All right, well let us begin. It is time to have some fun with psychopharmacology. Who's ready for some fun with psychopharmacology? Excellent. It's the end of the day. We hope to excite you, innervate you, energize you in thinking about tackling iterative psychopharmacology. Have you put those two words together before? What comes after the thing you just did before? How do you plan and organize in a longitudinal perspective what I'm going to do? A roadmap, a forethought and planning approach to treatments. Because how often does the first thing you do work? Nothing? Well sometimes it does, sometimes it does. And then when the second thing, does that work? Hopefully. And if by the third time nothing? So today is a chance for us to talk with you about how to think through that reality, how to embrace it, how to work with it in the patients that you're seeing from a pharmacological perspective but also from a decision-making perspective. So you'll be hearing from me first. I'm Joe Goldberg. I'm a clinical professor of psychiatry here in New York at the Icahn School of Medicine in Mount Sinai. I also want to acknowledge I'm involved with an organization called the American Society of Clinical Psychopharmacology, ASCP. I encourage you all to have a look at our website. It is a wonderful organization for practical applications in psychopharmacology. I'm going to be joined shortly by my colleague and friend Stephen Stahl and by my colleague and friend Michael Face. So together we're going to be talking about how do you think through applying principles of clinical reasoning and decision analytics to everyday practice. If you like what you hear today, there is a new book out, literally just came out from Cambridge Press, that encompasses many of the ideas that I'll be talking about and we'll all be talking about today. So here are our overarching objectives. Don't treat diagnoses. Don't follow algorithms. Figure out what's going on with the given patient in front of you. As an N-of-one trial, deeply evaluate and assess as much as you can that you think is relevant to describe and understand the totality of the problems that you are seeing and from that devise a personalized treatment plan. It's kind of an artisanal approach to psychopharmacology based on their unique illness characteristics, their past treatment outcomes, a non-arbitrary, non-trial-and-error approach to intelligent decision-making. We'll talk about how to utilize patient-specific decision analytic matrices, decision trees, pro-con assessments, contingency planning when you're undertaking iterative or sequential pharmacology trials. We'll talk about the core components of shared decision-making and strategies to balance patient and provider identified treatment priorities. I want to reduce your depression symptoms. You want to not gain weight. I want to improve your overall psychopathology. You want to be able to work in function. How do we find convergence collaboratively with our patients? We'll say just a little bit about the notion of machine learning approaches that it's kind of a very fancy way of saying how do we take large databases and draw inferences from the many, many people that have been in these big databases to try to translate findings to the patient in front of you. So let's see. Seldom is there a definitive next step if we've learned nothing else from large-scale multi-site clinical trials as well as longitudinal studies. It's that if the first thing you try doesn't work, it's very hard to know what to do second. We'll suggest you today that that's not an impossible task. It just means that there's not an automatic what to do next and you have to think about a whole number of factors to help inform your decision-making. Should I keep what the person's on? Should I augment? Should I switch? What should I do? That's why we're here. So rules to guide successive steps really should depend on clearly identifying targets of treatment that goes beyond diagnoses. That's not just saying I'm treating major depression or I'm treating mania or I'm treating schizophrenia. Steve Stahl is going to talk a little bit more about some of the trans-diagnostic symptoms that should inform your thinking and some of the neuroscience-based thinking about say I want to target the the reward circuitry or the attentional processing network or impulsivity or the the ways to use pharmacology to really target what we think is going on in the brain rather than just well the patient's not better from their diagnosis. Aligning our own identified treatment goals with the patient's own goals and finding convergence there. Evidence-based, rationale-based, or both treatments make sense. Sometimes there's an evidence base. I'll talk about this. I can tell you here's what's been shown to work when other things don't. Or here are things that haven't necessarily been studied but there's a good compelling rationale to try them. Or here's a really bad idea or an utterly arbitrary idea. We should have clarity in our own minds about separating those kinds of things to tailor personalized treatments to construct backup and contingency plans in a personalized decision tree. I always ask patients whom I'm seeing in consultations is the person that you're working with articulating to you what they're going to do next. If this doesn't work what's the backup plan and normally what I hear is either I don't know or I'm sure she has something in mind. Rather than a real longitudinal sense you know what are we going to do plan A, plan B, plan C and then how do we balance those contingency plans with the risks and benefits because everything, everything we do is a risk-benefit analysis. I wish to offer up to you the idea of thinking about pharmacology from two standpoints what we'll call probabilistic versus deterministic thinking. What does that mean? Probabilistic is when you do studies on large groups of patients and you make inferences about groups based on how the group performs and then you try to extrapolate from the group to the individual. That means people that have a clozapine level over 350 are more likely to get better than those below 350. But what if my clozapine level is below 350? Should I automatically raise it? Well probabilistically speaking yeah you're statistically in a better category if you make the numbers look the right way and if the generalizable inferences from a group will pertain to your individual patient. But deterministic thinking means for this particular individual treating your patient almost like an n-of-one trial what works for them? So if the patient you're seeing for instance has bipolar disorder and probabilistically monoaminergic antidepressants are not thought to work terribly well for bipolar depression but in Mr. Smith he's having a terrific success with paroxetine, deterministically you're not going to say to Mr. Smith we have to take you off the medicine that's working for you and instead impose upon you a more probabilistic approach because clinical trials say the following. We infer from his own individual experience what works for him. But then we can't extrapolate that to the next person. We can't say I'll have what he's having unless that next patient's clinical profile aligns and conforms to what we saw in the prior patient. Let me share with you these two wonderful words. Moderators and mediators. If you're familiar with these concepts I've stolen them from a very important biostatistician named Helena Kramer at Stanford who talks about the idea that patients have baseline characteristics that inform if not predict outcomes. I've listed a number of examples on the left. Age, age at onset, chronicity, psychosis, suicidality, comorbidities, maybe even pharmacogenetics. These go beyond just the diagnosis to more deeply describe who the patient is. Baseline characteristics, moderators, can inform what you are likely to see. You might very differently treat a third episode comorbid ADHD substance abusing patient with bipolar disorder than a first episode pure euphoric non-rapid cycling non-mixed patient. They have the same diagnosis but their treatments are very different. The more you possess knowledge of the moderating factors in the patient that will influence outcome, the more you can tailor your treatment and feel much less arbitrary about what you're doing. Mediators are things that happen after the treatments begun. Non-adherence, side effects, drug-drug interactions, oops I started using substances, I got pregnant, I had a catastrophic life event, I can't find amphetamine salts anywhere, they're on back order. All the things that can impinge on an otherwise efficacious treatment that in the real world can really put the kibosh on successful outcome. So you can't always control many of these but you can account for them and be aware of them. And when you're evaluating patients and deciding you have treatment resistance, you'd like to not conflate real treatment resistance with a bunch of mediating factors mucked up the ability to determine if the treatment that someone gave you might have been efficacious. You took it for three days, you stopped it after you had some nausea, the side effects were overwhelming, you had a drug-drug interaction, a rash occurred. So a lot of what ends up getting called treatment resistant where patients who move on to the third and fourth and fifth treatment differentiate pseudo treatment resistance where these mediating factors muck things up from true treatment resistance. Adequate trials, adequate durations, the right diagnoses, etc. I invite you to think about the term candidacy for treatments. This is an example with lithium candidacy. The same way surgeons talk about are you a candidate for bypass surgery, coronary artery disease, are you a candidate for an aortic valve replacement? Maybe, yes, your valve is broken but your kidney functioning is really poor and your wound healing is not good and your platelet count is under 75. You're not a good candidate for what might otherwise be an efficacious treatment. How often do you ask the question are you a candidate for this treatment? And we can now take the literature, which I've done for you here, and distill it into, in the case of lithium, the more you have these things on your patients list, the more you can say, you know, you're really a candidate for lithium. Not to use phrases like it's the gold standard drug or you haven't tried it before so why not, but rather, as you can read, lithium tends to work really well in bipolar patients who have more highs than lows, who haven't gotten sick until a little later in life, that is not really prepubescent or adolescent onset, for whom many episodes have not gone by in time. Lithium works much better in the first few episodes than your umpteenth, so don't save it for the umpteenth thing you're going to do. It's not going to work as well, probabilistically speaking. Absence of mixed features, absence of rapid cycling, etc., etc. You could make a list like this for any drug that you want to the extent that the literature informs information or your own clinical experience. Now here you don't want to get, there's this term apophenia, it's kind of like delusional, where you think you see patterns. I think this drug works really well in left-handed people who wear a size 9 shoe and it may be a very spurious association. So to the extent your own clinical experience leads you to form a hypothesis, and that's another word I want to focus on today, is forming a hypothesis, I am inclined to say, you know, I am more likely to use this medication in an anxious depressed patient than a non-anxious depressed patient. Or after non-response to this and this, I have found this to be helpful. So you can use your own clinical experience to inform these decisions as well. I said earlier the idea about identifying the targets of treatment. So what does that mean? It's what we do everywhere in medicine. What are we trying to accomplish? You ask a patient, you know, what are your goals? I want to get better. We know that. But what does that actually mean? Does it mean that we're targeting mood dysregulation, your sleep-wake cycle, impulse control, cognition, motivation, hedonic tone? Sure, yeah, all that. Yeah, all that, all the above. What about if there's an urgency question? Vagus nerve stimulation is an excellent treatment for treatment-resistant depression. How long till it kicks in? About a year. Oh, I don't want to wait a year. Well, okay, but you know, if you've not responded to eight to twelve prior things, I don't have a lot else to offer you. So if urgency is a factor, VNS may be a really great idea to bank on for the future, but we need something in the meantime to help guide things along. Do we have viable alternatives? Some drugs do not have their rivals. Can you think of any examples of medicines where I don't anything else like it? Clozapine. Yeah, got anything like it? Not really. ECT, got anything like ECT? Oh, no, sure, let me reach in my pocket. Oh, no, it's empty. So part of our expertise, I think, is informing our patients the uniqueness of a particular treatment, its robustness of effect, and the probabilistic and deterministic factors that are likely going to make you a really good candidate for this. Now that doesn't mean selling someone on a treatment. If I say to someone, and I really think you're a good ketamine candidate, and here's why, and they say, I don't want to do that, then we've got to share decision-making, and we talk about what are the things you look to accomplish. Well, I like fast onset, few side effects, little hassle factor, no weight gain. Okay, got it, got it, got it, got it, and here are the things that can fit that bill in a very non-arbitrary kind of way. So, core tools of clinical decision-making. I said before, everything is a risk-benefit analysis. How often do you make formal pro-con lists? With patients, which is a very engaging tactic with patients. Here are the options that we can use for your ailment, for the totality of features you're presenting. Let's talk about their pros and their cons, because everything has pros and cons, and based on your own priorities, we can tailor the treatment to your needs. There may be particular characteristics about a given patient group that will especially inform some of these things. For instance, I mentioned youth. Suicide risk may be higher with monoaminergic antidepressants if you're under age 24. Is this drug safe in pregnancy? I saw a patient earlier in the week, we were going through a complicated history, and I was doing all this decision-making stuff, and I'll show you more data decision analytic tree. Oh, I was so proud of myself, and the last thing is, by the way, I'm trying to get pregnant. Are these, are your proposed treatments safe in pregnancy? We had to go back and start all over again to tailor an artisanal approach. If you're doing pharmacogenetics, Dr. Stahl's going to talk a bit about this, you know, how does it factor into your decision tree if the patients are known to 2d6 poor metabolizer? You're not going to want to give them things that you know are not going to work. Pharmacogenetics is a very powerful tool to help you know what not to use. It's a little more useful to know what not to do, perhaps, than what to do. Let's press on. So, gosh, how many of you have seen this slide ad nauseum? Yes, star D tells us you have two shots, more or less, at success in major depressive disorder, and after that, it doesn't mean nothing's going to work, but it also means whatever you want to do next doesn't really matter a whole lot, because all these switches or augmentations had about the same probabilistic likelihood of working. Why do I say probabilistic? Because what I said before, maybe venlafaxine will work better in patients with neuropathic pain who are 2d6 extensive metabolizers. Maybe bupropion will be especially helpful in patients who are overweight, smokers who have attentional processing problems. You delve a little deeper, and iterative studies, try A, try B, try C, don't necessarily take those things into account. So, one of my favorite concepts in evaluating complicated presentations is, how many failed treatments, or I should say, how many times has the treatment failed to alleviate your problems in the current episode? I don't know how often you ask that question yourselves overtly or even covertly, but we only have two FDA-approved treatments for treatment-resistant depression. That is esketamine nasal spray and olanzapine fluoxetine combination. Well, what if neither works, or what if the patient's priorities are, I don't want to risk a side effect, or a patient said to me, I say, I can't go to somebody's office and do a nasal spray. Why? Because I can't. Well, then let's explore it, because you have two FDA-approved options, and then if we say, what does the literature tell us, and I've canvassed this literature, you'd think it would be a much more extensive list than this. We have vagus nerve stimulation, the drug Pramapexil, MAO inhibitors. How many of you use MAO inhibitors? Oh, wow. See, they heard you last year. So, why do you use them? They work. They work. They work if you dose them properly. I think in the STAR-D study, where the response rate was kind of piddly, the dosing wasn't exactly optimized. Maybe Dr. Thais will have something to say about that. Putting together clever combinations, like mirtazapine, venlafaxine, deep brain stimulation. I haven't sent people for deep brain stimulation, but everything else that's not on this list is a gamble. So, someone comes to you, and they've been on an SSRI or two, and they're not better, and what are you going to pick for your third option? You could do one of two things, which is what I do. I say, we can certainly go through the arbitrary list of things that exist, and roll the dice, and see if they work for you, or we can go through all the moderators that we can concoct and think of for you, and then go to the literature as best we can, and say, these are things that have been shown to work. Has anyone talked to you about an MAO inhibitor? Has anyone talked about esketamine? Has anyone talked to you about combinations such as venlafaxine or mirtazapine? Because they're not arbitrary. We need a much longer list than this, though. Another thing to think about is how meaningfully impactful are our treatments, and here we have the dreaded placebo effect to contend with. So the popular literature that condemns psychotropics and antidepressants as not being very useful, don't often take into account the fact that placebo works in about 30 to 40 percent of people with major depression. I find that mind-blowing. I have a colleague named Walter Brown, who did an interesting study of open-label placebo in major depression. He told the patients, I'm giving you a placebo. Take it every day. Come in and see me. I will rate your symptoms. I will talk to you at length for the next 42 days. You know what his success rate was? I'll tell you. Seventy percent. You know, researchers try to minimize the placebo effect, and clinicians try to maximize it. So when you think about iterative pharmacology and things like placebo responsiveness, you have to beat the placebo. Effect sizes, this is from an earlier book that we have published called Practical Psychopharmacology. We've lined up for you the rank ordering of effect sizes, how clinically meaningful or impactful is something. So down at the bottom of the barrel, lamotrigine for bipolar depression didn't have a huge difference from placebo. It's significantly better, but not by a mile. Buspirone for generalized anxiety, SSRIs for negative symptoms. You can read on and on. Most of our treatments have small effect sizes or medium effect sizes like SSRIs for panic. It's okay. It's okay. If you had, I don't know, widely metastatic cancer and you went to your oncologist and he said, I have a treatment for you. It's okay. It's better than no treatment, but you know, it's not great. You might say, what else you got? So do you think we have a lot of high effect size treatments in psychiatry? I ask that as such a leading question. Yeah, I count four when I review this literature. So benzos for insomnia, patients know what they're asking for. They know their effect sizes. You don't have to question them about that. Clozapine and schizophrenia, yeah, someone said, because it works with a large effect size, 0.88. Anything over 0.8 is very clinically meaningful. We should all be so lucky. Intravenous ketamine over one, wow. And the king of the hill, everybody go on Lystex amphetamine for your ADHD. Again, patients know what they're asking for or translate for them. Yeah, this is a drug that's got a very meaningful difference from a placebo. So should you augment or switch in your decision-making? Well, it depends. It shouldn't be arbitrary. I take a page from the CanMat guidelines here which say, well, you ought to switch somebody if one thing clearly hasn't worked and you might augment if two or more things haven't worked because now you're getting into the trenches of we don't have as many things to offer you so we'll take what we can get. And with augmentation, you can make something out of something. You just can't make something out of nothing. If the patient's poorly tolerating something, okay, let's not use that. If they've had some partial response, let's capitalize on that. What's a partial response? At least 25% better than when you started out. So measurement-based care is your friend. If you're doing rating scales with patients, especially in a busy practice, wow, your PHQ-9 went from 20 to 17. Should we continue it? Maybe. We'll look at that more in just a second. It depends on how much magnitude of improvement there is. Symptom severity, patient preference. Here, let's take a case. Best 68-year-old woman, recently retired teacher, major depression, hypertension, hyperlipidemia, neuropathic pain, was on escitalopram and paroxetine, didn't like some side effects, now comes in with six months of some ongoing depressive symptoms. Oh, look, her PHQ-9 is a 16. That's moderately severe depression. How do you decide what to do next. Well, first, make a decision tree. Here's four options. You could think of some more. Should we give Beth a trial of CBT, an evidence-based depression treatment? Should we say, go back on your paroxetine or escitalopram? Even if it didn't work, it's like, you know, the food here is terrible and the portions are too small. Why would you go back to it? Well, you might. You could say, let's up it to an SNRI. Is that a broader spectrum drug? Dr. Stahl will tell us that. You could pick a newer generation drug. I propose to you, rather than which of these you do, you make your pro-con lists. I love pro-con lists. Just ask my kids. Every time they say, can we do X, well, let's make a pro-con list. I'm really popular at home. Ever since you wrote that book, Dad, you're no fun anymore. Well, fun has its place, but look, you can go through the pros and cons. So if we're going to do this, maybe the priorities are side effect mitigation or addressing pertinent psychosocial issues that might be more relevant for someone like Beth, who just retired, than someone where there's no obvious environmental factors. What if Beth's not a talker? How much does a psychotherapy address physical and cognitive symptoms? How about resuming something if it worked before? Well, but then you gotta counter the side effects. Are those worth countering? You can think this through with Beth together, and that's shared decision making. So the pros and cons right in front of you can help you decide what to do next. Breadth of spectrum, tolerability, you get the idea. Pro-con lists are fun. What if Beth has anxiety? If Beth has anxiety, I go read Michael Face, and Michael will tell me if I've misread him. But to my read of the literature, usually when anxiety accompanies depression, it's the kibosh on responsivity. It is much harder to treat an anxious than non-anxious depression. Unless, as Dr. Face showed, you tell me if I'm wrong in this trial he did of the lasodome, there's a reason I asked him to come today. It actually worked even better in anxious than non-anxious depression. I don't know any other psychotropic drug where you could say that. Now this is probabilistic. It doesn't mean everybody out there should do this. But it's a moderator that you need to know about. And you can probabilistically infer from a trial such as his whether this might be a better choice for Beth. And you can do this with every clinical characteristic that you're interested in. So let me skip ahead a bit, talk about a few more things with iterative decision making. There's something we call the 20% rule. You may know of this. Patients should be at least 20% improved after two weeks with whatever you're giving them, why? Well, it's not an indication of full response, but it's a pretty good harbinger of it. You're eight and a half times more likely to be a responder and six and a half times more likely to be a remitter. If Beth's PHQ score went from 16 to 12, not 13, not 14, 12 or better. It's like when you're in Las Vegas and you're deciding what do I do with a 17 in blackjack. You need to know the number because it's got predictive value. So with 85% sensitivity, which is really good, a little bit less specificity. So it means if she hasn't had that 20% improvement, it doesn't mean she won't get better. It means probabilistically, you're on the right track if you've seen a 20% improvement. Here's the study which showed that. You can't read this, but trust me, you can also get the slides. They're available. I think APA has the slides available, not email me. Which actually showed what I just told you is that if you don't see a 20% improvement by two weeks, do something. What? Anything. Something. Change your tie. Wear different socks. Raise the dose. Augment. Do something because the probability is kind of low that you're gonna see success unless you do that. Other factors to consider. I mentioned the number of failed trials in the current episode. You got viable alternatives, clozapine, ECT, not so much. Tolerability matters. Unique target symptoms, suicidality with lithium, pain, neuropathic pain with neurodegenergic drugs, particular comorbidities, timing of an intervention, cross taper decisions. We talk about this at length in the book, so that's a teaser for the book. How should I switch from one drug to another? It depends on particular moderators. Like what's the half-life of the outgoing drug? Is there a P450 interaction that you have to think about? Is there receptor redundancy? Are there rebound effects? Like if you're gonna stop an anticholinergic drug, you should taper it if the patient's going on a non-anticholinergic drug. Otherwise, they will have bowel problems and lacrimation and salivation. So these are some of the things that would inform that decision. Frame a hypothesis. Beth's getting more suicidal on venlafaxine. Should I switch? Should I change? Well, it depends. What's my hypothesis? Do I think she's getting worse because of my intervention or despite my intervention? Or maybe she hasn't had an adequate trial. Or maybe she's a P450 2D6 poor metabolizer and this drug's gonna go nowhere for her. Or maybe she's just not taking it. And I can posit these hypotheses to Beth and say, here are the four or five reasons that I could anticipate why things aren't getting better for you. Let's brainstorm this together as we decide what to do next. Beth now says she has memory and attention complaints. Is that because of a side effect? Is that because maybe she has untreated ADD? Is that because maybe I need to get neuropsych testing? Frame a hypothesis and vet it. And then you can test your hypothesis empirically. Think about implications of what you do next or what's gonna come next. So on this complicated looking puzzle, I've shown you some of the treatments that would count in hard to treat forms of depression. But there's implications. For example, olanzapine plus fluoxetine has a heck of a long half-life. And if I wanna tee you up for an MAOI trial, I'm gonna have to wash you out for five weeks from the olanzapine fluoxetine before I can do that. So maybe I want to insert ketamine or resketamine in between there, or consider a second generation intervention or lithium strategically as your bridge before you get to the next thing. Anyway, you can think about this. Dr. Saul's gonna talk a bit more about chess playing in psychopharmacology. Practice guidelines are useful, but they don't really tailor the treatment. So tailoring the treatment means accounting for moderators. Here I have a personalized decision tree. Should you use antidepressants in bipolar depression? Maybe, it depends. You have mixed features, no. Have you ever gotten worse with an antidepressant before? Don't use it. You have rapid cycling, don't use it. You have bipolar I, probably don't use it. But you got better with an SSRI before. There's no mixed features, no rapid cycling. You're bipolar II, not bipolar I. You can make an argument favoring the use of an antidepressant in that tailored way. Constructing a personalized decision analytic matrix. I'm gonna jump ahead a bit here. You can do your pro and con list for a given patient. These are all in the slides, so for sake of time, I'm just gonna say, here are all options you could consider for a hard-to-treat presentation. You could augment lamotrigine with lithium, that's evidence-based, but there's only one proof-of-concept study. You could use ketamine, you could use limeteperone, puriprazine. Rather than say, I could do all these things, you make your pro-con list. So here's what I'm gonna sort of leave you with, which is the idea of a personalized decision analytic matrix. Where it's said differently, everything I needed to learn about being a psychopharmacologist, I learned on HGTV, the garden channel. And if I had to do it over and I couldn't be a psychiatrist, I'd be a real estate agent. Why? Because the client comes to you and tells you what they want. They want four bedrooms, a fenced-in backyard, not on a busy street, with five bathrooms and a nice view. Okay, let me see what I got in the inventory and then I'll come back to you. You've told me what you want. You want fast onset, no weight gain, cognitive benefits, no cost, minimal hassle, and, right, is that what you want? That's what I want, okay. And how important are those things to you, and how important are they to me? You want fast onset, I want efficacy when other things haven't worked. So here's a fun little trick that's described in the book. You can actually assign a coefficient, numerically, to each of these priorities, and then you multiply it by how well each drug performs. So, go with me, lamotrigine and lithium gets a one for its importance, gets a three for its importance, gets a one for how well it performs. No weight gain, scores nicely for lamotrigine and lithium. Cognitive benefit, cost, you multiply these things out. So three times one, and four times two, and three times two, and two times four, and three times two gives you 31. Yes, it's laborious, but at the end of the day, you've gone through all five options, and the winner is Emma's gonna get a stimulant by her criteria and by ours. This is convergent decision analytic thinking. Shared decision making is just that, identify the patient's goals, your goals, and then use that to inform the next step based on priorities. Whether you use a decision analytic matrix like the kind I showed you, or even more informally, what are your top three priorities? Do they conform to mine? Well, I wanna be able to still use cannabis. Well, I don't wanna come off my stimulant, so an MAOI is a no. All right, let's broker the fit and play HGTV psychopharmacology. You can see how concordant Emma's wishes or your patient's wishes are with your own thinking. And then align here with the top two, and then we'll kind of vet it from there. So I am going to quickly talk about one other example, and then turn things over to Dr. Stahl. Imagine a patient, African-American male, longstanding history of schizophrenia, has persistent symptoms, didn't get better with iripiprazole, paliperidone, quetiapine, suicide attempts, hypercholesterolemia, hypertensive. As you're thinking about what to do and what your options are, consider using a decision analytic support tool. We used to use Framingham scores, now we use something called the pooled cohort risk equation. What are the chances that this fellow's gonna drop dead from a heart attack in the next ten years based on his age, race, cholesterol, blood pressure, etc? James has a 9.4% chance of dying from a heart attack when you plug those numbers in, this is free download on the Internet. That might configure into your decision making differently than if he had a lower score, 7.5 is considered the cut off, so it's less arbitrary. But another factor is what's the time frame? So here's an algorithm to predict weight gain with a lansipine. If you gain more than 4 to 5 pounds in the first few weeks, really high chance you're gonna get out of weight gain. If you don't gain more than 4 to 5 pounds in the first few weeks, pretty low chance. So you could take this data, and you could take James's pooled cohort equation, and then tell James, here's why I think we should proceed with lansipine for you. And if you gain more than 4 to 5 pounds, we have to intervene, and we have ways to intervene. I am going to skip ahead to the final slide, our takeaway points. I know I've covered a lot, but I want to give a chance for Dr. Stahl to give some perspective here too. Talk about the totality of what you are trying to treat in an almost artisanal, crafted approach for a given patient. When you are constructing iterative decision making plans, don't just say, let's start with this and see where it goes, and we'll cross that bridge when we get there, have something in mind based on the baseline characteristics and where you've been before. Frame and test hypotheses, I've given you some examples. Encourage joint, mutual thinking about pro-con lists for evidence-based options or rationale-based options. And really, truly share the process with the patient. We've arrived at this decision together, non-arbitrarily. It's not a bad thing for patient therapeutic alliance or adherence when the patient has that skin in the game. Don't think about just one treatment. You're in this for the long haul. If one or two things haven't worked, you by all means can construct a longitudinal plan. But you have to construct a longitudinal plan. So does broadening the mechanism of action give better outcomes? Are different receptor targets advantageous? When should you use decision analytic support tools, like therapeutic drug monitoring? I wish I had the answers to those questions, but since I don't, luckily, we have a secret weapon. We have a secret weapon. Do we have a secret weapon? We have a secret weapon somewhere here. Secret weapon, where are you? Here you are. And that secret weapon, his name's Steven Stahl. So, with that, someone who needs no introduction. Dr. Stahl is a preeminent psychopharmacologist, practical psychopharmacologist. He's gonna help us make sense of all this. Thanks, Joe. I think you found out that Joe is a meta-psychopharmacologist. He thinks about thinking, he thinks about psychopharmacology. And you've just seen it today, but mark my words, within 40 years, maybe four years, you'll have an app that'll do what he just did. Now, I myself am a little more seat-of-the-pants intuitive, if you will, and try to make my decisions. And what's fun, having written these books with him, is to see why I kind of did what I did intuitively, but to kind of bring it from, I don't know, shooting from the hips into consciousness. So it's been fun working on this, and we're gonna talk a little bit about iterative psychopharmacology. And the mindset of most prescribers is cross-sectional versus longitudinal planning. So in other words, you have a room full of depressed patients, or you're thinking about one patient over the course of the episode. Not necessarily thinking ahead to implications from the previous chest move. So those of us in this room are generally not first-line prescribers in primary care, and here are patients who have had other trials. And when we get them, you saw in star D, no thanks to Michael Thaise, actually, showing us how bad the drugs work, third and fourth. So now, smartass, what are you going to do? You're fifth. And so you have to not only think of fifth, but what are you going to do sixth, seventh, and eighth? And so that's how a good chess player plays. So we can fall victim to random trial and error, successive, you know, just throw something up there they haven't done before. But there really is a way to do, like Joe said, these various factors. And in some ways, you do that automatically, but not necessarily realize that you're doing it. But we need to do it as much as we can and capitalize on pharmacodynamics when we can, because I always like to throw into this mix bad math. At my core, I'm a pharmacologist, and I do believe in synergy, which means if you do this right, 1 plus 1 equals 10. And so that is what you need to sit upon to try to make a more better outcome. And utilizing tools that can help guide successive decision-making, which is really what Joe does for a living. And I'll just mention a little bit about laboratory-based. It's not really part of psychiatry yet, except in maybe one area. So one of the things I like doing is consider RDoC, if you will, which is a research diagnostic and domains kinds of criteria. The brain is not full of psychiatric disorders. The brain's full of symptoms. And you end up having six circuits. Now, in order to communicate, you can say, what's this? Depressed mood, apathy, problems concentrating, insomnia. What's that? Oh, that is a category. Syndrome, a set of symptoms put together create a diagnosis. Well, what's this? Anxiety and worry, irritability, problems concentrating, and insomnia. What's this? Inattention, impulsivity, problems concentrating, and hyperactivity. So if you play the diagnostic game, which we have to do to communicate with each other and to get paid on insurance forms, that's major depression. That's GAD. And that's ADHD. Now, we may think that way momentarily. But in real patients, what do you really say? You treat them, and they're still having problems concentrating. You think this way. This is dimensionally. And one thing in one syndrome that works tends to work in another. And that's why pro-cognitive agents that boost dopamine can work in areas outside of ADHD where they're approved. And this is looking in terms of too much activity or not enough activity. So that's the name of the game. And when you put that together with having read Dr. Stahl's textbooks cover to cover and studied them assiduously or wherever you get your information, you need to kind of know how drugs work. And we've been playing at the fountain of three monoamines for a long time. This is in Joe's book with me. And I don't need to tell you about it. But if you know this in mind, then you can go for bad math. Because what you do is you find out that if you put reuptake blockers for serotonin together with something that blocks an alpha 2 receptor, they work more better. And this is a meta-analysis. And at the bottom, it shows that if you put either myanserin, or tazapine, trazodone, myanserin is actually a European drug. Those of you who don't work in the United States will recognize that one. And they tend to work better. And so if you put it in an animal brain, you'll see the reason for that is it spits out more serotonin and norepinephrine together. And if you think serotonin and norepinephrine have anything to do with your patient's problem, it's going to spit it out more. Not everybody's going to respond to that. But this is synergy. And you look for as many of these as you can. We used to think that there was some synergy to an SS versus SNRI. But if you look at the responses in the gray box there, it says the 63% versus 59%, somewhat underwhelming. Now, there is basically some evidence that SNRIs do work a little bit better than SSRIs. And this is a little bit belied by the fact that the SNRI studies weren't done at the kind of doses, at least I use, 225 of vanillafexane. Wimp. And so nobody's really done it at higher doses. So I think there's a modest at best, but it sometimes works. And I often prefer to have, when I'm going for synergy, a platform of an SNRI, even at moderately high doses, to augment something on top of, just to give it a little bit more of a chance of bad math. But believe me, SNRIs at low doses are not much more than an SSRI. Now, I did this last year. You may know that the APAs asked me to do old drugs and try to help resurrect them a bit. And we talked about MAOs last year. This actually year, I'll give a plug to my talk tomorrow with Jonathan Meyer on lithium at 8 o'clock in the morning, if you're masochistic on a Sunday morning, to bring back these old drugs. But there's nothing like a triple drug. If you haven't used this, then you have not one of the greatest arrows in your quiver that you could. But MAOs, and then you can augment an MHL, yes. And although the books say you can't do it, baloney. TCAs plus MAOs all the time. And amphetamine augmentation is probably the most common augmentation of an MAOI. And if you have patients that are sick as hell, you go for it. And actually, it's interesting that Nardil, which is actually, did you know Nardil actually is a GABA-T inhibitor as well? It raises GABA as well as dopamine, norepinephrine, and serotonin. And Joe already told you, pop quiz, what's one of the worst types of depression to treat? It's if you've got anxiety with it. The SSRIs just don't work very well. And so velazodone, Mike showed that. But augmentation with the number of the atypicals do. Because the question is, do you want to avoid atypical augmentation? Not so much you want to give tranecypromine, which is more like giving amphetamine. But phenylzine is more like giving a drug for anxiety. So anxious, depressed patients that are really down the stream, believe me, I really believe that we can't lose these agents in the treatment of psychiatry patients. Are some antidepressants inherently more potent than others? Nah, not really. Three uptake inhibitors, if this is all a bunch of them. I mean, maybe one versus another. But part of the problem is that, let's give this part of the room. Everybody in here is depressed by now, I'm sure. So you're depressed. And you've got a Prozac. And you've got Paxil. And there's no difference between the two of you. However, there's somebody in this Prozac area that isn't getting very well that would have done better if he'd been sitting over here. And there's somebody over here the opposite. And that's how the problem is of doing these large trials. They don't really talk about the individual. And so part of what I believe in is to understand the evidence first and do evidence-based first. But then eventually, there's a such thing as not just evidence-based practice, but practice-based evidence. And that is what you all do well. And there is an intuitive reason why you think that something might work. So what we're trying to do in this book and also in this course is just to say, okay, think about why you wanna go next sometimes. And also think about what your move is gonna be after that, even at the time you're thinking about making that move. And you might be able to do better even. Someday there will be markers. I'll just briefly say that this is an inflammatory biomarker. It's thought that there's a lot of inflammation in the brain, depression. Maybe C-reactive protein says you'll do better on one than another. This hasn't been widely replicated. We don't actually use it yet. Other possible things, there's a couple of genes that screw it around with lithium response. Tomorrow we'll talk about how lithium works and you won't know anything at the end of that part of the talk because we still don't know how it works. But there is some evidence that there's a mild, favorable response with some genes. Here's Dawn, a 25-year-old graduate student with untreated ADHD. She's the met met homozygous. Does this information help you? Well, does it affirm the diagnosis of ADHD, favor a psychostimulant, favor an NRI, or discourage an alpha? And the answer is there's some slight favoring of an NRI as shown here. 1.4 times higher if you're Val-Val than if you're met met. But is it enough to actually pay the money to get the Val-Val and to make your decision? No, not yet. In fact, the APA's group of people has just gone through recently, I saw it just yesterday in the feed of the news that they say that pharmacogenomics for the treatment of depressions in terms of pharmacodynamic markers are not ready for prime time or reimbursement. Now, I do think, though, it helps you think. And so sometimes it's useful. And certainly it is useful for the pharmacokinetic drug interactions. But for dynamics, we still have this kind of thing as a hint, here's another hint. Dr. Goldberg's patient, James, has schizophrenia and heart attack and consults you about pharmacogenomics and he's GG at the DRD2 and he's CC at the 5-HT2C receptor 2C allele. What do you say? Well, you can prescribe olanzapine to this person with less fear of weight gain, prescribe olanzapine to the great trepidation of great weight gain, prescribe clozapine, he won't become neutropenic, prescribe anything you like, he probably won't develop tardive. And what you would say, officially, is that there is a slight improvement in the chance of tolerable weight gain. And this is some of the data for that. And you see people with different genes have different weight gains depending on the dopamine receptor gene you got on the left or depending upon the serotonin 2C receptor you have on the right. Will nobody gain weight with this gene? Not true. Will everybody gain weight with the other gene? Not true. Is it worth getting? Probably not yet. But I'm just showing you that there are things, there are hints, they're waiting for wide replication, it's not ready for prime time. Here's one that's a little more interesting to me even though I don't really work that much with these kinds of patients. You're devising a personal treatment decision tree for Ed, a 46-year-old construction worker with alcohol use disorder and he's got no complications and concurrent concomitant depression. He is an ASP40 carrier of the mu-opioid receptor. I just put this in here so you've heard about it today. It's not gonna be something you're gonna order you have to worry about. But based on this information, what would you recommend? Dasulfiram, naltrexone, acamprosate, olanzapine, samudorfam. And the answer is, if you're an opiate receptor mu type one, ASP40 homozygote, and you take naltrexone, they have shown a longer time until they relapse of alcohol and a longer time until they start heavy drinking in European descended cohorts with one or two of those alleles. Okay, interesting. That gives you a little more confidence the drug's gonna work. Are you gonna not give it to him if he doesn't have this? You'll still give it to him. Does it guarantee that if he has this, it's gonna get better? No. But those things are out there. But one of the things I wanna do pivot to is to say, are some antipsychotics better than others? And the answer is in mania, not so much. On an overall basis, maybe. But for treatment resistant mania or depression, of course the big guns are olanzapine and clozapine. Clozapine particularly. And maybe some year we'll do a course because Jonathan and I have also written a book on clozapine and we don't want people to stop using that drug. But really, what I think that we should be doing in the area of psychosis is therapeutic drug monitoring. Not in the area of depression, but why? So in other words, before moving on to the next intervention, how sure are the patients adherent? Of course my patients take their drugs, it's Joe's patients that don't take their drugs. And so that's what we all think. So what am I talking about? Here is if you prescribe antipsychotics, you must use plasma levels. In this room you have to, because you're saying patient's second, third, fourth line. And this is why. If you have an inadequate response to an antipsychotic at the top, you have these three choices. You can give the high dose monotherapy, you can switch, or you can go to clozapine. And then some people, after all that fails, will go to two antipsychotics, which is regrettable, but sometimes works. What I'm saying is that if you have inadequate response, particularly two or three times, do this. Turn your headlights on. Anybody here drive at night? How are you gonna see the road unless you turn your headlights on? Turning your headlights on when you have treatment-resistant psychosis is to get a plasma drug level. And a plasma drug level is interesting whether it's high or low. If it's low, the usual cause, of course, is the patient's not adherent. But not everybody. Some people have funny livers and some funny stomachs and they eat the drugs fast. And guess what you have to do to fix that? High dose antipsychotic monotherapy. Raise the dose, duh. And you don't look at the PDR's recommendation for what that dose is. You look at the plasma drug level, because it's not what is in the hand before you eat it, it's what gets into the brain. And at the state hospitals, where we have like 7,000 patients in five hospitals, we do this all the time, it's not common, but maybe 10%, 15% of the time, we will find that people are not responding because they don't have enough drug aboard. And they are not necessarily not adherent. And so you raise the dose, they get better. And it's so nice when something that simple happens. And you'll never know it because your headlights are off. On the other hand, what if the drug levels are okay? Well, if they don't have adverse effects, we could call that a PD or pharmacodynamic failure. The other one being a pharmacokinetic failure. What does that mean? There are some people whose antipsychotic needs are not the 60 to 80% D2 occupancy that patients have. Who are those patients? They're patients between 21 and 60, usually, who have somewhat worsening psychosis, who don't have drug abuse, who don't have concomitant medications, who do not have concomitant medical illnesses, who do not have substance abuse, and who are adherent, just like all the patients in your practice. And based upon that clinical trial population, they determine what the dose should be. Well, guess what? Some people are not as easy a patient as that. And there are people who respond to 81% D2 occupancy. Give me 83. 83% occupancy. So you inch it up, and there are people who will respond. And that is what you would do by watching blood levels. And in order to do that, you should know some of these ideas. And this comes from a book that Jonathan and I did. If anybody wants it, there's no reason to buy the book anymore. All you need is this table, which will gladly PDF you. But basically, if the antipsychotics are not at the therapeutic threshold, it's got almost zero chance of working. And guess what? How high do you go? It's not as many pills as you take. How high do you go in the brain? And the answer is, on the AGNP ASCP column, a group of people have said it's an alert level, which means that there's some evidence that if you go above that, there may be side effects that are problematic. It doesn't really say much more than that. And we've come up with prescribing-based evidence based on 20 years at the state hospitals of doing these levels in thousands of patients. You're gonna do randomized controlled trials in insane patients that are ethically impossible to give consent and double blind them? No, but you can actually just do it. And so the point of futility means that if you get to this point and they still don't have side effects and they're still not better, game over. Because there's no real evidence that we have found that going higher is even better, even if they can tolerate it. And the really big ones are Clozapine there, because a lot of people are wimps and they're afraid of going over 350. On the, I'm talking not 350 milligrams orally, I'm talking about 350 nanograms per ml. We push it to 1,000. Two old drugs, Lufenazine and Haldol are amazing for treatment resistant because you can go high, particularly with Haldol. And one of the other ones that we like a lot is Olanzapine. You'll see that most of the Olanzapine stuff that comes back on a quest form or whatever if you ever do this. By the way, these levels are costing between 36 and 80 bucks as opposed to a plasma pharmacogenomics panel which might cost $2,000. So, and it'll come back and it'll say at 75 that that's the highest you can go. We purposely go to 100 and possibly up to 150 on Olanzapine, but you have to tell your pharmacists about this, otherwise you'll get an alert at four o'clock on a Friday night saying, oh my God, Stahl, you're poisoning a patient, he's gonna die, he's level 76. It says so right here in Quest, this is horrible. So, if you know what you're doing is calming that person down, you give them a little Olanzapine and you say this patient's fine. So, the idea is that in treatment resistant, not first line use, treatment resistant psychosis, turn your headlights on, get some levels and go to the point of futility if the patient tolerates it. So, in summary, be mindful of pharmacodynamic properties and drug mechanisms of action so you get bad math out of it whenever you can. Play pharmacodynamic chess, which means think two steps ahead. Use laboratory measures, I think really therapeutic drug monitoring. Keep your eye on pharmacogenetic testing, certainly use it for drug interactions, but not necessarily for predicting therapeutic response from pharmacodynamics. And then we'll get the Goldberg app, which will basically, you put in your patient's stuff and you'll get this lecture, but for that patient on a piece of paper, I'm sure, be sure, I want the Goldberg app. And then successive pharmacotherapy should not be just trial and error, it should reflect the N01 thing that Joe talks about. I really do think it's a lot more fun to actually create a hypothesis. And you know what's cool about that? It's when you're right. It is really cool to say, you know, out of all this stuff, I predict that. Now, of course, I live in Vegas now, I don't know if you know that, I moved to Vegas, and so it's a little bit like gambling in the sense that, you know, you hit the slot machines, it doesn't always come out. But when it does, it is so sweet. And so that's what I would say. And those who did not learn from the pharmacological past are doomed to repeat it, and so keep up the good look, read Joe's books, and go for it. Thank you very much. So, the goal of good treatment is to activate your reward pathway. And to help us put all this together, we don't have any slides. Michael Face, also Shadino Introduction, Michael I've known forever, clinical psychiatrist, has been involved with just about every major study I can think of in my career in mood disorders and professor of psychiatry at Penn. So as one of the principal, co-principal investigators in the STAR-D study, involved in the STEPI-D study, the notion of iterative treatment is nothing new to him, and so I can think of no one better to help us put these comments into some context. So Michael. So it's a real treat to be here with you all today. This is my first live in person APA since 2019. So I'm glad to be back, glad to see so many of you back, and a standing room only room is kind of a dopamine augmenting enhancement for people that do the kind of things we do. But it was my pleasure today to listen to two masters give two complimentary talks on what I think the meta-message is, how can we maintain our morale as treating psychiatrists in our work with patients who have difficult to treat disorders. And Joe took a comprehensive approach that really talks about our relationship with our patient and our, in a shared decision-making model, how we bring our knowledge, our knowledge about predictors, our knowledge about some aspects of the pharmacology, and engage our collaborative patient in the process of making a plan and then working through that plan. Now a couple of things that are important parts of that that didn't get showcased today are keeping track of symptoms, keeping track of adherence, and engaging the participant's family in the process if possible, and then also helping them to keep their own morale up and do the kinds of active problem-solving, life-engaging activities that go along with well-being. And our people suffer, our patients suffer from illnesses, but they're not hapless bystanders in the process, and whether they exercise or not, whether they return the phone call that needs to be called, whether they make a plan to pay the taxes or file taxes and so forth, have to do with the probability of ambient negative thoughts that are factors that reinforce and probably even can undermine the chances of getting better. So let me start out with one provocative thing, and that is that we all know it's not provocative to say antidepressants have small effects. That's a known fact. But a small effect in group data is principally made up by a large effect for 10 to 20 percent of the participants. And so when you see this Cohen's d of .3, that's because 15 percent to 20 percent of the people who got the active drug were way better than the people who got the placebo. And so finding who's going to get that 15 percent benefit is the art of the psychopharmacology, and I think Joe's portion of this was providing us a road map to be better artists and make better use of our artistic craft. Now Stephen, on the other hand, is really me, always my favorite educator in terms of understanding complex psychopharmacology and then making it fun and interesting along the way, too. So when I agreed to do this discussion, I didn't realize these two had a new book that really bring their talents together. And one of the first things I do when I get home is to actually buy their book. So you all will be about $1.50 richer on the other side of this. Anyway, so Stephen's piece of this is to always keep thinking about how complex the brain is, how you have both regulatory and counter-regulatory effects, and along the way how you can make people worse rather than better with unanticipated drug-drug interactions, or when you're doing complex psychopharmacology, drug-drug, drug-drug, drug interactions potentially, for which there are no great data on four different drugs or five different drugs' additive components in terms of their possibility for doing harm. And so an additional piece I'd like to add to this consideration is de-prescribing, when our complex regimens aren't achieving the desired result, to kind of make a pack in your collaboration with your patient to identify the least helpful, least useful piece of this five-drug regimen, and if we're thinking about a replacement, actually thin the one out, get rid of it while you're initiating the newer medication. So I think that maintaining our morale in treating patients is one of the largest assets that we can bring to this, and I'll tell you a story. I was, as an early 30-something young zealot in clinical psychopharmacology, I got to work for five years on a treatment protocol for people with recurrent depression in which we had imipramine, and then we had five steps of alternate treatment strategies for people that didn't respond to imipramine, and at the end of the study, the patients that got assigned to me, not exactly random assignment, got assigned to me, had about a 68% chance of getting remitted. And one of my colleagues on the study, who's no longer living, was towards the end of his or her career, didn't have this zeal, didn't have this notion that we were doing great things and that there was an answer to this, that we could fix it, and if perfenazine didn't work, then lithium would. That person's treatment benefit rate was about 48%, so there's a 17% difference, or maybe 20% difference between two clinicians with the same protocol. And this wasn't talent, this was enthusiasm and morale and the caregiving that provided along the way. And so when you talk about placebo delivering 38, 40%, and drug delivering 52%, it's not really the placebo, it's the care that's provided. And so keeping our morale to continue to provide the best care while we have new ways of thinking about the collaboration, while we have new ways of thinking about how we can use predictors and the opposite of predictors, negative prognostic factors, and working on our treatment decision trees, and then looking for all the new and exciting discoveries that might provide another 8 or 10% potential benefit for another subgroup of people for whom we haven't figured out an effective treatment yet. So Joe, thanks for inviting me to do this, and I look forward to helping try to answer any questions you all might have. Thank you. Thank you. We have quite a lot of time for discussion and commentary. I'd invite anyone who has a comment, a question, a thought to come up to the microphone, and we have the three of us here happy to address any of the things that we brought up, or even things we may not have brought up, that you wish we had. Yes, sir. This was like Michael Hanau from Walden Behavioral Care, Eating Disorders. This was like hearing from the father, the son, and the Holy Ghost of Clinical Psychopharm, just very much blown away. You've got three Yiddish boys up here. Be careful. Go on. Well, I'm Jewish as well, so you can imagine the powerful impression it takes me to say such a thing. And I appreciate the touching on the issue of countertransference helplessness and how to manage and mitigate that, both for the benefit of the treater and the patient and raising spirits, not just neurotransmitter levels. As I say, I treat anorexia, and one of the heuristics I was introduced to in that line of work was, oh, you don't have enough nourishment to make serotonin for the medicine to work. I would wince when I hear perhaps a non-MD clinician say that, because I would presume having that little serotonin, I probably wouldn't have any functioning neurology at all. And yet, there does seem to be a higher level of treatment refractoriness in my patient population for comorbid depression, unipolar, bipolar depression than I've seen in, even in a tertiary care outpatient psychopharm clinic. And I'm wondering, I'm touching the mic, I'm wondering what's going on there biologically and how I can mitigate it before the patient is nourished such that my pharmacologic wares might actually be be helpful because it's going to take weeks, sometimes months, for my patient to be nourished. In the meantime, they are OCDing, they are deeply depressed, they may have agitated depression, they may have atypical depression. So, thank you. It sounds like you treat complicated patients and complicated presentations really demand going beyond just what's the diagnosis per se and what treatments could work. You know, I think eating disorders are a very difficult comorbidity to superimpose on mood disorders. And so, when there are more and more comorbidities, more difficult clinical profiles, it's to be expected on the front end. This is not going to be easy to do and I think it's useful to forecast for the patient what's in stores, that expectations are set. You know, as to the mechanisms here, and this is something Steve and I talk about a lot, is how much do mechanisms inform where to go next. We've been reshuffling the monoaminergic deck for years. We're just now starting to get some newer profiles of pharmacologies that will tap into the glutamate system, or the GABA system, or the cholinergic system. And, you know, hopefully as we are now broadening the breadth of spectrum beyond just the reshuffling of the monoamines, we'll be able to address some of the more complicated presentations like you're describing. Do you want to add anything, Steve or Michael? Yeah, I agree. I mean, it's pretty amazing. We've gone so far with drugs that really go after two monoamines. It's mostly serotonin and norepinephrine, a little bit dopamine, and we've created the whole field just on that. Watch out for this cholinergic story and psychosis that's coming down the road to us. That's going to be amazing. And you, of course, have the glutamate NMDA story and maybe the psychedelic story in the depression. So I think we're at a pivotal point to maybe help some of these patients. You can only whip the monoamine so hard and not everybody has a monoaminergic illness. Yes, sir. Thank you very much. I'm Jerry Cragan from the University of Toronto and insulin was discovered there and now we have glucagon-like peptide agonists. Antigen-best. That's right. We went from one metabolism to another and I'm interested in the side effects with respect to weight gain that we mentioned and the issues with bipolar patients and unipolar patients and trials that are starting to show with glucagon-like peptide agonists showing associate memory deficits that are disappearing with the trial of that. And I just wondered whether or not that's an area of research that is going to blossom as it is in other areas. As Daniel Drucker said when he first discovered this in Toronto, he did never think that heart failure, Alzheimer's, Parkinson's, and now I'm wondering in depression whether or not there's a role for that and what do you think of that? Yeah, so, you know, one way we think about treating mood disorders is how can we target the circuits that are involved? But then on another level we just think about what's neuroprotective and to what extent does the brain respond to anti-apoptotic, neurotrophic, synaptogenic treatments. I think if anyone heard the presentation Roger McIntyre and I gave this morning, we were talking a lot about the ways in which GLP-1 agonists may be neuroprotective for insulin receptors in the brain for the ways in which GLP-1 agonists might actually have direct effects on cognition and on mood. It's another exciting new vista that I think we're just beginning to talk about. When we talk about iterative pharmacology, you know, on the one hand we're saying given what we have to work with right now, how can we sort of not plod through randomly what to do next, but rather how to make the most sense of the playing pieces that we have on the chessboard. As new chess pieces become available, how about that, a new chess piece? Beyond bishops and rooks and castles, you know, can we actually further do, you know, greater impact in our treatments? That's going to be a very powerful tool. I sure hope that the neuroprotective aspects that people like Roger talk about with the GLP-1 agents and the incretins will expand our chessboard. Thank you very much. You know, there's never been a time in human history in which highly palatable food was available for very little cost and that when you look at animal models, animal chronic stress models, it takes almost life-ending stress before an animal will lose the taste for highly palatable sucrose water. And so with an early age of onset of depression and the access to calories and then a subset of medication that are pro-weight gaining, we've set up a really bad storm for weight gain being a consequence and a likely feature. It's very common in early onset mood disorders to see predisposition to weight gain. And now we have medications that promote it. So it's our job to have weight charts to manage this. I say this and it's really, I won't win the Nobel for this, but nobody gains 10 pounds without first gaining 5. And it's almost impossible just with weight gain from a psychotropic to gain 5 in 2 or 3 weeks. So if you're keeping track of this, it won't happen. And you just need the discipline to monitor it and to move on when a medicine is having this potentially life-ruining consequence. Question for Dr. Goldberg. How do you, in terms of style, how do you share with patients your plan, there's a plan B and a plan C, without undermining the placebo effect? Oh, if this doesn't work, we can have something else. I have in mind something, here's what we're going to do and here's what else we're going to do. Thank you for that question. So I've never yet said to the patient, I sure hope the placebo effect is going to work really well for you. If you have, but I know in my head the things that are more likely to make a placebo response. So, you know, low baseline severity, lack of chronicity, lack of treatment resistance, lack of suggestivity, lack of neuroticism, you know, it's odd, you know, the placebo effect is so high and yet the patients that we see tend to be so in tertiary care settings lacking in the things that would be predictive of placebo response. So I don't think my placebo response is necessarily as good as that of other people, but I will do my darndest to capitalize on it by really through the therapeutic alliance. I don't mean that in a kind of a fluffy way so much as to say it's you and me against the illness and so let's both step back from it. Chairs aligned and I have a white erase board and I will say, this is what we're up against. And then we make a battle plan, you and me together, sketching out the goals of treatment, the targets of pharmacotherapy, the limitations of pharmacotherapy, that is, I'd be lying to you if I said this drug is going to do more than we know it can do. We can certainly try to do some paradoxical things like say, I'm not really sure this is going to work. A lot of people tell me it's not that helpful for them, but, you know, maybe you're special or maybe there's something about your makeup that you can prove me wrong. And sometimes patients alight to the idea they have to prove something wrong. So give them something to prove wrong. Predict they're not going to get better and let them prove you wrong. So you can actually manipulate some of the interaction and some of the things that you can do. Some of the interaction in such a way as to capitalize on the placebo effect. And that's why I'd say you sneak in some psychotherapeutic principles to maximize the outcome. It's in our book. Thank you. Thank you. Thank you so much for a very stimulating set of ideas. This is probably not set right because I'm not hearing. Anyway, it does reinforce this long thing I write at the end when I have a plan. I have 16 ideas there, many of which I never get to because something else intervenes. But I do appreciate that there's all this ideation going on about where we're going to go. I have a group of patients that are trauma fast, mostly childhood. And their scores are inevitably, every time I see them and I do measurement-based care, my program allows me and I get it every time, which is probably too much. Anyway, I check it before I see them and I'm like going, oh, my God, they're terrible. And almost inevitably, I see their scores and I forget how high they always are. And I look back and I know they're always that bad. And they'll come, oh, I'm doing pretty good. And you're like, 21, you know, and the HQ and the anxiety, GAD, you know, way up there. And so I struggle with kind of like other than patient self-report, maybe just in a different self-report system. But by and large, it's really hard to make their measures come down, yet they might be better. And so just some thoughts about that as well as maybe with that particular group, do you have any thoughts? Yes, thank you. So speaking of moderators, I can't think of a more challenging moderator than childhood trauma. I once asked Charlie Nemeroff, so what's the best treatment for depression with severe histories of trauma? And he said there is none. I said, well, come on, what do you really do? He said, no, really, I mean, there's really no good treatment. You really just have to sort of get in there with them on this. So this is where I would invoke the targets of treatment and the goals of therapy. So we may not necessarily meaningfully lower your PCL-5 score or necessarily treat the totality of your ailment, but let's define some of the things that are important to you. You know, maybe it's your sleep-wake cycle. Maybe it's some aspect of quality of life. Maybe we can make an inroad here. And almost to divert the patient's attention away from the comprehensiveness by just plugging away in little bits and pieces, if I can make an inroad here. And then we celebrate the Dickens out of that. Oh, my God, you got up two hours earlier. Oh, my God, I didn't expect that would happen. Wow. And do everything you can to then sort of inspire some sense of enthusiasm on their part and generate or ignite some momentum on their part. These are very, very hard, I think, challenges to confront, but the piecemeal approach of identified targets of treatment is how I think of it. Do you want to add anything? No, that was wonderful. Okay, that was wonderful. There you go. Hard patience. And we acknowledge that. I have to be frank with you. This is a challenging problem. I had a patient I saw last week. I began by saying we have to consider the possibility of futility. There have been a number of things. They haven't worked. You may not get better from this, but I'm going to be here with you to guide you through. I then proceeded to enumerate seven treatment options. Every one she shot down for one reason or another. And at the end I said, you know, we have a dilemma. On the one hand, I'm talking about you live with this, and yet here's seven things, but you found things that were not ideal. They're very imperfect. We live with imperfect options. So let's try to broker the fit as best we can. Thank you. I enjoyed the presentation. Are any of you using artificial intelligence in terms of the possibility of predicting outcome? And I had heard that it may have some utility as far as predicting placebo effect. Yeah. I don't have access to it, no. My own intelligence is quite artificial, but I think where we're at at the moment is there are some studies that have tried to capitalize on big data to make inferences about groups. And that's the probabilistic model that I was talking about before. So, for example, there was a post hoc analysis using the STAR-D sample as what they call a training sample. You take a portion of those patients, and it's almost like a big regression model in statistics where you say, all right, what are the predictors that I can identify? Maybe what are the top three that I can infer from the group? And then say, well, now if the patient that I see tomorrow has those three characteristics, how fairly can I infer probabilistically? Since these are the three top predictors in STAR-D, maybe they'll be deterministic predictors for you. It's highly imperfect. I'd say another example is the Framingham system itself for predicting heart disease. The Framingham study has been collecting zillions of patients over decades from which they can say, here's the chance you're going to have a heart attack. I can't guarantee it, but I can enumerate all these risk factors. So, to me, the most honest thing we can do with patients at the moment with artificial intelligence is to say, what inferences can we probabilistically draw from a group, and how does that translate into you as the individual and fill in the missing pieces? And I hope we'll see more of that as the field expands. AI is meant to learn from itself over time. So as the collective database gets bigger and bigger, we will hopefully be able to make even better predictions. Thanks for an excellent talk. I wanted to ask about something that Professor Stahl touched on briefly, which was antipsychotic polypharmacy, which I think is a bit of a dirty word these days. But I guess when I first started out, I was a real monotherapy evangelist. Over the years, with patients who are very treatment-resistant, and clozapine is contraindicated, or they just don't want to be on it, you put them on two antipsychotics, and it does seem to help. So, I mean, I wonder if that's because, the reason that we don't really use that too much is because of lack of evidence. But, you know, it's also possible that there just aren't enough studies on it at this stage. So I guess, yeah, I was wondering what your position is on that. Well, there's more than one way to have your D2 receptors blocked. You can do 50% of them with drug 1 and 30% of them with drug 2, or you can use plasma drug levels to move drug 1 up so much that it actually gets 80%. And because that's simpler, we've tended to want to do that. Now, you can put two drugs together and, you know, you end up having sometimes more side effects because you have the side effects of both, but you can get the side effects from high dosing too. So I think that we do that. The combination we like the most is clozapine plus some other agent, usually not a pain because you can't tolerate it too well, but it's like either haloperidol or one of the risperidone, paliperidone, because clozapine doesn't really work, we don't think, by D2 antagonism. It's got some other mysterious effect. It doesn't really block that many D2 receptors, and whatever it does is great. But sometimes you don't respond to clozapine. And then our most common polypharmacy is to add, let's say, high-dose risperidone or high-dose haloperidol. And we also like giving it with Adipo to clozapine. Adding two drugs together, we usually say you add a pain and a pain and a don't and a don't, because they don't tend to tolerate together so much. So some of it has to do with tolerability. And we just know, believe me, I dare you to try six patients that are poor responders to standard doses, get a blood level, and move them on up to point of futility, and I want an e-mail if it didn't work in one or two of those cases. This works so well that we wrote a book about it. Then we actually have conferences, first of all, in the California prison system, because we're doing it in the state hospital system. Then we're having conferences to Australia. Everybody here from Australia? Actually, do you know Ellis and those guys in Sydney in the forensic units? So Australia is doing this. And when we started, they said we can't get a blood level in Australia. We can't even do them anywhere in the country, and they finally found it, and it's not very expensive, and they got it. So they're doing it there. We're doing another conference every quarter in Ireland, and we're trying to proliferate. It really does, I mean, it just works. And we're not talking about ambulatory patients that go to community mental health centers every month on time. We're talking about patients in jails and prisons and state hospitals and who are very sick and mostly noncompliant. And in that group, when simple drugs don't work, that's why we just use one, because we've got experience with it. We're confident with it. There's more than one way to skin a cat, and if it works, God bless you. I don't have any problem with it. But, you know, I just want the patient to get better, and so this is a tried and true pathway. That's why we say it. You've convinced me, so I'll give it a try and send you an e-mail if it works. I have a handful of patients that I have on two antipsychotics because I ran out of other things to possibly do. But we had several thousand patients in the STEP-BD study, and we looked over the five years we took care of them who got antipsychotic polypharmacy, and it really was an operational measure of futility on the doctor's part. And the best we could determine, it resulted in more side effects without increased chance of benefit. I do not have the experience in schizophrenia to talk with that kind of authority, but I can't say that outside of an occasional case I can justify it for people with mood disorders. Thank you, guys, for the presentation. One of the things that came up in the presentation was kind of time is brain, and there are some diagnoses where we classically miss the boat as docks, whether that's not using Clostridol in psychosis, lithium in bipolar, or plazodone in anxious depression. Are there any other common diagnoses where we, as doctors, end to fail to provide optimal therapy and then negatively impact their prognosis because we're delaying treatment? I think this notion of candidacy and sort of matching my HGTV approach to, you know, I have the right house for you. So the more characteristics you've identified that would conform to a profile, I don't know if we're missing or making an error of commission as much as omission. So if I see a pure euphoric expansive first episode manic, I think it would be an error to not at least bring up the role of lithium. If I see a melancholic depressed patient with neuropathic pain, I'd be in error to not at least talk about a tricyclic antidepressant as having particular value. If I get any patient from Steven Stahl, they're going on an MAO inhibitor based on that alone. So I would encourage more the dialogue rather than thinking is this an error as much as what are the characteristics that I can identify and cull together to really broker the fit, and then what are the options and their pros and cons. And this way you're sort of maintaining the fluidity of thinking as opposed to saying the message we don't want to convey is there's one and only treatment for this. Sometimes there is. There are certainly some patients where, you know, clozapine gets you better and nothing else or ECT works for you and that's all we've found. So I think if we reach that end point, I think we want to maintain the sense of fluidity based on those characteristics. Hi, my name is Tamiya Bhutani. I'm an inpatient geriatric psychiatrist at Mount Sinai. Thank you so much for the talk. I did appreciate kind of hearing some supportive words regarding morale because I'm in my first year of attending hood and I honestly feel like my morale has never been lower. So I was just curious, you know, how some of the take home messages from this talk could be adapted to the inpatient setting where we don't have the same luxury of time, you know, to even give something a medication a full two weeks to see if we're getting, you know, an adequate response. My opinion, based on being a recovering inpatient attending myself, is to sort of put yourself in the context of your role in the patient's life. So in some ways, you're a little blink in their trajectory of things and you may not at all be launching them on the trajectory for a long term treatment, but you might plant some seeds. For example, Mr. Smith, I'm sorry to see you here again. You were just here a month or two ago and by my appraisal, it sounds like you were having a hard time staying on the medicine that you were on and that led to your coming back in the hospital. Remember last time you were here, we talked about the idea of a long-acting injectable as maybe being a way to forestall a future hospitalization. I wonder if that might be something you want to think about now. No, thanks. Okay, maybe next time. But as a frame of reference or as a suggestion I might make to the outpatient clinician, again, with the notion of iterative treatment, it's not just you, it's your part of that system. So if I'm on the receiving end of your patient and you say, so we saw your patient, ooh, how difficult their problems are, here's what we thought about and you can treat the admission almost like a consultation rather than we definitively fix things. I hate it when a complicated patient gets hospitalized and I get a call the next day, they're so much better in the hospital because I don't know why that is, but it's not going to have much lasting value. So I, on the outside, would very much appreciate, you know, here are some things to think about implementing iteratively. So at the University of Pittsburgh, when I was a young attending, the more successful you got, the less often you came onto the unit. And one thing I would recommend for morale is to get your senior faculty involved in case conferences and rounding and doing consults on difficult patients and so forth. Another thing I think is your own peer support and peer supervision. And again, work on making this more enjoyable and more professionally satisfying because it can be back-breaking, I know. And then you don't want to transmit poor morale to the patient. So whatever you're saying, you want to maintain a sense of hope no matter what because there's no reason not to. I don't know what's going to exist two or three or four or five years from now, so being in the moment and being mindful can be very, I think, helpful in grounding that. I hate to say we're out of time. I always hated that phrase. But sadly, we must come to an end. Thank you all for coming, and we hope to see you again soon.

psychopharmacology

iterative treatment

personalized approach

N-of-one approach

decision-analytic matrices

shared decision-making

treatment resistance

machine learning

patient characteristics

pharmacokinetics

therapeutic drug monitoring

treatment outcomes

therapeutic relationship