Timothy C. Hain, M.D.• Page last modified: December 3, 2022 • You may also be interested in our many pages on migraine on this site
Before getting started on this discussion, not that there is considerable healthy skepticism in the neurology community, and some feel that the evidence is just not sufficient to recommend this drug. For example, Hovaguimian A, Roth J.(2022) wrote "A Cochrane review from 2015 reviewed the evidence for selective serotonin reuptake inhibitors and serotonin-norepinephrine reuptake inhibitors compared with placebo or amitriptyline in chronic migraine. This review noted that the studies were of poor quality, with incomplete data and design flaws, and found low or very low quality evidence for the efficacy or safety of these drugs." Still, the lack of evidence is not the same as evidence of lack.
A recent metaanalysis of Smyth et al (2022) wrote "Amitriptyline and venlafaxine may reduce vertigo severity and frequency to a similar degree as other medications, although like the majority of other treatments the evidence is primarily taken from before-and-after analyses. "
Venlafaxine (Effexor). Very effective for migraine and migraine associated vertigo.
This antidepressant medication, of the SNRI group, is very effective for migraine prevention and has relatively few side effects. (Diamond, Pepper et al. 1998; Nascimento 1998; Adelman, Adelman et al. 2000; Bulut, Berilgen et al. 2004; Ozyalcin, Talu et al. 2005; Tarlaci 2009; Salviz 2015; Li et al, 2017). We particularly favor this drug for the visual dependence symptom commonly seen in migraine but also found in persons with anxiety and in those who have adapted to a vestibular disorder.
Although we have found venlafaxine to be very effective, it tends to be less favored by other authors. For example, Dr. Stefan Evers from the Department of Neurology in Munster Germany (2008), stated in an "Expert Opinion" that venlafaxine was a "drug of second choice". Dr. Evers listed a drug as a first choice (Flunarizine), that has never been approved in the USA due to serious safety considerations. As we find venlafaxine very effective here in Chicago, perhaps the differences lie in the protocol. Dr. Evers recommends rather high doses - -75 to 150, while we use much smaller doses. Another possibility is that our patients have more dizziness (about 100%) than other trials for migraine, and it may be that it works better for dizziness with migraine. A recent study by Salviz et al (2015) suggested that venlafaxine was much more effective than Propranolol for "vestibular migraine". Li et al (2017) suggested that venlafaxine was superior to both flunarizine and valproic acid for vestibular migraine.
The usual dose of venlafaxine is small -- varying between 12.5 mg and 75 mg/day, taken in the morning. We usually start persons with 1/3 of the time release (37.5) and titrate upward every week. In other words, the usual prescription for generic venlafaxine looks like:
Venlafaxine 37.5 XR capsule.
Start with 1/3 capsule QAM for first week.
Increase to 2/3 capsule for second week.
Take full capsule from 3rd week and onward.This can also be prescribed as Venlafaxine ER.
To avoid conflict with well meaning but uninformed pharmacists, we put these instructions on a handout, and just put on the prescription: "Start up as directed on the handout". This avoids telephone calls.
Venlafaxine increases the blood pressure slightly, and is neutral in regards to weight.
Venlafaxine often reduces hot flashes in menopause. (Evans et al, 2005) On the other hand, we have encountered patients who feel "hot all the time". The evidence for other non-hormonal treatments, such as herbals, that reduce hot flashes is not strong (Franco et al, 2016)
We have encountered withdrawal problems in persons who stop effexor in larger doses than 37.5 mg, but almost none with the low dose of 37.5 generally used for migraine. The most disturbing withdrawal symptom is "head zaps". (Papp et al, 2018). This can nearly always be avoided by simply downtapering in the same way that it was uptapered.
We have also occasionally encountered suicidal thinking in adults on venlafaxine in spite of good headache control and without previous depression-- we stop the drug or reduce the dose in this situation (see also Todder and Baune, 2007). In higher doses (i.e. 150 mg), we have encountered patients that developed mania. For this reason, we think it is best to try to keep the dose/day at 75mg or less. If the patient might benefit from more venlafaxine than 75 mg/day, perhaps because of depression, we prefer that they be monitored by a psychiatrist. Venlafaxine also is sometimes associated with "strange dreams". This side effect is rarely intolerable, and presumably is related to the higher level of adrenalin associated with this family of medications.
There have been occasional reports of the "serotonin syndrome" provoked by the combination of effexor and triptans (see abortives), as well as the "SSRI" family of antidepressants. It seems unlikely that this would occur when using the small doses typical for migraine prophylaxis (37.5 to 75 per day), but caution is advised.
We do not know of any reports (as of 2014) that suggest that there are greater side effects when combining two noradrenergic drugs at the same time, such as an ADD type drug (e.g. an amphetamine), and venlafaxine. While there are no reports, this seems relatively unwise to us, and likely to result in problems such as weight loss and hypertension. If someone needs to do this, we suggest seeing a physician who is an expert on the pharmacology of these medications.
Venlafaxine has a complex metabolism ! An in-depth discussion of the metabolism of venlafaxine, from Colvard (2014) can be found here : https://cpnp.org/resource/mhc/2014/01/key-differences-between-venlafaxine-xr-and-desvenlafaxine-analysis".
Venlafaxine is metabolized in the liver to several metabolites including desvenlafaxine (O-desmethylvenlafaxine) as well as a few minor metabolites. Venlafaxine and its major metabolite (desvenlafaxine) do similar things to serotonin and norepinephrine, they are selective inhibitors at serotonin and norepinephrine transporters, so this conversion is not necessarily a matter of concern. However, desvenlafaxine has much higher potency than venlafaxine at these pumps (i.e. about a 4:1 difference at the norepinephrine pump as well as a roughly 2:1 greater affinity at the serotonin pump). In other words, the metabolism causes a delayed increase in effect on neurotransmitters. Additionally, it is reasonably possible that one of the ignored metabolites of venlafaxine, produced in the liver, is the one that has the anti-migraine activity, as some other SNRI-SSRI's are ineffective for migraine (e.g. duloxetine).
The main liver enzyme that does the conversion of venlafaxine to desvenlafaxine is the CYP2D6 cytochrome P450, but there are also some minor metabolism through the CYP1A2, CYP3A4, and CYP2C19 enzymes. According to Colvard (2014), roughly 55% of a single dose of venlafaxine is converted to desvenlafaxine. Desvenlafaxine is inactivated in the liver, and both venlafaxine and desvenlafaxine are also (mainly) eliminated through the kidneys. The CYP2D6 cytochrome enzyme varies in the population and about 7% of the Caucasian population are called poor metabolizers (PM) who have either increased or decreased levels of CYP2D6. African Americans and Asians have far less prevalence of poor metabolizers. CYP2D6 phenotyping can be done for roughly $250.
As venlafaxine and desvenlafaxine act similarly, but desvenlafaxine is more potent (especially at the noradrenergic site), this might result in relatively more or less effectiveness between venlafaxine and desvenlafaxine. Of course, this conclusion depends on one assuming that the effectiveness on migraine is related to either its effect on the serotonin or norepinephrine transporter -- which nobody has proven as yet. One would also expect that the liver metabolism of venlafaxine to desvenlafaxine might result in relatively more norepinephrine affinity, and more adrenergic side effects (such as tremor). Another way to put this is that one would expect more side effects (such as tremor) from anything that increases the ratio of desvenlafaxine to venlafaxine, such as being prescribed desvenlafaxine itself, or if one is a rapid metabolizer.
There are many other CYP2D6 antidepressants -- such as fluoxetine, paroxetine, fluvoxamine, amitriptyline, doxepin, maprotiline, clomipramine, imipramine, and trimipramine. Other interactors include many beta blockers and opiods. Venlafaxine and desvenlafaxine have relatively minor effects on the CYP2D6 enzyme -- causing other substrates (e.g. desipramine) to increase modestly.
Drugs that are metabolized by the liver may reduce the conversion of venlafaxine to desvenlafaxine, and reduce the net drug effect on serotonin and norepinephrine because desvenlafaxine is more potent at these receptors. This may be the reason that topiramate, added to venlafaxine for migraine management, sometimes seems to make the depression worse. This also means that drug-drug interactions with venlafaxine generally do NOT result in one getting too much drug effect, as most interactions reduce conversion to the more active form of venlafaxine (desvenlafaxine). This adds to the safety of this medication.
Drugs that increase serotonin or adrenalin may cause side effects due to too much of these neurochemicals when combined with venlafaxine. Practically, there are many people on venlafaxine combined with various other adrenergic agonists, and very few patients with any health problems traceable to this combination. The more critical literature supports the idea that this drug interaction is greatly overdiagnosed. Thus this "serotonin syndrome" seems to be overdiagnosed in emergency department settings looking for easy explanations, and by protective pharmacists who diligently check for drug interactions in their computer systems, but there are likely a few real cases out there too.
Venlafaxine, when given for migraine, should be stopped prior to becoming pregnant. There are several reasons for this. First, migraine usually remits during pregnancy. Second, there may a risk of birth defects. Briggs et al (2017) states "Human studies suggest risk in 3rd trimester". A recent study also suggested risk during early pregnancy (Anderson et al, 2020).
Venlafaxine is generally thought to be safe in nursing mothers (see drugs.com site https://www.drugs.com/pregnancy/venlafaxine.html). It is not approved for treatment of children.
Beta-blockers can be combined with venlafaxine with minimal side effects related to their interaction. As noted above, they could interact by decreasing conversion of venlafaxine to desvenlafaxine, which would perhaps reduce effectiveness of venlafaxine. However, recent data suggests that beta-receptors are critical for antidepressant efficacy in neuropathic pain (Yalcin et al, 2009), and for this reason, we reserve this option until side effects are noted.
| Drug | Effectiveness for migraine | Comment |
| venlafaxine (Effexor) ER | good | |
| desvenlafaxine (Pristiq) | good | more expensive metabolite of venlafaxine |
| duloxetine (Cymbalta) | poor | might work in higher doses |
| milnacipran (Savella) | unsure | mainly used for fibromyalgia, high cost |
| Fetsima | unsure | optical isomer of Savella, high cost, high side effect as well (nausea). |
We do not favor using non-time release venlafaxine (the cheapest form), due to side effects. It is much more common to encounter start-up tremors with the generic. This is likely due to a "spike" in drug levels compared to the extended release form. On the other hand, after the "start up" problems are over, many people do well on the non-ER variant of venlafaxine. One would also expect there to be less sleeping issues due to insomnia, with this medication being taken in the AM.
A close relative of Venlafaxine is is desvenlafaxine, discussed extensively above under the metabolism section. This drug can be substituted for venlafaxine, as a time release as well, and is associated with less interactions with other drugs as well as a more predictable response. The main issue with it is that it is not easy to up or down taper, because it doesn't come in a capsule with beads.
We suspect that desvenlafaxine has more "norepinephrine" type side effects than does venlafaxine, because it has a higher affinity for the norepinephrine transporter. Desvenlafaxine is the active metabolite of Venlafaxine -- the conversion occurs in the liver through CYP2D6. The bioavailability of desvenlafaxine is higher than venlafaxine, after you factor in the metabolism. Thus, the starting dose of desvenlafaxine (50 mg) is FAR higher than the starting dose of Venlafaxine that we recommend (12.5). For this reason we generally switch to desvenlafaxine ER only if venlafaxine fails at a lower dose. Logically however, just increasing venlafaxine might work as well.
We have had a few patients complain of withdrawal. This makes sense as withdrawal is dose dependent with venlafaxine, and 50 mg of pristiq has more "active ingredient" than 37.5 of venlafaxine. We simply do not believe the suggestions made that pristiq has no withdrawal syndrome.
desvenlafaxine is available only as a brand name drug, and unfortunately, only as a tablet rather than the simpler capsule with beads. This means that desvenlafaxine cannot be "split" so that people can gradually ramp it up. Another reason not to use desvenlafaxine. We will probably have to wait till the generic comes out, as the "bead" technology of venlafaxine would seem perfectly suitable to desvenlafaxine too -- they just chose an unfortunate delivery method.
The bottom line for desvenlafaxine is that we think it should be reserved as a fallback for venlafaxine, mainly for individuals who don't respond, or perhaps in people who are taking many other medications that potentially interact with the P-450 system and reduce conversion of venlafaxine into its more potent cousin, desvenlafaxine. It costs far more than venlafaxine, it probably has more side effects due to the increased affinity for norepinephrine, and it cannot be "split".
The other common SNRI, Cymbalta (duloxetine) is minimally effective for migraine (Taylor et al, 2007). However, it was reported as "effective" for chronic headache combined with depression (Volpe et al, 2008), and also reported as effective in rather high doses in non depressed individuals (Young et al, 2013). Our take on this is that it likely does work for migraine, in very high doses. This would suggest that it is the norepinephrine affinity of venlafaxine that makes it more effective than duloxetine for migraine, as duloxetine has low norepinephrine affinity. This logic means that if someone is already on high doses of duloxetine, there is no pressing need to switch them to venlafaxine.
So far, we are simply not sure about another SNRI - -Savella (milnacipran), which was approved for fibromyalgia. Few people take Savella due to its high cost and also because it is "approved" for fibromyalgia, which limits coverage. Engel (2014) reported that it was promising, but needed to be started in very low doses. This would fit into the pattern that we have noticed for venlafaxine (low doses are best on start up).
The optical isomer of Savella is Fetzima. Again, we are not sure if this drug works for migraine. Generally speaking, the optical isomers are "cleaner" than the originals. Fetzima has greater norepinephrine action than does venlafaxine, which generally translates into more "activation" type side effects. It may also translate into greater migraine effectiveness. It is too soon to say. As Fetzima is a variant of Savella, one would think that if Savella works, Fetzima would work too. However, again low doses at startup would seem crucial to avoid the norepinephrine type side effects (e.g. tremor). In the few people where we have tried Fetzima, it seems to be more nauseating than Venlafaxine. Fetzima was a new drug in 2014 - -which generally also means an expensive drug.
We are also uncertain about two other new antidepressant drugs. Viibryd (vilazodone), is an SSRI/5HT1A agonist. Vortioxetine has somewhat complex effects on multiple 5-HT receptors. We suspect that they do NOT work on migraine, but there is currently little evidence one way or the other. To put this into perpsective, the excellent migraine abortive drug, Sumatriptan, for example, is a selective 5-HT1B and 5-HT1D agonist. Vorioxetine is a 5HT1A agonist like sumatriptan, but a 5HT1D antagonist (doing the opposite). Vilazodone, having 5HT1A agonism, might have some positive effect on migraine.
Quetiapine (Seroquel), has also been suggested as potentially effective for sensory exaggerations. Weight gain can be a problem with seroquel.
While the tricyclics amitriptyline and nortriptyline are both excellent migraine prophylactic medications, both have very significant side effects. Expect weight gain in particular.
Other references can be found here: