You are here: Home / Courses / Jose de Leon: Training psychiatrists to think like pharmacologists / 27. Evidence based and personalized medicine. / Jose de Leon’s response 2 to Donald F. Klein’s response 2
Saturday, 26.09.2020

Jose de Leon’s response 2 to Donald F. Klein’s response 2

Jose de Leon: Training Psychiatrists to Think like Pharmacologists
27. Evidence-based versus personalized medicine.
Are they enemies?

 

            I would like to start by thanking Dr. Klein for taking time to keep discussing points with me. I am going to comment on seven issues he raised:

 

1) Being a utopian neighbor.  I have now read Dr. Klein’s 1974 article on the package insert suggesting making it more similar to a pharmacological monography. I apologize that I had not seen it before (I was in high school in 1974). I am grateful that Dr. Klein called me a “utopian neighbor” but, hopefully, Dr. Klein and I, as creators of utopian places, will have better luck than the creator of the first Utopia (1516), Thomas More, who was decapitated by the power of his time. If I get decapitated by the pharmaceutical companies or many of their subrogees who have infiltrated the current US academic psychiatric establishment, I am afraid that my head will not be a significant loss for the next important battle in US psychiatry: to convince the new NIMH director to drop the RDoC. I do not seem to have much influence since I tried to publish a letter to the editor in Acta Psychiatrica Scandinavica supporting a prior letter by Dr. Klein (2016) against the RDoC, but the editor rejected it. On the other hand, keeping Dr. Klein’s influential head in place as long as possible seems to me to be very important. As a matter of fact, I hope that Dr. Klein is able to convince the new NIMH director, who came from Columbia University, to drop the RDoC (de Leon, 2017).  

 

2) Shared admiration of Uhlenhuth. Dr. Klein expresses his admiration for his friend “Uhli.” I have to confess that until 2016 I did not know who Uhlenhuth was but, hopefully, I have partially resolved my ignorance by reading many of Uhlenhuth’s articles and including his name in a 2017 article (De las Cuevas and de Leon, 2017). The article is titled “Reviving Research on Medication Attitudes for Improving Pharmacotherapy: Focusing on Adherence” and includes in the abstract the following paragraph:

“There is little current interest in research into patients' attitudes toward medications. In the 1960s, psychiatric researchers including Uhlenhuth, Rickels and Covi focused on this area, but this research topic needs to be revived in the 21st century.” 

 

3) Giving Dr Klein too much credit as a mathematician.  I am afraid that I have to 100% agree with Dr. Klein this time; he does not understand the limitations of ANOVA or ANCOVA tests.  As my prior comments explain, these tests have two major limitations: 1) they are inadequate when dependent variables do not follow normal distributions, and 2) they provide significance results but no effect sizes. Effect size is the important piece for clinicians (Kraemer and Kupfer, 1996).

 

4) Stratification.  Dr. Klein says, “What raises my amateur statistical hackles is Dr. De Leon’s advocacy of initial stratification using pharmacological variables. It was unclear to me if that was in the context of Anova or Ancova.” Dr. Klein is going to have to forgive me for 100% agreeing with him a second time. I am afraid that, in effect, this is an amateur’s question. Stratification is not related to the statistical test; it is related to the design. One way of dealing with heterogeneity is by using modification of the statistical tests once the randomized clinical trial (RCT) has been completed. Stratification is a way of dealing with heterogeneity by modifying the design. Feinstein, a physician and the founder of clinical epidemiology, in his book “Multivariate Statistics” (Feinstein, 1966), has a section titled “Intrinsic value of stratification” (p. 505-6) which concludes with this paragraph: “Consequently, the main scientific point is usually missed in statistically oriented discussions of different methods and timing for multivariate analysis in randomized trials. If the results are to be used in the real world beyond the published document, clinicians must be able to determine what can be expected for individual patients who suitably resemble the pertinent clinical subgroups in reports of treatments. An appropriate, effective stratification is the best way to identify these groups.”   

            To provide an example of stratification, let’s assume that, for a narrow therapeutic drug, being a poor metabolizer (PM) is a major determinant of drug response. In this situation, PMs may be extremely prone to have adverse drug reactions (ADRs), while non-PMs may be at normal risk for ADRs. The correct way of handling this heterogeneity in drug response is to stratify according to PM status during randomization in an RCT. Two independent randomizations are needed: a randomization to the study drug or the control drug for the PM group and another randomization for the non-PM group. Moreover, it is possible to mask the effect of PM status by giving PMs half the dose of non-PMs by using pills with half the strength (0.5 mg instead of 1 mg). The goal of stratification according to PM status is to eliminate that variable as a confounder in this RCT.   

 

5) Relevance or lack thereof of the metabolic outliers (which Dr. Klein calls hyper- and hypometabolizers and pharmacologists call ultrarapid metabolizers [UMs] and PMs). As indicated in my second answer on April 29, 2017, the frequency of outliers varies per drug based on its metabolism, so I cannot give a specific answer that will address all drugs. I will use an example with which Dr. Klein is familiar, since he did some pioneer imipramine studies. Imipramine is mainly metabolized by CYP2D6 and CYP2C19. If Dr. Klein had done blood levels, which pharmacologists call therapeutic drug monitoring (TDM), I could have estimated the prevalences of UMs and PMs in Dr. Klein’s imipramine studies. In the absence of TDM and assuming that his studies 1) were done mainly in Caucasians and 2) imipramine treatment was not contaminated by other co-medications, one should expect to find around 10% of imipramine PMs (7% of CYP2D6 PMs and 3% of CYP2C19 PMs) who metabolize imipramine poorly and around 1.5% of CYP2D6 UMs who would metabolize imipramine abnormally fast.  To explain to Dr. Klein the imipramine dosing relevance of being a CYP2D6 PM, giving one of them 75 mg/day is roughly equivalent to giving a normal metabolizer 150 mg/day, or giving one of them 150 mg/day is roughly equivalent to giving a normal metabolizer 300 mg/day. The CYP2D6 UMs are much more complicated. The best pharmacogenetic guideline recommends not prescribing TCAs in CYP2D6 UMs (Hicks et al., 2013).

In naturalistic studies with high levels of co-medication, one should expect a higher percentage of outliers and a lower percentage of normal metabolizers. Risperidone is metabolized by CYP2D6 and CYP3A4. In a US risperidone naturalistic study with 537 patients (de Leon et al., 2005) which used CYP2D6 genotyping and paid close attention to co-medication, the prevalences of outliers for risperidone metabolism were high (Table 1).

 

Table 1. Prevalence of metabolic outliers in 537 risperidone patients                                      

Outliers                       Mechanism                             Prevalence                                                    

UM    

  Genetic reasons        CYP2D6 UM                           2% (13/537)

  Co-medication          CYP3A4 inducers                    8% (45/537)

PM     

  Genetic reasons        CYP2D6 PM                            7% (37/537)  

  Co-medication          CYP inhibitors                       31% (165/537)                                             

 

Some patients had more than one of these major metabolic abnormalities and after taking that into account only 57% (307/537) were normal in their metabolism with no major disturbance due to genetic or environmental causes, while 43% (230/537) were PMs or UMs. Regarding the clinical impact of metabolic outliers, after correcting for confounding variables, genetic risperidone PMs (CYP2D6 PMs) had an odds ratio (OR) of 3.4 of having relevant adverse drug reactions (ADRs) on risperidone and an OR of 6.4 of having a risperidone discontinuation due to ADRs. After correcting for confounding variables, environmental risperidone PMs (taking CYP inhibitors) had an OR of 2.4 of having a risperidone discontinuation due to ADRs (de Leon et al., 2005).  

       

6) Regarding data on drug-drug interactions (DDIs) in antidepressants and antipsychotics. There are two types of DDIs: pharmacokinetic and pharmacodynamic.

6.1. Pharmacokinetic DDIs. The data on pharmacokinetic DDIs is very limited because the DDIs of drugs approved before 1996 were not studied since the FDA did not ask for it. For the drugs approved after 1996, the data depends on the year of approval (the more recent drugs were better studied for their metabolism and DDIs) and the willingness of the company to carry out studies, whether clinically relevant or not. There is DDI data that can be used to orient clinical practice only when the company was willing to conduct studies using 1) clinically relevant dosages of inhibitors and inducers and 2) sufficient time to reach steady state on the studied drug. On the other hand, there are no good studies for orienting clinicians on how to manage DDIs when the company designed a DDI study to be negative by using low doses of inhibitors or inducers and/or short durations for effectiveness of the inhibitors and inducers (de Leon, 2014a;b).

By reviewing all data from in vitro to clinical data, we have developed tables for pharmacokinetic DDIs using correction factors. Inhibitors require a correction factor <1, which is equivalent to lowering the dosage. A correction factor of 0.5 is equivalent to dividing the dose in half. Inducers require a correction factor >1, which is equivalent to increasing the dosage. A correction factor of 2.0 is equivalent to doubling the dosage. The most updated table for the effects of pharmacokinetic DDIs from other psychotropic drugs on second-generation antipsychotics is freely available at

https://www.researchgate.net/publication/299978576_Assessing_drug-drug_interactions_through_therapeutic_drug_monitoring_when_administering_oral_second_generation_antipsychotics_Updated_Table_4_for_brexpiprazole_and_cariprazine.

            This table is an update of Table 4 from a 2016 article (Spina et al., 2016a).  The most updated table for the effects of pharmacokinetic DDIs from antiepileptic drugs on antidepressants is freely available at

https://www.researchgate.net/publication/312120498_Clinically_significant_pharmacokinetic_drug_interactions_of_antiepileptic_drugs_with_new_antidepressants_Table_4_updated_with_TCAs.

            This table is an update of Table 4 from another 2016 article (Spina et al., 2016b).  Dr. Klein mentions ketoconazole, one of the most powerful inhibitors among non-psychiatric drugs. We have no recent tables with correction factors for the pharmacokinetic effect of non-psychotropic drugs with inhibitory or inductive properties on antidepressants or antipsychotics, although Box 1 from a recent article on second-generation antipsychotics (Spina et al., 2016a) provides a narrative description of DDIs with non-psychiatric drugs. A 2005 article has some correction factors for the effect of non-psychotropic drugs with inhibitory or inductive properties on six second-generation antipsychotics (de Leon et al., 2005).

6.2. Pharmacodynamic DDIs. The pharmacodynamic DDIs of psychotropic drugs have never been systematically studied and, for the most part, clinical data does not exist, but these DDIs can be predicted by knowing pharmacodynamic mechanisms. Recently, we have been able to get an article accepted for publication (Spina and de Leon, 2017) which includes summary tables for clinicians describing pharmacodynamic DDIs between antiepileptic drugs and other psychotropic drugs. Two of the three reviewers of this accepted article correctly accused us of speculating since most of these DDIs have never been published. Unfortunately, when DDIs kill patients, it does not matter if the DDI has been previously published or not. We have published a case of a lethal pharmacodynamic DDI combining intramuscular haloperidol and ziprasidone; these antipsychotics have additive effects on heart potassium channel blockade (Wahidi et al., 2016). The worrisome issue is that if the company which produced thioridazine had listened to Tom Ban’s suggestion to better study the risk of Torsades de Pointes in the 1960s (Shorter, 2013), it might have been possible to develop the pharmacological science to possibly prevent deaths secondary to Torsades de Pointes during thioridazine treatment. As second-generation antipsychotics block the same heart potassium channels as thioridazine, the development of pharmacological science regarding thioridazine-induced Torsades de Pointes would have prevented deaths by the same pharmacodynamic mechanism caused by other antipsychotics (Wahidi et al., 2016). 

In summary, my impression after reviewing hundreds of deaths spanning 15 years in my state’s public mental hospitals and community programs is that pharmacokinetic and/or pharmacodynamic DDIs occasionally do definitively contribute to the deaths of psychiatric patients (Wahidi et al., 2016). My clinical impression after conducting TDM in hundreds of patients and analyzing many TDM databases from several countries is that some pharmacokinetic DDIs are clinically relevant in dosing. Co-prescription with potent inducers such as carbamazepine, phenytoin and phenobarbital requires significant dose increases for many antiepileptics, antipsychotics and antidepressants to avoid lack of efficacy. Co-prescription with potent-to-moderate inhibitors such as paroxetine, fluoxetine, fluvoxamine, bupropion and duloxetine requires dose decreases for many antiepileptics, antipsychotics and antidepressants to avoid ADRs.      

 

7) Regarding Dr. Klein’s case of a DDI between ketoconazole and bupropion. I could not find in PubMed any article on an interaction between ketoconazole and bupropion. That is not a problem if you believe that mechanistic thinking is an important means of applying scientific principles to psychopharmacology as I do (de Leon and De las Cuevas, 2017).  Ketoconazole is a potent CYP3A4 inhibitor, but in high doses it can inhibit other CYPs including CYP2B6 (Khojasteh et al., 2011). This is important because bupropion is mainly metabolized by CYP2B6 (Italiano et al., 2014). Knowing about bupropion metabolism is important for psychiatrists. For example, most psychiatrists are also not aware that bupropion should not be combined with carbamazepine, phenytoin or phenobarbital because they are extremely potent CYP2B6 inducers and CYP2B6 is extremely sensitive to induction (de Leon, 2015b). Imagine that your patient is on 150 mg/day of bupropion and you add a therapeutic dosage of carbamazepine. Once the induction has reached steady state (several weeks; up to 2 months), carbamazepine induction would have reduced serum bupropion and metabolite concentrations by 90%, leaving 10% of the baseline serum concentrations. This means that if the psychiatrist wants to keep the same effects of 150 mg/day of bupropion on carbamazepine, he/she needs to prescribe 1500 mg/day, which may give the dispensing pharmacist a myocardial infarct. This is why it is better to avoid co-prescribing bupropion with potent inducers (Italiano et al., 2014).  To conclude, oral ketoconazole use has been restricted in the US, but as Dr. Klein described, it was not a good drug to combine with bupropion. 

 

References

De las Cuevas C, de Leon J. Reviving research on medication attitudes for improving pharmacotherapy: focusing on adherence. Psychotherapy and Psychosomatics 2017; 86: 73-79.

 

de Leon J. False-negative studies may systematically contaminate the literature on the effects of inducers in neuropsychopharmacology. Part I: focus on epilepsy. Journal of Clinical Psychopharmacology 2014; 34: 177-183.

 

de Leon J. False-negative studies may systematically contaminate the literature on the effects of inducers in neuropsychopharmacology. Part II: focus on bipolar disorder. Journal of Clinical Psychopharmacology 2014b; 34: 291-296.

 

de Leon J. Jose de Leon’s Final Comment. Thomas A. Ban: RDoC in Historical Perspective. INHN web page 2017 (in press).

 

de Leon J, Armstrong SC, Cozza KL. The dosing of atypical antipsychotics. Psychosomatics. 2005; 46: 262-273.

 

de Leon J, De Las Cuevas C. The art of pharmacotherapy: reflections on pharmacophobia. Journal of Clinical Psychopharmacology 2017; 37:131-137.

 

de Leon J, Susce MT, Pan RM, Fairchild M, Koch WH, Wedlund PJ. The CYP2D6 poor metabolizer phenotype may be associated with risperidone adverse drug reactions and discontinuation. Journal of Clinical Psychiatry 2005; 66: 15-27.

 

Feinstein A. Multivariable Analysis. An Introduction. New Haven, CT: Yale University Press; 1996.

 

Hicks JK, Swen JJ, Thorn CF, Sangkuhl K, Kharasch ED, Ellingrod VL, Skaar TC, Müller DJ, Gaedigk A, Stingl JC; Clinical Pharmacogenetics Implementation Consortium. Clinical Pharmacogenetics Implementation Consortium guideline for CYP2D6 and CYP2C19 genotypes and dosing of tricyclic antidepressants. Clinical Pharmacology and Therapeutics 2013; 93: 402-408.

 

Italiano D, Spina E, de Leon J. Pharmacokinetic and pharmacodynamics interactions between antiepileptics and antidepressants. Expert Opinion on Drug Metabolism and Toxicology 2014; 10: 1457-1489.

 

Khojasteh SC, Prabhu S, Kenny JR, Halladay JS, Lu AY. Chemical inhibitors of cytochrome P450 isoforms in human liver microsomes: a re-evaluation of P450 isoform selectivity. European Journal of Drug Metabolism and Pharmacokinetics 2011; 36: 1-16.

 

Klein D. RDoC is adverse to scientific creativity. Acta Psychiatrica Scandinavica 2016; 134: 452-454.

 

Kraemer HC, Kupfer DJ. Size of treatment effects and their importance to clinical research and practice. Biological Psychiatry 2006; 59: 990-996.

 

Shorter E. The Q–T interval and the Mellaril story: a cautionary tale. INHN web page 2013 inhn.org/controversies/edward-shorter-the-q-t-interval-and-the-mellaril-story-a-cautionary-tale.html

 

Spina E, de Leon J. Potentially clinically relevant pharmacodynamics drug interactions between antiepileptic and psychotropic medications: an update.  Current Pharmaceutical Design 2017 (in press).

 

Spina E, Hiemke C, de Leon J. Assessing drug-drug interactions through therapeutic drug monitoring when administering oral second-generation antipsychotics. Expert Opinion on Drug Metabolism and Toxicology 2016a; 12: 407-422.

 

Spina E, Pisani F, de Leon J. Clinically significant pharmacokinetic drug interactions of antiepileptic drugs with new antidepressants and new antipsychotics. Pharmacological Research 2016b; 106: 72-86.

 

Wahidi N, Johnson KM, Brenzel A, de Leon J. Two sudden and unexpected deaths of patients with schizophrenia associated with intramuscular injections of antipsychotics and practice guidelines to limit the use of high doses of intramuscular antipsychotics. Case Report in Psychiatry 2016; 2016: 9406813.

 

 

August 24, 2017