Organoid screens before surgery: the POLO trial design

What the work claims

This is a trial registry entry, not a result, and everything below should be weighted accordingly: it documents intent, design, and institutional commitment, not efficacy. The claim embedded in the design is that a drug sensitivity screen on a patient-derived organoid (PDO) can be inserted into the clinical pathway for low-grade non-muscle invasive bladder cancer (NMIBC) early enough to change treatment, and that this can be done in a community-hospital setting rather than a dedicated precision-oncology center.¹

The design is a prospective, open-label, single-arm, single-center phase II trial of 28 evaluable patients. Bladder tumor tissue is sampled, in the registry's description during transurethral resection and, for the trial proper, by outpatient biopsy. Organoids are generated and screened against four established intravesical agents: epirubicin, mitomycin C, gemcitabine, and docetaxel. The patient then receives six instillations of the screen-selected agent as neoadjuvant treatment intended to achieve chemoresection, that is, chemical destruction of the tumor in place. Three months after diagnosis, the standard transurethral resection (TURBT) is performed both as treatment and as the response readout.¹

The primary endpoint is pathological response at 15 weeks. The secondary endpoints are telling in what they reveal about the team's own risk assessment: feasibility of the drug screen within 4 weeks, safety at 12 weeks, tolerability of instillation at 24 weeks, and recurrence-free survival at one year.¹

How it works

The clinical logic rests on a genuine unmet need that the registry states plainly. Low-grade NMIBC is treated by TURBT, recurs in roughly half of patients within two years, and requires cystoscopic surveillance for at least five years. Progression from low-grade disease to higher grade or stage is infrequent to rare, which makes an aggressive surgical-first pathway look disproportionate for selected patients, particularly the elderly. Intravesical chemoresection has been attempted before, but, as the registry notes, all patients in those attempts received the same chemotherapeutic agent, and the anticipated response rates were missed.¹

The scientific bet is that the misses were a selection failure rather than a drug failure. NMIBC is molecularly heterogeneous in both mutation pattern and transcriptomic character, and the four agents used for instillation have never been administered on the basis of a biological likelihood of response. The POLO design converts that hypothesis into a workflow: organoid established from the individual tumor, four-drug screen, most effective agent carried into the bladder. The team reports prior groundwork, having validated that PDOs from about 30 NMIBC samples reproduce the molecular landscape of the parent tumors and having standardized a drug-screen protocol on them.¹

Two operational numbers define the whole enterprise. The screen must complete within about four weeks, because a neoadjuvant window cannot stretch indefinitely in a patient with an active tumor, even a low-grade one. And the organoid must establish from an outpatient biopsy, not just from resection material, because the trial's premise is to guide treatment before surgery, not after it.

Where a skeptic should push

The single most load-bearing assumption is that in vitro organoid sensitivity ranks intravesical response. This is a stronger assumption than the one behind most PDO-guided systemic therapy trials, and it deserves more scrutiny than it usually gets. Systemic therapy exposes tumor cells to drug concentrations approximated, however imperfectly, in culture. Intravesical therapy is pharmacologically different in kind: the drug is delivered topically at very high concentration, for a dwell time measured in hours, into an environment whose urine chemistry, tumor contact geometry, and urothelial barrier an organoid plate does not reproduce. An organoid measures intrinsic cellular chemosensitivity. Whether it ranks response to a topical, dwell-limited exposure is an open question the trial's design cannot cleanly answer, because there is no comparator arm receiving physician-choice or uniform-agent therapy.

Second, the evidentiary frame. Twenty-eight patients, single-arm, non-randomized, with pathological response at 15 weeks as the primary endpoint and recurrence-free survival at one year as a secondary. Low-grade NMIBC has a slow natural history; one year is short, and a single arm cannot separate chemoresection benefit from the selection of patients with indolent disease, since the entry criteria (suspected low-grade, negative urine cytology) already filter toward favorable biology. The trial can demonstrate feasibility and a response signal. It cannot demonstrate that organoid-guided selection beats the historical standard, and its own cited history, uniform-agent chemoresection missing its response targets, is the cautionary tale for reading single-arm response rates optimistically.¹

Third, the operational chain has several silent failure points that the registry does not fully describe: what happens to a patient whose organoid fails to establish or whose screen fails the four-week window, whether screen-selected agents ever tie, and how the analysis handles screen-outcome discordance. The feasibility endpoint suggests the team is watching establishment and turnaround; the reader should treat the primary endpoint's denominator as an open question until results appear. Timing is also worth flagging: the registry currently lists the trial as not yet recruiting, with an estimated start of December 2026 and completion in late 2031.

What organoid screens must prove before surgery moves

For organoid models of human organs and the drug-discovery work built on them, POLO is significant less for what it will find than for where it sits in the causal chain of treatment. Most PDO-guided therapy trials use organoids as a rescue instrument, ranking options for patients who have exhausted standard care. POLO puts the organoid upstream of the first definitive treatment, in a position to displace surgery in a low-risk cancer. That is the highest-leverage and highest-scrutiny location an in vitro assay can occupy, and if the workflow performs, it becomes a template for any organ cancer where a topical or localized therapy exists and where surgery carries real morbidity.

The non-obvious opportunity is in the secondary endpoints. A 4-week feasibility endpoint and a tolerability endpoint turn the trial partly into a measurement of the organoid assay itself as a clinical-grade instrument: turnaround, establishment rate from outpatient biopsies, and reproducibility under time pressure. If POLO publishes those numbers honestly, the field gets operational benchmarks that matter more to adoption than any response rate, because every hospital considering PDO-guided therapy is silently asking the same questions this trial pre-registers.

The threat is reputational and structural. If a 28-patient single-arm trial produces a modest or ambiguous response rate, the result will be read, in headlines and in hospital committees, as organoid-guided therapy failing in bladder cancer, when the design would not have been able to show benefit even if it existed. Conversely, a striking single-arm response rate in selected low-grade disease will be read as validation, when the same selection that produced it guarantees nothing about high-grade or muscle-invasive disease. Either misreading slows the field. There is also a governance point: this is precisely the stage at which a confirmatory randomized comparison against TURBT-first should already be sketched, and nothing in the registry suggests it is.

For the organoid platform vendors and core facilities watching this space, the durable lesson is about assay specification. A screen intended to rank intravesical agents should, ideally, be validated against intravesical pharmacology: dwell-time-limited exposure, urine-like medium, and contact geometry. An assay validated only on continuous-dose plating may rank the right drugs for the wrong reason. POLO will not test that distinction, and the next trial in this space should.

The bottom line

Established: a credible team with prior PDO validation work in roughly 30 NMIBC samples has designed, and is preparing to run, a phase II trial that inserts a four-drug organoid screen ahead of surgery in low-grade bladder cancer, with pre-registered feasibility and safety endpoints. Not established, and not testable by this design: that screen-selected intravesical therapy outperforms uniform or physician-choice therapy, or that organoid sensitivity predicts response to topical dwell-limited drug exposure at all. What would confirm the model: a randomized comparison showing screen-guided selection beats the historical response benchmarks the registry itself cites, with screen turnaround and establishment rates high enough to treat the pre-surgical window as routine. What would break it: low establishment rates under the 4-week feasibility window, or a response rate in line with the uniform-agent attempts the trial is designed to improve on. Until those data exist, POLO is best read as a carefully instrumented feasibility probe, and its most valuable results may be about the assay rather than the cancer.