listed in a general way according to field F-3, as a specific action in field F-2, or as acting against some other chemical H in field F-li or acting like some other chemical H in field F-l. In the present problem, the general action "antihistaminic" is found in F-3 punch 206. Compounds are also listed as having an antagonistic action, punch 3 or L in field F-l, against the secondary chemica'l histamine (found in the name file to be CBC 800, 199) listed in field H. There is the possibility that a compound may be listed as having activity similar to some other well-known antihistaminic which would be listed in field H but with a different verb in F-l. Compounds with related activity and possibly related structure may be found in field F-2 under 8A1 or 8A11 for anaphylaxis, or in F-3 under 208 for antispasmodic since histamine is often considered a spasmogenic agent. Several sorts of the biological cards will be necessary to isolate all the possible antihistaminic entries from the general file. .3_^ As a result of sorting for these headings, 3,447 cards are found. These involve the results of biological tests of 1182 parent compounds, which with the various salt modifications actually represent 1315 compounds. By use of the collator it is very simple to remove the duplicate numbers or eliminate the salts. 4.^ Unfortunately these compounds have been tested by many procedures and on.several different animal species. As a preliminary breakdown, the cards have been separated into groups according to animal species (field A-l). Code Designation of Test Organisms Number of Cards Common Name (No test organism given) 197 A42010I Rana pipiens 1 Frog A610201 Gallus domesticus 1 Chicken A710101 Bos primegius 3 Cattle A720101 Canis 26., Dog >140 A7201011 Canis familiaris mj Dofe A720102 Urocyon cenereo 4 Gray fox A720201 Felis 3f| Cat i 107 A7202011 Felis domestica 7oJ Cat A730201 Mus 9 Mouse A7302011 Mus musculus 49 Mouse A730202 Rattus 12 Rat A7302022 Rattus norvegicus 36 Rat
146 (Cont. ) Code Designation of Test Organism Number of Cards Common Name A730401 Cavia A7304011 Cavia porcellus 1957' 527, [2484 Guinea pig Guinea pig A7306011 Cricetus f rumentarius 2 Hamster A740101 Homo 4IT 204 Man A7401011 Homo sapiens 158, Man A770101 Lepus 1 Rabbit A7701011 Lepus americanis i 198 Rabbit A770102 Oryctolagus 115 Rabbit A7701021 Oryctolagus cuniculus 81. Rabbit Perhaps the most startling find of this particular count or sort is that 197 cards, apparently prepared from scientific journals, purport to describe the antihistaminic activity of various compounds but do not state what experimental animal was used.1 The count also indicates that a comparative survey of antihistaminic activity in various species is not yet complete. 5. The groups of cards under guinea pigs, rabbits, and man are sufficiently large to warrant further sorting. Since several different procedures have been used, the groups under dogs and cats do not yield enough cards of a single type to be interesting. Dividing the guinea pig cards according to the test procedure used, yields four definite groups: gl Action against histamine on isolated intestine 1206 (F-2 813, C-l 66 67 or 68, F-l 3 4 L M or 5) gll Protective action against histamine aerosol 190 (H 800199, F-l 3 4 L M or 5, F-2, 11 111 or 16, Y i) gill Protective action against parenterally administered histamine 82 (Same as group II except Y 5) giV Protection against anaphylactic shock 11 (F-2 8A11, F-l L or M) The remaining 995 guinea pig cards were rejected since they did not fit one of these four classes, or had inadequate data, or could not be grouped together. The cards in the four groups were classified according to the quantitative activity in field G (column 76), with 0 active but amount not stated, 1 relatively inactive and increasing up to 9 as most active.
147 Rank gl g" gill giv 0 83 - 69 20 - 20 1 - 1 1 - 1 1 527 - 396 34 - 29 10 - 9 - 3 292 - 256 37 - 24 23 - 18 - 5 202 - 155 64 - 39 26 - 17 2 - 2 7 67 - 39 20-11 11 - 6 2 - 2 9 48 - 27 18 - 15 15 - 11 6 - 4 The left hand columns are the total number of cards in this activity class; the right hand columns have had the duplicate compounds removed. These are not true duplicates, but are instances in which different workers, or the same workers in different references have found the same activity for a given compound by the same experimental procedure. Two reasonably sized groups of cards were found among the rabbit entries. These included 9 items tested on uterine muscle and 63 items tested on intestinal muscle against histamine produced spasm. The 204 cards referring to antihistaminic activity in man were rather discouraging in that a common method had been used in only 21 instances and 11 of these were found to be listed as inactive. b^ As a check on the relative uniformity of the assay procedures, the number of duplicating cards in various activity levels was determined for the largest group of items (those listed under gl): gl CARDS COMMON TO OTHER ACTIVITY LEVELS Rank 1 3 5 7 9 0 20 6 10 2 1 1 396 45 25 1 3 3 45 256 28 5 3 5 25 28 155 11 7 7 I 5 11 39 8 9 3 3 7 8 27 It is readily apparent from this table that there is some overlapping between adjacent activity levels, but that the number of items reported as having both very high and very low level activity is relatively small. This is probably some tribute to the majority of pharmacologists that have used this method. The duplicate entries of the same activity were eliminated before this table was constructed. There were thirteen CBC numbers that appeared in three activity classifications lf 3, 5 or 1. 3, 7 as determined by the guinea pig intestinal strip method. These compounds were taken back to the original data sheets, and as the cause for the multiple listing did not appear obvious from these, the original references were consulted. The compounds all originated in
148 the same laboratory, were listed as slightly active as antihistaminics in a Federation Proceedings abstract, listed with various activities on an arbitrary scale in a paper describing their chemical synthesis, and finally listed in terms of activity against definite amounts of histamine in a pharmacological paper. The interpretation of their activity by various abstracters and coders led to their multiple listing. This is really a semantic error, if it is an error at all, and can be expected to occur in any system that tries to reduce biological data to symbols. Some of the duplicate entries may have occurred in a similar manner, since many of the compounds are described in both "chemical" and "pharmacological" papers. The other aspect of the assay correlation problem, which compounds appear as high activity compounds by more than one investigative technique, is encouraging. If one combines the 7 ("high activity") and 9 ("most active") levels of each group, but subtracts the common compounds of each group, then: gl 39 + 27 - 8 = 54; gll 11 + 15 - 3 = 23; glll 6 + 11 -4=13, and giV 2+4-2 = 4. Of the 23 gll items, 10 are common to gl; of the gill items, 10 are common to gl and 1 additional is found in gII; and all of the items in giV are listed in gl. On a percentage basil this is a remarkably good correlation among the active compounds. As a total check of all the active and inactive compounds, a table has been constructed of the gl, gll, and gill entries. CORRELATION OF QUANTITATIVE ACTIVITY VERSUS TEST METHOD Number of Common Compounds gl 1 gll 1 gill 1 II 1 1 III 1 5 111 III 1 0 i i i II i Si 3 6 3 6 3 0 3 0 3 1 3 J_ 5 2 5 2 5 0. 5 Â£ 5 A 5 1 7 4. 7 1 7 0. 7 _P_ ^ A 7 .2 9 Â± 9 _2 9 0. 9 J7 9 .2 9 _I gi3 gll 3 giI I 3 II 1 0 III 1 4 ill iH 1 1 i i A II 1 jp. 3 !i 3 10 3 11 3 1 3 10 3 _L 5 9 5 6 5 _i 5 _i 5 3. 5 i 7 _4 7 I 7 A 7 _3 7 2 7 1 9 J. 9 J. 9 _3 9 A 9 0. 9 _L gl 5 gll 5 gill 5 II 1 0 III 1 _3 ii7 iH 1 J. i i 2 II 1 JJ 3 5 3 3 3 _2 3 _a 3 _i 3 Ji 5 & 5 n_ 5 _& 5 _2 5 11 5 i 7 5 7 4 7 âL 7 _i 7 1 7 1 9 1 9 A 9 10. 9 1 9 J. 9 1 gi7 gll 7 gill 7 II 1 0 III 1 1 114 III 1 i I 1 1 II 1 0, 3 A 3 j2 3 A 3 J. 3 J_ 3 3 5 5 5 a. 5 S. 5 8 5 _4 5 J. 7 A 7 A 7 _5 7 L 7 4 7 2 9 3. 9 A 9 JL 9 i. 9 i 9 J. g'9 gll 9 gill 9 II I i III i A 114 III 1 J. I ' 1 II 1 0. 3 2 3 & 3 1 3 1 3 A 3 .2 5 10 5 _6 5 _fi 5 4 5 4 5 3 7 J 7 2 7 _2 7 2. 7 i 7 A 9 _3 9 1 9 J 9 J 9 JL 9 _L
149 Since all the duplicating entries are listed, some activity groups appear to have more compounds than the total number of compounds listed for that activity group. The overall correlation is fair, with more entries listed in low-low and high-high in general than in crossed divisions as low- high. The gll class has a lower over-all activity rating than gl and gill or else the gl and gill are biased toward the higher activity levels. One compound (800,422) is listed as having activity at the following twelve levels: gl 3, 5, 7, and 9; gll 3, 5, and 7; gill 5, 7, and 9; and giV 5 and 9. One other compound is listed nine times and one eleven times. This is the result of using these compounds for comparative purposes or standard reference chemicals as disclosed by sorting field H for compounds listed as using -03 or -04 criteria in field T. The rabbit data is not correlatable with the guinea pig data. Only one compound is listed in both groups. The name file discloses that it is atropine. The formula file shows that the compounds are almost exclusively esters. Reference to data sheets indicates that the compounds were primarily tested for atropine-like activity and for antihistaminic activity secondarily. Since there is no overlapping of active compounds, it is impossible to decide if the method is adequate for determining useful antihistaminic activity. As a tentative conclusion, based on the punch cards alone, the guinea pig test methods most commonly used do adequately separate the various chemical compounds, with the most active compounds as determined by one technique being most active by other techniques, and the less active compounds almost always being less active. This is true within the normal limits of operator error. ,7. For visual examination of the active compounds, it is necessary to refer to the chemical formula file. The gl 7 and gl 9 punch card numbers have been removed from the chemical file. Of this group of 54 cards, 11 have formulas that correspond to R-O-C-C-N= with large groups for R, such as benzhydryl, minor substituted benzhydryl, or benzylphenyl, no sub- stituents on -C-C-, and two methyl groups on the N or a very small heterocyclic group. Twenty- three have formulas that correspond to Ar-C.H^-N-C -C-N with aromatic ring groups such as Ar1 benzene, thiuphene, or pyridine for Ar and Ar', with minor ring substituents in some instances, no substituents on -C-C-, and two methyl groups on the N. The remaining compounds comprise i very miscellaneous group as far as structural chemistry is concerned. The gII high activity group only contributes two more ethylenediamine derivatives, two more ether derivatives and three compounds that are related to Ar-N-C-C-X with X either ft chlorine or bromine. The glll compounds that are different only contribute one ethylenediamine compound. All high activity compounds in giV are listed in gl. Reference to the CBCC name file, for the names of active compounds of ether or ethylene- diamine derivatives listed above, reveals that 16 compounds are listed in gl that are commercially available on the open market under a variety of trade and common names. Of the miscellaneous compounds, the name file lists theophylline, epinephrine, nicotin- amide, meperidine, ephedrine, papaverine, arterenol, Orthoxine, prophenpyridamine, and chlorcyclizine as highly active. Most of these miscellaneous compounds are not usually considered antihistaminics by definition, and the two that are increase the number of active commercially available compounds to 18. For many purposes, this type of correlation between chemical structure and high activity would be sufficient. Another aspect of the problem is the determination of the correlation between chemical structure and low activity. For this, the chemical structure file was consulted and the cards corresponding to the 396 gl 1 numbers found. Of these formulas, 196 were RR=N-C-C-N=RR, 108 were R-O-C-C-/-N=RR, and the rest miscellaneous. The ethylene- diamine derivatives included several acetamidine and imidazoline compounds, some compounds with very low molecular weight or very large molecular weight or size, with very large aromatic substituents or large alicyclic substituents, and with these exceptions were similar to the high activity compounds in gl 7 and 9. The same remarks may be made for the ether derivatives. Consulting the name file for the miscellaneous compounds revealed that the list
150 included such items as quinidine, atropine, nicotinamide, meperidine, Trasentin, procaine, papaverine, Dibucaine, cocaine, Benodane, methadon, alpha-fagarine and lysolecithin. Two of these miscellaneous compounds were among those listed as having high activity as antihistaminics. Consulting the name file for the low activity ethylenediamine and ether derivatives did not reveal that any of these relatively inactive agents have a common, public, or trade name, although many of them have serial or laboratory numbers of one kind or another. Since most of the active compounds that are commercially available are of the type Ar-CH2-N-CH2CH2-N(CH3)2, (Ar)2=C-O-CH2CH2-N(CH3)2, or with a phenothiazine ring Ar1 R replacing the Ar-CH2-N- , or with minor substitutions in one of the Ar rings, and since most of Ar' the apparently possible major modifications of these two types of compounds are listed in the lesser active groups, further endeavor along these two chemical lines might not be very profit- able. More promising leads for further synthetic work might be found by critical examination of the miscellaneous group in the lesser active classes with the possibility that new synethetic relatives might be more active, or else to investigate types of compounds that have not been listed at all as having been investigated. JT In order to find if compounds with antihistaminic activity may be in other classes of, pharmacologically active compounds, it is necessary to know what other types of activity anti- histaminics may have. Such a search or correlation might be useful to find other uses for both old and new antihistaminic agents. Consequently, all of the punch cards having CBC numbers that have been found as having any antihistaminic activity have been separated. Preliminary sorting of these cards reveals that antihistaminics are listed as having apparently been tested 3427 times for some 81 other types of activity. Of these, 255 are not definitely listed on the punch card under the common F-2 heading, 1544 are listed as having some effect on muscle tone, 858 refer to toxicity determination (667 "acute" toxicity), 260 on histamine shock, 160 on blood pressure, 67 on the toxicity of a secondary chemical, 30 on body secretions, 26 on relief of action of a secondary toxic chemical, 26 on anaphylaxis, 15 on permeability, 13 on growth, 13 on ability to produce ulceration, 12 on respiration, 11 on sedative activity, and the balance with only a few chemicals being tested for a particular activity. It should be remembered that these groups include both negative and positive results. With a particular card selected at random, the top of the punch card reads 500,990 005 A7401011 2 329 6 W 842 and use of the biological code indicates that man (A7401011) was given CBC 500,990 orally (2) and that the permeability (842) of the capillaries (329) of the skin (6) was not affected (W) by the dose given. The reference is on data sheet 005 of compound 500,990 which, according to the name file, is diphenhydramine. The machines in this instance could serve for a very rapid cross search of activity. Compound 500,990 is listed as being tested 121 times of the total 3427 entries for secondary activity. The correlation of the secondary actions and antihistaminic activity is not very good. In the first place, some of these cards are additional entries of material that has already been covered. Of the 1544 cards referring to action on muscle tone, 1496 refer to actions on muscle that has been made spastic with histamine or acetylcholine and have already been considered in gl. Similarly the 260 histamine shock items and 67 and 26 items involving a secondary chemical have been considered in gll and gill and the 26 anaphylaxis items have been considered in giV. The 160 items involving blood pressure include 84 items in dogs and cats relating to the ability to overcome histamine vasodepression and the other items are insufficiently limited to one animal species to indicate any possible correlation. These duplicate entries of activity have to be rejected for this type of correlation, although they have a very useful purpose in that they insure that any misinterpretation of the biological code will not invalidate the entire system. The only reasonably sized group of secondary actions remaining is the 667 "acute" toxicity items. Most of these involve rats as the species and can be broken into two approximately equal sized groups according to route of administration since they are almost all oral or intra- peritoneal with only a few intravascular administrations listed. Unfortunately, correlation of these compounds with those listed in gl does not yield any positive results, since there are almost as many high rodenticidal activity as low rodenticidal activity cards in both the high and low antihistaminic activity entries. This simply means that rodenticidal activity and antihistaminic activity are not related.
1S1 Therefore, while it is obviously desirable to test antihistaminics for rodenticidal activity before testing in man, it does not seem that all efficient rodenticides should be tested for anti- histaminic activity, either in guinea pigs or man. Consequently, there does not seem to be any secondary activity that is directly related to antihistaminic activity (other than antianaphylactic), nor is there any other general type of activity that might be searched for compounds that would be expected to have antihistaminic activity. .9^ The most desirable test of the data summarized in answers 6 and 7 is its applicability to results found in man. The available published data of the investigation of antihistaminics in man is inadequate, discontinuous, and non-uniform as indicated by the 21 cards in which the anti- histaminic had been tested for the ability to prevent the local effects of histamme in the skin out of the total 204 listed tests in man. The 11 compounds listed as inactive in man were also listed ingl 1, 3, or 5. Two cards with 7 activity in man were listed in gl 9 as was one card with only 1 activity in man. One card with activity not stated (0) wns listed in gl 7. The remainder of the cards were duplicate entries. The gl 9 card that was listed as only a 1 activity in man was referred to the data sheet. . Apparently a high enough dose to produce a high level of activity could not be tolerated in man. With this expected type of exception, the data in man, as little as it is, does correlate well with the data obtained by the most commonly used test procedure. The two high activity compounds by test in man are com me rcially available; it is unfortunate that comparable testing of the other commercially available compounds has not been undertaken. By the tune this symposium is published, such a study may have been undertaken with the common commercially available compounds, but I doubt if adequate data will ever be published for the other gl 9 compounds. Many of these are available to only limited numbers of individuals; in general, if a compound is tested in man by a commercial house, and not found superior to some comparison compound that is widely used, there is not much incentive to publish the results. For chemical-biological correlation to have any eventual application, the results of such tests of compounds that appear only mediocre or even relatively inactive must be made available. 10. It is scarcely reasonable to carry out any profound chemistry sorts when the cards have already been put in small groups by the biological sorting. lf all the antihistaminic agents had been investigated by one common technique, or all the techniques could be related in any way, then the machines could be used for chemical type and frequency analysis. A few trials have been run to indicate the possibilities. A frequency analysis of the number of carbon atoms versus the relative activity of the CBCC compounds in gl has been made.
152 NUMBER OF CARBON ATOMS IN ANTIHISTAMINIC â¢ Â« 1n Â£ JS E J h u o o < 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 0 6 1 3 4 6 6 3 7 7 10 6 4 5 1 0 0 1 i 18 10 12 10 17 32 18 41 55 57 42 32 17 10 7 9 4 3 10 6 8 9 12 15 9 16 30 43 24 26 10 7 8 10 9 5 4 1 2 4 5 14 5 10 26 23 15 14 5 2 3 5 4 7 3 0 1 1 0 9 1 4 6 4 3 0 0 0 0 0 2 9 6 2 0 0 0 8 0 5 9 2 2 2 0 0 0 0 0 The analysis is biased since some of these compounds contain -S- or -N= or -O- as isosteric replacements for -CH= or -CH=CH- in some aromatic rings. On this basis, most of the compounds with 14 carbon atoms belong either in the 15 or 16 class. Most of the compounds investigated have 14 to 18 carbon atoms, including most of the relatively active and inactive agents. There is only a very minor shift of activity toward the lower number of carbon atoms as the activity increases. A more interesting analysis is based on the frequency distribution of the chemical code groups in the gl compounds. By use of the rotating file it is possible to place any group first in the chemical code designation. The 155 compounds in gl 5 require 82 code groups for all of the chemical structures although the total number of separate groups is 924 or an average of about 6 groups to name a compound. The incidence of common groups is even greater with gl 7 and 9. With gl 7 that has only 20 compounds after the gl 5 entries have been removed, only 40 characters are required and these have occurred only 153 times with 6 characters accounting for 71 of these. With gl 9 the number of groups has dropped to 38 with 107 entries, and 5 characters account for 51 of these. Analysis of these code designations for the gl 9 entries in the chemical code indicates that some type of aliphatic substituted tertiary amine has occurred 29 times, that a pyridine ring has occurred 6 times, pyrimidine 2, ether oxygen 8, alcohol 2, furan 1, benzene or phenyl 12, four carbon atom chain 2, three carbon atom chain 5, two carbon atom chain 11, and one carbon atom methyl or methylene 12. These are the findings one would expect from the visual examination of the data in T. If the effect of a certain substituent on antihistaminic activity is desired, it can be readily found. For example, the effect of introducing a non-ionic halogen atom into an aromatic ring can be found from columns 55 and 56 of the chemical cards. Of the 108 such compounds found in the gl series only 4 are found in the highly active group. It would probably be easier with this small group of active materials to find the non-halogenated congeners in the formula file than to run the collator on the parent groups of 1182 compounds. Visual examination of the formulas of the compounds against the non-halogenated parents indicates that halogenation may increase the activity or decrease it, and that this is a function of the position of the halogen atom on the aromatic ring. This type of position isomerism is a feature that the CBCC chemical code is not equipped to handle; i. e. , the code does not differentiate between ortho, meta, and para substi- tution of a benzene ring, since the number of such isomers is always small enough that they caD be
153 separated visually. SUMMARY J_. The answer to the fundamental question, "Can the CBCC system show any correlation between chemical constitution and antihistaminic activity by use of the IBM machines" is generally in the affirmative as developed in section^. It has been pointed out that the biological data is limited by lack of use of a universal test procedure for all compounds and that the relationship of effects in man to effects in small animals is limited by lack of published data of experiments in man, although there is no lack of general clinical data. There is good correlation between man and animals in the data available, and the data obtained by one test procedure is generally equivalent to that obtained by other test procedures. The results can only be as good as the original data. If the data is not uniform, or based on uncontrolled experiments, clinical impressions, or non-comparable methods, then the Center cannot function. It may be able to indicate groups of data, or trends, so that future work may be better carried out, or in rare instances to find an isolated compound with a desirable action so that families of compounds can be prepared to find better compounds with this action. Certain aspects of the problem have not been considered. The third component of chemical biological correlation, that of time, whether expressed as duration, uptake, fate, or excretion, has been ignored, since there is so little available data on this subject as far as the antihistaminics are concerned. Another problem, that of position isomerism and biological activity, is beyond the scope of the machines, the code, and our present chemical knowledge. For example, by any classification system you wish to use CBC 801,638 and 800,422 have the same chemical components, the same size and shape, and very similar chemical properties. r"2 .N-CH2-CH2-N(CH3)2 ^ 7-N-CH2-CH2-N(CH3)2 800,422 801,638 The both have the same molecular weight, two aromatic rings, the same constituents, the same shape, and even the Hirschfelder models look identical. The 800,422 is a highly active anti- histaminic that is commercially available as t ripelennamine and is widely used in man; the 801,638 is an inactive, or relatively inactive, antihistaminic that is worthless. This is one instance when we do not know enough of the underlying chemistry of the two molecules to know how they differ, and until we have this information, we cannot expect to know why the biological activity differs. As abstract journals do not list all actions, particularly minor or secondary actions of compounds in their indices, the CBCC serves as a reservoir of such information and .secondary activity studies may lead to many investigations, either to develop compounds with better primary
154 activity, or with more desirable secondary activity in some instances. This demonstration should not be considered complete as many items are not listed for lack of space. For example, the Center has a combination chemical-biological card which will be valuable for rapid searches of a limited nature, and the true value of the chemical code has probably not been emphasized. Much data about intermediate activity compounds has not been included in this brief presentation, and only limited examples of the numerous inactive compounds. Such information would be valuable to an investigator, and it is available in most of the modern review articles of the subject. Examination of the finally selected formulas by visual inspection indicated that high activity antihislaminics belong to three classes: a group of ethylenediamine derivatives with specific limitations as to substituents, a group of 2-dialkylaminoethyl ethers with specific limitations as to substituents, and a miscellaneous group of compounds that had no common chemical features. It must be remembered that such correlations are non-limiting in a dynamic field such as biological science. Other compounds, which are entirely different in type of structure, may have very similar activity or be highly active. Similarly, all compounds with the same or similar structural components and similar size and shape as one of the two limited groups above will not necessarily be active. However, all highly active compounds of these basic types will have this size and shape, and when all the possible ones of this size and shape are made, it will not be necessary to make all of their derivatives or derivatives outside of these dimensional limits.