Five Years or Ten: Betting Against a Recurrence Curve That Never Flattens
Five years of letrozole are almost finished, and the recurrence curve for hormone-receptor-positive breast cancer famously refuses to flatten at that mark. Extending to ten years lowers that ongoing risk — at the cost of five more years in a bone and joint side-effect profile she has already found hard to tolerate.
Diane K., a 61-year-old new grandmother, has spent the last several months organizing her Tuesdays and Thursdays around a two-year-old who calls her "Gigi" and has strong opinions about which park has the better slide — the kind of ordinary domestic reorganization that has nothing to do with her cancer and everything to do with why she is now asking pointed questions about how much longer she actually has to keep taking a pill that leaves her hips aching by the end of those park visits. She was diagnosed five and a half years ago with a node-positive, strongly hormone-receptor-positive tumor, completed four cycles of AC followed by twelve weeks of paclitaxel and then radiation, and started letrozole about eight months after diagnosis; with four years and ten months of it behind her, the decision about whether to stop at five years or continue to ten is now genuinely in front of the team.
The reason this decision resists a clean answer is that hormone-receptor-positive breast cancer's own recurrence curve does not do what a five-year survivor might reasonably expect it to do — it does not flatten out near zero at the five-year mark the way some other cancers' curves do. A large EBCTCG meta-analysis following node-positive, ER-positive patients out to twenty years found recurrences continuing to accrue steadily well past year ten, meaning five years of endocrine therapy leaves real, ongoing risk on the table rather than resolving it. MA.17R tested exactly her situation — four and a half to six years of prior aromatase-inhibitor therapy, and she is at four years and ten months — and found a real disease-free survival benefit from extending. Two things travel with it: no overall survival difference at five years, 93% against 94%, and much of the disease-free gain coming from prevented contralateral primaries rather than distant recurrence. The node-positive survival finding usually attached to extended letrozole is not MA.17R's at all; it belongs to the earlier MA.17, where letrozole after tamoxifen improved survival specifically in node-positive women. So her two positive nodes place her in the population that benefited in MA.17, not in a subgroup MA.17R itself identified. Set against that benefit is her own DEXA scan, already showing osteopenia at the hip after four years of aromatase inhibition, and a cardiovascular risk profile that another five years of estrogen deprivation will not improve. What tips it is not that both arguments are real but that they are arguments about different endpoints: the recurrence curve she would be treating is largely a curve of second primaries, and the bone she would be spending is the bone she stands on.
Survivorship clinic, the five-year conversation
I want to extend her to ten years, and I want to be exact about which trial is carrying that, because the two get run together constantly. MA.17R is the one that actually tested her situation — a full course of an AI, then five more years — and it showed a real disease-free gain. The node-positive overall survival signal is MA.17's, out of the tamoxifen-first population, and I'm not going to borrow it for her. What carries her case is the EBCTCG data: node-positive ER-positive disease keeps accruing recurrences past year ten, so five years is a round number, not a biological endpoint.
I don't dispute the recurrence data, and osteopenia at a T-score of -2.1 after four years is a trajectory rather than a one-time finding — five more years of aromatase inhibition will plausibly carry her into osteoporosis before she's seventy.
But I want to press on the framing you were careful enough to offer, because it cuts harder than you may intend. If MA.17R's disease-free gain came substantially from preventing new contralateral primaries, and there was no overall survival difference at five years, then what we are offering her is a reduced chance of a second, separately-treatable breast cancer — not a demonstrated reduction in dying of this one. That is still worth having. It is not obviously worth a hip fracture at sixty-eight, and a disease-free survival curve doesn't price that trade for her.
Both of you are right about your own numbers, which is why I don't think this is actually an "extend or don't" decision — it's a "extend with what protection" decision. Zoledronic acid, given for bone protection during continued AI therapy, has real trial support for slowing exactly the density loss the endocrinologist is worried about, without requiring Diane to give up the recurrence-risk reduction the oncologist is arguing for.
That doesn't make the tradeoff disappear — she'd be adding a second drug with its own infusion schedule and a small renal-monitoring burden — but it changes the actual choice in front of her from a binary to a three-way one, and I think she deserves to hear it framed that way rather than as a straight yes-or-no on letrozole.
Agreed: letrozole extended to ten years with zoledronic acid added for bone protection, and a repeat DEXA scan at eighteen months rather than the standard two-year interval, specifically to catch an accelerating decline early.
Not agreed, and named as such rather than smoothed over: the endocrinologist remains uncomfortable extending AI therapy in any patient already showing osteopenia this early in the course, zoledronic acid or not, and asked that this be documented as a standing reservation rather than a resolved question, in case the eighteen-month scan doesn't look the way the group hopes.