HDR brachytherapy with custom-made surface molds for facial skin carcinomas
ForFacial BCC/SCC, primary or post-surgical recurrence, node-negative
TL;DR5y actuarial LC 98% overall (99% primary, 87% recurrent) in 136 facial BCC/SCC treated with HDR Ir-192 surface molds, no severe late complications.
The RT read is geometry plus fractionation: a custom PMMA mold held dose at 5 mm depth within 5% of prescription over curved facial surfaces, with 180 cGy fractions to 6000-6500 cGy and no severe late effects. Whether that conformality keeps its late-toxicity profile with fewer, larger fractions was not tested here.
In node-negative BCC/SCC on curved facial skin (nose, eyelid, ear) where an electron or orthovoltage field fits poorly, this supports HDR surface-mold brachytherapy as a conformal RT option; it does not inform hypofractionated mold schedules or deeply invasive lesions.
A custom PMMA mold held dose at 5 mm depth within 5% of prescription over curved facial anatomy with a sharp edge gradient, giving 99% 5y LC in primaries and no severe late effects at 180 cGy per fraction. It supports molds for non-flat sites; transfer to short-course schedules is untested.
14 details
Prospective single-institution series, 136 consecutive pts treated 1992-1997 in Barcelona. Minimum follow-up 12 mo; median follow-up not reported.
Pathologically confirmed facial BCC (102) or SCC (34), 73 primary and 63 recurrent after prior local therapy. Gross tumor left in place in most; margins positive in 128. No nodal or distant disease on CT head/neck and chest X-ray.
HDR Ir-192 via 117 custom PMMA molds (3-7 parallel catheters, 5 mm bolus layer) or 19 Brock applicators. 6000-6500 cGy in 33-36 fx of 180 cGy at 5 mm depth; lesions ≥4 cm boosted 1800 cGy/10 fx after a 3-wk rest to 7500-8000 cGy. Eye shielded with a 5 mm lead sheet in the mold.
All pts complete responders; 3 local recurrences (1/73 primary, 2/63 recurrent). 5y actuarial LC 98% overall, 99% primary, 87% recurrent; no regional or distant failure.
| Group | N | Local recurrences | 5y actuarial LC |
|---|---|---|---|
| All pts | 136 | 3 | 98% |
| Primary tumors | 73 | 1 | 99% |
| Recurrent tumors | 63 | 2 | 87% |
| Previously untreated >4 cm | 16 | 1 | 91% |
| Timepoint | Good/excellent cosmesis |
|---|---|
| 2 wk | 38% |
| 3 mo | 92% |
| 6 mo | 98% |
Erythema G1 in 117 and G2 in 19; 14 (10%) had G1 ulceration at 6 wk, all healed by 4 mo. Good/excellent cosmesis 98% at 6 mo; no severe late complications.
Primary-tumor LC of 99% sits alongside the 95% superficial X-ray result in the Lovett EBRT series and above Lovett's 77% with electrons. Daly's Ir-192 interstitial implants reached 96% LC but with 18% severe late effects, the toxicity gap the mold approach is designed to close.
Dose per fraction (180 cGy) far from modern short-course skin brachy, so the late-effect profile may not transfer. Recurrent-tumor LC timepoint reported inconsistently (5y vs 4y), and cosmesis was graded by the treating team.
The contribution is technical: a reproducible, afterloaded surface applicator that conforms to non-flat facial anatomy with a sharp edge gradient. It shows high control with conventional fractionation but does not settle how it compares head-to-head with superficial X-ray, electrons, or Mohs.
Single-arm single-institution series without a comparator; follow-up minimum only 12 mo with median unreported; comparisons rest on historic EBRT and implant series.
- Does late cosmesis hold with hypofractionated HDR mold schedules?
- Head-to-head vs superficial X-ray or electrons on curved facial sites
- Durability of control beyond 5 years in recurrent lesions
📚 Sources · 📄 1 paper
The longer read
This is a technique paper wearing the clothes of an outcomes paper, and it should be read that way. The clinically interesting claim is not that radiotherapy controls facial basal and squamous cell carcinoma; that was established long before this series. The claim is that a custom-built acrylic mold, carrying afterloading catheters over an impression of the patient's own face, can deliver a homogeneous dose at 5 mm depth across curved anatomy (nose, eyelid, ear, preauricular skin) with a sharp falloff at the applicator edge. The dosimetric evidence for that claim is direct and checkable: variation at depth within 5% of prescription, and thermoluminescent readings that matched the plan within 5%. That is the part of the paper most likely to transfer.
The control figures, 99% at five years for primaries and 87% for recurrences, look excellent, but their footing is softer than they appear. The cohort is single-institution, without a comparator, and follow-up is described only by a 12-month minimum; the at-risk counts at five years are a minority of the starting cohort. The recurrent-tumor control rate is attributed to five years in the Results and four years in the Discussion, and the reported lesion size distribution is internally inconsistent. None of this makes the result wrong, but it means the headline numbers should be held loosely and the comparisons to other series even more loosely.
The comparisons the authors draw are to Lovett's external-beam series, where superficial X-rays reached 95% and electrons only 77%, and to Daly's interstitial iridium implants, which matched on control at 96% but at the cost of 18% severe late effects. The more defensible reading is that the mold approach plausibly combines implant-level conformality with the noninvasive, low-toxicity profile of superficial X-rays, especially on surfaces where an electron or orthovoltage field cannot sit flat. The cross-series comparison cannot prove superiority: different eras, different case mixes, and different definitions of control.
The fractionation is the biggest barrier to transfer. Thirty-three to thirty-six fractions of 180 cGy, plus a boost after a three-week break for larger lesions, is a seven-week-plus commitment for a population whose mean age was 67. The authors attribute the absence of severe late complications to the small fraction size, contrasting it with the 300 to 400 cGy fractions in Lovett's series. That explanation is reasonable but untested here, and it cuts against the direction contemporary skin brachytherapy has taken toward short courses. A reader should not assume the late-cosmesis profile carries over to hypofractionated mold treatment; this series simply did not test that.
What would make the result wrong? Selection toward lesions likely to do well, unmeasured depth of invasion (the paper states it was not determined), and team-assessed cosmesis could all inflate both control and cosmetic outcomes. What the paper does settle is feasibility: a mold could be made in three working days with two patient visits, sessions took 3 to 8 minutes, and no treatment was interrupted by applicator failure. For a radiation oncologist weighing how to treat a lesion on a curved facial surface, it supports surface-mold HDR as a geometrically sound option; the dose schedule to pair with it is a question this series leaves open.