Adrenal SBRT Tumor Control Probability Modeling (HyTEC)
ForAdrenal metastases treated with SBRT, no prior adrenal RT
TL;DRPoisson TCP fit to pooled adrenal SBRT series: 1-yr LC >95% at BED10 116.4 Gy (EQD2 ≈97 Gy), fit P<.0001.
The 95% LC point sits at BED10 116.4 Gy (EQD2 ≈97 Gy), reachable with 45 Gy/3 fx (112.5) or 55 Gy/5 fx (115.5); NRG-BR001 already prescribes 45 Gy/3 fx. Dose inputs were IDL-corrected median prescriptions, not GTV DVHs, and every series used motion management, so the target transfers only with ITV, compression, breath-hold or gating in place.
For a previously unirradiated adrenal metastasis where adjacent OAR tolerances allow, this supports planning above the 40 Gy/5 fx series median toward the modeled LC target; it does not extend to adrenal reirradiation or to choosing SBRT over adrenalectomy.
The 95% LC point sits at BED10 116.4 Gy (EQD2 ≈97 Gy), reachable with 45 Gy/3 fx (112.5) or 55 Gy/5 fx (115.5). Inputs were IDL-corrected median prescriptions, not DVHs, and every series used motion management, so the target transfers only with ITV, compression, breath-hold or gating.
10 details 1 trial watching
HyTEC literature review plus dose-response model. PubMed search Jan 2007 to Mar 2017 returned 250 results; included series ranged 6 to 49 lesions (6-48 pts). 1-yr LC fit to a Poisson TCP model by maximum likelihood, CIs by profile likelihood.
Adrenal metastases, lung the dominant primary, plus kidney, liver, melanoma and others. Mostly single-institution retrospective series. All but 1 series excluded prior adrenal RT.
Median dose 40 Gy (mean 37.2 Gy), median 5 fx (range 1-27), prescribed to the 70% to 100% IDL. PTV margins 0 to 10 mm. All series reported motion management: 4DCT/ITV, compression, breath-hold, gating or CyberKnife tracking.
Modeled endpoints were 1-yr LC and 1-yr OS. LC was RECIST-based in most series; 3 used PET/CT progression.
LC fit P<.0001; 1-yr LC >95% at BED10 116.4 Gy (EQD2 ≈97 Gy). OS also showed a dose-response (P<.001) with wide CIs.
| Parameter | Local control | Overall survival |
|---|---|---|
| D50 (Gy EQD2) | 34.6 (95% CI 27.1-45.4) | 44.7 (95% CI 33.5-60.6) |
| γ50 | 0.4996 (95% CI 0.248-0.814) | 0.4546 (95% CI 0.157-0.796) |
| Fit χ2 | 57.5, P<.0001 | 44.7, P<.001 |
| Schedule | BED10 |
|---|---|
| 45 Gy / 3 fx | 112.5 Gy |
| 55 Gy / 5 fx | 115.5 Gy |
| Casamassima: 36 Gy / 3 fx to 70% IDL | just over 139 Gy at isocenter |
A later meta-analysis (39 studies) predicted 1-yr LC of 70.5%, 84.8% and 92.9% at BED10 60, 80 and 100 Gy, within this model's CI. Direction matches the BED10 ≥100 Gy LC association reported for SBRT in early-stage NSCLC. NRG-BR001 (phase 1) prescribes 45 Gy/3 fx.
Competing distant-death risk in small cohorts degrades the 1-yr LC estimates. Neither PTV margin nor IDL could be tied to LC, so the model gives no margin guidance. Circle-weighted study points, not patient-level data.
Durability is the open issue: two later series at BED10 85.5 and 72 Gy lost LC at 2 and 3 years, suggesting dose affects durability. The OAR feasibility of reaching BED10 116.4 Gy near bowel and stomach is left to the planner.
Dose-response fit to small retrospective series using study-level median prescriptions, not patient DVHs; no prospective validation, so the BED10 116.4 Gy target is model-derived.
- Durability of LC beyond 1 year at BED10 116.4 Gy n=397 · primary completion 2028-06 · MR-guided SBRT, adrenal met cohort, tumor control endpoint
- SBRT vs adrenalectomy for adrenal oligometastases
- Optimal PTV margin and motion management for adrenal SBRT
📚 Sources · 📄 1 paper
The longer read
The useful output here is a number adrenal SBRT practice lacked: a modeled dose at which 1-year local control clears 95%, placed at BED10 116.4 Gy. The series it pools prescribed across a wide range (median 40 Gy, 1 to 27 fractions) and reported 1-year LC anywhere from 44% to 100%, with no way to tell how much of that spread was dose and how much was selection. The Poisson fit says a real share of it tracks with dose, and the authors read a dose response for both LC and OS despite the heterogeneity.
The direction fits what the field already believed from SBRT in early-stage NSCLC, where BED10 at or above 100 Gy has been tied to better LC. The later meta-analysis the paper cites, built on 39 studies, predicts 92.9% 1-year LC at BED10 100 Gy, inside this model's confidence band. That agreement reassures about the fit but is not independent validation: both analyses draw largely on the same retrospective literature.
The methodological choice that should most temper confidence is the dose input. Without DVHs, each study contributes one point, its median prescription corrected for IDL when the IDL was reported. Within-study dose heterogeneity collapses, and an IDL-corrected prescription is a coverage estimate rather than a true minimum or mean GTV dose. The Casamassima series shows how wide that gap can be: 36 Gy in 3 fractions to the 70% IDL corresponds to BED10 just over 139 Gy at isocenter, with BED10 79.2 to 139 Gy across the PTV. A model with that much uncertainty on its x-axis can place the curve's midpoint reasonably, but its upper tail, where the 95% target sits, is the least constrained region of the fit.
Two further caveats shape transfer. Every included series used motion management, so the target presumes ITV, compression, breath-hold or gating; it says nothing about delivery without accounting for the craniocaudal excursion adrenal tumors show. And the endpoint is 1-year LC in small cohorts with heavy competing mortality, which censors late failures. The observation that two later series at BED10 85.5 and 72 Gy lost control at 2 and 3 years hints that the dose needed for durable control may be higher than the dose needed for 1-year control, which would push the target up, not down.
What the report leaves untouched is the surgery question. Adrenalectomy cohorts in the cited comparison carried more isolated adrenal disease and fewer lung primaries, so the reported surgical advantage is confounded, and no randomized comparison exists. It also offers no OAR-constrained feasibility data: whether 45 Gy in 3 fractions or 55 Gy in 5 can be delivered beside bowel, stomach and kidney is the planner's problem. NRG-BR001, prescribing 45 Gy in 3 fractions, is the nearest prospective test of the dose this model favors, and its LC data are what would confirm or move the target.