Atypical Parkinsonism: Defining an Adequate Levodopa Trial
A single patient with two years of progressive parkinsonism and a levodopa trial that never reached a dose or duration anyone would call definitive. The disagreement isn’t about which drug comes next — it’s about whether this trial has actually been run.
G.R., a 66-year-old man, spent his first two years of retirement finally tending the vegetable beds his wife had wanted for a decade — until the digging got harder to explain away. What he first called stiff knees turned, over eighteen months, into a stooped, shuffling walk with three witnessed falls, none of them backward onto stairs the way his father's Parkinson's disease had presented. His primary care physician started carbidopa-levodopa 25/100 three times daily eight weeks ago, working up to three tablets a day, and reported back only that he “didn't seem any different.” The outside referral note that came with him today says, verbatim, “poor response to levodopa, favor atypical parkinsonism.”
That conclusion is premature by the numbers alone. Three hundred milligrams of levodopa a day for eight weeks does not meet the dose or duration most movement-disorder practices use to call a trial adequate — closer to 1,000mg a day, or the maximum a patient can tolerate, held for at least four to eight weeks, with carbidopa dosed high enough to actually block peripheral decarboxylation rather than let most of the dose get metabolized before it ever crosses into the brain. That figure is not arbitrary: the MDS-PSP criteria define levodopa resistance as 30% or less improvement on the MDS-UPDRS motor scale after at least 1,000mg a day for a month. His 300mg for eight weeks reaches neither half of that definition. Williams and colleagues' 2005 pathological series is often invoked here, having found that the parkinsonian subtype of progressive supranuclear palsy meets the UK Brain Bank levodopa-response criterion often enough to be mistaken for idiopathic Parkinson's disease early on — but that subtype is defined by asymmetric onset with tremor, and his onset is symmetric with no rest tremor at all, so the confound Williams described is not the one he presents. What does apply is narrower and harder: the akinetic-rigid, predominantly axial PSP category his exam actually points toward has levodopa resistance built into its definition. Declaring non-response on an untested criterion doesn't just delay a diagnosis; it manufactures one.
On exam today his vertical saccades are slowed more than his horizontal ones, his rigidity is symmetric rather than the asymmetric onset more typical of idiopathic PD, and orthostatic testing is borderline — a 15mmHg systolic drop, not yet frankly abnormal. None of that settles PSP against MSA on its own. What it does settle is that the levodopa question is still open.
At the movement-disorders clinic, reviewing an outside trial
Three hundred milligrams of levodopa for eight weeks isn't a negative trial — it's an untested one. The dose most of us use to actually rule a response in or out is closer to a gram a day, or whatever a patient can tolerate short of that, held for at least a month, with the carbidopa dosed high enough that most of the drug isn't being decarboxylated in his gut before it ever reaches his brain. And I want to be careful about which literature I'm leaning on. Williams and colleagues' 2005 pathology series gets cited constantly here — patients with the parkinsonian variant of PSP met the UK Brain Bank levodopa-response criterion often enough that response alone doesn't separate them from Parkinson's disease early on. But that variant is asymmetric-onset with tremor, and he is neither, so I'd be overreaching to hang his case on it. The tighter point is that the MDS-PSP criteria define levodopa resistance as 30% or less MDS-UPDRS motor improvement after 1,000mg a day for a month, and the akinetic-rigid axial category his exam points toward requires that resistance to be shown. If we write “poor response” into his chart today off a trial that never approached that dose, we've recorded a criterion as met that was never tested.
You're right that a gram a day for a month is the actual standard, and I'm not disputing the pharmacokinetics or the target dose. But he's already fallen three times in eighteen months and his standing blood pressure is drifting the wrong direction before we've given him another milligram of anything — both of those get worse, not better, as the dose climbs toward a gram a day. His falls happening this early in the course are themselves a core clinical feature for PSP in the MDS criteria — repeated unprovoked falls within three years, in the postural instability domain, not one of the supportive clues — independent of what the drug does to him. I'm not willing to accept a real, near-term fall risk just to settle a diagnostic question the exam is already pointing toward.
Pointing at Williams 2005 explains why an inadequate trial is uninformative — it doesn't make escalating his dose to a gram a day free of the fall and blood-pressure risk we're already watching climb.
Neither of you actually disagrees about what an adequate trial requires or about the real risk of getting there carelessly — you disagree about whether we can afford to find out. We can structure the trial instead of choosing a side. Titrate over the next three to four weeks toward 800–1000mg a day of levodopa in divided doses, keep him in physical therapy through the escalation specifically for fall mitigation, check standing blood pressure at every visit rather than waiting for symptoms, and set an explicit stopping rule now — a systolic drop past 20mmHg or a fourth fall halts the trial at whatever dose he's reached, and we treat that as his real answer rather than pushing further to hit an arbitrary number. If the falls and orthostatic changes are already doing the diagnostic work my geriatrics colleague is worried about, this protocol will make that clear within a month without our having pretended the trial never needed to happen at all.
Agreed: titrate carbidopa-levodopa toward 800–1000mg/day of levodopa over three to four weeks, with physical therapy and standing vitals at every visit and an explicit stopping rule set today rather than decided reactively later.
Not agreed, and left explicitly open: whether a partial response at a lower dose than the full target should count as informative on its own, or whether only reaching the full target (or the stopping rule) qualifies as a real answer — the geriatrician wants any meaningful gait improvement, even partial, treated as diagnostically useful; the movement disorder neurologist worries a partial read at half-dose invites the same premature-conclusion problem that brought him to this visit in the first place. They agreed to revisit that specific disagreement once dosing data exists to argue about, rather than resolve it hypothetically today.