630 GADDUM: HORMONE ASSAY: INTRODUCTION [Vol. 75 Hormone Assay The following four papers were read at a joint meeting of the Biological Methods Group and the Society for Endocrinology on Thursday, October ZOth, 1949. Introduction BY J. H. GADDUM CANDIDATES in pharmacological examinations are sometimes asked why certain substances are assayed by biological methods. The correct answer is, of course, that no one would do bio-assays if he could help it, but that some drugs, such as insulin, cannot be controlled in any other way, and some drugs, such as acetylcholine, are commonly present in concentrations that are too low to be detected in any other way. The motives that drive us to bio-assays can, however, also be classified pragmatically. Some of us wish to estimate the potency of conimercial preparations, so that they may be safe to use and so that we may assess their comniercial value.For these purposes, methods of bio-assay have been developed that can give almost any required degree of accuracy. The members of the Biological Methods Group of the Society are expert in such matters. The members of the Society for Endocrinology, 011 the other hand, are interested in bio-assays mainly as a means of discovering new facts. 'Their problems are complicated and their arguments often involve assumptions and approximations that are open to criticism. The object of both should be that the endocrinologists should learn accuracy and that the analysts should learn something of the difficulties of the endocrinologists. Methods of hormone assay can be classified according to the kind of effect observed, though some hormones can be assayed by means of several different kinds of effect.Table I summarises most of the established methods of assay for hormones and gives some indication TABLE I[ Adrenaline, etc. Thyroid . . Androgens . . Oestrogens . . Progesterone. . Corticoids . . Insulin . . Parathyroid . . Secretin . . Oxytocin . . Vasopressin . . Antidiuretic . . Growth . . Prolactin . . Thyrotrophin Gonadotrophins Corticotrophin .. .. .. . . . . 1 . .. .. . . . . .. .. .. . . . . . . Smoked Animal drum weight 1 1 Organ Fluid Bio- weight volume chemical 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 quantitative. 2 quantal. Mor- Death, phology etc. 2 1 2 1 2 1 2 2 2 2 1 2 2 1 2 of the effects. A few of these are quantal in the sense that the response of each animal is not measured, but merely recorded as positive or negative.These latter are denoted in the table by the figure 2 ; other things being equal, they require for an'y given degree of accuracy about twice as many animals as the first type of test. Under the heading of morphology are included observations of the presence of legs in tadpoles, of the appearance of vaginal smears and of histological preparations of sections of tissues. Such results are difficult to assess quantitatively and the test is generally quantal. Most of the tests are based on a quantitative measure of the effect.October, 19501 GADDUM: HORMONE ASSAY: INTRODUCTION 531 The heading, “death, etc.,” includes the symptoms produced by insulin in mice.It may be asked,how death can be quantitative. The answer is that some tests depend on the duration of survival. For example, when gonadotrophins are administered to males they act on the testes, which release androgens, and these androgens then act on the prostate. It might be expected that assays based on the weight of the testes would be more accurate than assays based on the weight of the prostate, but this is not necessarily so. We probably agree about the principles upon which assays should ideally be founded- (1) Each assay should involve a comparison between an unknown preparation and a standard preparation. (2) Each of these preparations should contain the same single active principle and no other substances that modify its action in the test. (3) Each assay should be complete in itself; it should involve no arbitrary assumptions about such things as the slope, or even the shape, of the dose - effect curve.(4) The limits of the error should be calculated from the internal evidence of each assay. If the first two conditions are fulfilled, the test must give an estimate of the concentration of a single substance, and the results of parallel quantitative tests of the same solution by a series of different methods will all agree, within the errors of the tests. If the substance used as a standard is not the same as the substance present in the unknown solution, it is generally found that the results of parallel quantitative tests do not agree. This circumstance forms the basis of a valuable method for identifying substances and distinguishing them from other closely related substances.By means of such tests it has been shown in recent years that adrenaline is generally accompanied, or replaced, in the body by the closely related substance noradrenaline. It follows from these facts that the biological response on which the result of an assay depends need have no relation to the effect that the preparation is expected to produce. The requirements of an ideal assay are seldom, if ever, completely fulfilled. Even in careful assays of insulin the assumption is generally made that the dose - effect curve can be transformed into a straight line by mathematical devices that have been successful in the past. In practice, it is necessary to compromise in order to make any progress at all.We may discuss how far this compromise should be allowed to go. A standard preparation is not always obtainable. There is no international standard preparation for many of the hormones, and in many instances it is not yet clear how to make a standard preparation. A later paper will tell, for example, something about human pituitary gonadotrophins, which are only available at present as variable mixtures of unknown substances that increase one another’s effects to unknown extents and cannot be separated from one another. Oestrogens should present a simpler problem since their chemical formulae are known, but even so it is impossible a t present to make satisfactory assays of mixtures of oestrogens. The reasons for this have been discussed fully by Emmens,l but let us consider some of the facts. Biological fluids may contain variable proportions of three oestrogens, oestradiol, oestrone and oestriol.The first of these substances is the most active and the last the least active, but the ratios of the activities depend on the details of the experimental technique and may vary enormously. For example, Emmens quotes figures varying from 1 to 250 for the ratio of the activity of oestrone to that of oestriol. In his own experiments, the ratio was 2-4 when the drugs were given in four doses, and 70 when they were given in two doses. These figures alone are enough to show that it is impossible to attach any real meaning to the results of assays of unknown mixtures of oestrogens. Estimates which purport to give “oestrogenic activity’’ of such mixtures are liable to vary over a 30-fold range or more, according to the way the test is done.A control sample of urine contains 1 mg. of oestrone and 10 mg. of oestriol. The patient is then given a dose of medicine, and another sample of urine is collected containing 2 mg. of oestrone and 4 mg. of oestriol. Two observers both estimate the oestrogenic activity of these samples of urine in terms of oestrone. Observer A gives the oestrogens in two doses and observer B gives them in four doses. In the first test, according to Emmens, the ratio of activity of oestrone to activity of oestriol is 70, and it may be calculated that the conclusion will be that the medicine causes The last three hormones in the list act primarily on other glands.Consider an imaginary example.532 GADDUM: HORMONE ASSAY : INTRODUCTION [Vol. 75 a rise in the oestrogenic activity of the urine from 1-14mg. of oestrone equivalent to 2.06. In the second test, the ratio is 2.4, and the conclusion will be that the medicine causes a fall in the oestrogenic activity from 5.17 to 3-67. The two methods give opposite results and they are both wrong; their results have no meaning. The variation of ratios can, of course, be reduced by standardising the technique, but if this is done the interpretation of the results becomes very artificial. In the example just discussed, any conclusions about the effect of the medicine would depend on which system of doses had been arbitrarily chosen as the standard one. The position is further complicated, as Emmens pointed out, by the presence in urine of variable quantities of substances, not oestrogenic themselves, but having the property of increasing the effect of substances that are.The discovery of the synthetic oestrogens has made it unnecessary t o use biologically standardised mixtures of oestrogens in therapeutics. If it had not been for this discovery, as much time and labour might have been wasted in the study of methods of standardisation of mixtures of oestrogens as has been wasted on the same problem in connection with extracts of digitalis. These are known to contain a variable mixture of active substances, and enormous amounts of scientific energy have been devoted to the search for a reliable method of predicting their effect on man.Such experiments serve a practical purpose, but rest on an insecure logical basis and scarcely deserve the name of assays at all. It is hoped that they will soon become unnecessary because doctors will use pure active principles. The only possible way of getting results with any intelligible meaning in bio-assays of mixtures of oestrogens seems to depend on removing the interfering substances and separating the oestrogens from one another before the biological tests are made. Chemical methods give results that are easier to interpret, since the three main oestrogens have equal effects in the Kober test generally used, so that it is possible to estimate the total amount of oestrogen present. This test is, however, not sensitive enough to estimate the oestrogens in non-pregnant blood.If a biological test could be discovered in which all the oestrogens were equally effective, progress could be made, but sufficiently sensitive chemical methods would be even more valuable. Accuiate tests for single oestrogens have, of course, been available for years, but they have played only a small part so far in advancing knowledge. The researches that have led to our present detailed information about the chemistry of these substances were controlled by biological tests in which all the laws of bio-assay were neglected. This shows that it sometimes pays to take risks, but does not justify all the risks that have been taken by endocrinologists. In some instances it may still be necessary to speak of mouse units and to assume that all the mice in the world are equal, but every effort should be made to get away from such dangerous assumptions as soon as possible.There can be no justification for speaking, as some still do, of mouse units of the chorionic gonadotrophin present in the urine of pregnant women. An international standard preparation of this substance is available and should be used. Results should be given in international units. Unfortunately, this standard cannot be safely used for assays of other gonadotrophins, such as those present in the urine after the menopause. Apart from the provision of reliable preparations of hormones, the contribution that bio-assays can make to clinical endocrinology depends on the estimation of hormones in blood and urine. The concentration in the blood presumably gives an indication of the concentration with which the tissues are in equilibrium. The concentration in the urine may provide an indirect estimate of the daily output of hormone, provided that all the urine excreted during 24 hours is collected.If, for example, it is found that 5 per cent. of a hormone appears in the urine when it is injected, it may be justifiable to make the provisional assump- tion that the daily output by the gland is equal t o the amount excreted in the urine multiplied by 20. In such instances it is important to remember that changes observed in the urinary output may represent either a change in the amount of hormone released into the blood stream or a change in the proportion of the hormone reaching the urine. So long as this fact is borne in mind, estimations on the urine may be expected to give valuable information not given by estimations on blood.The time will probably come when provision for assays of this type will be an essential part of the equipment of all large hospitals, but that time is not yet. Very few methods of assay available at present are sensitive enough for this purpose; the discovery of more sensitive methods is a most important objective. Chorionic gonadotrophin is present in such high concentrations during pregnancy that its concentration in both blood and urineOctober, 19501 SCHILD: GENERAL APPROACH TO BIOLOGICAL ASSAYS 533 can be accurately estimated by bio-assay, but this is not true of any other hormone. Until more sensitive tests are discovered the amount of hormone available is too little for a perfectly designed assay.REFERENCE 1. Emmens, C . W., ( ‘ Variables affecting the Estimation of Androgenic and Oestrogenic hctivity,” Spec. Rep. Ser., No. 234, Medical Research Council, London, 1939. PHARMACOLOGICAL LABORATORY UNIVERSITY NEW BUILDINGS EDINBURGH, 8 General Approach to Biological Assays BY H. 0. SCHILD SYNopsrs-The term ‘‘ bio-assay ” is used for two different procedures. In the first, an unknown sample is assayed in terms of a standard of the same composition, the aim being to determine the amount of some chemical substance present; this may be termed “ analytical assay.” In the second, the test sample is assayed in terms of a substance or mixture of different composition, the aim being to determine the biological activity of the test sample.The paper briefly reviews some of the principles underlying the analytical type of assay. BIOLOGICAL assays may be used for two purposes, which must not be confused. Either they may be used analytically, as substitutes for a quantitative chemical analysis, or they may be used to determine the biological activity of a substance or a mixture of substances. In the first case the aim of the assay is to detect the concentration of a certain chemical substance in a sample. The only essential condition is that the standard and unknown should have the same composition. If they are chemically identical, then it does not matter what species or what reaction is used in the test; the final result is always the same and the best method is that by which an accurate result is produced with the greatest economy of time and labour.An entirely different situation arises where the two samples differ in chemical composition. The assay then ceases to be an analytical method and becomes a comparison of biological activity in which species, end-point and experimental conditions become all-important. When two substances that are chemically different, say cocaine and procaine, are com- pared, it is impossible to assign to them an activity ratio in general terms, since their relative activity depends on experimental conditions. Further, even under constant experimental conditions, the results of successive experiments are not necessarily identical, as the relative sensitivity of a preparation to different substances is not constant.Not only major chemical differences, but even minor differences in composition and purity may completely distort an assay and change it from an analytical method into a comparison of activity valid only for the particular species and method of administration used. For example, Gold, Cattel, Kwit, Kramer, Modell and Zahml found that by oral administration in man, 1.25 mg. of digitoxin produced the same effect as 1.25 g. of digitalis, when tested by a fall in heart rate in patients with auricular fibrillation. By the intravenous cat method, however, the former dose corresponds to 3 to 4 lethal “ cat units” whilst the latter corresponds to 12.5 cat units. Although digitoxin is the main active constituent of digitalis, yet in this instance differences in purity, in the method of administration and in the species used led to entirely different results by the two methods.COMPARATIVE ASSAYS NECESSITY FOR A STABLE STANDARD- It is generally believed that biological assays, at least those of the analytical kind, should be comparative and should be made by reference to a stable standard. This belief is based on two assumptions: first, that it is not possible to keep a biological reaction under534 SCHILD : GENERAL APPROACH TO BIOLOGICAL ASSAYS [Vol. 75 “statistical control’’ in different laboratories and that statistically significant variations in the activity of so-called animal units occur between different laboratories and also within the same laboratory at different times; secondly, that international standard preparations are stable and do not deteriorate. A good example of the variability of animal units is provided by the collaborative digitalis assay done for the U.S.P.under the guidance of B l k 2 In this assay the intravenous lethal cat dose of digitalis was determined, using the same standard preparation in various laboratories and at various periods. The assay technique was completely standardised and in each experiment the standard was compared with an unknown consisting of a dilution of the standard preparation. When the results were analysed it was found that the lethal dose of standard-the cat unit-would stay constant, within the limits of error of the assay, only for periods of about 8 or 10 days in any one laboratory. During that period there was therefore no need, in assaying the unknown, to refer back more than once to the standard preparation.When, however, the assay was repeated in the same la‘boratories a few months later, the cat unit had altered significantly in about half of the laboratories. Table I, which gives some data TABLE X Degrees of Variance ratio, Variation freedom Mean square F Between collaborators . . . . . . 9 0,03162 4.99 Between preparations . . .. .. 1 0.1 1491 18.15 Collaborators x preparations . . . . 9 0-00463 0.73 Error . . . . . . .. . . . . 160 0.00633 1.00 taken from Bliss’s paper, shows that significant variation in the lethal dose occurred between different laboratories (collaborators), but it is of interest that the mean square for “ interaction” of laboratories on preparations was no greater than that for error.As in this instance inter- action measures the relative potency of standard and unknown as determined in different laboratories, the result signifies that although the absolute activity as determined in different laboratories varied, the relative activity of standard and unknown remained the same within statistical limits in the hands of different investigators. In other words, whilst the absolute values could not be brought under statistical (control the relative activities had been so controlled. This is a good demonstration of the superiority of comparative assays over assays relying on animal units. DETERIORATION OF STANDARD- In using comparisons with a stable standard, rather than animal units, for biological standardisations, one assumes of course that the variation in activity of the stable standard is negligible compared with the variation of the animal unit.Unfortunately, there is no way of determining whether this is true or whether a so-called stable standard is deteriorating. This problem arose, for example, in 1942, when the standard for posterior pituitary at Hampstead was replaced by a fresh standard. The new standard was compared in a number of laboratories against the old standard by the three methods, oxytocic, vaso-pressor and antidiuretic, and was generally found to be about 15 per cent. more active than the old one. The Biological Standards Commission, however, decided to adopt as a unit the same weight of the new standard as had been used for the old standard, taking no account of the higher activity of the new one.Both standards had been prepared by the methods of Voegtlin, who had himself assumed that an extract prepared from fresh glands strictly according to his instructions would always have the same activity. This may well have been so, and, in that event, one has to assume that the old standard had slightly deteriorated. I do not wish to suggest, however, that everyone should now prepare his own standards by Voegtlin’s methods for fear that the international standard may have deteriorated. It may be expected that workers less skilled than Professor Voegtlin would produce extracts varying in activity by much more than 15 per cent. STANDARDISATI~ON ON MAN In many biological standardisations one: is dealing with impure extracts whose identity with the standard cannot be guaranteed.Since relative activity may then depend on the species and end-point used for the assay, should these standardisations be done on man?October, 19501 SCHILD : GENERAL APPROACH TO BIOLOGICAL ASSAYS 535 The answer can only be found experimentally by comparing standardisations done on animals and on man. The most thorough attempt at standardisation on man has probably been that of Gold, Cattel, Otto, Kwit and KrameI.3 in New York who tried to standardise digitalis on man. These workers used two reactions: the slowing of the heart rate in auricular fibrillation and an inversion of the T-wave in the electrocardiogram. The first method is preferable because it measures a therapeutic action, but the second gives more accurate results.In one series, using the slowing of the heart rate, these workers concluded that the result of comparisons for activity of the standard powder with an unknown in man agreed fairly well with those of the cat assay, although the agreement between the human assay and the frog assay was not as close. In another series, however, using the T-wave test, they found marked discrepancies between the human assay and the cat assay, amounting sometimes to over 50 per cent. Kamm separated the oxytocic and the vaso-pressor principles, using the guinea-pig uterus as test object. Pitressin had only a very small action on the guinea-pig uterus, and this was usually attributed to contamination with pitocin. Later on it was found that pitressin did affect the uterus in some species; for instance, in the cat it may produce a strong inhibition of the uterus (Robson and Schild*) and more recently it has been shown by Chassar MoiI-5 that in man pitressin produces under certain conditions a greater contraction of the uterus than pitocin.Obviously a pituitary extract standardised by the guinea-pig method gives no adequate indication of activity in the human uterus. The routine assay of pituitary extract on the parturient human uterus is probably impractical, but it would be interesting to know whether at that stage an extract of pituitary standardised by the usual methods for oxytocic activity would be more active than purified oxytocin standardised in the same way. It would therefore seem that in dealing with mixed extracts, animal standardisations are a useful safeguard rather than a reliable method of assessing activity.It is important that animal standardisations of these extracts should occasionally be checked by human standardisation, but even this is no complete safeguard, as pointed out by Gaddum (previous paper), since all human beings do not react alike. Apparently these experiments have not been repeated. Posterior pituitary hormone is another interesting example. STATISTICAL METHODS The simplest type of graded assay is one for which no statistics are required, inwhich the standard and unknown are matched by a trial-and-error process until they produce equal effects. This method has one great advantage: it does not depend on the assumption of a mathematical relation between dose and effect.However, it also has several disadvan- tages. First, it is inefficient because the preliminary effects are not utilised in the final assessment. Secondly, it is subjective, depending on the judgment of the experimenter as to what constitutes a satisfactory match. Thirdly, its experimental error cannot be determined from the assay itself but only by separate experiments. And fourthly, since the effects are matched at only one dose level, it gives no indication of whether the con- centration - action curves of standard and unknown are parallel. It thus fails to show up qualitative differences between standard and unknown. Assays that involve statistical methods are usually based on the assumption of some mathematical relation between dose and effect.Experience has shown that frequently the relation between dose and effect is expressed by an S-shaped curve when the dose is plotted on a logarithmic scale. Thus, if the reaction is of the all-or-nothing type (quantal), an S-shaped curve results if the percentage of animals reacting is plotted against log dose, whilst if the reaction is of the graded type (quantitative) an S-shaped curve results when the size of the effect is plotted against log dose. Usually they are interpreted as integrated normal curves when the reaction is quantal and as Langmuir adsorption curves when it is quantitative. In order to obtain approximately straight lines only the middle part of the graded response curve may be utilised. With all-or-nothing reactions the ordinates may be transformed to normal equivalent deviations or probits.The transformed curves are usually approximately linear, especially in the range above 20 per cent. mortality. Other transformations have been used, such as the logit transformation derived from the logistic curve. Finney6 has shown that practically identical results may be obtained in an assay using any one of four different transformations. These S-shaped curves may be variously interpreted.536 SCHILD: GENERAL APPROACH TO BIOLOGICAL ASSAYS [Vol. 75 Gaddum has outlined three characteristics for a satisfactory assay, viz., (1) a linear relation between x and y over the relevant range, (2) a standard deviation (this applies to graded responses only) independent of effect and (3) a small value of s/b.The latter is a good measure of variability, since it measures the variability of response in terms of the dose. Bliss has suggested another important characteristic of a good assay, namely, that it should furnish an estimate of its variability as an integral part of the assay. The statistics of quantal and quantitative assays have evolved on rather different lines. This is due to a mathematical peculiarity of quantal assays, the need for weighting the response. In quantitative assays, on the other hand, it can usually be assumed that all points on the curve have the same variance and therefore the same weight. Owing to this fact, Fisher’s analysis of variance could be readily applied to quantitative assays with all its implications of planning experiments in such a way that sources of variation not attributable to experimental error may be eliminated from the final comparison. Statistical methods are becoming increasingly complex, and we must leave it largely to the statisticians to work out their mathematics.Biologists should, however, keep in mind the basic assumptions on which these mathematics are founded. For example, the limits of error of an assay involving two straight parallel lines may be calculated by one of two formulae, one giving what are usually called the approxirnate limits of error, and the other giving exact fiducial limits. The reason why one is called “approximate” and the other “exact” is that in the derivation of one formula a mathematical approximation is used, namely, the variance of the ratio of two variables, whilst the derivation of the other is mathematically exact.It would, however, be wrong to assume t‘hat, because the formula is mathematically exact, the actual fiducial limits as calculated in an individual assay are necessarily correct. This would only be so if the assumptions on which the calculations are based were exact. These, however, we know to be only approximations; for example, we know that the assump- tion of linearity in quantitative assay is an approximation. One may, of course, test the data for deviations from linearity and parallelism, but lack of statistically significant deviation from linearity does not prove a linear relation in a particular assay. Thus the fiducial limits as calculated from a given experiment must always be taken with a grain of salt.This applies particularly to research work where new situations constaqtly arise and where the basic assumptions have not been tested by extensive previous experience. REFERENCES 1. 2. 3. 4. 6. 6. UNIVERSITY COLLEGE, LONDON Gold, H., Cattell, McK., Kwit, N. T., Kramer, IM. L., Modell, W., and Zahm, W., J . Pharmacol., Bliss, C. I., J . Amer. Pharm. Assoc. Gcielztific Edition), 1944, 33, 225. Gold, H., Cattell, McK., Otto, H. L., Kwit, N. T., and Kramer, M. L., J . Pharmacol., 1942.75, 196. Robson, J. M., and Schild, H. O., J . Physiol., 1938, 92, 1. Chassar Moir, J . Obstet. and Gyn., 1944, 51, 247. Finney, D, J., J . Roy. Statist. Soc., 1947, 9, Supplement, p. 46. 1944, 82, 187. DEPARTMENT OF PHARMACOLOGYOctober, 19501 SOMERS : THE MEASUREMENT OF THYROIDAL ACTIVITY 637 The Measurement of Thyroidal Activity BY G.F. SOMERS SYNoPsIs-After an account of methods used or proposed for evaluating thyroid activity in natural and synthetic products, this paper includes a brief description and comparative assessment of the method based upon comparing the resistance of mice to anoxia, which is reduced by thyroid treatment. Assay methods based on the antagonistic action of anti-thyroid drugs and thyroid preparations are also discussed. r. I H E evaluation of thyroid preparations is a problem not yet satisfactorily solved. The official method of standardisation (British Pharmacopoeia, 1948)l is based on Harington and Randall’s estimation2 of acid-insoluble organically combined iodine. The United States Pharmacopoeia, 1947,3 makes use of the total iodine present in organic combination.Neither method gives values in close accord with physiological results4 ; on chemical considerations the B.P. method is to be preferred, on the assumption that thyroxine is the only source of acid- insoluble iodine. The thyroid contains two distinct organic iodine compounds, thyroxine and di-iodo- tyrosine; only the former is physiologically active. The relative amounts of the two com- pounds differ considerably in different samples of thyroid. The method of Harington and Randall2 appears satisfactory for those thyroid preparations in which all the acid-insoluble iodine is in thyroxine, but it is unsatisfactory for iodinated proteins.These contain a large amount of acid-insoluble iodine, of which only a small and variable proportion has thyroid activity.5 Chemical analysis of these preparations being therefore of little value, biological assays are necessary. A modification of Harington and Randall’s method was developed by Leland and Foster6; in this, after the preliminary alkaline hydrolysis, the thyroxine fraction is extracted with butyl alcohol. Reineke, Turner, Kohler, Hoover and Beezley7 have claimed excellent agreement between the results of biological assays and a chemical extraction method modified from that of Blau.8~9 Other chemical methods suggested involve absorptiometric measurementslO of the red colour formed with sodium nitrite in ammoniacal solution,ll of the red colour formed with diazobenzene sulphonic acid in sodium carbonate solution,l2 or of the purple colour formed with diazotised N1-diethylsulphanilamide in alkaline solution.13 Polarographic methods have also been emp10yed.l~ J5$16 They not only determine the total iodinated diphenyl ethers present, but can also distinguish between thyroxine and 3 : 5-di-iodotyrosine. It should be remembered that the optical form of thyroxine is all-important, the physiological activity of DL-thyroxine residing largely, if not entirely, in the laevorotatory component.Thyroxine in thyroid extracts and in iodinated proteins is solely the L-form. BIOLOGICAL ASSAYS Although chemical methods are reasonably satisfactory for standardising gland extracts, there is occasionally a need to evaluate the thyroid activity of new materials, such as iodinated protein, thyroxine analogues and other organic compounds.In testing for thyroid activity, the accepted principles of biological assay must be applied. Simultaneous examinations of a standard preparation and the test material should be made; accurate assays are only possible when the active principle of both is the same. In comparing new compounds with old, quantitative assessments should be made with caution and results obtained with one species should not be assumed to apply to another. METHODS Numerous biological methods have been suggested for the assay of thyroidal activity ; some of them bear no obvious relationship to the physiological action of thyroxine in mammals, e.g., the acetonitrile test in mice.17 No one of them has been generally accepted, possibly because of their practical difficulties and inherent inaccuracy.Most procedures have not been developed on a reasonably quantitative basis; few workers have accompanied their reports on the comparison of two thyroid preparations with information about the precision538 SOMERS : THE MEASUREMENT OF THYROIDAL ACTIVITY [Vol. 75 of the estimates attainable. it has often been found that in fact the precision claimed has not been achieved. The commoner methods employed will now be briefly considered. INCREASE OF METABOLISM- Thyroid preparations cause an increase in metabolic rate ; the associated changes occurring in oxygen consumption and carbon dioxide production can be used for the estimation of thyroidal activity.Such methods are most closely related to the physiological actions of thyroid in man, but they tend to be too laborious and troublesoqe for routine assays. For the measurement of oxygen consumption the animal is usually enclosed in a chamber immersed in a constant-temperature water-bath.ls 9 1 9 Expired carbon dioxide is absorbed in soda-lime and the oxygen consumed is measured by a valve or from a Benedict spirometer. By this method, Wokes2* compared thyroid extracts on guinea-pigs, and obtained limits of error, P = 0-95, of & 30 per cent., using two dose levels of the standard and test preparations with six guinea-pigs on each dose level. The guinea-pig is very suitable for this type of assay because of its sensitivity and rapid response. The apparatus is possibly over-elaborate, and a simpler method for measuring the oxygen consumption of mice, which should be readily adaptable to the assessment of thyroid preparations, has recently been described by Maclagan and Sheahan.21 is usually employed.Mice are placed in a closed container through which air is drawn after previous removal of carbon dioxide with soda-lime. The carbon dioxide evolved by the animals is absorbed in a train containing standard sodium hydroxide, the amount being subsequently determined by titration with standard hydrochloric acid. According to March, the method is one of the most accurate for determining thyroidal activity, but a simpler and less time- consuming method is desirable for routine assays. Gaddum and Hethe~ington~~ compared the activities of a number of thyroid preparations by this method, but the results were only roughly proportional to the thyroxine contents determined by the method of Harington and Where data susceptible to statistical analysis have been given , For the estimation of carbon dioxide production the method of INCREASED SUSCEPTIBILITY TO ANOXIA- The thyroid gland has an influence on the susceptibility of mammals to anoxia, their resistance being decreased by previous administration of thyr0xine.2~ y25 p26 $27 Smith, Emmens and Parkes28 have developed a method of assay for thyroid preparations based on this decreased survival time of mice enclosed in jars.We have utilised the method extensively for com- paring the activity of thyroxine and a number of its derivatives.29 For calculation of the results we used the logarithms of the survival times, because the distribution within each group of the survival times themselves was skew.The relative activities were calculated by the method of Bliss and Marks30 and errors in the slopes provided for according to 1rwi1-1.~~ The slope, b, for thyroxine sodium by the subc:utaneous route was -20.8 with a standard error (P = 0.95) of s 3 . 1 , b being calculated as the increase in log survival time per 10-fold increase in the dose. The value of s/b = 0.35; to attain a precision of 5 20 per cent. (P = 0.95) it would therefore be necessary to ernploy 240 animals on both the standard and test preparations and to use three dose levels for each. The slopes were even less steep by the oral than by the subcutaneous route (b = 14-8, s/b = 0.48).The precision of the method therefore leaves something to be desired, but the assays are relatively easy to carry out. LOSS OF BODY-WEIGHT- The administration of thyroid to mice, rats and guinea-pigs causes a loss in body- weight.2°~22~32~33~34 Gaddumlg found that adult rats lost weight pari passu with the rise in their metabolic rate. WokesJ20 however, found wide variations in the susceptibility of mice to thyroxine and considered that large numbers would be required for an assay based on weight changes. Hutcheon,% using rats, showed that the slope of the line relating the loss in weight to the log dose of thyroxine was 9.1 per unit log dose with a standard deviation of & 4.4 g. The precision of the estimates is thus not good; further, the method cannot be specific, the presence of toxic factors influencing the result.There is also a wide variation in the response of guinea-pigs to a given dose, so that large numbers must be employed under carefully controlled conditions, or the results obtained will be far from precise.3qeas The value of s/b is therefore 0.49.October, 19501 SOMERS : THE MEASUREMENT OF THYROIDAL ACTIVITY 539 DECREASE IN SUSCEPTIBILITY TO TOXIC SUBSTANCES- Hunt37 observed that previous administration of thyroid by mouth increased the resistance of mice to acetonitrile poisoning. A method for assaying thyroid preparations, based on this observation, was developed by Hunt and Seidell,38 and this was improved by Hunt39 and by Haffner and K ~ n y a m a .~ ~ Unfortunately, the susceptibility of mice to acetonitrile is extremely variable and is affected by numerous known and unknown factors. Gaddum and Hetheringt~n~~ controlled these factors in some experiments, but the variation in the response of the mouse to acetonitrile alone was so great that they rejected the procedure as a possible test method. VVokes4O found that at ordinary room temperatures the dose - mortality curve for acetonitrile became very flattened. A t higher temperatures the dose - mortality curve was steeper, but thyroid no longer afforded any protection. Laland and St0a41 obtained more favourable results with the method and considered it serviceable for routine tests, but they have not stated its precision. INCREASE IN THE RATE OF AMPHIBIAN METAMORPHOSIS- G~dernatsch~~ showed that a thyroid preparation accelerated the rate of metamorphosis of amphibian larvae, and caused a rapid shrinking in size, if suspended in the water in which they were living.Several of the effects on tadpoles have been utilised for testing the activity of thyroid preparations, These include: time to first appearance of the leg buds, time to produce death, length of gut, body-weight, body volume and total length. A number of Anura have been employed, Rana pipiens and Rana temporaria being mostly used. The chief difficulty with the tadpole test has been a regular supply of tadpoles; this can be over- come by the use of Xenopus Zaevis if they are bred in the laboratory.43 As a method of assay, the use of amphibian larvae can be criticised on the grounds of unspecificity, particularly because of the wide difference in species from mammals and in the mode of administration Di-iodotyrosine and D-thyroxine are active in amphibia43 but relatively inactive in mammals.Quantitative methods have been described by numerous workers. woke^,^ using tail length in tadpoles from Rana temporaria, obtained results giving an s/b value of 0.38, and sub- stantially similar values are shown by the results of Deanesly and Parkesm using eruption of the front legs in XenopzGs laevis. RESTORATION AND MAINTENANCE OF GROWTH IN YOUNG THYROIDECTOMISED RATS- The retardation of growth in young thyroidectomised rats is prevented by the daily injection of a thyroid powder suspension or thyroxine. The possibility of using the response for measuring the biological activity of iodinated caseins and other allied substances was investigated by Rowlands.@ The method, however, as an assay procedure, presents difficulties in practice.The curve relating dose to response is so flat that a very large number of animals would have to be used, 6100 for an error of & 10 per cent. at P = 0.95. The time needed for the test, 90 days, is also too long. ANTAGONISM OF THYROXINE TO ANTI-THYROID DRUGS- Certain anti-thyroid compounds, such as ally1 thiourea, phenyl thiourea, sulphaguanidine, thiourea and thiouracil, prevent the formation of the active thyroid p r i n ~ i p l e . ~ g ~ ~ 947 The subsequent decrease in the amount of circulating hormone in the blood causes an increased production of thyrotrophic hormone by the anterior lobe of the pituitary; this, through its stimulating action, causes enlargement and histological changes in the thyroid gland.The hyperplasia and histological changes induced by the feeding of anti-thyroid drugs is inhibited by simultaneous administration of thyroxine, but not by and the ability of thyroidal substances to reduce the enlargement of the thyroids of chicks49 and rats48t60 treated with thiouracil may be used for the assay of thyroidal activity. The method gives results in accord with those obtained by the standard metabolic method.5* According to Hutcheon,% the method with rats is reasonably precise, the s/b value being 0.16. The chick has advantages over the rat in its cheapness, greater sensitivity, availability and ease of handling.The anti-thyroid drugs also cause a decrease in the iodine content of the thyroid glands, which can be prevented by the simultaneous administration of thyroxine.51 The iodine concentration in the glands can be measured chemically or by radio-active tracer techniques using 1311, and can therefore be employed as an index of thyroidal activity.540 SOMERS: THE MEASUREMENT GF THYROIDAL ACTIVITY [Vol. 75 CONCLUSIONS In spite of considerable effort, no wholly satisfactory method for measuring thyroidal activity has yet been evolved and discrepancies arise in comparing biological with chemical results. While chemical methods based on the estimation of acid-insoluble iodine (British Pharmacopoeia, 1948) may be preferred to those based on total iodine, neither appears to give a reliable indication of the physiological activity of various thyroid preparations.* 923 Discrepancies may be large, and indeed evidence is accruing that the published estimates for the thyroidal activity of iodo-casein preparations, which has been stated to be equivalent to about 3 per cent.of L-thyroxine,52 were in ‘The chief difficulty with biological assays is the extent of variation in the response of the individual animals, which necessitates the employment of relatively large numbers before an adequate degree of precision can be attained. While thyroxine is undoubtedly the active principle present in thyroid extractsa and in iodinated proteins,= the form in which it is administered plays an important part in its activity, especially by the oral Differences in the availability of thyroxine in various preparations may account for the discrepancies.Similarly, the same calorigenic substances may be utilised differently by different organisms and results obtained with one species may not apply to another. We can say that neither chemical analysis nor physiological assay will give a true index of the activity of a preparation on man; this can only be settled by clinical trial. In any event the development: of a new method for the synthesis of L- thyroxine60 may reduce considerably the need for bio-assays. ,67 y5* 959 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.REFERENCES “British Pharmacopoeia,’’ General Medical CoLtncil, London, 1948. Harington, C. R., and Randall, S. S., J . Yharw. Pharmacol., 1929, 2, 501. United States Pharmacopoeia, U.S. Pharmacopoeia1 Convention, Washington, 1947. Wokes, F., J. Pharm. Pharmacol., 1938, 11, 521. Deanesley, R., and Parkes, A. S., J . Endocrinol., 1945, 4, 324. Leland, J. P., and Foster, G. L., J . Biol. Chern., 1932, 95, 165. Reineke, E. P., Turner, C. W., Kohler, G. O., Hoover, R. D., and Beezley, M. B., Ibkd., 1945, Blau, N. F., Ibid., 1933, 102, 269. -, Ibid., 1935, 110, 351. Reineke, E. P., and Turner, C. W., Agv. Exp. Sta. Mo., Res. Bull., 1942, 355. Koche, J., and Michel, R., Biochem. Biuphys. Acta, 1947, 1, 335. Moser, H., Experientia, 1947, 3, 119. Winikoff, D., and Trikojus, V.M., Biochem. J., 1948, 42, 475. Simpson, G. K., and Traill, D., Ibid., 1946, 40, 116. Simpson, G. K., Johnston, A. G., and Traill, I)., Ibid., 1947, 41, 181. Borrows, E. T., Hems, B. A., and Page, J. E., J. Chem. Soc., 1949, S 204. Hunt, R., Arch. Intern. Med., 1925, 35, 671. Richards, A. N., and Collison, L. W., J . Physiol., 1928, 66, 299. Gaddum, J . H., Ibid., 1930, 68, 383. Wokes, F., “A Comparison of Different Methods of Estimating the Biological Activity of the Thyroid Hormone,” Ph.D. Thesis, University of London, 1938. Maclagan, N. F., and Sheahan, M., Proc. 1st Internat. Congr. Biochem., 1949, 413. Mclrch, J. R., J . Physiol., 1929, 67, 221. Gaddum, J. H., and Hetherington, M., J . Pharm. Pharmacol., 1931, 4, 183. Duran, M., Biochem. Z., 1920, 106, 254.Rydin, H., C.R. SOC. Biol., Paris, 1928, 99, l(585. Campbell, J. A., J . Physiol., 1938, 92, 2 9 ~ . Leblond, C. P., Proc. SOG. Exp. Biol., N . Y., 1944, 55, 114. Smith, A. U., Emmens, C. W., and Parkes, A. S., J . Endocrinol., 1947, 5, 186. Basil, B., Somers, G. F., and Woollett, E. A., Brit. J . Phavmacol., 1950, 5, 315. Bliss, I. C., and Marks, H. P., J . Pharm. Pharmacol., 1939, 12, 82 and 182. Irwin, J. O., J . Roy. Statist. SOG., 1937, 4, Supplement, 1. Cameron, A. T., and Carmichael, J., J . Biol. Chem., 1920, 45, 69. Haffner, F., and Konyama, T., Arch. Exp. Path. Pharmak., 1925, 107, 69. Kreitmair, H., 2. ges. exp. Med., 1928, 61, 202. Hutcheon, D. E., J . Pharmacol., 1948, 94, 308. Rotter, G., and Mecz, M., Arch. Exp. Path. Pharmak., 1932, 166, 649.Hunt, R., J. Bid. Chem., 1905, 1, 33. Hunt, R., and Seidell, A., Hyg. Lab. Bull., 1909, 47. Hunt, R., Amer. J . Physiol., 1923, 63, 257. Wokes, F., J . Pharm. Pharmacol., 1935, 8, 54. Laland, P., and Staa, K. F., Acta Pharmacol. Toxicol., 1949, 5, 1. Gudernatsch, J. F., Amer. J . Anat., 1912, 15, 431. Deanesly, R., and Parkes, A. S., J . Endocrinoi., 1945, 4, 324 and 356. 161, 599.541 October, 19503 SOMERS: THE MEASUREMENT OF THYROIDAL ACTIVITY 44. Rowlands, I. W., Ibid., 1945, 4, 305. 45. Mackenzie, C. G., and Mackenzie, J. B., Endocrinology, 1943, 32, 185. 46. Astwood, E. B., Sullivan, J., Bissell, A., and Tyslowitz, R., Ibid., 1943, 32, 210. 47. Astwood, E. B., J . Pharmacol., 1943, 78, 79. 48. Dempsey, E. W., and Astwood, E. B., Endocrinology, 1943, 32, 509.49. Mixner, J. P., Reineke, E. P., and Turner, C. W., Ibid., 1944, 34, 168. 50. Reineke, E. P., Mixner, J. P., and Turner, C. W., Ibid., 1945, 36, 64. 51. Astwood, E. B., and Bissell, A., Ibid., 1944, 34, 282. 52. Reineke, E. P., “Vitamins and Hormones,” Academic Press, New York, 1946, 4, 207. 53. Bailey, G. L., Bartlett, S., and Folley, S. J., Nature, 1949, 163, 800. 54. Kendall, E. C., Trans. Assoc. Amer. Phys., 1915, 30, 420. 55. Reineke, E. P., and Turner, C. W., J . Biol. Chern., 1943, 149, 563. 56. Schittenhelm, A., and Eisler, B., 2. ges. ex?. Med., 1932, 80, 569. 57. Thompson, W. O., Thompson, P. K., Dickie, L. F. N., and Alper, J. M., Arch. Intern. Med., 1933, 58. Elmer, A. W., “Iodine Metabolism and Thyroid Function,” Oxford University Press, London, 59.Monroe, R. A., and Turner, C. W., Amer. J . Physiol., 1949, 156, 381. 60. Chalmers, J. R., Dickson, G. T., Elks, J., and Hems, B. A., J . Chem. Soc., 1949, 3424. 52, 809. 1935. RESEARCH DIVISION GLAXO LABORATORIES LIMITED GREENFORD , MIDDLESEX DISCUSSION DR. H. 0. J. COLLIER said that for the test of thyroidal activity, using protection against toxic agents as a criterion, it might be expected that certain toxic agents would be more suitable than others. For example, some of the substances that were known to undergo a marked chemical detoxication in the body might be suitable for this test, since the test material could be expected to stimulate the general level of activity of the cells. Had Dr. Somers conducted any comparative experiments on different types of toxic agents for use in this test of thyroidal activity, and, if so, did they lend any substance to the foregoing suggestion ? DR.SOMERS replied that other chemical substances could be and had been tried, but he had only used acetonitrile. He thought Dr. Collier’s suggestion a reasonable one that probably explained the mode of action of thyroxine in lowering the toxicity of acetonitrile. MR. W. A. BROOM said that in his laboratory they had had experience with three of the methods described by Dr. Somers, namely, the method based on the change in metabolism in guinea-pigs, the closed vessel mouse method and the tadpole method. In their hands, the metabolism method was most unreliable and time consuming, and the mouse method, though workable, was less precise than the tadpole method.They preferred the tadpole method, which gave results that agreed very well with the experience of veterinary workers studying the effect of iodinated casein on the milk yield of cows. DR. R. PITT-RIVERS pointed out that experimental evidence was available that D-thyroxine was active when tested on myxoedematous patients; it had one-eighth to one-tenth of the activity of L-thyr0xine.l She did not consider it justifiable to compare a D- with a DL-amino acid and to deduce the relative activities of D- and L-forms because, in the presence of the natural isomer, the very slight activity of the unnatural isomer might be masked or lost owing to experimental error. REFERENCE TO DISCUSSION 1. Pitt-Rivers, R., and Lerman, J., J . Endocrinol., 1948, 5, 223.542 STEWART : THE ASSAY OF POSTERIOR PITUITA4RY LOBE EXTRACTS [Vol.75 The Assay of Posterior Pituitary Lobe Extracts BY G. A. STEWART SYNOPSIS-A review is given of the methods of assaying the oxytocic, vasopressor and antidiuretic hormones of posterior pituitary lobe extracts. The difficulties of assaying the oxytocic hormone by the guinea-pig uterus method are described in detail, and descriptions are given of the rat uterus and chicken blood-pressure assays, which enable the hormone t o be determined with a much greater degree of accuracy. Assays of the vaso- pressor hormone on the blood pressure of the spinal cat and anaesthetised dog are described, and mention is made of the more recently developed assays of this hormone on the blood pressure of the rat.Assays of the antidiuretic hormone on the rat and dog are described. The relative merits of the assays referred to are compared and discussed. The variability of commercial extracts in respect to their contents of these three hormones is considered. THREE of the active principles of posterior pituitary lobe extracts are used clinically, namely, the oxytocic, vasopressor and antidiuretic hormones, and of these, as a rule, only the oxytocic and vasopressor hormones are assayed commercially. Generally an extract containing all three hormones is assayed for its oxytocic activity only. THE ASSAY OF THE OXYTOCIC HORMONE- The “in vitro” method of Dale and Laidlawl using the guinea-pig uterus for the assay of the oxytocic principle is beset by many practical difficulties, notably that of obtaining a suitable uterine preparation.The uterus usually employed is that from a virgin guinea-pig in dioestrus. Some uteri contract very slowly when stimulated by posterior pituitary lobe extract, some fail to give good differentiation between graded doses of the same extract, whilst others may react to the same dose of the pituitary extract with alternate large and small contractions. As a result of these many difficulties, different workers have sought to find conditions under which the guinea-pig uterine assay may work to advantage in their own laboratories, whilst others have investigated the “in vitro” assay using the uteri of different animals. Morell, Allmark and Bachinski2 described a method using eight small longitudinal strips of uterus, obtained from both uterine horns of a guinea-pig, suspended in Van Dyke and Hastings’ s~lution.~ The number of strips responding to a dose of pituitary extract was recorded on a smoked kymograph paper. Three doses of the standard solution and three doses of the unknown solution, whose potency was assumed to be equivalent to that of the standard, were administered in various volumes.The larger doses produced responses from between 50 and 87.5 per cent. of the strips. The medium doses produced responses from about 50 per cent., and the smaller doses, responses from about 25 per cent. of the strips. Each dose was given three times in a restricted randomised design. The injections of the standard doses alternated with those of the unknown solution. They observed that it was preferable to give the higher concentrations first.From these quanta1 responses the potency of the unknown solution could be calculated and a standard error determined. There was no significant difference in the results obtained with uterine strips from the uteri of guinea-pigs in oestrus and in dioestrus. Later, these workers4 transformed the same technique into a quantitative assay by measuring all responses over 1 mm. produced by the eight uterine strips. Of 23 assays on solutions whose potencies were known, 16 assays produced a range of percentage standard errors that did not include the known potency and thus showed the fallibility of the method. Hamburger5 studied the assay of the oxytocic hormone on the isolated uterus of avirgin guinea-pig and used four concentrations of both the standard and the unknown solutions.The ratio between successive concentrations was 2 to 1. Two experimental designs for the injection of doses into the bath were considered, one in which the four concentrations of standard were given in diminishing order, followed likewise by the four doses of the unknown solution. The second design was to give The response was related to the logarithm of the dose. This dose sequence was given three times in all.October, 19501 STEWART : THE ASSAY OF POSTERIOR PITUITARY LOBE EXTRACTS 543 doses of the standard alternately with doses of the unknown twice at each of the four dose levels, the latter being randomised. Responses were measured and related to the logarithm of the dose. From these data the potency of the unknown solution could be determined.No complete analysis of the results was given, but Hamburger claimed that a posterior pituitary lobe extract could be standardised with " sufficient accuracy" by the employment of three or four suitable guinea-pig uteri. Some workerss s 7 have attempted to improve the guinea-pig uterine assay by modifying the magnesium and calcium ion concentrations of the Ringer's solution. They claim thereby to have improved the assay. Fig. 1 shows the tracing of a test carried out by the assay procedure adopted in our laboratories. The uterus of a virgin guinea-pig was suspended in the physiological salt solution described in the British Pharmacopoeia,s maintained at a temperature of 37" C.and aerated with compressed air. Recordings were made on a smoked kymograph paper with a frontal writing lever giving a magnification of x 2. Two different submaximal responses to an extract of the standard were obtained. Next a dose of the unknown solution giving a response intermediate between the two standard responses was found, and the three doses were repeated in the reverse order. The maximum ratio we allow for the two doses of standard is 1.33. The interval between doses was 10 minutes. GUINEA-PIG 210 G. S = 0.0167 UNIT/rnl. DOSES IN mi. T = UNKNOWN 11600 0.8 0.6 0.7 0.7 0.6 0.8 S U S T T S S MIN Fig l.* Typical example of the isolated guinea-pig uterine assay of a posterior pituitary extract. S = ml. of Extract of the International Posterior Pituitary Standard 2 I.U.per ml.; dilution 1 : 120. T = ml. of Unknown; dilution 1 : 600. * Fig. 1 and all the tables in this paper are reproduced from the Journal of Pharmacy and Pharmacology, 1949, 1, 436-453, by permission of the Editor. Gaddums studied the precision of the guinea-pig uterine assay when four doses were employed, e.g., standard, unknown, unknown, standard, until the unknown is found to have an activity greater than one particular potency and less than another potency. He discarded results in which the ratio of the two limiting activities was greater than 1.5, and found that the standard error on 43 assays was 7.73 per cent. HoltonlO described a method employing the isolated uterus of a non-pregnant rat of weight 120 to 200g., suspended in Locke's solution having one-quarter of the calcium and one-half of the glucose concentrations normally employed.Recordings of the uterine con- tractions were made with an isotonic and linear lever of Schild's design.ll The bath was aerated with a mixture of 95 per cent. of oxygen and 5 per cent. of carbon dioxide and kept at a constant temperature of about 32" C. A four-point assay was conducted by the method described by Schild.12 Doses were administered regularly at intervals of three or four minutes and the mean experimental time of an assay was 3.75 hours, with a mean percentage error of544 2.16 on eight satisfactory assays: We have confirmed that this is the most precise of the isolated-ute:rus methods for the assay of the oxytocic hormone. STEWART THE ASSAY OF POSTERIOR PITUITARY LOBE EXTRACTS Wol.75 Fig. 2 is a tracing of such a test. RAT 175 G. A = 0.06 UNI B = 0.045 UN F* 40 r 8. 4. 49 DOSE INTERVAL 4 MINS. C = 0.3 ml. UNKNOWN /300 D -- 0.3 ml. UNKNOWN /400 A BC D BA DCC DA BDC B A Fig. 12. Bachinski and Allmark13 have conducted “in. vitro” assays on the uterus of the guinea-pig, rat and rabbit and on strips of human uterine muscle excised during Caesarean sections at or near full term. They found the sensitivity of the rabbit uterine muscle to “Pitocin” and “ Pitressin” (Parke Davis preparations of the relatively pure extracts of the oxytocic and vaso- pressor principles) more closely resembled that of human uterine muscle than did those of the uterine muscles from other animals. A sample of Pitocin assayed against an extract of the Canadian standard posterior pituitary powder gave results on the normal guinea-pig uterus that differed significantly from the assay conducted on the normal rabbit uterus or human uterine muscle.They employed Van Dyke and Hastings’ solution3 throughout their assays, because its magnesium ion concentration approximates to the serum magnesium ion level of man as well as of pregnant woman. that Pitressin exerts a variable oxytocic activity depending upon the concentration of magnesium and calcium ions in the physiological salt solution : the observations on variations in the concentration of magnes- ium confirmed the results of Fraser.15 It was also shown14 that the sensitivity of guinea-pig uteri to stimulation by Pitocin and Pitressin depends upon the hormonal state of the guinea- pig, Table 1-shows that the sensitivity of virgin guinea-pig uteri to stimulation by Pitocin TABLE I It has recently been THE SENSITIVITY OF UTERI TO PITOCIN AND PITRESSIN FROM VIRGIN GUINEA-PIGS TREATED WITH PROGESTERONE AND STILBOESTROL Uterine sensitivity r Sensitivity ratio, Pitressin, J- Pitocin/Pitressin rnilIiunits/ml.milliunits/ml. Progesterone-treated guinea-pigs . . . . 0.04-0.33 0.08-0.26 1.08 Stilboestrol-treated guinea-pigs . . . . 0*01-0*04 0*002-0.018 2-60 Sensitivity ratio of progesterone-treated guinea-pigs to stilboestrol-treated guinea - .. .. .. .. 0.138 0.058 - pigs . . .. is much greater when they have undergone a treatment with stilboestrol than with progesterone. The uteri of The same was found for the oxytocic effect of Pitressin.October, 19501 STEWART: THE ASSAY OF POSTERIOR PITUITARY LOBE EXTRACTS 545 stilboestrol-treated guinea-pigs are more sensitive to Pitocin than to Pitressin, whilst the uteri of progesterone-treated guinea-pigs are almost equally sensitive to both.Because of the complex nature of a pituitary extract and because the ratio of the amount of the oxytocic hormone to the amount of the vasopressor hormone in the extract may not be the same as in an extract of the international standard, exact details of the conditions of assay should be stated when a sample is being compared in different laboratories. Con- fusion has often arisen over the potency of a particular commercial sample assayed on the guinea-pig uterus in different laboratories. Table I1 shows the results of the assay of commercial extracts prepared by six different firms and examined for oxytocic activity under the same conditions on the guinea-pig uterus TABLE I1 RESULTS OBTAINED FOR THE OXYTOCIC ACTIVITY OF COMMERCIAL POSTERIOR PITUITARY PER ML., WHEN ASSAYED BY THE GUINEA-PIG UTERUS METHOD USING RINGER’S SOLUTION EXTRACTS, LABELLED TO CONFORM WITH THE B.P.REGULATIONS OF 10.0 OXYTOCIC UNITS B.P. 1948 Sample Potency, Limits, u./ml. u./ml.* A 10.0 8.G11.4 B 8.8 7.5-10.0 C 7.8 6.7- 8.9 D 11.0 10.0-12.2 E 8.7 7.5-10.0 F 7.7 6.7- 8.9 * Range of upper and lower doscs used in the bracketting arrangement described for the uterine assay. in the physiological salt solution described in the British Pharmacopoeia.8 All the ampoules were labelled 10 units per ml.The samples did not possess equal oxytocic activity. Table I11 shows the results obtained for the assays of six different mixtures of Pitocin and Pitressin in which the ratio of Pitocin to Pitressin ranged from 4.0 to 0.125. The assays TABLE Irr TYPICAL DIFFERENCES IN GUINEA-PIG UTERINE ASSAY RESULTS USING TWO USUAL FORMULAE FOR THE PHYSIOLOGICAL SALT SOLUTION Ratio, Pitocin Pitressin 4.00 2.00 1.00 0.50 0.25 0,125 Actual total oxytocic potency, * u./ml. 10.1 10.2 10.4 10-8 11.6 13.2 Ringer’s solution, B.P., 1948, potency found, u./ml. 8.0 9.2 10.0 9.3 11.2 13.0 Ringer’s solution, with added magnesium (Hsu, 1948)t 10.6 10.6 10.6 13-6 13.6 19.2 * Allowance has been made for 4 per cent. of oxytocic hormone impurity in Pitressin and for 4 per cent.OlE vasopressor hormone impurity in Pitocin in computing the actual total ox*ocic potencies of the solutions. t Hsu (1948) Ringer contains 0-045 per cent. of magnesium chloride. were conducted in two physiological salt solutions, one of which contained the magnesium ion concentration stated in the British Pharmacopoeia8 and the other a much higher con- centration.’ The assays conducted in the latter solution produced much higher results than those expected when the ratio of Pitocin to Pitressin was 0.5 or less. Gaddumle studied under various conditions the stability of the oxytocic hormone of the British standard of 1926, which later became the 1935 International Standard Powder. We have confirmed his results for powders of the potency required by the United States Pharmacopoeia XIII, i.e., not less than 1000 oxytocic units per gram.In recent years, however, commercial samples have very rarely attained this minimum unitage, and in a comparison of the stability of commercial powders of oxytocic potency less than 1000 units per g. with powders of potency similar to that of the International Standard Powder we obtained different results. Also the ratio of the vasopressor hormone to the oxytocic hormone was not always unity in the less potent powders. McClosky, Miller and Le Messurier17 have found that in whale hypophysis the ratio of the vasopressor hormone to the oxytocic hormone546 STEWART : THE ASSAY OF POSTERIOR PITUITARY LOBE EXTRACTS is greater than unity. For various reasons, therefore, assays are sometimes attempted in which the ratio of the oxytocic hormone to the vasopressor hormone in the standard differs from that in the sample under test.Because of these many difficulties with “in vjtro” assays we decided to study an “in vivo” assay. GaddumlS showed that the depressor effect of posterior pituitary extract on the blood pressure of the fowl was due to the oxytocic hormone. Coonlg conducted assays on the blood pressure of a cockerel anaesthetised with sodium phenobarbitone. The procedure was to match a response of the standard with an equal response of the unknown solution. Tachyphylaxis gradually develops and thus any method based on matching responses will entail certain inaccuracies. Blackwell Smith, jun., and Vos, jun.,20 employing the cockerel, performed, according to the design of Bliss and Marks,21$22 a four-point assay with a dose injected every 10 minutes in a random arrangement.The latent sources of variation due to the influence of the height of the blood pressure immediately preceding a given response and the size of the preceding dose were ignored. An analysis of the data from several assays by partial regression co- efficients showed that small fluctuations in blood pressure and the effect of the preceding dose could be ignored without markedly influencing the accuracy of the assays. Errors up to 18.2 per cent. with an average error of 6.9 per cent. were obtained in their assays of posterior pituitary solutions of known strength. Thompson,23 however, pointed out that the main difficulty was to eradicate from experi- mental comparisons the effect of changing sensitivity and this was only achieved in part by the four-point random dose method, as any change in sensitivity occurring during the time required for the injection of, and response to, the four doses causes variations in the responses, which would contribute to the error of the assay. This can be great in the chicken blood-pressure assay.Thompson therefore used a statistical design similar to that used by Vosa for the assay of ergometrine. When White Leghorns were in short supply we tried Rhode Island Red cockerels, but the sensitivity of these birds to posterior pituitary extract was extremely small. On two occasions no response was elicited. Light Sussex cockerels appeared to be suitable subjects and we finally adopted them for our assays.The weight range of these birds was from 1.8 to 2-3 kg., and the birds were injected intravenously with 180 mg. of sodium phenobarbitone per kg. We tried other barbiturates but none was as good as sodium phenobarbitone in maintaining a high blood pressure in the bird. The cockerels were prepared for assay according to the description given by Coon. The blood pressure was recorded from the ischiadic artery by means of a capillary mercury manometer, and 8.5 per cent. of sodium citrate was used as anticoagulant. An extract of the International Standard, 2 oxytocic units per ml., was generally diluted to 1 in 10 and occasionally to 1 in 20 with 0.9 per cent. saline. Doses not exceeding 0.4 ml. were injected rapidly into the crural vein regularly every 3, 4 or 5 minutes.The unknown was diluted to a strength assumed to‘be that of the standard dilution, and assays were carried out on the same bird according to the Schild12 experimental design and also that described by V O S . ~ ~ If the standard and test materials are similar to each other, the experi- mental design qf VosM gives the more accurate results. Should the standard and unknown materials be dissimilar then by the analysis of variance possible with Schild’s experimental design12 one can say whether the assay is valid or not. Fig. 3 illustrates the V o P experimental design. The dose of standard is kept constant whilst the three different doses of the unknown are given, so as to produce one response greater than, one equal to, and one less than that produced by the standard dose.We have found that falls in blood pressure of 30 to 60 mm. of mercury give responses linearly related to the logarithms of the doses. When the blood pressure falls to a low level, ephedrine hydrochloride, 4 to 8 mg. per kg., may be administered to raise it. After this has risen it will remain elevated for a period of 1 to 3 hours and falls only very gradually. If the bird becomes insensitive to posterior pituitary extract then it may be rested for an hour or two, after which the sensitivity has usually returned. Table IV shows a comparison of the results of six different solutions when assayed by the guinea-pig uterine method, using the physiological salt solution described in the British Pharmacopoeia,8 and by the chicken blood-pressure method.The column on the extreme right shows the good agreement between the two methods. Fol. 75 > Both Coon19 and Thompson23 used White Leghorn cockerels.October, 19501 STEWART: THE ASSAY OF POSTERIOR PITUITARY LOBE EXTRACTS TABLE IV COMPARISON OF THE GUINEA-PIG UTERUS AND CHICKEN BLOOD-PRESSURE METHODS FOR THE MEASUREMENT OF OXYTOCIC ACTIVITY 547 Chicken depressor u./g. & S.E. (C) Sample method, A 677 -J= 52 B 872 f 70 C 762 f 49 D 792 f 68 E 921 & 49 F 785 f 65 G 467 f 44 Guinea-pig uterus, 700 875 637 700 900 824 600 u-/g. (GI Ratio, C/G 0.967 0.997 1.196 1.131 1.023 0.950 0.945 Coon,l9 using White Leghorn cockerels, showed that the vasopressor hormone produced a potentiating effect on the depressor response of the bird to oxytocic hormone when the ratio 5.7. 49 COCK 2.3 Kg. DOSE INTERVAL 4 MINS. S = 0.06 UNIT B = 0.3 ml. UNKNOWN 1/50 A = 0.2 ml. UNKNOWN 1/50 C = 0.4 ml. UNKNOWN 1/50 S A S B S C S B S C S A S C S [ZERO 6.r Fig. 3. of the oxytocic to the vasopressor hormone was less than 0.4. Table V shows the results obtained on the assays of mixtures of Pitocin and Pitressin in Light Sussex cockerels with the TABLE V THE INFLUENCE OF DIFFERENT RATIOS OF PITOCIN TO PITRESSIN ON THE ASSAY OF THE OXYTOCIC PRINCIPLE OF THE POSTERIOR PITUITARY LOBE GLAND BY THE CHICKEN DEPRESSOR METHOD Ratio, Pitocin Pitressin 4-00 2-00 1.00 0.50 0.25 0.125 0.100 Actual total oxytocic potency,* u./ml. 10.1 10.2 10.4 10.8 11.6 13.2 14.4 Potency found, u./mI. 10.3 11.2 9.7 10.8 11.1 11.1 18.6 Limits of error ( p = 0.95) 9.3-1 1.3 8.9-13-5 8 .6 1 1-0 10.0-1 1.6 10- 1-1 2.1 8.3-13.9 16.0-2 1.6 * Allowance has been made for 4 per cent. of oxytocic hormone impurity in Pitressin and for 4 per cent. of vasopressor hormone impurity in Pitocin in computing the actual total oxytocic potencies of the solutions ratio of Pitrocin to Pitressin from 4.00 to 0.100. Only when the ratio of Pitocin to Pitressin is 0.100 or less does the vasopressor hormone produce any appreciable potentiating effect. The chicken assay is more reliable and more informative in that fiducial limits can be calculated; it requires less skill and experience on the part of the laboratory assistant and less time, 1 to 1.75 hours for an assay involving 10 doses of the unknown solution, compared with 1 to 2 days for the ordinary guinea-pig uterine assay, and it is less expensive than the guinea-pig uterine assay.548 STEWART : THE A S S U OF POSTERIOR PITUITARY LOBE EXTRACTS Pol.75 THE ASSAY OF THE VASOPRESSOR HORMONE- Two methods have been widely used for the assay of the vasopressor hormone. Hogben, Schlapp and MacDonald25 used the spinal cat method. Burn26 described the preparation of the cat by Elliot’s method. The blood pressure was recorded from the left common carotid artery and injections of the drug were made into the right external jugular vein. The blood pressure falls about an hour after the operation and can be maintained at that level for long periods of time. We have found that it is best to leave the preparation at least 2 hours after the operation before beginning the assay.The standard extract of 2 units per ml. is diluted to 1 in 10 with 0.9 per cent. saline and the unknown solution is diluted to a potency assumed to be equal to that of the standard dilution. Injections do not exceed 0.5ml. Although Burn stated that in some cats the interval between doses can be shortened to half an hour or even 20 minutes, we have found that hourly intervals are better, since this reduced the likelihood of tachyphylaxis. After 10 hours the preparation becomes more sensitive to the vasopressor hormone. Doses used are equal to half what is required to produce maximum response, and responses are matched by the null method. Doses differing by 10 per cent. or more can be distinguished. Since the time required to assay a sample by the cat method is long, and since there is an acute shortage of experimental cats, we have used the anaesthetised dog in the method described by Kamm, Aldrich, Grote, Rowe and Bugbee.27 Medium-sized healthy dogs are deeply anaesthetised with an intraperitoneal injection of 0.4 g.of chlorbutol per kg. Cannulas are inserted in a femoral vein and a carotid artery. Injections are made every 15 minutes and doses are kept as small as possible. The method proved to be as accurate as the spinal cat method. Fig. 4 shows the tracings obtained in assays of a solution by the cat and dog methods. Simon2* used the effect of the vasopressor hormone on the blood pressure of the decapi- tated rat and found this to be an extremely sensitive preparation, responding to 2.5 to 5.0 milliunits. Shipley and Tilden29 have used the pithed rat with satisfactory results. Stephenson30 described a preparation of the hind limbs 6f the rat perfused a t room temperature with Ringer - Locke solution through the abdominal aorta. He designed a simple and sensitive outflow recorder for measuring the changes in pressure produced by the pressor hormone.After 4 to 5 hours the preparation is said to respond to 2 milliunits or even 1 milliunit of posterior pituitary extract. Some preparations are known to last 24 hours. In our experience it was found after a few experiments that the preparation failed to recover after the second or third injection of the drug into the perfusate. Histamine in small doses was without. action. THE ASSAY OF THE ANTIDIURETIC HORMONE- Although most types of animals have been used at one time or another for assay of the antidiuretic hormone, the method described by Burn31 has been widely employed. Sixteen male rats, weighing 140 to 240g., are divided into two groups.Each rat is given 5ml. of water per 1OOg. of body-weight by stomach tube. The posterior pituitary extracts, standard and unknown, are injected subcutaneously into the two groups of rats respectively, and the urine excreted is collected. The assay is based upon the time interval between the administration of water and the maximum excretion of urine during periods of 15 minutes. The time elapsing is greater the larger the dose of the extract. The potency is calculated from a previously determined curve relating dose to time elapsing before the maximal rate of excretion occurs.The estimated results, according to Burn, differed from the true results as much as 23 per cent. and the average variation was 12 per cent. Recently, Theobald, Graham, Campbell, Gange and Driscoll32 have made reference to assays of the antidiuretic hormone by the method described by Verney and his Theobald has shown that the intravenous injection of from 0.5 to 10 milliunits of “ Infundin” (Burroughs Wellcome preparation) inhibited water diuresis in the dog and man and in woman during the last few weeks of pregnancy. The amount of antidiuretic activity necessary to inhibit water diuresis was remarkably constant for each dog, and the water-diuresis curves obtained over a period of several months could be almost superimposed.He adds that the antidiuretic activity affords the most delicate and possibly the most precise method of assaying posterior pituitary extracts, and he assumes that the vasopressor hormone and the antidiuretic hormone are one and the same principle.October, 19501 STEWART : THE ASSAY OF POSTERIOR PITUITARY LOBE EXTRACTS 649 Heller,= ?36 however, has prepared from the vasopressor fraction an extract that contains a high proportion of antidiuretic activity and very little vasopressor activity. Wokes37 found in a series of assays of commercial posterior pituitary lobe preparations, assayed for oxytocic potency on the guinea-pig uterus and for antidiuretic potency on the rat, that four out of 20 assays gave results showing a disproportion of the two active principles that he claimed was significant.We have found disproportionate amounts of the oxytocic and vasopressor hormones in commercial pituitary extracts. DOG 8 5.0 Kg. B.P. 70 m.m. Hg. DOES IN mls. 3. 5. 49 DOSE INTERVAL 15 MINS. S = 0.2 UNITS/rnl. T = UNKNOWN 1/50. Q*4 0.5 0.3 0.25 0.4 0.3 0.3 0.4 0.4 0.5 S S T T S S 7 T s S 0.25 T. E 0.4 S. CAT $2 2.8 Kg. B.P. 70 m.m. Hg. DOSES IN mls. 5. 5. 49 DOSE INTERVAL 60 MINS. S = 0.2 UNITS/ml. T = UNKNOWN 1/50. 0.5 0,4 0.3 03 0.45 0.3 0.2 S 5 T T S T 5 0.3 T. E 0.45 S. Fig. 4. When the potency of one of the posterior pituitary hormones in a pituitary extract is required, a method should be employed that measures only that active principle. It should not be assumed, for example, that if the extract contains 10 vasopressor units per ml.when assayed against the International Standard, it will necessarily contain an equivalent unitage of the oxytocic and antidiuretic principles. I wish to thank the Directors of the Wellcome Foundation Ltd. for permission to publish this paper. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. Dale, H. H., and Laidlaw, P. P., J . Pharmacol., 1912, 4, 75. Morell, C. A., Allmark, M. G., and Bachinski, W. M., Ibid., 1940, 70, 440. Van Dyke, H. B., and Hastings, A. B., Awier. J . Physiol., 1928, 83, 563. Bachinski, W. M., Allmark, M. G., and Morrell, C. A,, Canad. J . Res., 1945, E23, 126. Hamburger, C., Acta Pharmacol. Toxicol., 19-15, 1, 112. Garcia de Jalon, P., Farwzacoter. Act., 1947, 4, 177. Hsu, Yu-C., Quart.J . Pharm. Pharmacol., 1948, 21, 146. “British Pharmacopoeia,” 1948, p. 812. Gaddum, J. H., Quart. J . Pharm. Pharmacol., 1938, 11, 697. Holton, P., Brit. J . Phannacol., 1948, 3, 328. Schild, H. O., Ibzd., 1947, 2, 189. Bachinski, W. M., and Allmark, M. G., J . Amer. Pharm. Ass., Sci. Ed., 1947, 36, 73. Stewart, G. A., J . Pharm. Phavmacol., 1949, 1, 436. Fraser, A. M., J . Pharmacol., 1939, 66, 85. Gaddum, J. H., Biochenz. J., 1930, 24, 941. McClosky, W. T,, Miller, L. C., and LeMessurier, D. N., J . Pharmacol., 1936, 57, 132. -, J . Physiol., 1942, 101, 115.550 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. STEWART : THE ASSAY OF POSTERIOR PITUITARY LOBE EXTRACTS [Vol. 75 Gaddum, J . H., J . Physiol., 1928, 65, 434. Coon, J. M., Arch. Int. Pharmacodyn., 1939, 62:, 79. Blackwell Smith, R., jun., and Vos, B. J., jun., J . Pharmacol., 1943, 78, 72. Bliss, C. I., and Marks, H. P., Quart. J . Pharna. Pharmacol., 1939, 12, 82. Thompson, R. E., J . Pharmacol., 1944, SO, 373. Vos, B. J., jun., J . Amer. Pharm. Ass., Sci. Ed., 1943, 32, 138. Hogben, L. T., Schlapp, W., and MacDonald, A. D., Quart. J . Exp. Physiol., 1924, 14, 301. Burn, J. H., “Biological Standardisation,” Oxford University Press, London, 1937, p. 67. Kamm, O., Aldrich, J. B., Grote, I. W., Rowe, L. W., and Bugbee, E. P., J . Amer. Chem. SOC., Simon, A, Arch. Exper. Path. Pharmak., 1937, 187, 678. Shipley, R. E., and Tilden, J. H., Proc. Soc. Ex$. Biol. Med., 1947, 64, 453. Stephenson, R. P., J . Physiol., 1948, 107, 162. Burn, J. H., Quart. J . Pharm. Pharmacol., 1931, 4, 517. Theobald, G. W., Graham, A., Campbell, J., Gange, P. D., and Driscoll, W. J., Brit. Med. J., Klisiecki, A., Pickford, M., Rothschild, P., and Verney, E. B., Proc. Roy. SOC., B., 1933, 112, 496. Verney, E. B., Ibid., B., 1947, 135, 25. Heller, H., J. Physiol., 1939, 96, 337. Wokes, F., Quart. J . Pharm. Pharmacol., 1932, 5, 390. , , Ibid., 1939, 12, 182. -- 1928, 50, 573. 1948, 2, 123. -, Ibid., 1940, 98, 405. WELLCOME BIOLOGICAL CONTROL LABORATORIES DARTFORD, KENT DISCUSSION PROFESSOR GADDUM enquired whether it was possible to obtain preparations of human uterus that were more sensitive to the vasopressor than to the oxytocic hormone. MISS MACAULAY enlarged on Professor Gaddum’s mtmtion of the anaesthetised rat method1 for assaying the vasopressor activity of the posterior lobe of the pituitary. In their experience the chief advantages of the preparation were (a) its sensitivity-it responded t o 0.001 i.u. of vasopressor activity, while 0.005 i.u. gave a rise in blood pressure of 30 to 40 mm.; (b) it did not respond to the oxytocic fraction: (c) 10 per cent. discrimination was easily obtained; ( d ) no tolerance cleveloped to injections made every 10 minutes- usually 30 to 40 injections were made into the same animal; (e) it was completely insensitive to the histamine concentrations encountered in the worst commercial powders. DR. F. J. DYER asked the author if he would favour making routine assays on low potency commercial pituitary powders by two methods, oxytocic and vasopressor (or antidiuretic) and expressing the answer as a ratio of the one to the other. The International Standard was assumed (by B.P. definition) to possess an oxytocic to vasopressor ratio of 1 to 1. MR. STEWART said that Russell2 had claimed that Pitressin had a much greater oxytocic effect than Pitocin on human uterine strips excised a t term. More recently, Bachinski and Allmark conducted assays on a statistical basis and found that Pitocin and Pitressin were equally active in oxytocic potency on an isolated uterine strip taken a t the peak of pregnancy. He himself had not conducted assays with this type of preparation. Replying to Dr. Dyer, Mr. Stewart said that the use of the rabbit uterine assay and the expression of the answer in terms of one hormone only was possibly the most suitable method. At least Bachinski and Allmark had shown that assays conducted on the normal rabbit and parturient human uteri were comparable. However, the best method could only be stated after further clinical researches had been made into the mode of action of these hormones. REFERENCES TO DISCUSSION 1. 2. Landgrebe, F. W., Macaulay, M. H. I., and Waring, H., Proc. Roy. SOC. Edin., B., 1946,62, 202. Russell, C. S., J . Obstet. Gynaecol. Brit. Empire, 1943, 50, 287.