This commentary covers Selegiline, rasagiline, and moclobemide which is far as I know are the only such drugs are available, although others, such as ladostigil, are on the horizon.

First, one should appreciate that some ‘selective’ drugs are only slightly selective — that is to say their potency of one effect over another effect might only be threefold, as in the case of selegiline, or nearer tenfold, like rasagiline.

Furthermore, drugs are only ‘selective’ in relation to the properties they possess that have actually been measured, or talked about.  For instance, an SSRI drug might be highly selective for that property over NRI potency, but not very selective for that property over inhibition of cytochrome P-450 enzymes (cf. fluoxetine).  The pharmaceutical company is likely to choose to forget that issue.

Because the levels achieved by the same dose in different people can vary, commonly by a factor of 10, this means that drugs with those kinds of lowish levels of selectivity occasionally produce non-selective effects in a proportion of people.

Second, remember that truly selective inhibition of MAO-B almost certainly produces a smaller increase of dopamine than full inhibition of both the A and B subtype and, such as tranylcypromine achieves.  This explains why a number of people have contacted me to tell me I don’t know what I’m talking about and that rasagiline is good for serious depression.  The above reveals that is highly likely to be correct, providing you are very wealthy and can afford the dose of 3 to 6 mg a day needed to make it non-selective, and which will cost you a small fortune.  At that level it will have the same potential interactions as tranylcypromine i.e. it will cause serotonin toxicity if combined with a serotonin reuptake inhibitor.

Selegiline in combination with (S)SRIs

Selegiline (deprenyl) is selective for MAO-B at the usual dose of 10 mg daily and may thus be less likely to precipitate serotonin toxicity because 5-HT is metabolised mostly by MAO-A.  There have been reports that are probably serotonin toxicity.  Gitlin’s report demonstrates typical serotonin toxicity features.  This reaction occurred 16 days after ceasing selegiline (50 mg daily- more usual dose is 10 mg daily) when the patient took venlafaxine 37.5 mg x 1 dose.  It did not recur on re-challenge one week later {Gitlin, 1997 #870}.  Because the margin between a selective and a non-selective dose is narrow it is difficult to be confident in normal clinical practice that serotonin toxicity can reliably be avoided in such situations.

The Zornberg report is the epitome of an incompetent, unrevealing and misleading case report: it is explained by the enormous (toxic) dose of imipramine (a total of 700 mg daily — no blood levels documented — and did not show serotonin toxicity features {Zornberg, 1991 #475}.  Not only was the patient being given 700 mg daily of imipramine, but this was supplemented with 100 mg of desipramine plus heaven knows what else including hydroxyzine.  The main clinical picture was that of delirium and there were no signs of serotonin toxicity.  How a reputable journal like the Lancet came to publish something like this is almost beyond belief.

I have to say that I have seen an awful lot of appalling case reports in my role as a journal referee and expert on serotonin toxicity, but this one is definitely a contender in the first division of extraordinary treatment and bad case reports.

I would be terrified if I thought doctors like that were looking after one of my nearest and dearest.

There are 3 cases reported from a review of the French pharmacovigilance database; details are presently unavailable.  I lectured for this group in Toulouse in 1997 and they conducted the review following that, so they were informed of what symptoms to note {Bagheri, 1998 #12426}.  There is a Spanish report of an apparent fatality that I cannot comment on: if anyone can translate it let me know please {Bilbao Garay, 2002 #12489}.

There are a couple of retrospective chart reviews: one found 7/28 patients, on selegiline and any class of antidepressant drug, had taken an SSRI type drug; one patient on fluoxetine was diagnosed as serotonin toxicity.  The other review found no serotonin toxicity in 23 patients on selegiline and fluoxetine.  In my opinion we can be sure that selegiline is less of a risk than tranylcypromine (or other old non-selective MAOIs): I am certain that if you did the same with tranylcypromine deaths would occur, as indeed they did, as reported by Beasley et al in the early fluoxetine trials {Beasley, 2000 #12453;Beasley, 1993 #302;Ritter, 1997 #986}.  

In 1997 Richard et al published a thorough follow up of the Parkinson Study Group (PSG) data; they also reviewed the FDA data.  This Parkinson Study Group data showed no severe serotonin toxicity at all, but there was some small suggestion of serotonergic side effects like agitation and restlessness.  In an older sample of patients with brain disease, increased susceptibility to serotonergic side effects is to be expected and does not constitute evidence for an interaction with selegiline.  This sample had ~4500 patients treated with selegiline and an antidepressant; the proportion treated with a serotonergic antidepressant is not known but is likely to have been substantial.  There were no deaths from serotonin toxicity.  There is little evidence for severe serotonin toxicity and little evidence for a significant increase in serotonergic side effects {Richard, 1997 #14163}.

I corresponded with the Parkinson Study Group, but they were unable to say what portion of the patients had a tricyclic antidepressant (TCA) or a selective serotonin reuptake inhibitor (SSRI):  so their estimate of serotonergic symptoms in 0.24% of cases may reflect a significant risk of serotonin toxicity in such patients.  It would be reasonable to suppose that about half were on non-serotonergic antidepressants, and half were on sub-therapeutic doses, so the real risk might equate to nearer 1%.  Those estimates are reasonable from my experience and the literature on cytochrome P450 enzyme genetic variability.  When extrapolated to higher doses in higher risk patients it may be that a 5% risk of significant serotonin toxicity with selegiline and selective serotonin reuptake inhibitors (SSRIs) is more realistic estimate: that would fit better with the reported cases.

The spectrum concept of serotonin toxicity is the key to assessing this risk.  Pharmacokinetic and pharmacodynamic interactions are likely to be crucial in tipping the balance; selegiline is selective at lower levels but drugs, especially the selective serotonin reuptake inhibitors (SSRIs), that act as cytochrome P450 enzyme inhibitors will probably induce toxicity in some patients in some circumstances.  Since we do not yet fully understand some of the intricacies of these drugs’ interactions and metabolism even the best-informed specialists can get into difficulty.  Providing the dose is not excessive and significant pharmacokinetic interactions can be avoided serotonin toxicity is uncommon with SRIs combined with selegiline: however, one or two deaths may have occurred.

However, there can be a rapid transition from side effects to toxicity, as other sections of this review reveal.  Clinicians, who have prescribed such combinations, and observed patients experiencing that transition, tend to go to bed wondering whether they really know as much as they thought they did, and contemplating what they might say to the relatives and their medical defence insurer.

Here are all the other references I have found concerning serotonin toxicity and selegiline {Andreu, 1997 #6966;Heinonen, 1998 #13262;Hinds, 2000 #13284;Montastruc, 1993 #626;Starr, 1991 #1620}.

Moclobemide

Moclobemide does confer risk of serotonin toxicity with SRIs although the history and evidence relating to its interactions is clouded with misleading information put out by the drug company. Since this is all past history and very few people use this drug any more I will post the detailed analysis about moclobemide & ST separately at some later date, because the original section in the serotonin toxicity document posted years ago needs tidying up and re-referencing.

Suffice it to say that even at therapeutic doses, and most definitely following overdoses, one can get quite severe serotonin toxicity reactions and deaths have occurred. Yet again, venlafaxine seems to be particularly implicated. I summed it up in this brief comment {Gillman, 2004 #3277} in a way which I have not since bettered:

“The above data indicates clearly that moclobemide/SRI combinations represent a predictably risky strategy and constitute a balancing act, between efficacy and fatality, of such delicacy as to be unattainable in clinical practice.”