We enrolled BC patients (pts) treated with hypofractionated-RT after conservative surgery (42.4 Gy + 10 Gy boost to tumor bed), PC pts receiving radical-RT (65 Gy / 25 fractions), HNC pts treated with conventional RT (54-70 Gy @ 1.8-2 Gy/fr) or moderately hypofractionation (56.1-69.96 Gy @ 1.7-2.12 Gy/fr). We assessed baseline sublingual-microvasculature (MV) health status before RT using a sidestream dark-field camera coupled to the GlycoCheck™ software. The system records videos showing the live movement of red blood cells (RBCs) in the microvessels. It computes 5 functional parameters: (a) perfused boundary region (PBR) estimating the penetration of RBCs into the permeable part of the endothelial glycocalyx (higher PBR values result from damaged glycocalyx, indicating impaired microcirculation); (b) density of capillaries (1/mm2); (c) blood flow in the recorded area (103 μm3/s/mm2); (d) recruitment capacity (quantifying the ability to recruit additional capillaries); (e) a global MV_HealthScore (MVHS), higher values indicate healthier MV. MVHS is computed by weighting information from the previous parameters. The computation of MVHS needs a prolonged video acquisition not reached by less compliant patients. Endpoint for this analysis was acute tox. We selected a clinically relevant tox endpoint for each cancer site: grade≥2 erythema (G2+E) for BC, any grade≥3 tox (G3Tox) for HNC and persistent grade≥1 tox simultaneously present in gastrointestinal and genitourinary domains (G1+GIGU) for PC. We used logistic regression to assess MV functional parameters' association with toxicity.
We evaluated 63 BC, 38 PC, 39 HNC pts. 77 (55%) had an evaluation for MVHS. 23/63 BC pts presented with G2+E, 12/38 PC pts with G1+GIGU, 12/39 HNC pts with G3Tox. Average MVHS was 3.6, 3.2, 2.7 in BC, HNC and PC, respectively (P>0.15). The MVHS associated with tox: healthy MV protects from tox (continuous, OR = 0.66 for 1 point increase, p = 0.01, AUC 0.69). Categorizing pts as low MVHS (<2.5), average MVHS (2.5-6.5) and high MVHA (>6.5), the observed toxicity rates in the 3 classes were 50%, 26%, 0% (p = 0.02). Also, a 3-variate model including the separated functional parameters predicted tox (AUC 0.67) on the 144 pts: higher PBR is a risk (OR = 7.6), higher capillary density protects (Logarithm, OR 0.002), higher blood flow is a risk (Logarithm, OR = 14.2).
These data suggest an interaction of compromised MV with repair mechanisms. The MVHS model predicts acute toxicity in BC/PC/HN pts proving (i) a quantitative relationship between MV state and radio-susceptibility and (ii) in an organ-agnostic way. The systemic functional information derived by the sublingual microscope could boost the personalization of predictive models and tailor them to the single-patient functional status. Funded by AIRC IG 21479.