Fibromyalgia Pharmacotherapy: Duloxetine, Milnacipran, or Pregabalin First?
Three drugs carry an FDA indication for fibromyalgia, and none has ever been tested head to head against another. The disagreement here isn't whether to treat — it's which of her specific symptoms should decide which one goes first.
Dara M., a 39-year-old hairdresser who spends most of ten-hour shifts on her feet, has had widespread pain for fourteen months, beginning a few weeks after a bad flu she never fully bounced back from. She now meets the 2016 ACR fibromyalgia criteria on widespread pain index and symptom severity, with diffusely tender soft tissue on exam, normal inflammatory markers, and normal thyroid function. She has quietly stopped going to her Sunday recreational soccer league, something she used to plan her whole week around, and describes waking up most mornings feeling like she never slept at all. Asked directly whether she feels depressed, she says no — but her PHQ-9, administered the same visit, comes back at 11, moderate range, and the score is driven almost entirely by the fatigue and sleep items rather than the mood or anhedonia items themselves, a distinction that turns out to matter for which drug's own trial history actually points at her.
All three FDA-approved agents work through the same broad central-sensitization framework, but their own development histories tested different symptoms as the thing that mattered most. Duloxetine's regulatory path ran through a population enriched for depressive symptoms — Arnold and colleagues' 2005 trial found its pain benefit held independent of mood improvement, but the drug still carries a separate, primary indication for major depression, and her flat mood/anhedonia subscore is a weak argument for reaching for it on that basis specifically. Milnacipran sits at the opposite end of the same three drugs' pharmacology: its reuptake-inhibition ratio runs roughly one part serotonin to three parts norepinephrine, the most NE-weighted of the three, and its registration program refused to count anyone a responder unless physical function and patient global impression improved alongside pain, not pain by itself — a composite bar set around what a patient can do rather than how she feels. Clauw and colleagues' 2008 trial, run on that design, also picked up significant improvement in fatigue on the Multidimensional Fatigue Inventory, though as a secondary outcome rather than a primary one, which makes it the softest of the three evidentiary claims on this table. Pregabalin's case rests on a third axis: Crofford and colleagues' 2005 trial, and the sleep-focused analyses that followed it, found pain improvement tracking closely with improved sleep continuity, plausibly because reduced excitatory release at the dorsal horn's α2δ subunit blunts the nighttime arousals a pain-disrupted sleep cycle produces — and unrefreshing sleep is the one symptom she names before anyone asks about pain.
Clinic visit, choosing the first prescription
Start with milnacipran. She's not describing a mood problem — she's describing a fatigue and function problem, and milnacipran is the one agent here whose registration program refused to call anyone a responder unless physical function and global impression moved, not just the pain score. Clauw and colleagues' 2008 trial ran on that design and picked up a significant fatigue improvement too, on the Multidimensional Fatigue Inventory — a secondary endpoint, I'll grant you, but at least a measured one, which is more than either other program managed. Milnacipran's own reuptake profile, roughly three parts norepinephrine to one part serotonin, is the most NE-weighted of the three approved agents, which lines up with a patient whose chief complaint is that she can't get through a shift, not that she feels low.
I'll grant that the trial-population match for milnacipran is real, and I'm not arguing she's clinically depressed by any formal definition. But a PHQ-9 of 11 isn't nothing, and duloxetine is the only one of the three agents in front of us with its own separately demonstrated efficacy for that exact symptom cluster.
Arnold and colleagues' 2005 trial is being read here as if it just confirms duloxetine's pain effect is independent of mood, which is true as far as it goes — but that trial specifically enrolled patients with and without major depressive disorder because the investigators already expected mood to be doing real work in some of them. And I'd note what just got conceded: the fatigue signal is a secondary endpoint, which is exactly the kind of finding that looks load-bearing right up until someone builds a prescription on it. Building the plan entirely around her stated symptom order treats a screening-tool result as noise the moment it stops matching the story she's telling.
You're right that a PHQ-9 of 11 shouldn't just be waved away — that's fair, and it's worth tracking regardless of what we start today. But she told us, unprompted, that the thing she notices first every single morning is that she doesn't feel like she slept, before she's even registered the pain. Crofford and colleagues' 2005 trial, and the sleep-architecture work that followed it, found pregabalin's pain benefit tracking with improved sleep continuity specifically — a mechanism that maps onto exactly the symptom she leads with, in a way neither of the other two drugs' own trial designs were built to test.
I don't think this has to be adjudicated by evidence alone, because the evidence genuinely doesn't pick a clear winner among three drugs that have never been compared to each other directly. It's reasonable to let what she actually says first carry some real weight here.
Milnacipran started at 12.5mg daily for the first several days, titrated per label to 50mg twice daily, with the PHQ-9 and FIQR both repeated at eight weeks rather than pain alone.
Not agreed: whether her current PHQ-9 already warrants its own parallel workup now, rather than waiting on the fibromyalgia reassessment to reveal whether it moves.
Repeat the PHQ-9 at the eight-week visit alongside the FIQR. If it hasn't improved as milnacipran's own noradrenergic effect might be expected to lift it, address mood as its own separate problem then.
A score of 11 is real today, not hypothetical, and a brief same-visit depression conversation and follow-up plan costs little against the risk of one intervention quietly masking another for two months.