Resectable Stage III Melanoma: Treating Before Surgery, Not Just After It
A single patient with a growing, biopsy-proven nodal metastasis and a genuinely new question about when checkpoint therapy should start relative to the knife. The disagreement is about whether a real trial's population advantage should override a real, individual worry about one specific node's own trajectory.
G.F., a 63-year-old man, has rewired more houses in this county than he can count and was mid-project on a client's kitchen panel when his wife noticed a firm lump in his groin that hadn't been there a month earlier. He'd had a thin melanoma removed from his lower back eight years ago and was told at the time it was fully excised with clear margins — a diagnosis he'd genuinely forgotten about until a core biopsy of the new inguinal node came back as metastatic melanoma. Staging imaging found disease confined to a single enlarged node cluster, 3.8cm at its largest, with no distant metastases: resectable stage IIIC disease.
The node has grown measurably in the four weeks between his primary care visit and this consultation — from an estimated 2.6cm on his doctor's exam note to today's 3.8cm on CT, a rate of change that is exactly what makes the timing question here more than academic. SWOG S1801 randomized patients with resectable stage III/IV melanoma to either three doses of pembrolizumab before surgery followed by fifteen more after, or eighteen total doses of pembrolizumab starting only after surgery — and found two-year event-free survival of 72 percent with the neoadjuvant approach against 49 percent with adjuvant-only, a gap driven largely by fewer patients relapsing in the weeks immediately surrounding surgery itself. The mechanism the trial's own investigators point to is straightforward: treating while the full tumor and its draining lymphatics are still present exposes the immune system to more antigen than treating a surgical bed with disease already removed.
Nothing about that argument assumes the node holds still for three weeks while neoadjuvant dosing runs its course, and his node's own measured growth rate over the past month is the fact working against a straightforward extrapolation from the trial's overall result. S1801's own reported pathologic-response data give the team something concrete to watch for regardless of which timing wins the argument: patients who achieved a major pathologic response at surgery — less than ten percent viable tumor remaining — had a two-year recurrence-free rate above 90 percent, a number that, if his own surgical specimen reaches it, would answer the adjuvant-duration question almost by itself.
Whether a growing node changes when to treat
S1801 gives us a genuine, randomized answer here: three doses of pembrolizumab before surgery, fifteen more after, against eighteen doses starting only post-operatively. Two-year event-free survival was 72 percent with the neoadjuvant sequence against 49 percent adjuvant-only — not a marginal difference, and the mechanism makes sense: the immune system sees more of the tumor and its draining lymphatics intact than it would see a surgical bed alone. I'd start neoadjuvant dosing this week.
I've read the same trial. What I'm looking at that the trial's average patient doesn't hand me is a node that grew from 2.6 to 3.8 centimeters in four weeks. If that rate holds through a three-dose neoadjuvant course, we could be operating on a considerably larger, potentially less cleanly resectable field than the one in front of us today — or watching it extend beyond the node cluster we can currently see clearly on imaging.
I take the population-level argument seriously. I don't think it answers what this specific node is doing right now, and 'the trial's average outcome was better' isn't the same claim as 'every individual node's growth trajectory is safe to wait on.'
The distinction I'd draw is that S1801's result isn't just an average across patients who all held perfectly still — it's an intention-to-treat result that already includes the real minority who progressed during the neoadjuvant window before ever reaching surgery, and the neoadjuvant arm still won. The trial has already absorbed a version of exactly the scenario you're worried about into its own numbers.
So I'd proceed with neoadjuvant dosing, but I don't think that has to mean proceeding blind for three weeks. Interval exam and imaging at two weeks gives us a real checkpoint — if the node is clearly outpacing what the trial's own progressing subgroup looked like, that's new information justifying an earlier move to surgery. That's not abandoning the evidence; it's using the same standard the trial itself used to decide who benefited and who didn't.
Agreed: begin neoadjuvant pembrolizumab this week, with an explicit interval exam and imaging checkpoint at two weeks to reassess whether the node's growth is tracking within the range S1801's own progressing subgroup showed, or clearly outpacing it — in which case surgery moves up rather than waiting for the full three-dose course.