1 00:00:00,150 --> 00:00:06,600 I've put this up as a joint presentation with my colleague from Birmingham, Robyn Fenner, 2 00:00:06,600 --> 00:00:14,350 that's our department on the left and he did our work at the city hospital in Birmingham as well as other places. 3 00:00:14,370 --> 00:00:20,310 I'll sit down if that's okay. You're happy with that? Nobody going to throw rotten eggs at me. 4 00:00:21,270 --> 00:00:31,140 Good. Yeah, right. Well, that is a man called Philips Aurelius Theophrastus from Bastos von Hohenheim. 5 00:00:32,010 --> 00:00:38,130 And to make life easy for all of us, he called himself Paracelsus, which makes life a lot easier. 6 00:00:38,640 --> 00:00:49,950 You can see his dates there from the 16th century when he called himself Paracelsus because he was interested in this man who is Celsus, 7 00:00:50,370 --> 00:01:00,420 a very famous Roman physician who was highly regarded and whose name is Celsus means high or noble or high minded. 8 00:01:01,260 --> 00:01:17,790 So our friend decided to call himself Paracelsus, which is sort of noble ish or high ish or high minded ish, but it's also a pun on his surname. 9 00:01:18,810 --> 00:01:25,470 Which is Hockenheim. And Hockenheim means the hi ho, literally a place near heaven. 10 00:01:26,250 --> 00:01:29,820 Which Celsius is as well a high place. 11 00:01:30,270 --> 00:01:36,840 So Paracelsus is a pun on Hohenheim and is in homage to his hero Celsius. 12 00:01:37,440 --> 00:01:42,540 Hmm. And in 1564, Paracelsus wrote this book. 13 00:01:42,540 --> 00:01:47,340 It's called Septem Defence Bonus, although its proper title is in German. 14 00:01:47,670 --> 00:01:53,430 Surprisingly, for those times, you'd expect a medical text to have been written in Latin. 15 00:01:53,910 --> 00:02:03,880 But he wrote in the vernacular Sieben Tag Diggings Raiden and the subtitle means Paracelsus is the reply. 16 00:02:03,900 --> 00:02:15,090 If you like two criticisms from his opponents which have demeaned him in the eyes of his patrons, 17 00:02:15,900 --> 00:02:21,600 he was rather arrogant man and his theories built not on the classical theories 18 00:02:21,600 --> 00:02:26,770 of the day which were dependent on the Hippocratic Theory of the four humours. 19 00:02:27,330 --> 00:02:34,500 So which I guess many people are familiar. Hot and cold, but dry and wet. 20 00:02:34,720 --> 00:02:43,950 And the balance in the humours in the body, the choleric, the melancholic, the the sanguine in the splenetic humours. 21 00:02:44,190 --> 00:02:49,860 The idea was that there was an imbalance in the humerus in the body causing disease. 22 00:02:49,860 --> 00:02:57,960 If you could rebalance the body, the humerus would come into perfect harmony and you'd be cured of your disease. 23 00:02:58,200 --> 00:03:06,239 And this was, as I say, a Hippocratic theory, which Galen had propounded from the seventh second century onwards and which 24 00:03:06,240 --> 00:03:11,360 actually dominated medical thinking until even the middle of the 19th century. 25 00:03:11,490 --> 00:03:15,780 Very powerful theory. But Paracelsus had a different theory. 26 00:03:15,790 --> 00:03:23,790 He had a tripartite theory based on things that he used to treat based on the Trinity, the Holy Trinity, and so on. 27 00:03:24,210 --> 00:03:27,750 So he wasn't very popular among his colleagues and was highly criticised. 28 00:03:28,110 --> 00:03:35,340 So in 1564 was published after he died, of course he wrote it before his death, his seven defences. 29 00:03:35,580 --> 00:03:38,190 And this is a page from the first defence. 30 00:03:38,640 --> 00:03:48,560 As you can see, it's written in German in typical Gothic print of the time and I've blown up the top bit so you can see it a bit more clearly. 31 00:03:49,650 --> 00:03:55,860 And I've highlighted the most famous sentence from this book, which is still quoted today, 32 00:03:55,860 --> 00:04:06,390 and there it is for us is thus net gift est allerdings in gift and Nixon gift align the doses marked ascending kind gift ist. 33 00:04:06,600 --> 00:04:10,860 It is written out. It means what is a poison? 34 00:04:13,100 --> 00:04:17,450 Everything he says. Allerdings in gift. Gift is a poison nowadays felt with one ask. 35 00:04:17,810 --> 00:04:21,110 Everything is a poison. And then he repeats. 36 00:04:21,110 --> 00:04:24,280 It admits own gift, and nothing is not a poison. 37 00:04:24,290 --> 00:04:32,869 He really wants to tell you this. Everything is and nothing is not only the dose determines that something is a poison. 38 00:04:32,870 --> 00:04:39,140 And he's absolutely right. And this is a message that has been forgotten for 400, 500 years or more. 39 00:04:40,580 --> 00:04:45,530 Not only does he tell you it twice in German, but he tells you it in Latin as well. 40 00:04:46,190 --> 00:04:51,170 In the marginal note, nothing is without poisonous effect. 41 00:04:52,770 --> 00:04:57,030 Except for the dose. That determines it. 42 00:04:58,350 --> 00:05:01,410 So 1564 posthumously published. 43 00:05:01,830 --> 00:05:06,750 We now move forward 300 years to 1864. 44 00:05:07,260 --> 00:05:15,870 And these two gentlemen, Cato Goldberg. On the left, there's a Norwegian professor of chemistry and his colleague, if you can spot him there. 45 00:05:16,380 --> 00:05:22,590 Pater Varga surprisingly had time for anything in the scientific world, was a mathematician. 46 00:05:23,550 --> 00:05:29,040 And these two together discussed the question of chemical interactions. 47 00:05:29,040 --> 00:05:35,520 And they gave a talk on the 11th of March 1864 to the Norwegian Academy of Science and Letters. 48 00:05:35,820 --> 00:05:42,300 This is the first page of the published paper that came from that talk written in Norwegian. 49 00:05:43,020 --> 00:05:53,790 They are the title, says Varga. On behalf of his friend or his colleague, Goldberg presents their joint work studies on affinity. 50 00:05:54,030 --> 00:06:03,090 And for those of you whose Norwegian is not as good as mine and mine is nonexistent, there's a translation of the first paragraph. 51 00:06:04,700 --> 00:06:07,800 And what they say is we're studying chemical forces. 52 00:06:07,820 --> 00:06:15,440 There are lots of theories about them, but we doubt if anybody will ever understand the laws whereby chemical forces interact. 53 00:06:16,010 --> 00:06:18,260 This is a little disingenuous, actually, 54 00:06:18,680 --> 00:06:27,590 because in this paper they go on to say exactly how chemical forces interact and propound a law which is now known as the law of mass action. 55 00:06:28,610 --> 00:06:31,819 They didn't actually use that term there so themselves. 56 00:06:31,820 --> 00:06:36,170 But I'll show you in a minute. Well, not surprisingly, nobody read this paper. 57 00:06:36,860 --> 00:06:43,850 It's in Norwegian. And at that time, maybe one and a half million people lived in Norway. 58 00:06:45,020 --> 00:06:54,470 So three years later, Golberg and Varda republished their work now increased in volume in this French paper called H2, 59 00:06:55,070 --> 00:06:59,060 Xfinity, XI MC, which was taken notice of. 60 00:07:00,110 --> 00:07:05,180 Even so, it took a little time for the new ideas to take time to emerge. 61 00:07:05,540 --> 00:07:15,200 And the idea of the law of Mass Action wasn't formally stated until about, what, 35 years later. 62 00:07:15,530 --> 00:07:21,650 This is from 1879. This is an extract from the Oxford English Dictionary and it defines mass action. 63 00:07:22,160 --> 00:07:24,709 You can see the law of mass action. You read it there. 64 00:07:24,710 --> 00:07:33,170 The principle and this is important, the rate of a reaction, that equilibrium is proportional to the concentrations or activities of the reactants. 65 00:07:33,860 --> 00:07:38,840 That is the law of mass action. It has nothing to do, incidentally, with Extinction Rebellion, 66 00:07:39,320 --> 00:07:46,220 which is a type of mass action of a different sort, as this latter definition shows you. 67 00:07:46,490 --> 00:07:52,100 But there it is quoted in 1879, the first time that the OED lists it. 68 00:07:52,490 --> 00:07:59,330 And you can see that they these authors of the Journal of Chemical Society refer to Goldberg and Vargas French paper. 69 00:08:01,410 --> 00:08:08,550 And in case you're wondering why it's called mass action and not concentration action, because it's concentration is not mass. 70 00:08:08,970 --> 00:08:17,910 It's because in those days, the term active mass which go back and forth they used in their French version, mass active meant smaller concentrations. 71 00:08:18,990 --> 00:08:25,160 And again, there you are in 1879, this times in the London, Edinburgh and Dublin Philosophical Magazine, 72 00:08:25,980 --> 00:08:30,210 another reference to go back and Varga in relation to active mass. 73 00:08:30,600 --> 00:08:36,209 So a hugely important and influential paper demonstrating the relationship at 74 00:08:36,210 --> 00:08:41,340 equilibrium between the concentrations of substances and the rates at which they react. 75 00:08:42,630 --> 00:08:48,960 Actually in passing. It doesn't matter if you've got as many intermediates in the reaction as you like. 76 00:08:49,410 --> 00:08:52,010 The rate at which the two, 77 00:08:52,020 --> 00:09:00,300 let's say two compounds are converted into one at the end depends on the concentrations of those reactions those two reactants. 78 00:09:00,750 --> 00:09:06,630 It doesn't matter how complicated the reaction is. So let's have a look and see what's actually going on here. 79 00:09:07,620 --> 00:09:12,870 If we imagine two substances, A and B, which I imaginatively call them. 80 00:09:13,900 --> 00:09:17,000 They react and they produce a compound. 81 00:09:17,020 --> 00:09:22,440 Abby. That is. And that's the law of mass action. 82 00:09:22,650 --> 00:09:26,130 The rate of reaction depends in this case on the chance of collision. 83 00:09:27,570 --> 00:09:31,950 What does the chance of collision depend on? Well, it depends on the amount of molecules there are. 84 00:09:31,950 --> 00:09:42,510 The more molecules, the more chances there are for collision. And so the probability of a collision in the first case on the left is one, one chance. 85 00:09:43,050 --> 00:09:46,890 In the middle case, it's three chances, and here it's four chances. 86 00:09:46,890 --> 00:09:50,580 So it's both substances that matter. 87 00:09:50,640 --> 00:09:57,450 Concentrations of both increase the risk of the chance of a collision and therefore the chance of a reaction. 88 00:09:58,380 --> 00:10:03,030 Now, if we look at what Goldberg and Vega published that say, yes, 89 00:10:03,030 --> 00:10:07,590 this is their French edition referring to, incidentally, these slides come from Robin. 90 00:10:09,360 --> 00:10:11,520 They are looking at reversible reactions. 91 00:10:13,170 --> 00:10:20,310 And what they say is that if A plus B gives a B, there will be a forward rate constant and a backward rate constant. 92 00:10:22,180 --> 00:10:26,440 Denoted by k forward rate constant kb u backward rate constant. 93 00:10:27,820 --> 00:10:32,530 And the velocity of the reaction can be described as the product of these things. 94 00:10:32,800 --> 00:10:40,959 So the velocity, the rate at which this reaction goes forwards is the product of the forward rate constant. 95 00:10:40,960 --> 00:10:46,600 And the two concentrations of the substances, the square brackets indicate concentrations. 96 00:10:48,730 --> 00:10:55,960 The backward velocity, similarly is the product of the backward rate constant and the concentration of the product. 97 00:10:56,590 --> 00:10:59,850 For those who are interested in these kinds of things, this is a second order. 98 00:10:59,860 --> 00:11:03,190 Reaction depends on two concentrations. 99 00:11:04,220 --> 00:11:13,340 And this is a first order reaction to the solution of the differential equations are different but at equilibrium these two. 100 00:11:15,050 --> 00:11:25,209 These two velocities are equal. So the rate at which the two compounds are combining is the same as the rate at the speed. 101 00:11:25,210 --> 00:11:31,900 I should say the rate. The speed is the same as the speed with which the single compound is dissociated into the other two. 102 00:11:32,080 --> 00:11:38,850 So there's complete equilibrium. And so these two equations, these two things become equal. 103 00:11:40,600 --> 00:11:50,790 We can equate those two things. And if we say that that equals that and then rearrange it, 104 00:11:51,240 --> 00:11:58,140 we get that the ratio of that to the product of the two compounds is the ratio of the two rate constants. 105 00:11:58,350 --> 00:12:02,220 And we can now replace that constant by a single constant, which I've called K. 106 00:12:04,400 --> 00:12:08,240 And this, in effect, is the law of mass action. 107 00:12:10,750 --> 00:12:14,650 The rate of the reaction is proportional to the concentrations. 108 00:12:16,620 --> 00:12:26,480 Now that's chemistry. In pharmacology, we deal with receptors, primarily other things, tube receptors and the receptacle. 109 00:12:26,510 --> 00:12:29,150 This is a paper by my colleague Humphry Rang, 110 00:12:29,510 --> 00:12:36,680 who started life in the Department of Pharmacology in Oxford with Bill Patten and then moved to become Professor of college at UCL. 111 00:12:38,000 --> 00:12:41,960 And in this paper, Humphry writes about the history of the receptor concept. 112 00:12:43,160 --> 00:12:51,320 And the idea is that you have a drug which reacts with proteins usually in the body, and the protein is that receptor for the drug. 113 00:12:51,950 --> 00:12:59,420 And that protein then causes so-called second messengers to act producing effects in tissues. 114 00:13:00,230 --> 00:13:04,430 So the interaction of a drug with its receptor is an important concept. 115 00:13:04,670 --> 00:13:11,110 And it followed Erlich's idea of a magic bullet by this idea from a man called Jan Langley. 116 00:13:12,330 --> 00:13:19,110 Who was a professor of physiology in Cambridge and who published this seminal paper in the Journal of Physiology in 1905. 117 00:13:19,500 --> 00:13:25,560 And he was studying the effects that nicotine and curare had on skeletal muscle, striated muscle. 118 00:13:26,410 --> 00:13:37,290 What he discovered was that there were substances in the muscle that recognised these drugs, nicotine and curare, and caused the muscle to contract. 119 00:13:38,130 --> 00:13:43,880 And so he proposed. That this was an accessory substance in the muscle. 120 00:13:44,240 --> 00:13:48,980 Recipient of stimuli transferred the contraction to the muscle. 121 00:13:49,340 --> 00:13:55,060 And he talked about it as a receptive substance. And you'll see he puts that in the running head. 122 00:13:55,540 --> 00:13:59,050 Although he didn't include the term receptive substance in the title of the paper. 123 00:13:59,950 --> 00:14:05,440 Only in the running head and in the body of the paper. This idea is now what we call receptors. 124 00:14:06,550 --> 00:14:10,510 The receptors are generally proteins, usually in the plasma membrane of a cell, 125 00:14:10,810 --> 00:14:16,720 sometimes intracellular on which drugs act to produce effects via second messengers. 126 00:14:17,050 --> 00:14:20,080 This is a hugely important principle in pharmacology. 127 00:14:21,280 --> 00:14:27,650 And the more drug you have. The faster the reaction occurs and the bigger the effect. 128 00:14:28,160 --> 00:14:30,260 And that is the principle of the dose response curve. 129 00:14:30,620 --> 00:14:37,579 So let's look at the mathematics of the receptors, which is very similar to what we've just been talking about here, 130 00:14:37,580 --> 00:14:42,229 the chemicals, A-plus, b, reacting to produce a B for a receptor. 131 00:14:42,230 --> 00:14:45,350 We just replace the letters with different letters. 132 00:14:45,650 --> 00:14:51,680 We have a receptor plus a drug gives a drug receptor complex, and that's a reversible reaction. 133 00:14:52,490 --> 00:14:59,340 The drug can dissociate from the receptor. If it didn't, the receptor would be destroyed and you'd have to start making new receptors. 134 00:14:59,360 --> 00:15:02,690 There are, of course, some reactions that are irreversible, but very few. 135 00:15:04,390 --> 00:15:14,200 So let's have some definitions. We call our D, which is on the right side there the number of receptors that the drug is occupying. 136 00:15:15,240 --> 00:15:18,600 Drug will not necessarily occupy all the receptors available. 137 00:15:19,720 --> 00:15:23,650 It may or may not, depending on how much drug there is and what kind of drug it is. 138 00:15:25,260 --> 00:15:29,309 We'll call our F the number of free receptors, the receptors that are not bound by the drug. 139 00:15:29,310 --> 00:15:32,760 And so the total number, our T is our D plus our L. 140 00:15:35,070 --> 00:15:43,229 Now we can write our equations. So we know that by the law of mass action, it's the ratio of the concentrations that are constant. 141 00:15:43,230 --> 00:15:46,920 We call that K before. If you remember, that's the law of mass action. 142 00:15:47,400 --> 00:15:51,510 If you multiply the concentration, if you like, of three receptors, 143 00:15:51,510 --> 00:15:55,680 but the concentration of the drug divided by the concentration of bound receptors 144 00:15:55,680 --> 00:16:01,080 that is constant in a reversible reaction at equilibrium by the law of mass action. 145 00:16:03,590 --> 00:16:06,980 So we rearrange it. We move the RF to the other side. 146 00:16:07,970 --> 00:16:12,260 And since we know that the total number of receptors is already plus RF, 147 00:16:12,260 --> 00:16:19,700 we can substitute for RF and we get this thing at the bottom and that it rearranged. 148 00:16:20,790 --> 00:16:30,869 Gives you this. Where the proportion occupied now is the number of receptors bound to the drug over the total number of receptors. 149 00:16:30,870 --> 00:16:35,519 And so what we get at the end is this proportion occupied is the concentration of the drug 150 00:16:35,520 --> 00:16:41,850 divided by the constant K plus the concentration of the drug relatively simple equation. 151 00:16:42,840 --> 00:16:49,560 What does this mean? What is this constant? Well, when the constant equals the concentration of the drug. 152 00:16:51,150 --> 00:16:55,410 That thing becomes on the left becomes that thing on the right. 153 00:16:57,600 --> 00:17:01,460 And that's just a half. So we've now defined our constant. 154 00:17:01,820 --> 00:17:07,910 The constant is the concentration of drug at which occupancy is half maximal. 155 00:17:09,410 --> 00:17:13,400 And that's commonly called the K-T or the dissociation constant. 156 00:17:13,730 --> 00:17:20,240 And the lower the dissociation constant, the more potent the drug is or the higher affinity it has for the receptor because you 157 00:17:20,240 --> 00:17:26,690 need less drug to saturate or to occupy half of the receptors in this case for CD. 158 00:17:27,740 --> 00:17:35,750 So the lower the CD, the more affinity, the more efficacious, if you like, technical term, the drug is at the receptor. 159 00:17:39,440 --> 00:17:48,710 And here I've put it in a different way. The amount of drug bound there is set to be over the amount that can maximally bend that the b max. 160 00:17:49,620 --> 00:17:55,050 Which is the proportion of receptors occupied as drug over CD plus drug concentration. 161 00:17:56,250 --> 00:17:59,250 And this is the equation of a rectangular hyperbola. 162 00:18:00,220 --> 00:18:02,620 Like that. That's what it looks like. 163 00:18:03,130 --> 00:18:10,720 So if you draw B over B max on the horizontal axis, on the vertical axis, the concentration, the numbers don't matter. 164 00:18:11,140 --> 00:18:15,520 On the horizontal axis, you get this rectangular hyperbola. 165 00:18:16,270 --> 00:18:20,260 And what this shows you is that when you don't have any drag, there's no binding. 166 00:18:21,010 --> 00:18:27,909 As you increase the amount of drag, there's more and more binding, but you can't occupy more receptors than there are to be occupied. 167 00:18:27,910 --> 00:18:32,500 And so you reach a maximum now without going into the details. 168 00:18:32,740 --> 00:18:35,860 Oh, yes. And when you draw, this is a logarithmic. 169 00:18:36,940 --> 00:18:41,680 Concentration curve, you get what's known as the sigmoid curve. It looks like the shape of a long X. 170 00:18:43,940 --> 00:18:48,680 And that is a typical curve for the binding of drugs to receptors. 171 00:18:49,610 --> 00:18:54,380 Notice that that's pretty linear in the middle section from about 20 to 80%. 172 00:18:54,860 --> 00:19:00,440 This is long linear and that's useful for modelling responses in that range. 173 00:19:02,690 --> 00:19:07,729 This is also true without going through all the intermediate steps for the effect 174 00:19:07,730 --> 00:19:14,300 that binding to the receptor produces and we can change B to E binding to effect, 175 00:19:15,080 --> 00:19:18,739 and we find that the ratio of E to Emacs, 176 00:19:18,740 --> 00:19:23,930 in other words, the proportional effect you can produce with a drug that's binding to receptor is 177 00:19:23,930 --> 00:19:29,120 exactly the same in the relationship to the amount of drug that you're using. 178 00:19:31,000 --> 00:19:34,320 And if we rearrange that, we put the emacs on top. 179 00:19:34,330 --> 00:19:39,310 We change constant now instead of 2kd, which is the 50% binding. 180 00:19:39,670 --> 00:19:46,030 We call it D 50, which is the concentration of the drug that produces 50% of the maximum possible effect. 181 00:19:46,390 --> 00:19:49,720 Then that is the dose response curve equation. 182 00:19:51,850 --> 00:20:00,190 And again, if we draw of a re max against log concentration, we get a sigmoid curve, typical dose response curve. 183 00:20:01,090 --> 00:20:06,730 And again. 20 to 80% of it is on the long linear part of the curve. 184 00:20:09,020 --> 00:20:14,930 So the law of mass action applies to receptor drug interactions, it predicts dose responsiveness, 185 00:20:15,440 --> 00:20:21,050 and it's relevant to both benefits and harms of medicines, adverse drug reactions and the like. 186 00:20:21,890 --> 00:20:30,379 The law of Mass Action has many, many more later with many more, many more applications. 187 00:20:30,380 --> 00:20:37,670 For example, enzyme kinetics have exactly the same equation in the same rectangular hyperbola transport, 188 00:20:37,670 --> 00:20:40,489 active transport of ions across cell membranes, 189 00:20:40,490 --> 00:20:50,570 exactly the same equation the control of acidity and alkalinity in cells, the Henderson Halbach equation exactly the same equation. 190 00:20:50,810 --> 00:20:58,070 Wherever you have things going on chemically in the body, the law of mass action applies and you get the same equation out at the end. 191 00:20:58,370 --> 00:21:04,849 It's universal. Now let's move on to some real dose response curves here. 192 00:21:04,850 --> 00:21:11,070 This paper by Steve Stephenson, published in what used to be called British Journal of Pharmacology and Chemotherapy. 193 00:21:11,090 --> 00:21:16,430 Now, just British Journal of Pharmacology in the 1950s shows the effects of a series 194 00:21:16,430 --> 00:21:22,070 of salts of tri missile ammonium on the contraction of guinea pig ileum. 195 00:21:22,490 --> 00:21:26,870 So you put up a guinea pig piece of guinea pig gut in a water bath and you 196 00:21:26,870 --> 00:21:30,740 add these things to it and you measure the contraction using a tension gauge. 197 00:21:32,160 --> 00:21:35,400 And the more you put in, the bigger the effect dose response curve. 198 00:21:36,000 --> 00:21:39,180 But what you can see here is there are three different classes of drug. 199 00:21:40,170 --> 00:21:44,640 There's these three beautiful style in detail. Notice they're all parallel. 200 00:21:46,250 --> 00:21:50,930 They have the same effect exactly on whatever it is that the receptor. 201 00:21:52,230 --> 00:21:56,010 But they're apart because some are more potent than others. 202 00:21:56,010 --> 00:21:59,850 So you need less of the beautiful to produce the same effect. 203 00:21:59,940 --> 00:22:05,370 So there's a difference between potency and efficacy, which is the effect you can produce. 204 00:22:06,660 --> 00:22:13,080 What you see is that these three produce the maximum effect. You can't get the gut to contract any more than that. 205 00:22:13,380 --> 00:22:17,010 They are what are known as full agonists. They produce the full effect. 206 00:22:17,550 --> 00:22:21,060 But these three peptide cocktail and. No, no, don't do that. 207 00:22:21,090 --> 00:22:25,050 You can occupy as many receptors as you like. You never produce the maximum effect. 208 00:22:25,080 --> 00:22:30,150 These are known as partial agonists. And that's an important principle in pharmacotherapy. 209 00:22:30,600 --> 00:22:34,510 Choosing drugs that do or do not have partial agonism effect. 210 00:22:34,580 --> 00:22:37,660 I'm not going to talk about that. This is an odd one. 211 00:22:37,680 --> 00:22:41,670 The D'asile, it looks as if it's contracting and then relaxing. 212 00:22:42,210 --> 00:22:48,840 And that's a phenomenon known as harms, which is rare, but has been described and does sometimes occur. 213 00:22:49,800 --> 00:22:52,890 But mostly we have full agonists or partial agonists. 214 00:22:53,490 --> 00:23:02,010 Now, this is fine in vitro. We can do this. We can string up some gut and drop some drugs on them, and we know the concentrations we're using. 215 00:23:02,010 --> 00:23:05,610 It's relatively easy, although very elegantly done. 216 00:23:06,420 --> 00:23:10,980 But you can and you can do this with other systems in vivo, in vitro. 217 00:23:11,820 --> 00:23:19,170 Here's an experiment with adipocytes fat cells, and you incubate them with glucose and you add insulin. 218 00:23:19,590 --> 00:23:23,100 And the more insulin you put in, the more glucose is taken up. 219 00:23:24,540 --> 00:23:28,709 And this is linear scale, and here's the low dose response. Q So that's fine. 220 00:23:28,710 --> 00:23:32,610 Another in vitro experiments this time using human cells, not guinea pig, 221 00:23:33,750 --> 00:23:41,250 but getting in vivo examples is difficult because we can rarely measure the actual 222 00:23:41,250 --> 00:23:46,139 concentration of drug at the site of action for giving a drug that affects the heart. 223 00:23:46,140 --> 00:23:49,260 For example, how do we know what the concentration is in the heart? 224 00:23:51,140 --> 00:23:55,560 Take a biopsy may be difficult. Occasionally you can do it. 225 00:23:55,580 --> 00:24:00,890 And here's an example. These two drugs be met tonight in furosemide are loop diuretics. 226 00:24:01,760 --> 00:24:07,100 You take them totally. They circulate in the blood and they are excreted via the kidneys. 227 00:24:07,970 --> 00:24:15,770 They don't work as diuretics, increasing urine flow until they get into the lumen of the of the kidney tubule. 228 00:24:15,800 --> 00:24:22,660 In other words, they have to be in the urine before they increase the urinary flow, which is a bit of a [INAUDIBLE], really. 229 00:24:22,660 --> 00:24:27,110 You'd want them to do it before they get out. So if there's no urinary flow, they don't work. 230 00:24:28,450 --> 00:24:31,929 Can't kick urinary flow with these drugs. They don't work. 231 00:24:31,930 --> 00:24:40,030 They have to be in the urine before they'll increase the urine. It's a bit of a catch 22, but once they get in, you can measure the concentration. 232 00:24:40,180 --> 00:24:45,730 This case rate of excretion and that's the concentration at the site of action because you're measuring them in the urine. 233 00:24:46,330 --> 00:24:51,190 So here we have dose response codes for the concentration at the site of action versus the rate of it, 234 00:24:51,520 --> 00:24:54,940 in this case, sodium excretion, beautiful dose response curves. 235 00:24:54,940 --> 00:24:58,000 These are American data, really elegant experiments. 236 00:24:58,210 --> 00:25:04,270 So you can sometimes show in vivo that there are dose response scores of the kind that you expect. 237 00:25:04,480 --> 00:25:12,250 And notice I didn't point it out in the previous slide, but here as well, typically dose response curves occur over about two orders of magnitude. 238 00:25:12,970 --> 00:25:23,260 That is very typical. Some are have a different slope, but typically a 100 fold difference is the range over which you expect changes to occur. 239 00:25:24,730 --> 00:25:27,880 Now here's an indirect method of doing it. 240 00:25:28,630 --> 00:25:34,960 Instead of measuring the concentration actually at the site of action, you measure the plasma concentration as a surrogate for that, 241 00:25:35,380 --> 00:25:41,350 and you assume that the plasma concentration somehow is proportional to the site of action concentration. 242 00:25:41,770 --> 00:25:49,629 And this is an experiment that was done some years ago now, a long time ago in 75 for everybody was born, I guess. 243 00:25:49,630 --> 00:25:57,430 Is that right? No propranolol. The beta blocker, which you can give orderly measure the plasma concentration, 244 00:25:57,430 --> 00:26:05,440 and then you exercise the individual and see how much the tachycardia, the increase in heart rate is prevented by the beta blocker. 245 00:26:05,690 --> 00:26:13,630 You can see there's a very good relationship. These are five different patients, each studied three times three symbols for each one. 246 00:26:13,840 --> 00:26:21,190 And you can see there's a very good linear relationship on the log scale with the inhibition of exercise induced tachycardia. 247 00:26:21,430 --> 00:26:24,309 So this is the 20 to 80% linear log, 248 00:26:24,310 --> 00:26:31,300 linear bit of the dose response curve using plasma concentration as a surrogate for the concentration at the site of action. 249 00:26:31,810 --> 00:26:38,080 We call this typically the dose response curve. It's not it's a concentration response curve. 250 00:26:39,190 --> 00:26:46,390 But we call it the dose response curve because we assume that the dose you give is related to the concentration results. 251 00:26:46,630 --> 00:26:52,660 The bigger the dose, the bigger the concentration. Now, these are beneficial effects I've just shown you. 252 00:26:52,870 --> 00:26:56,950 Diuresis slowing the heart rate when you've got a tachycardia. 253 00:26:57,310 --> 00:27:04,000 What about some adverse effects? Well, this is an early classification of adverse drug reactions. 254 00:27:04,810 --> 00:27:14,440 Eddie Wain, who was originally Professor of therapeutics in Sheffield and when I was a medical student, was the Professor of Medicine in Glasgow. 255 00:27:16,620 --> 00:27:21,120 Classified adverse drug reactions as being either predictable and unpredictable. 256 00:27:23,080 --> 00:27:33,040 That's a funny classification, but it lasted until Ruth Levine, an American pharmacologist, talked about dose related and non dose related reactions. 257 00:27:33,550 --> 00:27:40,180 Then on Wade and Linda, Billy Own Wade was professor of clinical pharmacology in Birmingham, 258 00:27:40,840 --> 00:27:44,650 also talked about dose related and non dose related adverse reactions. 259 00:27:46,330 --> 00:27:49,780 And this idea was reinforced by Mike Rollins, 260 00:27:49,780 --> 00:27:59,590 who at that time was professor of clinical pharmacology in Newcastle and later became well known as chairman of Nice and other big committees. 261 00:27:59,980 --> 00:28:06,280 He and his colleague Thomson invented this imaginative classification and B. 262 00:28:07,880 --> 00:28:11,720 Lots of imaginative classifications like that. Type one and type two. 263 00:28:13,450 --> 00:28:16,570 Or. And B or ABCD sometimes. 264 00:28:17,770 --> 00:28:22,059 And they said there are type A reactions and type B they meant dose related and dose related. 265 00:28:22,060 --> 00:28:26,980 But then they realised nobody could remember which was which. This is a bit of a bind. 266 00:28:26,980 --> 00:28:31,400 So they invented a mnemonic. They called them augmented and bizarre. 267 00:28:33,040 --> 00:28:37,150 Augmented means increased, but the dose bizarre means no relation to those. 268 00:28:38,110 --> 00:28:43,899 And they later on, some years later, in an article in the BMJ, they talked about dose related, 269 00:28:43,900 --> 00:28:49,360 non dose related, exaggerated or a bad, predictable or unpredictable, preventable and so on. 270 00:28:51,250 --> 00:28:59,740 This to me is odd. And this this idea lasted for about 35 years and nobody challenged it. 271 00:29:01,580 --> 00:29:04,940 But it is very odd. Why should anything be not dose related? 272 00:29:05,510 --> 00:29:09,690 Everything is surely related to dose. Let's just see. 273 00:29:10,850 --> 00:29:16,070 Test this question about known dose related adverse reactions. 274 00:29:16,580 --> 00:29:20,550 Here's a statement. That you might or might not agree with. 275 00:29:21,420 --> 00:29:25,740 Toxicologists need many molecules for. They're going to study toxicology. 276 00:29:26,400 --> 00:29:29,670 Pharmacologists study the right amount of molecules. 277 00:29:30,990 --> 00:29:39,389 Immunologists need only one molecule. And well, you know that I'm going to test these assumptions. 278 00:29:39,390 --> 00:29:45,600 At least I'm going to test the immunological assumption. Does anybody here suffer from hay fever? 279 00:29:47,310 --> 00:29:51,690 Yeah, a few. Not as many as I'd expect. Some of you hiding it, maybe, anyway. 280 00:29:52,530 --> 00:29:56,320 You all know about hay fever, too? 281 00:29:56,720 --> 00:30:06,820 Yeah, I have it. Itchy eyes, runny nose, sneezing, bronchospasm, wheeze. 282 00:30:07,590 --> 00:30:13,730 It's not pleasant. Now I'm going to make some statements about hay fever, and I want you to raise your hands. 283 00:30:13,740 --> 00:30:17,250 Don't be shy if you think the statement is true. 284 00:30:18,430 --> 00:30:23,530 Right. Is this statement true? So here's the first statement coming up. 285 00:30:25,380 --> 00:30:31,380 The risk of an attack of hay fever is not related unrelated to the pollen count. 286 00:30:33,780 --> 00:30:38,760 No takers. Not just shy. Nobody believes that statement. 287 00:30:38,780 --> 00:30:46,430 Okay, I'll try a second statement. The risk of an attack of hay fever increases as the pollen count falls. 288 00:30:48,860 --> 00:30:54,650 Still no takers. It's very disappointing, really hoping somebody would raise their hand. 289 00:30:55,070 --> 00:30:58,730 The risk of an attack of hay fever increases, the pollen count rises. 290 00:30:59,420 --> 00:31:04,940 Oh, right. Everybody seems to think that that might be true. 291 00:31:06,380 --> 00:31:14,000 Well, it is true. Of course it's true. The higher the pollen count, the greater the risk of an attack of hay fever. 292 00:31:15,120 --> 00:31:18,900 And you hear on the weather forecast, the pollen count is rising. 293 00:31:18,940 --> 00:31:25,740 It's a warning to all you guys that, you know, take your steroid or your optical or whatever you use. 294 00:31:28,290 --> 00:31:35,440 And here's the evidence. This paper by friends in the Annals of Allergy and Asthma and immunology. 295 00:31:37,480 --> 00:31:42,610 Shows a dose response curve. It's the wrong way round, if you like. 296 00:31:42,630 --> 00:31:50,720 Here's the log pollen count on the vertical axis and the symptom score, but you can see a clear dose response curve. 297 00:31:50,730 --> 00:31:56,910 The higher the pollen count, the bigger the risk of an attack of hay fever in the population is. 298 00:31:56,910 --> 00:32:01,530 Notice that it goes over 1 to 3 or four orders of magnitude. 299 00:32:01,530 --> 00:32:06,720 It's quite a shallow dose response curve, but then pollen counts vary enormously. 300 00:32:08,640 --> 00:32:11,880 And they even say this in their text. 301 00:32:12,930 --> 00:32:17,520 Quantitative Dose Response Models, Nonlinear relationships. 302 00:32:17,550 --> 00:32:26,640 These are classical dose response curves. So the immunological idea, the idea that people have about immunology, that it's not dose related is wrong. 303 00:32:28,250 --> 00:32:37,810 Everything is dose, really. Everything is a poison. So let's examine this in relation to people's ideas about non dose related adverse reactions. 304 00:32:38,830 --> 00:32:44,200 I'm going to show you some theoretical dose response curves just for illustrative purposes. 305 00:32:44,230 --> 00:32:50,230 Here's the dose response curve. Beneficial dose response curve, whatever it is, doesn't matter. 306 00:32:50,620 --> 00:32:55,330 Plotting the response against the low dose concentration. And it has a sigmoid shape. 307 00:32:55,720 --> 00:33:02,590 No, no drug. You get no effect. Linear effect between 20 and 80% and the maximum possible effect achieved. 308 00:33:03,070 --> 00:33:08,980 You wouldn't want to bother using a dose in this range because you don't get any extra benefit. 309 00:33:09,790 --> 00:33:19,640 You'd probably choose a dose in this sort of range here. For maximum benefit that you can achieve, but you have to be worried about adverse reactions. 310 00:33:21,270 --> 00:33:27,360 Now the idea that adverse reactions are of two types A and B is naive and simplistic. 311 00:33:27,810 --> 00:33:29,790 Robin and I suggested three types. 312 00:33:30,920 --> 00:33:37,610 Depending on the dose response relationship, which you can actually measure something you can actually measure and quantify. 313 00:33:38,730 --> 00:33:45,670 And the first time we described what we call hyper susceptibility reactions, people are more susceptible than others are. 314 00:33:45,930 --> 00:33:51,000 These are typically hypersensitivity reactions, allergic reactions, but they don't have to be. 315 00:33:51,270 --> 00:33:55,889 There are other types, so we call them hyper susceptibility rather than hypersensitivity. 316 00:33:55,890 --> 00:34:02,250 And the whole mark of these is that they occur at doses of the drug, which are lower than those responsible for benefit. 317 00:34:02,760 --> 00:34:07,339 Penicillin. Allergy. The problem is, of course, 318 00:34:07,340 --> 00:34:14,600 you give somebody a dose that you think is beneficial and they're at the top of the dose response curve for the adverse reaction, 319 00:34:14,600 --> 00:34:21,110 for the allergic reaction. That's why it appears to be non dose related because it occurs at only one dose. 320 00:34:21,350 --> 00:34:27,830 And if you increase the dose here, you don't get any different effect here because you're already at the top of the dose response curve. 321 00:34:29,640 --> 00:34:34,380 So that's hyper susceptibility. The second type are what we call collateral. 322 00:34:35,220 --> 00:34:40,410 They occur right beside in the same dose range as the beneficial responses. 323 00:34:40,410 --> 00:34:45,300 And these are difficult to handle because means you can't you can't avoid them. 324 00:34:45,450 --> 00:34:49,469 These you can avoid by not giving the drug a pity. 325 00:34:49,470 --> 00:34:51,240 But that's what you have to do. 326 00:34:51,680 --> 00:34:59,700 These if you want to use the drug, then you have to wear the fact that there's going to be likely going to be an adverse reaction. 327 00:35:00,180 --> 00:35:04,590 The things you can do about that, and I'm not going to discuss how you handle that problem, 328 00:35:05,070 --> 00:35:09,660 but collateral reactions do occur and can be difficult to treat. 329 00:35:09,660 --> 00:35:11,440 And then, of course, there are toxic reactions. 330 00:35:11,490 --> 00:35:18,300 These are reactions that occur at doses that are higher than beneficial and should be entirely avoidable by not using high doses. 331 00:35:19,380 --> 00:35:22,680 And they can occur either on a separate mechanism. 332 00:35:23,900 --> 00:35:30,050 Or via the same mechanism as benefit, but giving too much, too much of a good thing. 333 00:35:30,560 --> 00:35:36,620 If you give warfarin to prevent the blood from clotting and you give too much, you'll get a risk of bleeding. 334 00:35:37,990 --> 00:35:42,520 So that's a harmful reaction at the top of the same dose response curve as the benefit. 335 00:35:44,330 --> 00:35:48,830 But. Well, what else? That might be, for example, 336 00:35:48,830 --> 00:35:58,220 that the toxic effects of penicillin on the central nervous system are mediated by an action that has nothing to do with its antibiotic effect. 337 00:35:58,640 --> 00:36:03,620 There's something there, but you only get it at very high doses. So some examples. 338 00:36:03,620 --> 00:36:07,790 Well, here's a real example. This is a curious syndrome called the lupus syndrome. 339 00:36:07,790 --> 00:36:09,000 A lupus like syndrome. 340 00:36:09,440 --> 00:36:17,030 It's similar to a disease called lupus erythematosus, but it's induced by drugs such as this one, hydrazine we don't use anymore. 341 00:36:17,420 --> 00:36:23,890 And if you look at the data, you see that it's dose related. So it's a limited allergic reaction, but it's dose related. 342 00:36:23,910 --> 00:36:28,820 You increase the dose. You increase the risk. And there's the sex difference as well. 343 00:36:28,830 --> 00:36:38,220 Women are more susceptible than men. This is another allergic reaction, which, as it happens, Rabin's wife studied when she was a Ph.D. student. 344 00:36:39,170 --> 00:36:45,590 And what she did was to challenge the skin with a substance called di nitroglycerine, benzine. 345 00:36:45,590 --> 00:36:51,530 When you inject not clear benzene, then you get an allergic reaction, which you can measure by the thickness of the skin. 346 00:36:52,400 --> 00:37:00,470 Then you can see that when you challenge the skin, you get increases into each one of these as it is a separate patient, and that's the control. 347 00:37:01,190 --> 00:37:06,650 So you challenge the skin with this stuff, then you get a dose related increase in thickness of the skin. 348 00:37:08,590 --> 00:37:15,180 And the response also depends on the sensitising dose that you used in the first place. 349 00:37:15,190 --> 00:37:25,239 So it's two types of dose response reactions. But despite this, people keep on talking about non dose related side effects and why? 350 00:37:25,240 --> 00:37:29,590 Why do they do that? What makes them think that things are not dose related? 351 00:37:29,890 --> 00:37:38,700 Well, the first problem is they have this idea is that allergic reactions are not dose related, which is wrong, but they think it is the case. 352 00:37:38,710 --> 00:37:42,100 And so they call anything that's an allergic reaction, non dose related. 353 00:37:43,090 --> 00:37:48,309 It's a misnomer. But this side effect of valproate paper isn't in that region. 354 00:37:48,310 --> 00:37:51,940 It's these are probably collateral or toxic reactions. 355 00:37:52,120 --> 00:37:57,069 So why do they call them non dose related? Well, it's because they see them at only one dose sometimes. 356 00:37:57,070 --> 00:38:00,930 Or there may be other reasons of maybe they've only used one dose. 357 00:38:00,940 --> 00:38:03,630 I mean, let's look at some of the possible reasons. 358 00:38:03,640 --> 00:38:09,700 Well, the first reason is that they've seen an event they've associated with the drug, and there's no association at all. 359 00:38:10,390 --> 00:38:15,970 It's just not so. So they've made a mistake in calling it an adverse effect in the first place. 360 00:38:16,420 --> 00:38:22,690 And that could be just because it's a chance finding or there may be methodological problems, confounding factors, biases. 361 00:38:22,690 --> 00:38:30,310 And so here's an example. This is a drug called into Catterall, which dilate shore bronchioles and asthma. 362 00:38:31,000 --> 00:38:34,780 And what these people showed was that there was. 363 00:38:37,850 --> 00:38:41,360 Headache occurring in these patients would be given in the Catterall. 364 00:38:41,810 --> 00:38:47,810 They thought it might be due to vessel Dilatation maybe, but it was also present in the placebo group. 365 00:38:47,810 --> 00:38:56,420 And when you look at another study, you can see that the risk of headache is actually pretty much the same across all doses and all non doses as well. 366 00:38:56,510 --> 00:39:04,730 This is just an it's a non-event. People get headaches and if you're taking into control, you may or may not get a headache. 367 00:39:05,630 --> 00:39:11,630 The fact that it's not dose related makes you assume it's a non dose related effect, but it's not an effect at all. 368 00:39:12,320 --> 00:39:18,649 It never was. So that's one reason why people get it wrong is another reason. 369 00:39:18,650 --> 00:39:23,390 This is a drug called disulphiram. And liver damage was reported. 370 00:39:24,530 --> 00:39:28,040 What's Disulphiram use to treat anybody? 371 00:39:28,040 --> 00:39:36,900 No alcoholism. And it's not surprising that one sees liver damage in people who are taking excess alcohol. 372 00:39:38,280 --> 00:39:44,580 This is confounding by indication. It's not that the drug is causing liver damage. 373 00:39:45,060 --> 00:39:50,220 It's that the drug is associated with another factor that causes liver damage. 374 00:39:50,310 --> 00:39:54,060 Nothing to do with the drug at all. So it's confounding by indication. 375 00:39:55,380 --> 00:40:01,560 Now, here's another reason. The effect has its dose response curve at lower doses than you studied. 376 00:40:01,590 --> 00:40:05,400 This is what I was talking about before. Hyper susceptibility reactions. 377 00:40:05,730 --> 00:40:08,760 So here we're back to our hyper susceptibility curves. 378 00:40:08,790 --> 00:40:13,230 Here's the benefit curve. Here's the harm curve, the susceptibility reaction. 379 00:40:13,590 --> 00:40:16,980 And of course, you give the drug in a beneficial dose. 380 00:40:17,430 --> 00:40:22,800 And so the adverse effect appears to be non dose related because you're at the top of the dose response curve. 381 00:40:25,040 --> 00:40:32,059 There is a dose response curve can easily be shown by giving a test dose, a very low dose of the drug. 382 00:40:32,060 --> 00:40:35,750 And we do this for some drugs has been done for penicillin, for example. 383 00:40:36,020 --> 00:40:44,000 It's not a very good test. We still do it for amphotericin before we give it as an antifungal drug, give a small dose. 384 00:40:44,450 --> 00:40:53,330 If you get an adverse reaction to the test dose, a small reaction that predicts a large reaction, that's a beneficial dose since you don't use it. 385 00:40:54,270 --> 00:40:57,839 If there wasn't a dose response curve, test doses would be totally useless. 386 00:40:57,840 --> 00:41:02,550 You couldn't use them because they'd produce the same effect if there was no dose relationship. 387 00:41:03,360 --> 00:41:10,230 But they do produce different effects. The test dose produces a small effect and if you gave the beneficial, do she get a big effect? 388 00:41:11,100 --> 00:41:19,739 So the test dose predicts the risk. This also reminds us that you can desensitise people with hypersensitivity 389 00:41:19,740 --> 00:41:23,910 reactions by giving them small doses of the drug to which they're sensitive, 390 00:41:24,420 --> 00:41:27,810 and you can then build up the dose gradually and desensitise them. 391 00:41:28,170 --> 00:41:32,640 Couldn't do that if there wasn't a dose response curve. So that's the second reason. 392 00:41:33,840 --> 00:41:38,970 Here's another case of a so-called non dose dependent adverse drug reaction. 393 00:41:39,000 --> 00:41:45,150 It was a very widespread error. Throw down to red skin with flaking of the skins. 394 00:41:45,510 --> 00:41:51,210 Quite unpleasant. You can lose a lot of water through the skin surface in this way, become dehydrated. 395 00:41:51,720 --> 00:41:55,050 These types of rashes can be fatal in some cases. 396 00:41:56,180 --> 00:42:03,360 And this these authors say it's an unusual non dose dependent, 397 00:42:03,360 --> 00:42:11,610 but they say the literature says that adverse reactions are dose dependent with mild reactions at low doses and severe eruptions at high doses, 398 00:42:11,610 --> 00:42:16,130 they set themselves. And they say, we gave this man 800 and he got a rash. 399 00:42:16,140 --> 00:42:21,150 We gave him 600 and he got a rash. So we think it's non dose dependent on this guy. 400 00:42:21,150 --> 00:42:24,720 The dose response curve was up to 600, clearly. 401 00:42:25,200 --> 00:42:30,000 And in fact, there is other evidence here that these reactions are dose dependent. 402 00:42:31,230 --> 00:42:34,500 So there is this misunderstanding. The third. 403 00:42:37,340 --> 00:42:45,559 Reason that we've given here is that there is huge variability in the population studied so that you don't see the dose relationship. 404 00:42:45,560 --> 00:42:49,190 It's hidden in the variability of the results. 405 00:42:49,220 --> 00:42:58,879 Here's a good example. This is a drug called Ember sent, an Amber Sent and is an endothelial receptor antagonist that's used to treat lung, 406 00:42:58,880 --> 00:43:03,260 arterial hypertension, a very unpleasant disease, difficult to treat. 407 00:43:03,920 --> 00:43:09,200 And in this study, there were a whole range of adverse events reported. 408 00:43:09,530 --> 00:43:13,350 I've highlighted two of them nasal congestion and abdominal pain. 409 00:43:13,370 --> 00:43:20,120 Here are the data to contrast. So here's the placebo data in increasing doses of the drug. 410 00:43:21,650 --> 00:43:25,970 These are the numbers of people here, the numbers affected and the percentage affected. 411 00:43:26,180 --> 00:43:30,290 So for nasal congestion, the percentages increase with dose. 412 00:43:31,340 --> 00:43:39,230 Very nice dose response curve. This is probably very likely an adverse effect of the drug, which is dose related. 413 00:43:40,310 --> 00:43:46,350 Here's the abdominal pain. You can see there looks as if there's an effect, but it's very flat. 414 00:43:46,370 --> 00:43:49,930 There's no dose relationship. What's the explanation of this? 415 00:43:49,940 --> 00:43:53,300 Is this a non dose related reaction? Well, I don't think so. 416 00:43:54,140 --> 00:43:57,960 It could be that this is the top of the dose response curve for this adverse reaction. 417 00:43:57,980 --> 00:44:06,140 That's possible. But I think there's a better explanation than that here of the data that I've just shown you for headache now plotted on a graph. 418 00:44:06,740 --> 00:44:11,930 And what I've done now is to add the confidence intervals to those estimates. 419 00:44:13,220 --> 00:44:17,570 Then you can see that you can't tell anything from this information you really don't know. 420 00:44:17,900 --> 00:44:22,760 And if you fit a dose response curve to it, it could well be just like that. 421 00:44:24,740 --> 00:44:33,620 So the study was not big enough to rule out what is probably the true dose relationship between this drug and the headache, 422 00:44:34,010 --> 00:44:41,210 assuming that it's an effect at all, which is suggested by the difference between the placebo and the treated, 423 00:44:41,450 --> 00:44:47,719 although there is quite a bit of variance there as well, which might mean that it's not a true effect. 424 00:44:47,720 --> 00:44:56,120 We can't really tell. Point is that the study wasn't big enough and that's a major problem for pre licensing drug studies. 425 00:44:56,420 --> 00:44:59,450 They're generally not big enough to detect adverse reactions. 426 00:44:59,750 --> 00:45:04,920 They're good at detecting benefits. Because you don't need such big studies. 427 00:45:04,920 --> 00:45:08,190 Most people will respond to the beneficial effect. 428 00:45:08,190 --> 00:45:15,419 That's what you're looking for. But when it comes to harms, where maybe only a few percent are affected, you need much, 429 00:45:15,420 --> 00:45:22,650 much bigger studies and nobody does them pre licensing because they're not really interested in finding out about drugs as adverse effects. 430 00:45:23,130 --> 00:45:25,500 They want to market them for their beneficial effects. 431 00:45:27,460 --> 00:45:35,920 So now a main problem with all of this is that it's very difficult to find data on dose response curves. 432 00:45:37,210 --> 00:45:43,540 Nobody publishes dose response curves that study have just shown you with three doses is very unusual. 433 00:45:44,560 --> 00:45:47,630 Most studies involve one dose or maybe two at most. 434 00:45:48,160 --> 00:45:52,000 And even though they do, they won't you know, they won't publish dose response curve, 435 00:45:52,000 --> 00:45:56,830 certainly not with one or two points only, not even often with three. 436 00:45:57,730 --> 00:46:01,930 And I've never seen a drug study with more than three arms to it. 437 00:46:01,960 --> 00:46:14,500 Maybe there are, but it's very uncommon. So we thought that it would be useful to try and use modern meta analytical techniques in a systematic 438 00:46:14,500 --> 00:46:21,970 review to generate dose response curves by looking for data in the literature from published studies. 439 00:46:23,200 --> 00:46:28,940 A dose here, a dose there, and put them all together in a single model and generate a dose response curve. 440 00:46:28,960 --> 00:46:34,860 So that's what we've done in this paper. And Richards here, he's the chief mover in this, 441 00:46:34,870 --> 00:46:43,480 although Langford was a mathematician who actually did the work and get advised, I think, on the interpretation, very helpful. 442 00:46:43,780 --> 00:46:46,900 And I stood back and let it all happen. Great. 443 00:46:48,220 --> 00:46:57,310 And here's the data we analysed. We looked for trials of a drug called Alogliptin, which is an antidiabetic drug. 444 00:46:57,310 --> 00:47:01,480 It lowers the blood sugar and a measure of the effect of lowering blood sugar. 445 00:47:01,750 --> 00:47:07,420 Is this thing here? HB one C, which is a measure of the efficacy of treatment in diabetes, 446 00:47:07,930 --> 00:47:12,640 gives you long term measure of overall exposure to glucose over weeks or months. 447 00:47:13,510 --> 00:47:23,200 We found 14 studies. And here are all the data from the different studies showing the relationship between dose and change in HB one C. 448 00:47:23,830 --> 00:47:29,390 And you can see it's a hyperbolic curve probably. This is a linear. 449 00:47:31,180 --> 00:47:38,320 Scale. And here's the effect. And it's upside down because we're reducing something rather than increasing it. 450 00:47:38,740 --> 00:47:45,200 But it still looks as if it has the same shape. And so we used a pharmacological model. 451 00:47:45,740 --> 00:47:49,390 You recognise this bit of it I showed you before. 452 00:47:49,400 --> 00:47:53,090 The effect equals the imac's times dose over the 50, 453 00:47:53,090 --> 00:48:01,400 the constant plus the dose so that we added a baseline effect because there's always some HP H1 C in the system. 454 00:48:01,730 --> 00:48:05,330 Most systems you look at have nothing to start with and then it changes. 455 00:48:05,870 --> 00:48:10,550 For HP one C, there's always something there. So you put in a baseline and there's an error term. 456 00:48:10,790 --> 00:48:17,839 And so the response was modelled using this equation and a lot of fancy computing and this is what we came up with. 457 00:48:17,840 --> 00:48:21,950 We used four different methods and pretty much the same really. 458 00:48:21,950 --> 00:48:25,070 There's the, there's the fit. And you can see we've got. 459 00:48:26,070 --> 00:48:30,840 Confidence intervals on the likelihood that that is the true mean. 460 00:48:31,740 --> 00:48:36,670 And so we were able to. Generate dose response codes. 461 00:48:36,670 --> 00:48:43,300 These two here show you the rarity of high dose experiments and the huge variance on those. 462 00:48:43,960 --> 00:48:48,070 We can ignore that this is the main main bit of the business. 463 00:48:49,180 --> 00:48:55,629 So we thought that was a useful technique. And now I'm going to finish off by showing you some work that we've been doing recently. 464 00:48:55,630 --> 00:48:59,740 It's very early stages, but it's quite interesting already. 465 00:48:59,740 --> 00:49:05,830 And this was a young student who came to work with us, Evie Monahan, who turned out to be an absolute whiz. 466 00:49:06,700 --> 00:49:14,740 She bashed her way through a whole load of data in double quick time, demonstrating it in ways that I couldn't begin to understand. 467 00:49:15,340 --> 00:49:19,809 And I'll show you one of her results. This is a drug called Atorvastatin. 468 00:49:19,810 --> 00:49:28,240 It's a statin that's used to lower blood cholesterol, a particular form of cholesterol called LDL, low density lipoprotein cholesterol. 469 00:49:28,900 --> 00:49:32,560 And it's the statins are widely used. Everybody knows about the statins. 470 00:49:33,070 --> 00:49:38,379 They're used to prevent heart attacks and strokes by lowering cholesterol and they have adverse effects. 471 00:49:38,380 --> 00:49:43,840 They can cause muscle pain and sometimes diabetes if you use too much. 472 00:49:44,530 --> 00:49:47,410 So the question of what the correct dose should be is quite important. 473 00:49:48,250 --> 00:49:53,770 Can you find a dose that satisfactorily lowers the cholesterol and avoids the adverse effects? 474 00:49:54,730 --> 00:50:01,180 So what we did was she went through clinicaltrials.gov looking for trials containing atorvastatin. 475 00:50:01,180 --> 00:50:06,759 She found 600 or so, she screened them, excluded 526. 476 00:50:06,760 --> 00:50:10,180 I think I'll show you the exclusion criteria, but you're not going to read them. 477 00:50:10,690 --> 00:50:15,160 She came up with 73 that looked as if they might be useful and 55 of them were. 478 00:50:16,580 --> 00:50:22,060 The other statines we've studied weren't quite as good as that. This is probably the best I think of the bunch. 479 00:50:22,070 --> 00:50:26,480 That's why I'm showing it. Of course, you always show the best data and call it typical. 480 00:50:28,340 --> 00:50:36,470 Those are the exclusion criteria. So you see why she excluded 500 odd various reasons for not including the data. 481 00:50:36,770 --> 00:50:45,310 But here's her dose response curve. Using the program, the kind of approach that we used for the other Lipton work. 482 00:50:45,520 --> 00:50:48,999 And it's pretty good actually. I was really surprised. 483 00:50:49,000 --> 00:50:56,410 We got such good dose response data and the variances are quite small. 484 00:50:56,860 --> 00:51:04,210 But you'll notice two things about this. First, that the maximum effect occurs at about 50 milligrams. 485 00:51:06,120 --> 00:51:11,069 I shouldn't have to use more than 50 on average and the variance is a very tight. 486 00:51:11,070 --> 00:51:15,360 So it's unlikely that anybody in the population would need more than that. 487 00:51:16,410 --> 00:51:27,750 But the second thing you notice is that nobody, it appears, has studied anything, even approaching the edifice that all using high doses. 488 00:51:28,920 --> 00:51:36,450 And the variance at five mg shows you that there are relatively few studies, even as low as five and none below that. 489 00:51:36,780 --> 00:51:40,840 There's the ED 50, which is around four, by my calculation. 490 00:51:40,860 --> 00:51:45,990 I think it's maybe slightly different. I'm using 50 as the maximum, which may not be the case. 491 00:51:47,540 --> 00:51:50,989 But here's the manufacturer's dosage recommendations. 492 00:51:50,990 --> 00:51:56,870 The bottom line usual starting dose is ten and the maximum is 80. 493 00:51:57,860 --> 00:52:02,360 So are we using doses that are too high? Should we be using lower doses? 494 00:52:03,140 --> 00:52:09,230 Is that why people are now beginning to wonder about diabetes as an adverse effect of the statins? 495 00:52:10,130 --> 00:52:12,709 And could we have less in the way of muscle pain? 496 00:52:12,710 --> 00:52:19,220 If we used lower doses, we might get smaller beneficial effects, of course, and one would have to balance that up. 497 00:52:20,180 --> 00:52:24,080 But the lower your cholesterol is, the better, whatever it is. 498 00:52:25,070 --> 00:52:28,640 And so perhaps we should be thinking again about the doses. 499 00:52:29,270 --> 00:52:37,370 Now, before we did this study, I did a preliminary analysis and I got 8050 values from the literature. 500 00:52:37,400 --> 00:52:41,120 I'm not sure how well defined they have been in the past, 501 00:52:41,120 --> 00:52:51,170 but I plotted them against this measure of how effective the drug is in inhibiting its target enzyme, the HMG, COA reductase. 502 00:52:51,560 --> 00:52:58,070 And you can see there's a reasonably good relationship between those two things and this is a logarithmic scale, of course. 503 00:52:58,070 --> 00:53:01,430 It's like this is linear, it's not bad. 504 00:53:01,850 --> 00:53:09,890 And it does suggest that the ID 50 for reducing LDL cholesterol is somehow related to the mode of action of the drug. 505 00:53:10,460 --> 00:53:18,740 We need to confirm this with the data we've got at the moment, and it may not be as good if only because of regression to the mean I guess. 506 00:53:18,740 --> 00:53:24,170 I don't know. But there's a hint here that there may be something worth thinking about. 507 00:53:25,310 --> 00:53:29,420 So that's my review of some aspects of dose responsiveness. 508 00:53:30,260 --> 00:53:36,350 There are other things. It's useful when you're teaching medical students and training doctors how to use drugs properly to 509 00:53:36,350 --> 00:53:44,629 introduce the idea that things are graded in looking for signals of harms and benefits in big databases, 510 00:53:44,630 --> 00:53:51,650 for example, of anecdotal reports. If a report suggests that there's no difference across a range of doses, you think, well, 511 00:53:52,310 --> 00:53:57,740 is that truly in effect, you expect things to be dose related and benefits as well. 512 00:53:57,740 --> 00:54:01,459 Test doses. I've mentioned a dose titration. 513 00:54:01,460 --> 00:54:09,260 The principle of increasing the dose, reducing the dose, depending on the type of adverse reaction is to that side of the beneficial curve. 514 00:54:09,260 --> 00:54:11,090 Is it in the middle or is it toxic? 515 00:54:11,360 --> 00:54:20,090 The three dose adverse reaction approach and monitoring therapy, biomarkers and so on, and of course in drug development. 516 00:54:22,000 --> 00:54:25,740 So. Don't let anybody tell you. 517 00:54:26,730 --> 00:54:32,070 But any biochemical biomedical phenomenon is not dose related. 518 00:54:33,060 --> 00:54:38,100 And remember, Paracelsus, everything is a poison and nothing is not a poison. 519 00:54:38,700 --> 00:54:43,650 Only the dose determines whether something is or is not a poison. 520 00:54:44,430 --> 00:54:44,730 Did you?