KEY POINTS
- This retrospective matched case–control study examined 145 consecutive breast cancer patients who underwent mastectomy, implant-based reconstruction and pencil-beam scanning proton PMRT between 2015 and 2021. All received conventionally fractionated 50 Gy(RBE) in 25 fractions; 8 patients developed Baker grade III–IV capsular contracture and were matched 1:2 with 16 controls.
- Contracture occurred at a median of 23 months after treatment. Cases and controls were matched across demographic, tumor and surgical factors including age, BMI, smoking, menopausal status, laterality, stage, histology, mastectomy type and axillary dissection; median follow-up was 71.5 months in cases and 65 months in controls.
- Rather than analyzing dose or LET separately, investigators evaluated their joint distribution within a manually edited 5-mm peri-implant tissue shell. All 4,096 combinations of physical dose and dose-averaged LET were screened using a mixed-effects regression model accounting for the matched design.
- Three independent dose–LET combinations were associated with capsular contracture: V(55.8 Gy, 2.2 keV/μm), V(50.3 Gy, 5.4 keV/μm) and V(32.8 Gy, 0.9 keV/μm), all with p<0.01. After false-discovery-rate correction across the 4,096 tested indices, the signal remained borderline-robust at q=0.052.
- Exploratory cutoffs suggested limiting the peri-implant volume receiving ≥55.8 Gy with LET ≥2.2 keV/μm to <0.0017% and the volume receiving ≥50.3 Gy with LET ≥5.4 keV/μm to <0.0033%, while keeping >96.98% of tissue within the favorable ≥32.8 Gy/≥0.9 keV/μm metric. The first two thresholds represented only approximately one and two calculation voxels, respectively, emphasizing how preliminary these numbers are.
- Used individually, the three candidate constraints classified contracture status with accuracies of 83.3%, 87.5% and 87.5%. A support-vector-machine model combining them achieved AUC 0.867, 91.7% accuracy, 87.5% sensitivity and 93.8% specificity in leave-one-out internal validation.
- One patient with capsular contracture satisfied all three proposed constraints and received an estimated complication probability near zero. This false-negative case highlights that capsular contracture is not explained by proton dose–LET distribution alone and that surgical, biological or other unmeasured factors remain important.
CLINICAL TAKEAWAY
This study suggests that where high physical dose overlaps with elevated LET around an implant may matter more than dose alone for reconstruction toxicity after proton PMRT. The concept is clinically interesting, but constraints derived from only eight events—and in some cases representing one or two voxels—should not yet be used as validated planning limits.
SOURCE
International Journal of Radiation Oncology, Biology, Physics