Three Years Normal on Cabergoline: Is a Residual Shadow Reason Enough to Keep Treating
A single patient, three years into normal prolactin on cabergoline, asking to stop. The disagreement isn't about whether withdrawal is ever reasonable — it's about whether a residual sliver still visible on her MRI is the specific fact that should keep her on the drug longer than the guideline's own general timeline suggests.
L.F., a 36-year-old woman, manages produce ordering for a regional grocery chain, work she describes as "spreadsheets and trucks," and has spent the last three years quietly counting down to this appointment the way some people count down to a lease ending. She was diagnosed with a 14mm macroprolactinoma four years ago after eighteen months of irregular periods and unexplained galactorrhea, with a prolactin at diagnosis of 340 ng/mL. Cabergoline, titrated to 1 mg weekly, normalized her prolactin within four months and has held it normal — most recently 8.2 ng/mL — for the three years since, with her periods regular and the galactorrhea long resolved. She has read, correctly, that current guidance considers dopamine-agonist withdrawal after two years of normoprolactinemia and substantial tumor shrinkage, and she wants to try.
The complicating fact is her most recent MRI, obtained six weeks ago specifically to inform this conversation: the original 14mm lesion has shrunk to a 4mm remnant, more than the 50% reduction the withdrawal criteria call for, but it is still visible, not the complete radiographic disappearance some of her original tumor-outcome data described. That distinction carries real weight. The pivotal withdrawal study — Colao and colleagues' 200-patient prospective cohort in the New England Journal of Medicine in 2003 — found recurrence at five years in 78% of macroprolactinoma patients with a small remnant still visible at withdrawal, against 33% in those whose MRI showed no tumor at all. Remnant size was the predictor that analysis actually isolated, at roughly a 19% increase in recurrence risk for every additional millimeter. L.F.'s own numbers put her on the wrong side of that specific line, even though every other criterion for withdrawal is met. Her own baseline prolactin at diagnosis, 340 ng/mL, sits well above the range some of the withdrawal literature has separately flagged as an independent predictor of recurrence, a second data point that argues the same direction as her remnant finding rather than offsetting it.
Stopping a drug that's working, because it's working
I'd offer a supervised withdrawal attempt. She meets the two-year normoprolactinemia threshold with room to spare, her shrinkage clears the 50% bar comfortably, and the Endocrine Society's own hyperprolactinemia guideline treats withdrawal after two years of successful therapy as reasonable practice, not an experimental step.
She's also been asking about this for a year — medication burden and the cost of quarterly monitoring are real considerations for her, not trivial ones, and a guideline built around meeting stated criteria doesn't have a separate, harder bar for "small but present" residual tissue.
The guideline's criteria are a floor, not a guarantee, and the actual outcome data draws a much sharper line inside that floor than "met the threshold or didn't." The NEJM cohort found 78% five-year recurrence in macroprolactinoma patients with any visible remnant at withdrawal, against 33% with a fully clear MRI — that's not a marginal difference, it's more than double, and it's the single strongest predictor the study found.
I'd read her chart the same way that data does: she's cleared the shrinkage percentage, but she hasn't cleared the actual finding the study identified as decisive, and offering withdrawal today without naming that gap to her directly risks her hearing "you qualify" when the more honest read is "you qualify on the criterion that turned out to matter less."
That's a fair correction to how I'd frame the offer, not a reason to withhold it. She can still choose withdrawal with that specific number in hand — 78%, not the vaguer "recurrence is possible" — and choose it as an informed decision rather than a default one. A failed attempt isn't harmful in itself; restarting cabergoline after recurrence has consistently normalized prolactin again in essentially every reported case, hers included the first time.
Agreed: offer L.F. a supervised withdrawal attempt, but with the remnant-specific recurrence figure (78% at five years with any visible residual, versus 33% with a fully clear scan) stated to her explicitly as part of informed consent, rather than folded into a general "you meet criteria" conversation. Prolactin every 3 months for the first year if she proceeds, MRI repeated at 12 months regardless of biochemical outcome.
Not agreed: whether the visible remnant itself is a relative contraindication that should have been raised as a reason to defer withdrawal by another year of monitoring rather than proceed now with a clearer disclosure. The reproductive endocrinologist would have preferred to wait for a cleaner scan before offering the choice at all; the endocrinologist views an informed patient's own decision, made with the accurate number in hand, as the more defensible endpoint than deferring the choice on her behalf.