Focal Ablation vs IMRT Toxicity (SEER-Medicare)
ForLocalized prostate cancer, fee-for-service Medicare, mean age ~73
TL;DRGI toxicity favored focal therapy at 24mo (11.0% vs 21.5%, OR 0.45) but GU toxicity favored IMRT (41.7% vs 29.3%, OR 1.69).
The GI advantage rests on diagnosis claims alone: procedure-code-only GI events were too few to report and not significantly different, while the GU excess with FT was concrete (incontinence therapy 17.4% vs 7.5%). For an RT reader counseling an older man weighing ablation, that asymmetry is the point.
In an older man with localized prostate cancer choosing between focal ablation and IMRT, this supports counseling that the tradeoff runs GU-for-GI rather than uniformly lower toxicity with ablation; it does not extend to SBRT or proton patients, who were excluded.
The IMRT comparator is a 2010-2017, ≥20-fraction cohort with SBRT, protons and rectal hydrogel excluded, so the GI toxicity attributed to radiation predates spacer-era rectal sparing. That caps how much of this GI gap should carry into a current consent conversation, while the GU result (incontinence therapy 17.4% vs 7.5% favoring IMRT) transfers intact.
For ablation, the toxicity cost is concrete and periurethral: incontinence therapy 17.4% vs 7.5% and erectile dysfunction 18.2% vs 13.1%, both P<.01. The offsetting GI advantage collapsed in the procedure-only analysis, so the low-morbidity premise for cryotherapy and laser ablation in older men is weaker than the headline suggests.
10 details 5 trials watching
Retrospective SEER-Medicare claims analysis of localized prostate cancer diagnosed 2010-2017. 797 focal therapy patients Mahalanobis matched 2:1 to 1,594 IMRT patients on demographics, Medicaid eligibility, comorbidity, flu vaccination, primary care access, year, cancer characteristics and ADT. Logistic regression at 6, 12 and 24 months.
Fee-for-service Medicare beneficiaries with localized PCa as first and only cancer, continuously enrolled parts A and B from 12 months pre-diagnosis through 24 months post-treatment. New York, Idaho and Massachusetts registries excluded for missing AJCC staging. Patients who died within 24 months of treatment were excluded.
IMRT defined as external beam with no surgery plus ≥20 IMRT fractions and an IMRT planning code; a gap of ≥30 days between radiation dates counted as a separate course. SBRT and proton patients were excluded, as were patients with rectal hydrogel spacer, so the IMRT arm is conventional fractionation without a rectal spacer.
Presence of claims indicative of a GI or GU complication within 6, 12 and 24 months of treatment, assessed both as combined diagnosis and procedure codes and as procedure codes only. No patient-reported outcomes and no graded toxicity scale.
Direction reversed by organ system: GI favored focal therapy from 12 months onward, GU favored IMRT at every window measured. The 0-6 month GI comparison was null (OR 1.06, P=.81).
| Toxicity / window | FT | IMRT | OR (95% CI) |
|---|---|---|---|
| GI 0-6 mo | 3.6% | 3.5% | 1.06 (0.67-1.67), P=.81 |
| GI 0-12 mo | 6.2% | 9.5% | 0.63 (0.45-0.88) |
| GI 0-24 mo | 11.0% | 21.5% | 0.45 (0.35-0.58) |
| GU 0-12 mo | 34.6% | 15.8% | 2.69 (2.21-3.28) |
| GU 0-24 mo | 41.7% | 29.3% | 1.69 (1.42-2.01) |
GU excess with focal therapy was driven by incontinence therapy (17.4% vs 7.5%) and erectile dysfunction (18.2% vs 13.1%), both P<.01. GI events with IMRT were predominantly rectal bleeding and colitis; the authors note related procedures were rare and not significantly different.
Focal therapy patients were disproportionately rural (6.4% vs 1.9%) and lower-income (51.5% vs 38.7% in the lowest two quintiles) before matching, so claim-generating behavior may differ by arm independent of toxicity. Baseline continence and erectile function were unmeasured, which matters most for the two endpoints driving the GU result.
The GI result is fragile in a specific way the abstract does not foreground: it collapses to one diagnosis subcategory and disappears in the procedure-only analysis. The GU result is the more robust half, and it runs against the premise that ablation is the lower-toxicity option.
Claims-based retrospective matching, not randomised; toxicity defined by diagnosis codes, and the GI difference vanished when restricted to procedure codes. No PROs, no baseline function.
- Does the tradeoff hold vs SBRT or spacer-era IMRT? n=184 · primary completion 2027-06 · IRE vs prostatectomy or RT, side-effect endpointn=356 · primary completion 2029-08 · focal therapy vs usual care RCT, ISUP 2/3
- Do patient-reported outcomes track the claims-based GU signal? n=300 · primary completion 2030-12 · prospective validated urinary/sexual PROs after HIFUn=200 · primary completion 2032-03 · questionnaire outcomes across HIFU, cryo, laser, IRE
- Do HIFU and RFA carry the same GU profile as cryo/laser? n=354 · primary completion 2028-12 · HIFU hemi-ablation, continence + toxicity endpoints
📚 Sources · 📄 1 paper
Abstract
The longer read
The framing this paper answers, that focal ablation exists to deliver prostate cancer treatment with less toxicity than radiation, does not survive contact with its own data. The toxicity did not go down; it moved. Focal therapy patients generated fewer GI claims from 12 months onward and substantially more GU claims at every window, and the GU signal is not a diffuse claims artifact: it concentrates in incontinence therapy (17.4% vs 7.5%) and erectile dysfunction (18.2% vs 13.1%). Those are the two outcomes an older man is usually weighing when he is offered ablation instead of radiation, which makes this the most decision-relevant finding on the page.
The two halves of the result deserve unequal confidence, and reading them as symmetric is the main way to misuse this paper. The GI advantage for focal therapy rests entirely on diagnosis codes. When the authors restricted to procedure codes, GI events were too few to report and the difference was not statistically significant, and the entire GI signal traces to a single subcategory, "other gastrointestinal inflammation or unspecified," at 10.4% versus 21.1%. A diagnosis code for rectal bleeding in a man known to have had pelvic radiation is exactly the kind of claim that gets generated because the clinician is looking for it. The GU result has firmer footing because incontinence therapy is an act, not an impression.
Methodologically the matching is careful, and standardized differences below 10% across every matched variable is a real achievement, but matching can only balance what is coded. The pre-match distributions show focal therapy patients were three times more likely to be rural and considerably more likely to be in the lowest income quintiles, and none of the matched covariates capture baseline continence, erectile function, or prostate size, the three variables that most plausibly drive both treatment selection and the GU endpoints. The exclusion of anyone who died within 24 months of treatment is a defensible way to guarantee complete follow-up, but it selects toward healthier patients in both arms and limits how far the estimates travel.
For a radiation oncologist the practical caveat is what the IMRT arm is. Patients required at least twenty fractions, and SBRT, protons, and rectal hydrogel spacers were all excluded by design. That is a deliberately clean comparator, but it is also a 2010-2017 conventional-fractionation cohort without a spacer, which is not what most departments deliver today. The GI toxicity attributed to IMRT here is measured against a technique whose rectal dose profile has since changed, so the modest GI gap that survives is likely the ceiling of the advantage, not the current estimate. The GU comparison is less vulnerable to that drift, since the FT-side mechanism is periurethral and does not depend on which radiation era the control arm belongs to.
What this settles is narrow and worth stating plainly: in older Medicare men, ablation is not a lower-toxicity option in aggregate, and any consent conversation presenting it that way is unsupported. What it leaves open is whether the tradeoff looks the same against modern hypofractionated or SBRT delivery with a spacer, and whether patient-reported function tracks the claims signal in either direction.