External validation of an 18F-FDG-PET radiomic model predicting survival after radiotherapy for oropharyngeal cancer.

PURPOSE/OBJECTIVE: The purpose of the study is to externally validate published 18F-FDG-PET radiomic models for outcome prediction in patients with oropharyngeal cancer treated with chemoradiotherapy. MATERIAL/METHODS: Outcome data and pre-radiotherapy PET images of 100 oropharyngeal cancer patients (stage IV:78) treated with concomitant chemotherapy to 66-69 Gy/30 fr were available. Tumors were segmented using a previously validated semi-automatic method; 450 radiomic features (RF) were extracted according to IBSI (Image Biomarker Standardization Initiative) guidelines. Only one model for cancer-specific survival (CSS) prediction was suitable to be independently tested, according to our criteria. This model, in addition to HPV status, SUVmean and SUVmax, included two independent meta-factors (Fi), resulting from combining selected RF clusters. In a subgroup of 66 patients with complete HPV information, the global risk score R was computed considering the original coefficients and was tested by Cox regression as predictive of CSS. Independently, only the radiomic risk score RF derived from Fi was tested on the same subgroup to learn about the radiomics contribution to the model. The metabolic tumor volume (MTV) was also tested as a single predictor and its prediction performances were compared to the global and radiomic models. Finally, the validation of MTV and the radiomic score RF were also tested on the entire dataset. RESULTS: Regarding the analysis of the subgroup with HPV information, with a median follow-up of 41.6 months, seven patients died due to cancer. R was confirmed to be associated to CSS (p value = 0.05) with a C-index equal 0.75 (95% CI=0.62-0.85). The best cut-off value (equal to 0.15) showed high ability in patient stratification (p=0.01, HR=7.4, 95% CI=1.6-11.4). The 5-year CSS for R were 97% (95% CI: 93-100%) vs 74% (56-92%) for low- and high-risk groups, respectively. RF and MTV alone were also significantly associated to CSS for the subgroup with an almost identical C-index. According to best cut-off value (RF>0.12 and MTV>15.5cc), the 5-year CSS were 96% (95% CI: 89-100%) vs 65% (36-94%) and 97% (95% CI: 88-100%) vs 77% (58-93%) for RF and MTV, respectively. Results regarding RF and MTV were confirmed in the overall group. CONCLUSION: A previously published PET radiomic model for CSS prediction was independently validated. Performances of the model were similar to the ones of using only the MTV, without improvement of prediction accuracy.

Oral Candida spp. Colonisation Is a Risk Factor for Severe Oral Mucositis in Patients Undergoing Radiotherapy for Head & Neck Cancer: Results from a Multidisciplinary Mono-Institutional Prospective Observational Study.

BACKGROUND: This study aims to find a correlation between Candida spp. oral colonisation prior to radiotherapy (RT) and (i) the development of severe oral mucositis (OM) (grade 3/4) and (ii) early development of severe OM (EOM). METHODS: The protocol was registered on ClinicalTrials.gov (ID: NCT04009161) and approved by the ethical committee of the 'Fondazione Policlinico Universitario Gemelli IRCCS' (22858/18). An oral swab was obtained before RT to assess the presence of Candida spp. Severe OM occurring before a dose of 40 Gy was defined as EOM. RESULTS: No patient developed G4 OM, and only 36/152 patients (23.7%) developed G3 OM. Tumour site and lymphocytopenia were risk factors for severe OM (OR for tumour site: 1.29, 95% CI: 1-1.67, p = 0.05; OR for lymphocytopenia: 8.2, 95% CI: 1.2-55.8, p = 0.03). We found a correlation between Candida spp. and EOM (OR: 5.13; 95% CI: 1.23-21.4 p = 0.04). Patients with oral colonisation of Candida spp. developed severe OM at a mean dose of 38.3 Gy (range: 28-58; SD: 7.6), while negative patients did so at a mean dose of 45.6 Gy (range: 30-66; SD: 11.1). CONCLUSIONS: Candida spp. in the oral cavity appears to be a predictive factor of EOM.

A predictive nomogram for trismus after radiotherapy for head and neck cancer.

BACKGROUND: The aim of this study is to develop a prediction model for trismus (maximal interincisal distance equal to or less than 35 mm) based on a multivariable analysis of dosimetric and clinical factors. METHODS: The maximum inter-incisal opening (MIO) of hean and neck cancer (HNC) patients who underwent radiotherapy (RT) ± concurrent chemotherapy with radical intent, was prospectively measured prior to RT (baseline) and 6 months post-RT. The outcome variable is trismus. The potential risk factors (clinical and dosimetric) were first screened by univariate analysis and then by multivariate analysis. At the end of this process, we used the features identified as relevant, to fit a logistic regression model and calculate the probability of observed trismus during the 6-month follow-up after RT. RESULTS: One hundred and four consecutive patients were included (mean age 63 years, range 25-87), 68 males, 36 females. In the univariate analysis, the MIO at baseline, as an independent variable, and several Vdoses of different masticatory structures were found as significant. Additionally, using a bivariate model, a feature selection process was performed. Finally, we considered as best performing model the MIO at baseline and V42 at masseter muscles. The area under curve (AUC) of Receiver Operating Characteristic (ROC) curve value was 0.8255 (95% CI 0.74-0.9). The Hosmer and Lemeshow goodness-of-fit test, used to calibrate our model, was not-significant. CONCLUSIONS: A prediction nomogram was developed to assess trismus risk in planning process. An external validation of the model is required to apply it for current clinical use.

Accurate prediction of long-term risk of biochemical failure after salvage radiotherapy including the impact of pelvic node irradiation.

BACKGROUND AND PURPOSE: Explainable models of long-term risk of biochemical failure (BF) after post-prostatectomy salvage radiotherapy (SRT) are lacking. A previously introduced radiobiology-based formula was adapted to incorporate the impact of pelvic nodes irradiation (PNI). MATERIALS AND METHODS: The risk of post-SRT BF may be expressed by a Poisson-based equation including pre-SRT PSA, the radiosensitivity α, the clonogen density C, the prescribed dose (in terms of EQD2, α/ß = 1.5 Gy) and a factor (1-BxλxPSA) accounting for clonogens outside the irradiated volume, being λ the recovery due to PNI. Data of 795 pT2-pT3, pN0/pN1/pNx (n = 627/94/74) patients with follow-up ≥ 5 years and pre-RT PSA ≤ 2 ng/mL were randomly split into training (n = 528) and validation (n = 267) cohorts; the training cohort data were fitted by the least square method. Separate fits were performed for different risk groups. Model performances were assessed by calibration plots and tested in the validation group. RESULTS: The median follow-up was 8.5y, median pre-SRT PSA and EQD2 were 0.43 ng/mL and 71.3 Gy respectively; 331/795 pts received PNI. The most clinically significant prognostic grouping was pT3b and/or ISUP4-5 versus pT2/3a and ISUP1-3. Best-fit parameters were αeff = 0.26/0.23 Gy-1, C = 107/107, B = 0.40/0.97, λ = 0.87/0.41 for low/high-risk group. Performances were confirmed in the validation group (slope = 0.89,R2 = 0.85). Results suggested optimal SRT dose at 70-74 Gy. The estimated reduction of post-SRT BF due to PNI at these dose values was > 5 % for PSA > 1/>0.15 ng/mL for low/high-risk patients, being > 10 % for high-risk patients with pre-SRT PSA > 0.25 ng/mL. CONCLUSION: An explainable one-size-fits-all equation satisfactorily predicts long-term risk of post-SRT BF. The model was independently validated. A calculator tool was made available.

Refinement & validation of rectal wall dose volume objectives for prostate hypofractionation in 20 fractions.

BACKGROUND AND PURPOSE: Dose-volume objectives for the rectum have been proposed to limit long term toxicity after moderately hypofractionated radiotherapy (MHRT) for localized prostate cancer. The purpose of the present study is to validate and possibly refine dose volume objective for the rectal wall after 20-fraction MHRT. MATERIALS AND METHODS: All patients treated by 20-fraction MHRT at a single Institution were identified and relative rectal wall (%RW) DVH retrieved. The endpoint of the study is the development of grade 2+ late rectal bleeding (LRB) according to a modified RTOG scale. Clinical and dosimetric predictors of LRB were investigated at both uni- and multi-variable analysis. RESULTS: 293 patients were identified and analyzed. Of them, 35 (12%) developed the endpoint. At univariable analysis, antithrombotic drug usage (yes vs no), technique (3DCRT vs IMRT/VMAT) and several %RW DVH cut-points were significantly correlated with LRB. However, within patients treated by 3DCRT (N = 106), a bi-variable model including anti-thrombotic drug usage and selected %RW dose/volume metrics failed to identify independent dosimetric predictors of LRB. Conversely, within patients treated with intensity modulation (N = 187), the same model showed a progressively higher impact of the percent of RW receiving doses above 40 Gy. Based on this model, we were able to confirm (V32), refine (V60) and identify a novel (V50) cut-point for the %RW. CONCLUSION: We recommend the following dose volume objectives for the %RW in order to minimize the risk of LRB after 20-fraction MHRT: V32 ≤ 50%; V50 ≤ 25.8% and V60 ≤ 10%.