Abstract

This commentary explains the pharmacology behind the evidence that clomipramine is the most potent and effective of the so-called SNRI antidepressant drugs, and probably the most potent non-MAOI antidepressant drug available — many expert psychopharmacologists agree with that.  It can be effective at doses as low as 10-50mg.  PET studies indicate that even at 10 mg it produces 80% blockade of the human serotonin transporter (hSERT), thus reinforcing its potential to produce improvement at low doses.  It is regarded by many as the most effective drug for obsessive-compulsive disorder.  In many patients, when used in appropriately adjusted doses, it has a similar level of side effects to more recent antidepressants; however, the dose needs to be adjusted slowly and carefully, sometimes with the aid of therapeutic drug monitoring.  It should still have a prominent place in our therapeutic armamentarium.

Introduction: 50^th birthday tribute

[this commentary should be read in tandem with the commentary specifically devoted to TCAs]

Clomipramine (CMI) is usually thought of by doctors as a treatment for obsessive-compulsive disorder (OCD).  However, it is a potent antidepressant with true dual SNRI action (see tables of comparative receptor affinity data below, and my TCA review).  There is better evidence for clomipramine’s superior effectiveness for refractory depression than for most other drugs [1, 2].  The frequency with which it is ceased because of side effects (when used correctly) is little different to SSRIs or other new antidepressants [1], and probably less that venlafaxine (which has been the most widely used SNRI for nearly 3 decades).  However, the evidence for venlafaxine being clinically effective, as an SNRI, is in doubt.

CMI may be less toxic in overdose than venlafaxine (see my TCA review for further details [3]).  So much for the overhyped trope that new-is-better.

Therefore, with all those positive attributes, why do few doctors use clomipramine for depression?  First, ignorance about it; second, an accident of history; third, profits and advertising (see separate commentaries in ‘Bias in science’ section).

The endlessly repeated generalisation that TCAs have more side effects and are more toxic than SSRIs, or other newer drugs, is something of a misrepresentation: it is founded on a basic error of logic.  ‘TCAs’, which should not be lumped together as a class, are defined by their physical structure (tricyclic, three fused rings) not by their pharmacological properties, which are varied.  Therefore, to compare ‘TCAs’ with other similarly classes like SSRIs and SNRIs, is akin to trying to compare, not just apples and oranges, but a basket of mixed vegetables with a basket of mixed fruits.

[It is helpful to understand what one means by saying ‘it is the most effective antidepressant’. That applies to people who suffer from the endogenous, biological, melancholic, types of depression (and yes, it’s also effective for phobias and OCD).  The comment one sometimes encounters that ‘the best antidepressant is the one that works for you’ is unhelpful because it does not tell you in advance which one is most likely to work for you.  It is not logical to choose one at random, as if it was a lucky dip; there are too many possible choices.  Such a strategy would be expensive and time-consuming.  Sometimes people say doctors treat everyone the same, and sometimes they are thought to treat people as guinea pigs — and both of those statements are partly true.  I reframe this by explaining that there are general characteristics of illnesses that allow one to make a reasonable prediction that illnesses with certain characteristics will likely respond to a particular sort of treatment.  On the other hand, once you get to that stage, you find that, of those different treatments (say SSRIs or TCAs), some of them suit one person, and some another.  That is, as I used to say to patients, where you need to be your own guinea pig to find the medication is most efficacious for you.]

By 1980 there were many specialists in refractory depression in the UK who regarded it, along with tranylcypromine, as more potent than other drugs [8-11].  In his book ‘The Antidepressant Era’ Professor David Healy [12] makes the same observation, as did other eminent pioneers like Paul Kielholz, and the Noble prize-winning Arvid Carlsson.

I have had extensive experience using it where other antidepressants have failed and fully concur with the view that it has superior efficacy.

Without going into the intricacies of scientific methodology it is easy to appreciate that dealing with a purer sample of drug-responsive patients (i.e., people have already responded to clomipramine) produces a clearer picture of the relative potency of the two drugs, especially when the same patients regained their previous improvement after going back on clomipramine.  That initial experience, subsequently reinforced, persuaded me that venlafaxine was less effective.  That is the sort of information that RCTs simply cannot generate, which is one of the reasons I argue that RCTs are overrated and have attained too great an influence over treatment decisions [see my specific commentary on the epistemological problems of RCTs].

The top websites returned by a Google search are generally incomplete and misleading.  Wikipedia can be a satisfactory place to start, for non-medical people, but the Wikipedia entry for this drug leaves something to be desired.  Wikipedia does quote my review of the TCAs [3], but it then goes on to contradict some of the information with references that are inappropriate or out of date.  One can only try.

History

It seems ironic that despite the new drugs that have arrived in the last four decades the best evidence for superior efficacy and/or a unique profile of action is for the oldest ones (e.g. clomipramine, clozapine, tranylcypromine, valproate, and lithium).

Clomipramine was the first, and for some while the only drug (till zimelidine/fluoxetine ~1986), that was effective for obsessive compulsive disorder.

The fact that it is effective for OCD, where the previously available TCAs like amitriptyline and imipramine were not, is one important piece of evidence indicating clomipramine alone is a clinically effective SRI.  One still frequently reads the opinion that TCAs generally are dual-action SNRIs.  There are reasons to reject that proposition, these are discussed in detail in various of my published papers, especially [3, 14], and our most recent summary of the evidence is here [15].

Here is a list of some early references concerning clomipramine: [5, 7, 16-27].

Clomipramine has been in use for more than 50 years; for all that time it has possessed three of the properties that are promoted to increase the sales of newer, and expensive, drugs: viz. it is a serotonin reuptake inhibitor and a noradrenalin reuptake inhibitor, a 5-HT2A blocker, has a modest but useful sedative effect via H1 antagonism, and its little tickle of antimuscarinic activity may even be relevant also [28], Carlsson thought that.

As Carlsson said to David Healy in interview

Pharmaco-kinetics and ‘therapeutic’ plasma levels

DUAG CMI. [29] [30, 31] summarized by CMI better in melancholia [2]

Kragh-Sørensen [2]. 

Superior efficacy, which ones? [32]

All the TCAs are highly lipid soluble and are rapidly absorbed (peak levels are usually reached within 2 hours) with extensive tissue distribution (volumes of distribution from 5-20 L/kg. [33, 34].  Plasma concentrations for therapeutic effect are stated by various sources to be around 50-300 ng/ml (molecular weights range from 263 to 314) — although that range is probably a useful guide it should be recognised that the evidential basis for it is somewhat less scientific than we might wish it to be.

A 75 mg daily dose of CMI may be expected to produce a variable steady-state concentration of CMI between 30-200 ng/ml.  Older studies suggested an optimal therapeutic effect was with the sum of CMI and desmethyl-clomipramine between 200-400 ng/ml [35].  Elimination half-life estimates exhibit wide variation e.g. 19-37 hours (Drugbank), CYP1A2, 2C19 and 2D6 gene variations play a big part in that (see below).  [NB it is a potent hSERT inhibitor and recent PET studies demonstrate that it produces greater than 80% inhibition at plasma levels of around 5 ng/ml, see below.]

However, it is important to repeat, these therapeutic ranges are based on poor evidence and more recent data on CMI (and other TCAs) suggest at least some of them may be incorrect by an order of magnitude.  A simple practical way of assessing the therapeutic effect of these drugs is to measure the degree of postural hypotension in the same way as I advocate for monitoring the effect of MAOI antidepressants. [Link to BP measurement PDF].  The alpha1 antagonism of all the tricycles is very similar, Gillman [3], (Table 3), and, although the degree of postural hypotension is not often problematic, it is a useful index of the effect of the drug.  It is likely that if the alpha antagonistic effect is high that the hSERT antagonism will also be high.  This can be an excellent way of assessing the effect as a prelude to, or possible substitute for, plasma level measurement.

For clomipramine, the serotonin transporter (SERT) occupancy, using Positron Emission Tomography (PET) in humans, shows that with a plasma concentration of only 1.42 ng/ml, it produces about 80% occupancy of SERT [36], see ‘PET’ studies below.  Note the suggested therapeutic range of plasma levels is 50-400 ng/ml — 400 is possibly one, or even two, orders of magnitude more than required for some/many patients.

Szegedi et al. used fluvoxamine (as did I occasionally) in combination-treatment with clomipramine to ‘level the field’ re CYP450 variation and thus keep the ratio of CMI/D-CMI closer to unity [37].  It works, but it was troublesome, complex, and not without risk.  I found it preferable to use my good old sertraline + nortriptyline combination.  [Preskorn has written a prolix article about this more recently, parenthetically [38].]

Receptor profile and efficacy

The newer purported SNRI compounds, venlafaxine (and its ‘badge engineered’ derivative metabolite desvenlafaxine) and duloxetine may not be as effective at simultaneously blocking SERT and NAT.  That is because their ratio of potencies at those two sites is different (see table below), which indicates that, at any one dose, their effect at one site will probably be too great, and at the other, too little.  There is still doubt about how venlafaxine works and how well it works; it is a relatively weak noradrenalin reuptake inhibitor (see my TCA review [3]) that has only a weak effect on the tyramine pressor response [39] (75 mg does not alter the tyramine pressor response, and higher doses only a little).

Clomipramine is metabolised (mostly via 1A2, also by 3A4 & 2C19) into desmethyl-clomipramine which, like desipramine (aka desmethyl-imipramine), is a potent noradrenalin reuptake inhibitor. NB Clomipramine is the chlorinated analogue of imipramine (early publications referred to it as ‘mono-chlor-imipramine).

[We do have replicated hSERT data for CMI, but not for its metabolite DMCMI.  However, older studies did estimate IC₅₀ values for both, which give a good indication of the relative potencies, and thus allows us to extrapolate confidently.  That is what enables the statement I have made previously that DMCMI is a potent NRI (despite the absence of modern hSERT estimations), probably more potent than desipramine: these papers contain the data about that [40-47].  Therefore, clomipramine is definitely a potent SNRI.  It is important to adjust the dose carefully because a proportion of patients will get an excellent response at much lower doses than have often been used, and will therefore get a better balance of benefits versus side-effects.  Understanding this clarifies why I often state that some TCAs have similar, or even fewer, side-effects, when adjusted carefully, than many of the newer drugs.]

It is useful to further enumerate why clomipramine may be more potent than the newer SNRI drugs.  Apart from the fact that the ratio of the two re-uptake effects (SERT and NAT) is probably closer to 1 to 1 than the other drugs (see table), which is probably a good thing, it also has other effects which may be relevant; these are the sedative effects via H1 blockade, anti-muscarinic effects, and 5-HT2A antagonistic effects — there is reason to believe all of those effects may contribute beneficially to the overall effect, a view that Arvid Carlsson expressed many years ago.

It is worth observing also that the SNRI combination of sertraline with nortriptyline is in my view superior to sertraline with reboxetine (a ‘pure’, i.e. selective, NRI drug).  That suggests the extra properties enumerated above (and which are possessed to a similar degree by nortriptyline, but not all TCAs) do indeed make a difference and that in turn suggests another possible reason why clomipramine would be superior to venlafaxine.

These tables below are taken from my TCA review paper [3].  If comparing these figures with other data, consider only human cloned receptor (HCR) data, hSERT).  Much of the older data are not HCR and are not comparable.  Further discussion about this may be found in the commentary ‘Understanding Receptor Affinity Data’ [link]

Receptor profile, Ki in nmol/L, of TCAs and comparator drugs: uptake inhibition and receptor antagonism (human cloned receptor data)

Table legend

All data has been extracted from PDSP K_i data base, (except * Richelson [48]). Smaller K_i values represent greater potency. Note: where values are available from different laboratories and different experiments affinities can vary by about one order of magnitude, approximate mid-range values are given (Table 2 gives ranges). Receptors: H1 Histamine type 1, Musc acetylcholine muscarinic, A1 alpha1 adrenergic.

Human cloned receptor data, K_i in nmol/L, and TYR30, relating to dual action

Table legend

TYR30: tyramine pressor response at recommended therapeutic dose, degree of reduction placebo/NRI: +++ = nearly complete inhibition i.e. potent NRI effect. K_i data has been extracted from PDSP K_i data base (accessed June 2006) http://pdsp.cwru.edu/pdsp.asp , except: *approximation from Human cortex data, no HCR data available. **additional HCR values from [49]. For human in vivo considerations the TCAs are grouped as pairs (i.e. amitriptyline is metabolised into nortriptyline, and clomipramine to desmethylclomipramine and imipramine to desipramine: for TYR30 N/A indicates that in vivo, parent drug cannot be present without greater effect from more potent NRI metabolite). The NA/5-HT ratio is approximated because varying in vivo levels of metabolites occur. Clomipramine is the most potent, and the only available drug with combined affinities for both systems of less than one. It is unlikely that varying drug levels in different tissues would compensate for the extreme weakness of venlafaxine (and desvenlafaxine) as an NRIs.

***Only 1 set of values, from Deecher et al. [50].

Positron Emission Tomography (PET)

Positron Emission Tomography (PET) studies of binding to transporters (e.g. SERT & NAT) and receptors in humans [36, 46] confirm what the receptor potencies below indicate, that 10 mg clomipramine is effective at substantially inhibiting the serotonin transporter: however some of these results are inconsistent [51-54] and it is premature to draw firm conclusions.

Suhara et al. found

Occupancy of 5-HTT increased in a curvilinear manner. Even 10 mg of clomipramine hydrochloride showed approximately 80% occupancy, which was comparable with that of 50 mg of fluvoxamine maleate. Estimated median effective dose (ED50) of clomipramine hydrochloride was 2.67 mg for oral dose and 1.42 ng/mL for plasma concentration; those of fluvoxamine maleate were 18.6 mg and 4.19 ng/mL, respectively.

NAT occupancy: for Clomipramine (in non-human primates) 50% of NAT was occupied, at a dose of 0.44 mg/kg producing a plasma level 25 ng/ml, and for desmethyl-clomipramine 0.11 mg/kg and 4.4 ng/ml [46]. i.e., Clomipramine itself appears, in vivo, to be a significant NRI.

SERT occupancy in humans: clomipramine at a plasma concentration of 1.42 ng/mL produced 80% occupancy of SERT [36].

Efficacy

There is evidence from the ‘DUAG’ studies [2, 55-58]) that clomipramine is an antidepressant of superior efficacy.  It is also effective for obsessive compulsive disorder (possibly more so than SSRIs [59]), generalised anxiety disorder, phobias and panic attacks, narcolepsy/cataplexy [60-62], not to mention migraine.

Kragh-Sørensen’s aggregated analysis of the DUAG data [2] produces significant and striking findings when comparing patients with melancholic features, in their response to the SSRIs, paroxetine or citalopram, vs CMI. Using a more appropriate and rigourous standard of remission (defined by a HDRS score of less than three at six weeks), 25% of the SSRI patients vs 60% of those on CMI remitted (P .001).  This paper has not been cited even once in the 20 years since its publication, which is somewhat surprising.

The value of experts and experience is rightly regarded with caution; with that caveat in mind one may still note that most of psycho-pharmacologist that I have known over the years agree that clomipramine is an antidepressant of superior efficacy (refs above).

There is one further observation concerning the superior efficacy of clomipramine that I made many years ago, and for which there is not a convincing alternative explanation.  It is apparent that of the patients treated with antidepressants the chance of death from overdose varies by 5-10-fold between different TCAs (and other ADs), being substantially less for clomipramine alone.

When the number of deaths from ODs for every hundred thousand scripts issued is substantially less for a particular drug then one possible explanation is that drug is more effective: one might think that is an excellent property for an anti-depressant.  One explanation proffered was that it was because clomipramine was used mainly for obsessive-compulsive disorder, and such patients attempted suicide much less frequently than those with depression.  That is true, but OCD (as a primary disorder in the absence of depression) is about 5-10 times less common than depression.  During the time these statistics were gathered, in the UK, CMI was quite commonly used as an antidepressant, but it may still be that a sufficient proportion of the patients were being treated for OCD and that may account for the discrepancy.  I suspect the number of OCD patients represented in the clomipramine sample would be insufficient to account for this difference — it is unfortunate that we do not have better data for statistical analysis.

I am inclined to the explanation that if you take clomipramine, then you get better and are less likely to feel suicidal and act on that, which would be another piece of evidence to indicate that clomipramine is a better antidepressant.

Side Effects and Toxicity

Although TCAs can have troublesome side effects, as have all effective drugs, it is pertinent to note that there is, at worst, only a small difference in side effect discontinuation rates between TCAs and SSRIs (see Anderson [1]).  Some studies suggest no difference [63].  Paradoxically, for older patients any possible differences may be smaller.  Most data come from (SSRI) sponsored trials and are likely to be biased in favour of SSRIs — a fact strongly substantiated by the absurdly low estimates of sexual side-effects in company sponsored SSRI drug trials.  Indeed, since inhibition of orgasm is an effect of increasing 5-HT concentrations, [which is evident after a few hours, not days or weeks] it is not a ‘side effect’, but a reliable indicator of the drugs’ main therapeutic effect — it is in fact defined as a ‘type A’ adverse drug reaction, that is a reaction that is expected because of its main mechanism of action.

A key consideration which can sometimes be lost in this market-orientated side-effects debate is that a drug with less side-effects (or lower toxicity in over-dose) is not an advance, if it is less effective.  In my view, this sums up the position for many new drugs, including SSRIs — they are less effective for serious depression (see also extensive references about this in the TCA commentary).  It is not just my view; Professor Healy makes the same point and Carlsson, who was awarded the Nobel prize for medicine for his work, which included the discovery of the first SSRI, Zimelidine, did not even call it an antidepressant, he called it a serenic.  This is not the place for a detailed discussion of this question, but a reminder is relevant: it may be that SSRIs do help some symptoms in people with biological depression, as well as acting as a kind of anxiety reducing drug in others, but that is not the same as improving the core symptoms of depression (anergia and anhedonia).  Many researchers and experienced clinical psycho-pharmacologists would agree with that.

Antidepressants: A Minor contributor to suicide deaths

The key problem in serious depression is that illness severity and duration increase the risk of suicide.  The life-time risk of death by suicide is around 10% [64, 65].  The problem is the illness, not death caused by the toxicity of prescribed anti-depressants when taken as an overdose.

If patients get better from the depression, they are less likely to commit suicide.  The comparatively much lesser problem of the potential toxicity of drugs in overdose (and claimed differences between various drugs) is appropriately dealt with by good clinical management of any seriously ill patients. Choosing a marginally less-toxic-in-overdose drug is not the first-stop option.  Good clinical management involves seeing them more frequently and not prescribing potentially fatal quantities of drugs.  Incidentally, I am sure most doctors, including specialist psychiatrists, would not know what the potentially fatal dose of the particular drug they are handing out actually was.

A disingenuous argument

A major weakness of the argument that promotes preferential prescription of (supposedly) less toxic drugs is that a only a small minority of patients who commit suicide use their prescribed AD drugs to do so; studies show that less than 10% of deaths from suicide were from the prescribed AD drug [66, 67].  All studies show that drugs generally, whether prescribed or not, whether anti-depressants or not, make a minor contribution to the number of deaths from suicide in those suffering depression [68-70].

The notion that doctors should give over-riding consideration to prescribing less-toxic drugs is a disingenuous and specious argument based on poor suppositions.  It is a good scare tactic and sales pitch though — how could I even think that!  There is further discussion of this question in the TCA review.

Clomipramine may be less toxic in over-dose than most other TCAs (nortriptyline also seems to be a less toxic TCA) and about the same as venlafaxine [34, 71-73].

Clomipramine may be one of the least toxic TCAs, about the same toxicity as venlafaxine

Initial dose and rate of dose increase

Providing the dose of clomipramine is adjusted carefully it is tolerated well by most patients; plasma levels (of TCAs) are often useful in optimising the dose [74, 75], although it is a little surprising that many psychiatrists rarely or never measure plasma levels (e.g. see [76].

The dose should be judged according to the patients’ improvement and the degree of side effects experienced, not only the blood levels, and CYP450 status is only one of the factors influencing plasma levels.

Starting doses in the range of 10 to 25 mg daily (for ‘little-old-ladies’ even as little as 5 mg) are appropriate and should be taken as a single dose at night (because of its long half-life, and contrary to some extant recommendations, it is usually pointless to take it two or three times a day).

Experience and evidence suggest it is sometimes prudent and worthwhile to give smaller doses, 25-75 mg, plenty of time, 6-8 weeks, to show their full benefits before escalating further.

The old pharmacologists’ adage of ‘start low, go slow’ is especially appropriate.  Premature and hasty increases of dose are a common cause of therapeutic failure; consequently, persistence and patience are often rewarded.  However, vacillation and indecisiveness are equally to be avoided.  There is little excuse for the common story I hear of people with partial response who have been kept on a low dose for months on end without a trial of a higher dose.  It is surprising how many doctors are illogical, inconsistent, and careless concerning the monitoring, follow-up, and adjustment of drug doses.  Monitoring of blood pressure to assess the degree of postural hypotension may be helpful in guiding the dosage increases, especially if blood levels are not easily available or affordable.

Influence of CYP450 (P450) Enzyme Status on Plasma Levels and Dose

Amitriptyline, clomipramine, and imipramine are demethylated by CYP1A2, 3A4, and 2C19 into active metabolites.  These active metabolites are inactivated by via CYP2D6 (to less active or inactive metabolites).

Most P450 sub-types (‘isoforms’), including both 2D6 & 2C19 & 3A4, are polymorphic [77]: they have many mutations, with slightly different structures and levels of activity which affect rates of metabolism and blood levels.  As Zhou et al. state [78]:

CYP450s have a strong influence on levels of TCAs, including nortriptyline, desipramine, and desmethyl-clomipramine.

The Clinical Pharmacogenetics Implementation Consortium have recently updated guidelines for the dosing of Tricyclic Antidepressants using patients’ individual CYP2D6 and CYP2C19 genotypes [75]. And again in 2017 [79] and anyone who is interested in this should consult that document.

Beware of doctors, and the sales hype of companies, offering testing services (often expensive), who think they understand more about this than they do.  Simply measuring your plasma level may be simpler, cheaper, more accurate, and more informative.

If one is using nortriptyline, which is relatively unusual drug, in that it is almost entirely dependent on 2D6 for its metabolism, which produces an inactive metabolite; in this instance knowledge of 2D6 status is of more use [80, 81].

2D6 poor metabolizers (~10% of patients) are those individuals carrying two loss-of-function alleles.  These CYP2D6 gene deletions are indicated by the CYP2D6*5 allele and are usually reported as ‘CYP2D6*5/*5 or CYP2D6 (*5/*5)0N’, see Hicks p7:

These mutations are non-functional and that individual is thus unable to metabolise (break down) nortriptyline (or desmethyl-clompramine) via that pathway.  In these instances, starting with a dose of 10-25 mg would be prudent.

However, most drugs, including clomipramine, are not so simple.  They are often metabolized by more than one P450 isoenzyme, and the metabolites (sometimes more than one) may be active; in this case desmethyl-clomipramine; the complex interplay of different factors, in many instances, is not easily predictable.

The Clinical Pharmacogenetics Implementation Consortium recommendations are in the Hicks papers [79, 82], the content of which I suspect most clinicians are going to have trouble assimilating and utilizing; they are complex and would rarely alter clinical practice (as set out herein).

Good clinical practice already dictates ‘start low and go slow’ and the prominence of side-effects and response to a gradual increase of dose (allowing for the time taken to reach steady state), combined with monitoring the degree of postural hypotension, is what would guide me, and possibly prompt me to request plasma levels when appropriate.  That course of action is only rarely going to be altered by the results of genetic testing.

There is one (rare) situation where a genetic test result is useful: if an individual was an ultrarapid metabolizer for 2C19 they would have low levels of clomipramine (which would be rapidly converted to desmethyl-clomipramine); if they were also a 2D6 poor metabolizer that higher level of desmethyl-clomipramine would be metabolized slowly and lead to accumulation of higher than usual levels. Ultrarapid metabolizers for 2C19 constitute 5–30% of patients and the gain-of-function allele causing this is represented as *17: if there is one functional allele and one gain-of-function allele i.e. diplotype *1/*17, or two gain-of-function alleles, diplotype *17/*17, then that individual will be an ultrarapid metabolizer of clomipramine.

Thus, if the testing shows 2C19 diplotype *1/*17/ or *17/*17 and also 2D6 diplotype *2/*2, or *2/*3, or *3/*3 then there is a potential problem and care in dosing would be required.

See the Human Cytochrome P450 (CYP) Allele Nomenclature Database