Pitfalls in Antipsychotic Serum Level Interpretation

Goldberg and Meyer continue their discussion of therapeutic drug monitoring of antipsychotics in schizophrenia spectrum disorders, turning to the practical interpretation of reported results. Meyer explains why levels may be of limited value for agents such as iloperidone, ziprasidone, and quetiapine, and why evidence linking blood levels to clinical effects remains limited for several newer agents.¹ He describes how laboratory-reported ranges for antipsychotics can differ markedly from evidence-based thresholds, including a 5-fold spread in the upper limits reported for fluphenazine across laboratories. Unexpectedly high results, Meyer advises, call for assessing the patient, repeating the measurement, and documenting the clinical reasoning.² The discussion also covers clozapine assay error and the use of serial levels to interpret new adverse effects and long-acting injectable nonresponse.

Read part 1 of this conversation here.

Joseph F. Goldberg, MD: So suppose I am prescribing iloperidone and I am spooked by the box warning for QT prolongation. You mentioned earlier that we are not as concerned about QT—but would you say to the prosecuting attorney at the wrongful death suit, "Well, I did not get 2D6 genotyping on the deceased because I measured an iloperidone level and it was not high, so I had no reason to think that they would be at increased risk for QT prolongation"?

Jonathan Meyer, MD: So for iloperidone and most antipsychotics, there is really no literature about torsades. That is why I say we really just do not think about it too much. If there are other factors to consider, that is fine. But the one thing you can do is, of course, adjust dosing initially because you are starting, especially for iloperidone, at very tiny doses, 1 twice daily (BID), which even if they are a poor metabolizer, maybe you are getting the equivalent of 2 BID. We do not think that is going to be a big problem. The other issue, though, is does iloperidone have reliable levels out there? And there are only a couple of drugs I do not include in that plasma level handbook; there are agents I do not want people to use because the levels are not well established. So that includes iloperidone. It also includes ziprasidone, for which I cannot think of any reason to use it. And also quetiapine.

The issue with quetiapine is not only does it have a lot of metabolic harms, even at the low doses used for sleep, like 25 or 50, which we thought was a freebie, it is not, especially in middle-aged individuals. You keep them on those low doses; there is a metabolic price to pay. But also quetiapine has a very short half-life, and you get a level 12 hours later, it is often not detectable, and then it might lead you to wrong conclusions because it has all been converted to norquetiapine. Those are 3 drugs where I say levels are probably not going to be very helpful, and particularly if you are treating schizophrenia, I would go somewhere else. And that is an issue, of course, for very new drugs. I do not know that there is any lab out there doing lumateperone levels, for example, for those who might use it for schizophrenia, or cariprazine. It is a complex drug, with most of the work being done by the metabolite didesmethylcariprazine. I do not know if there is a lab out there which is doing didesmethylcariprazine levels, and it just becomes kind of a practical issue there, where you just practice. If you had to use cariprazine, if it is schizophrenia, it has a lot of reasons to use it, you just have to do it based upon clinical judgment, like we did in the old days.

Goldberg: So, speaking of this, we could talk about a laboratory reference range. I have always understood or inferred that a laboratory reference range is Quest Labs or LabCorp or some company in-house determines a reference range. I do not know how to do this. It is from here to here, whereas a therapeutic range is someone went out and did some correlate to pharmacodynamic efficacy and/or tolerability. Tell me if I am wrong, and if not, how should I?

Meyer: You are right, but the labs are often very confusing in how they communicate this. If you go to a lithium level, that is a therapeutic range, and they are basing that on the literature. For antipsychotics, though, they are often mixed. Sometimes they are reference ranges, sometimes they are therapeutic ranges, and that is why, when I talk about antipsychotics, I say the least reliable source of information on interpreting levels and how high you might go is not what you get from the lab. It is probably the one instance where I say what you get from the lab is often grossly in error.

And I wrote a paper on this about 6 years ago, and I also have that table in the plasma level handbook. For example, fluphenazine. Again, most of you do not use it, but it has been very well studied, and we have a very good sense that there are very few responders below roughly 1 nanogram per mL, and the point of futility is probably 4, meaning less than 5% of the population would either tolerate or respond to a level beyond 4. Okay, 1 lab says the maximum level should be 2, and yet another lab says the maximum level should be 10. Let me tell you, I do not know. I have not met too many people who can tolerate a level of 10. They do exist. There is probably 1 guy out there. In any other area of medicine would you see a 5-fold variation of what laboratories recommend as an upper limit for a range, however they define that range—and that is a big problem for our field. And when I wrote that paper, I actually called the labs and I said, "Where did you come up with this number?" And the most typical answer was, "We don’t know. One guy looked it up 8 or 10 years ago, and he’s retired, and we do not know. But this is a number we are sticking with." I was really surprised.

Goldberg: Is it based purely pharmacokinetic stuff? I mean, like preclinically or from rat studies, and 95% of exposed people?

Meyer: It’s a mixed bag, and that is what’s confusing. For clozapine, it looks to be a therapeutic range, like 350. Some say 600 or 700. I do not think that number is evidence based. I think 1000 is more evidence based. For olanzapine, it is often a population kinetic range. Like for people on a certain dose, this is the range we expect. It is extremely confusing and not helpful, and that is why I say I wrote the book in particular to try to synthesize that information.

When you talk about, "I’m worried about being sued," well, you cannot go to Quest and say, "how did you come to this level?" because they cannot tell you. They just say, "This is what we are giving you." And at least in the book, I walk you through my logic very clearly. This is why I say it is not unreasonable to go to a clozapine level of 1000 if tolerated. Here are the data. Here are the papers that support it. Here are real-world data from data sets in countries where they use clozapine, showing that there are people walking around with levels between 600 and 1000, and seemingly, Joe, they are not all dropping dead.

Goldberg: So, recapping, one reason for getting a level or levels, as you described so clearly earlier, is around adherence, not just at this static moment in time, but to get some sense over time. A second is, I guess, wiggle room. So you are above the threshold, but you are still symptomatic, and if you are tolerating it, there is room to go up.

How about thirdly, in terms of toxicity? Broadly speaking, I mean, you were kind of saying earlier most of our modern antipsychotics, the toxicities would be the drug-induced Parkinsonism (DIP) type of things, as opposed to you are going to drop dead from sudden cardiac death, right? So if I measure a ziprasidone or lurasidone level and it comes back through the roof, should I be worried that sudden cardiac death is right around the corner?

Meyer: No, because we just do not see that with our modern antipsychotics, even with clozapine. It is just not something we see, which is good, but there are other tolerability reasons. I mean, these drugs have been out for a long, long time. Even lurasidone has been out over a decade and a half, and ziprasidone, which was the one that kind of spooked people, has been out for a quarter century. Let me tell you, I do not think there are any credible reports of torsades with ziprasidone. So it is very reassuring.

On the other hand, let me give you a couple of comments. When you get levels that come back which are high, the first thing you do is assess the patient. Now, if the patient is not in your office—for example, they are on clozapine, you give a level back of 1500. This happened to me once, and I was at the VA. I called the guy up. I had just seen him in my office, you know, within a couple of weeks. But, you know, who knows what the hell he was doing? So I call him up and say, "Hey, Mike, how are you doing? Your level is a little high." He is like, "I’m fine." "What are you doing?" "You know, I am eating lunch." And I just went through the questions of what you might expect with somebody with a really high clozapine level. Dizzy? Are you having trouble, you know, moving your bowels? "Nope, I am fine." Like, okay, level is a little high. I am just going to check it again. And so you have to first assess the patient and see 1 thing: does it even make sense? And I will give you another example. A lot of labs have trouble doing clozapine levels because the reagents are homegrown, and you see a lot of noise. You, in fact, even see false positives for clozapine levels, unless you are getting the new modern assay. There is a company called Saladax, which has an assay which goes right on the laboratory analyzer. No human can screw it up, but that is not how most laboratories work. And often what happens is, well, I had this guy on 500. He was an inadequate responder. I went to 550. The level on 500 was, let us say, 700. The level on 550 comes back 1300. Well, that makes no sense, and somebody screwed up in the lab, most certainly. But the first thing you do is assess the patient. "Hey, Jose, how are you doing? Any side effects?" "I’m fine." "Okay." So you say the level does not correlate with what I am seeing clinically, and then your hypothesis is this does not make sense kinetically either, unless somebody slipped in some fluvoxamine on me when I was on vacation. So I am going to document—always document your logic. I assess the patient; he is having no adverse effects. The level does not make sense with the small change in dose. I think this is lab error, and reorder a level for tomorrow. So that is the most important thing, is you assess the patient.

We get these levels back high all the time on first-generation agents. We think the point of futility for haloperidol is 18. There are lots of people walking around with levels in the 20s and 30s, and the clinicians call us. And the first question we ask them is, "How do they look? Is he stiff as a board? Is he restless?" No, we do not say do not sweat it, but we say, okay, you probably do want to back out of this. Deprescribing is a very slow process because you may have induced, as we call it, some supersensitivity. You cannot just chop his dose in half; he will be very ill. But just document that he is tolerating that. We are not particularly concerned about the high level because he is clearly tolerating it. We are going to try to remediate this, but we have to do this solely for his psychiatric benefit. But your documentation should talk about your logic. That I assess the patient. Now, if this guy had a lot of side effects, I would say, okay, go on. We have overshot his tolerability threshold. We are going to have to move down a little more briskly, but it is often the person who is completely asymptomatic. And then if they are an outpatient, you want to figure out, "Hey, maybe they took their dose just before they went to the lab."

This happens all the time. It is one reason we try to avoid BID antipsychotics, because they cause more side effects, for one thing, for no therapeutic benefit, and you have trouble getting levels. So maybe this person who had forgotten to take his dose the night before, then took it an hour before their appointment, because he knows Dr Goldberg will yell at him if his level is not therapeutic. So, and then you get this level which does not make sense, and you just call him up and say, "Hey, what is going on?" You know, you just have to go through the checklist.

Goldberg: I do this with lithium all the time. If I have not gotten a level on someone in a while, and I say let us get a level, I kind of look them in the eye and say, you know, if by chance you have been missing your dose for a while, do not double up on your lithium the day or 2 before. It is not like you are going to get in trouble for this, but I want to make sure I can interpret it correctly. So, between you and me, people often do not take their medicines the way they are prescribed.

Meyer: Yeah, but that is the whole point. When they are outpatients, you have no idea what is going on. You know what is going on in a hospital. For outpatients, when you see high levels, especially where you have a prior level which was kind of reasonable, the first thing you do is call the patient. You know, occasionally somebody was taking the lithium correctly, and somebody gave them something which kinetically interferes with lithium clearance. They got a new antihypertensive, and they did not call you. They were taking a bunch of nonsteroidals, and the renal function was not great to begin with, and the level comes back 1.5. And then you call them up and say, "Hey, level is a little high," and they are like, "Yeah, you know, Dr Goldberg, I am a little shaky here." Like, okay, this is probably a real thing. But if it does not make sense and they are completely asymptomatic, then there is either a dosing error or, for clozapine, which is really the poster child for that, there is laboratory error. Document that and say I am going to repeat that.

I will give you folks another little pearl. This has to do with lithium. Lithium has to be drawn in a red-top tube. If they draw it in a tube which has heparin in it—you know that heparin has several forms. There is sodium heparin. Also, there is lithium heparin in that tube. And every once in a while, you are going to get a lithium level. It is going to come back like 3, and you will freak out. I will tell you the first thing you are going to do is call the patient and say, "Your level came back kind of high." Do not tell them they are going to drop dead. How do you feel? If you are walking around with a level of 3, you feel really ill. You will not be asymptomatic, and then you will document, "I think the lab screwed up with the tube that was drawn." Usually, it is a new phlebotomist who grabbed the wrong tube. It was a green-top tube instead of a red-top tube, and it had some lithium heparin along with the other forms of heparin, and obviously gave you a falsely high level.

Goldberg: Wow! All these years. So I should be writing "lithium level (red)."

Meyer: A competent phlebotomist looks it up. They know the righttube. The last case I heard was somebody who was brand new, perhaps not well trained.

Goldberg: When a side effect emerges that we know is dose related, the tremor or otherwise, how much should I, as a clinician, be reflectively thinking, "Gee, I am seeing this side effect. I had better check a level to see if it has been drifting up"? And you are going to say it depends on the drug. You are going to say it depends on the patient. So, with that in mind, should my mindset about the emerging adverse effect be, "Let us see where the level is before I go further"?

Meyer: Well, for medications with relatively short half-lives of a day or similar, I increase the dose. It is at steady state usually within a week. And if you are going to get, let us say, drug-induced Parkinsonism, if it is going to happen, over a relatively short time frame, it would be uncommon for oral antipsychotics for somebody to develop DIP after many weeks or months. Now, if they do and the dose has not changed, one thing I will make sure is they are taking the right dose. I will make sure there have been no kinetic issues, and then maybe I will get a level, especially if they have a baseline, just to see what is going on. And that is why, typically, if I have people on oral antipsychotics, just like for folks on lithium, I might get a level every 6 months because occasionally people are not adhering. Occasionally, as with lithium, you can kind of nip it in the bud and say, "You know, your level is half what it used to be. What is going on?" You can get on top of it. There are some older individuals on a fixed dose of a D2 blocker who may age into their drug-induced Parkinsonism, and it is not a kinetic thing per se. It may just be pharmacodynamic. I have lost some of my dopamine neurons because I am not 40 anymore. I am now 65, and I am now more sensitive to dopamine blockade. That is fine. At that point, I probably would talk to them about shaving back their dose a little bit and see if we can deal with this adverse effect. For lithium, we do see some cognitive complaints. They are not always related to the level. I mean, there are some people who complain with levels of 0.6, and some people do not get them until a level of 0.95. The reason you want to know the level is to kind of get a sense of how much margin of error I have in shaving down the dose, which is probably what I am going to do, assuming there are not other meds on board causing the problem.

You know, if they are on quetiapine and clonazepam, I think I might get rid of those first. If they are depressed, I would probably treat that. But there are people who get cognitive complaints on lithium. It happens, and I would get the level just to see where it is at. If it is kind of where it has been—and it is usually people who have been newly started who are complaining. Now, again, if somebody has never had a problem with a level of 0.75 and now they have a new problem, I would certainly check the level just for the reasons you say. Maybe their renal function has gone down, and the level is now 0.9. But if the level remains the same, again, there are changes in your blood-brain barrier as you get older, and what you see is the brain penetration for lithium in particular may go up as people age, even if they tolerate it at 0.75 when they are 40. Well, now they are 65, and they are getting more bang for their buck upstairs. I will explain it to the patient and say, "Now that you are older, and obviously you are starting to complain of some things, let us dial back your dose because we are having a different dose-response curve than when you were younger."

Goldberg: It is nice having data over some number of years to compare you to yourself. So if you have always lived more or less within, say 10% of some level, and I am seeing a deviation, it invites inquiry. Maybe we need to look more carefully at kidney function. Maybe we want to do an albumin-creatinine ratio, etc.

Meyer: Yes, if the level is the same, but the complaint goes up, what is the only thing that has changed? Their blood-brain barrier, and that is my point. So the patient says the level is always 0.7-something, and they are very adherent, kidney functions remain the same. But now they have a new cognitive complaint because their blood-brain barrier has changed as they have gotten older, and you will sometimes see this occasionally, and this is lithium specific, really. Where somebody is already at 0.7, they are fine. Now, as they have gotten a little older, the level is still 0.7, but now they are developing a little bit of a tremor. Why is that? They are getting more central nervous system penetration. That is it. That is the only variable that has changed, assuming the level is the same.

Goldberg: Unless I have a spectroscope sitting around, they are not going to really measure your brain lithium.

Meyer: Yeah, it is hard to do the magnetic resonance spectroscopy, you know, unless you have a cyclotron around. I cannot fit one in my office. But the nice thing about having serial levels is, if the level stays the same and there is a new complaint, something else has to be going on, and that is why, for the clinicians out there, it makes your life easier. Yes, it is about the patients, fine, but it is about making your life easier because you no longer have to guess, and especially when you are taking care of people who are severely mentally ill, especially schizophrenia spectrum disorders, you do not want to guess. If someone is not doing well or someone is not getting the response you expect for a given dose, get data. It allows you to make so much better decisions, and it makes your life easier because now you have the information to inform your decision as opposed to guessing about why they are not doing well.

And I will say this: this is also true for long-acting injectables. I am less concerned about adherence, but if you came to me and said, "Hey, I have this person on max-dose subcutaneous risperidone, and they are having inadequate positive symptom response, and you know, they are at steady state, they have been on for a few months," I would say, "Let us get a level and see what is going on. Maybe they are clearing it more quickly than the average person." Once I have that number back, then I can make a good decision. You know, if the level is relatively low for the dose, I will say, look, they are just clearing it, and we will have to find a way to get them more. Maybe we will sneak up the interval. We will find ways to give them more exposure. If the level is what we expect, then the question is, all right, what are we going to do as our next step? And then that depends on the patient and the circumstance. But at least now you know exactly what is going on, and you do not have to guess.

Goldberg: Wow. So, last question before I let you go, Jonathan. Whom should you not get an antipsychotic level on? I am going to guess your answer is going to be someone who is clinically stable, for whom there is no chance of indication to measure a level.

Meyer: I will say if I am taking care of people living with schizophrenia who are on oral meds, I probably get a level every 6 months just to see if they are taking it. Sometimes clinicians think, "Well, they were not taking it. I know immediately." Not true. The American Psychiatric Association, way back in the 1970s and 1980s, actually took people off of meds as part of dedicated studies to see if they could minimize the risk of tardive dyskinesia. Well, they did not minimize it really, and they made a lot of people relapse. But the point is that even at the end of a year, there were people on no drug. There were still 20% to 25% who had not relapsed, and for people who are very stable, sometimes they can go without a drug for long periods of time, weeks, months, and that is why I say having, you know, every 6 months, even on an oral antipsychotic, just kind of gives you that baseline. If the clinical status changes, now I know what their baseline is, and I can make sense of the situation. If they are adherent, then maybe there is something else going on. They relapsed on stimulant use. There is a major psychosocial stressor. So that is my comment about that. Now, if you are using antipsychotics for other purposes like bipolar depression, I would say maybe there is not as much of a reason to do that. Those people are still nonadherent. You could use levels to track adherence if you wanted to, but I cannot give you as much dose response. It would be more: Are you taking it or not? Which is sometimes fine, and that can occasionally be useful for people who have a history of nonadherence.

Goldberg: But if the laboratory's reference range, as you said before, comes out of who knows where, it would be wrong for the clinician to say, "Oh my gosh, I got your cariprazine level back and it is low. You had better raise your dose." That would be a misuse. That would be a misuse of interpretation.

Meyer: Well, regardless of the indication, as I said before, those therapeutic response thresholds are just population data based upon mathematical analyses. No one calls me, as I said, at the state hospital saying, "Hey, my clozapine patient did great with a level below 350." Nobody cares. I would like to know what the level is because if they do not do well in the future, like when they get out and do not take their meds or start smoking cigarettes, now I know this is the level that kept you stable. If you are no longer the same clinically, let us figure out what is going on.

Goldberg: Any final thoughts for us?

Meyer: Do not guess, measure.

Dr Goldberg is a clinical professor of psychiatry at The Icahn School of Medicine at Mount Sinai in New York, NY and the immediate-past president of the American Society of Clinical Psychopharmacology.

Dr Meyer is a voluntary clinical professor of psychiatry at the University of California, San Diego.

References

1. Schoretsanitis G, Kane JM, Correll CU, et al. Blood levels to optimize antipsychotic treatment in clinical practice: a joint consensus statement of the American Society of Clinical Psychopharmacology and the Therapeutic Drug Monitoring Task Force of the Arbeitsgemeinschaft für Neuropsychopharmakologie und Pharmakopsychiatrie. J Clin Psychiatry. 2020;81(3):19cs13169.

2. Meyer JM, Stahl SM. The Clinical Use of Antipsychotic Plasma Levels. Cambridge University Press; 2021.