BCG-Unresponsive Disease: Bladder Preservation or the Operation That Cures Today
Two patients with BCG-unresponsive disease sit on opposite sides of the same approved indication — one just outside it, one squarely inside it — and neither position turns out to be the whole answer.
T.M., a 57-year-old man, has coached his son's high school wrestling team for eleven seasons and still steps onto the mat himself to demonstrate a takedown, a habit his wife has been asking him to retire for years since a torn rotator cuff two winters ago. He completed a full six-week BCG induction and a full year of maintenance for high-grade Ta bladder cancer, and his three-month surveillance cystoscopy after maintenance ended found the same lesion in the same location — a pattern that reads less like a fresh recurrence than like disease that never actually cleared, and meets the formal definition of BCG-unresponsive disease since it persisted through an adequate induction-plus-maintenance course. Random biopsies of the rest of his bladder are negative, and there has never been any carcinoma in situ component at any point in his disease course, a distinction that will end up mattering more than the recurrence itself.
Both FDA-approved bladder-preservation options for BCG-unresponsive disease — pembrolizumab and nadofaragene firadenovec — carry a labeled indication specifically for carcinoma in situ, with or without accompanying papillary tumors. T.M. has papillary tumors and no CIS at all, which places him outside the population either trial's approval was actually built on, even though each trial's own secondary papillary-only cohort reported real activity: nadofaragene's CS-003 enrolled 50 papillary-only patients in that secondary cohort, fewer than half the 103 in the CIS-based cohort its approval actually rested on, and 35 of the 48 evaluable — roughly seventy-three percent — were free of high-grade recurrence at three months, sixty percent at twelve. That cohort also used a recurrence-free-survival endpoint rather than the complete-response endpoint the label was built around, so the two populations differ in what 'activity' even measures as well as in how many patients stand behind the number. Pembrolizumab's KEYNOTE-057 cohort B, papillary disease without CIS, is the larger of the two off-label datasets at 132 patients, but its twelve-month disease-free survival was 43.5 percent — a real signal, and a markedly more sober one than the three-month nadofaragene figure it usually gets quoted alongside. The off-label evidence for T.M. is not thin, but it is smaller and less mature than the on-label evidence, and the two numbers most often cited for it are not measuring the same thing.
In clinic, at the edge of an approved indication
I'd recommend radical cystectomy now. Neither pembrolizumab nor nadofaragene is actually approved for what he has — both labels are written for CIS, with or without papillary tumors, and he has papillary tumors with no CIS at all. BCG-unresponsive high-grade Ta carries real progression risk the longer it sits untreated, and he's fit enough to tolerate the operation well. I don't think we should be extending an approved indication past its own edge on a patient who can have the curative surgery today.
I'd want to offer him a bladder-preservation attempt before we take that step. Both trials' secondary cohorts — papillary disease without CIS — showed real activity: nadofaragene's papillary cohort had roughly seventy-three percent free of high-grade recurrence at three months, sixty percent at twelve. I'll concede the size point before you make it — that cohort is fifty patients against the CIS cohort's hundred-plus, and pembrolizumab's larger papillary cohort reported a more modest forty-three percent disease-free at a year. He's fifty-seven, active, and a permanent urinary diversion is a real and lasting cost. I think it's reasonable to name this honestly as an off-label extension and still offer it, the way oncology routinely does when a secondary cohort's data is this encouraging.
Calling it 'the edge of an approved indication' undersells how much real data exists specifically for his phenotype. It isn't invented off-label use — it's a secondary, prespecified cohort in the same pivotal trial, just not the cohort the label was written around.
Both of you are right about different pieces of this, which is exactly the problem. It's true the papillary-only data is real and prespecified, not a small case series someone is stretching to fit him. It's also true that neither agent's FDA label covers his disease phenotype, which matters for more than paperwork — the papillary cohorts are smaller, less mature, and weren't the population either trial's approval decision actually rested on. I don't think the honest answer is 'approved' or 'not approved' as a stand-in for the real question, which is whether HIS disease — papillary-only, no CIS, a single prior BCG course — looks like the patients in that secondary cohort who actually responded. If he wants to try it with that framing made explicit, I don't think that's unreasonable; I don't think it's obviously the safer choice either.
T.M. chose a nadofaragene trial with full understanding that the drug's label doesn't specifically cover his disease phenotype, with a firm agreement to proceed to cystectomy at the first sign of persistence or progression rather than cycling through further bladder-preservation attempts.
V.K., a 63-year-old woman, retired from three decades as a court reporter last spring and has spent most of her newfound time restoring a vegetable garden that her work schedule never let her tend properly, though a recent flare of hematuria interrupted a full weekend of transplanting. Her bladder cancer history began two years ago with carcinoma in situ, treated with a full BCG course; her most recent surveillance cystoscopy, six months after maintenance ended, found both persistent CIS and a new high-grade T1 lesion invading the lamina propria — a second BCG-unresponsive recurrence, now with a component her first diagnosis never had, and a meaningfully different biological picture than the pure CIS she started with.
That T1 component changes the calculus in a way her CIS alone would not. Both pembrolizumab and nadofaragene are squarely on-label for her: their pivotal trials enrolled CIS with or without accompanying high-grade Ta or T1 disease, and each reported meaningful complete-response rates — pembrolizumab's KEYNOTE-057 cohort A at forty-one percent with a median response duration of sixteen months, nadofaragene's CS-003 at fifty-one percent with roughly forty-six percent of responders still recurrence-free at a year. But T1 disease carries a real risk that pure CIS does not: lamina propria invasion is frequently understaged on transurethral resection alone, and a meaningful fraction of clinical T1 disease turns out to be occult muscle-invasive cancer once a cystectomy specimen is examined in full. Her own repeat resection specimen, notably, contained no detrusor muscle at all — the single tissue layer that would actually let anyone rule out invasion beyond the lamina propria — which means the T1 designation she's carrying right now is provisional, not confirmed. A bladder-preservation attempt that fails here has not just delayed a difficult decision; it has spent months of a window in which her disease may already have been more advanced than an incomplete biopsy could show.
In clinic, a curable window with T1 in it
She's exactly the population both pivotal trials were built on — CIS with a high-grade T1 component, BCG-unresponsive after an adequate course. Pembrolizumab's response rate is durable in a meaningful fraction of responders, median sixteen months in KEYNOTE-057; nadofaragene's CS-003 showed similar numbers with a quarterly instillation schedule she might tolerate more easily. A real shot at keeping her bladder, on an agent actually approved for her exact disease, is a legitimate offer to make her.
Being on-label doesn't change what T1 invasion means biologically. Lamina propria invasion is understaged often enough that a meaningful fraction of clinical T1 turns out to be occult muscle-invasive disease once you actually open the bladder, and her repeat resection didn't even capture detrusor muscle to rule that out properly. If we spend three to six months on a bladder-preservation trial and she doesn't respond, we haven't just lost time — we may have let an already-invasive cancer sit untreated through a window where cystectomy today would have been curative.
Roughly half of responders in each trial stayed recurrence-free at a year, which is a real number, not a coin flip dressed up as one — but a real response rate in the trial population isn't the same guarantee for a patient whose T1 component specifically raises the stakes of being wrong.
I'd frame the actual disagreement narrower than approved-versus-not, since neither of you is disputing the label here. The real question is whether her T1 component, specifically, predicts poorly enough to override an on-label option most patients in her exact disease category are legitimately offered. Neither trial's subgroup data lets us confidently say her T1 makes her meaningfully less likely to respond than a CIS-only patient — T1 was part of both trials' enrolled population, not a post-hoc addition. What I can say is that a repeat resection specifically aimed at ruling out occult muscle invasion, done properly this time with detrusor muscle in the specimen, would answer the question that actually matters here before either of you commits her to a plan built on an incomplete restaging.
Agreed: a proper repeat resection with dedicated detrusor sampling before committing to either bladder-preservation agent or to cystectomy — addressing the urologic oncologist's central objection directly rather than arguing around it.
Nadofaragene or pembrolizumab remains a reasonable on-label bladder-preservation attempt, with a firm plan to proceed to cystectomy at the first sign of non-response.
The bladder-preservation question is moot; she proceeds directly to cystectomy and staging for muscle-invasive disease.