Aim: To evaluate the tolerability and efficacy of sorafenib in patients with thyroid carcinoma.Methods: Patients with progressive locally advanced/metastatic medullary thyroid carcinoma (MTC), or differentiated thyroid carcinoma (DTC) with non-radioiodine-avid disease, were treated with sorafenib 400 mg twice daily until disease progression. The primary endpoint was the radiological response rate (RR) at 6 months. Secondary endpoints were RR at 3, 9 and 12 months, biochemical responses, toxicity, biomarker analyses and progression free and overall survival (OS).Results: A total of 34 patients were recruited to the study (15 medullary and 19 differentiated). After 6 months, the RR rate was 15% and a further 74% of patients achieved stable disease in the first 6 months. After 12 months of treatment, the RR was 21%. In the MTC patients, the RR at 12 months was 25% and OS was 100%. In DTC patients corresponding rates were 18 and 79% respectively. Median overall and progression-free survival points were not reached at 19 months. Commonest adverse events included hand–foot syndrome, other skin toxicities, diarrhoea and alopecia. Dose reduction was required in 79% patients. Median time on treatment was 16.5 months.Conclusion: This study demonstrates that sorafenib is tolerable at reduced doses over prolonged periods of time in patients with thyroid cancer. Sorafenib leads to radiological and biochemical stabilisation of disease in the majority of these patients despite dose reductions.

Handcrafted and deep learning (DL) radiomics are popular techniques used to develop computed tomography (CT) imaging-based artificial intelligence models for COVID-19 research. However, contrast heterogeneity from real-world datasets may impair model performance. Contrast-homogenous datasets present a potential solution. We developed a 3D patch-based cycle-consistent generative adversarial network (cycle-GAN) to synthesize non-contrast images from contrast CTs, as a data homogenization tool. We used a multi-centre dataset of 2078 scans from 1,650 patients with COVID-19. Few studies have previously evaluated GAN-generated images with handcrafted radiomics, DL and human assessment tasks. We evaluated the performance of our cycle-GAN with these three approaches. In a modified Turing-test, human experts identified synthetic vs acquired images, with a false positive rate of 67% and Fleiss’ Kappa 0.06, attesting to the photorealism of the synthetic images. However, on testing performance of machine learning classifiers with radiomic features, performance decreased with use of synthetic images. Marked percentage difference was noted in feature values between pre- and post-GAN non-contrast images. With DL classification, deterioration in performance was observed with synthetic images. Our results show that whilst GANs can produce images sufficient to pass human assessment, caution is advised before GAN-synthesized images are used in medical imaging applications.

Objective:To evaluate three image acquisition presets for four-dimensional cone beam CT (CBCT) to identify an optimal preset for lung tumour image quality while minimizing dose and acquisition time.Methods:Nine patients undergoing radical conventionally fractionated radiotherapy for lung cancer had verification CBCTs acquired using three presets: Preset 1 on Day 1 (11 mGy dose, 240 s acquisition time), Preset 2 on Day 2 (9 mGy dose, 133 s acquisition time) and Preset 3 on Day 3 (9 mGy dose, 67 s acquisition time). The clarity of the tumour and other thoracic structures, and the acceptability of the match, were retrospectively graded by visual grading analysis (VGA). Logistic regression was used to identify the most appropriate preset and any factors that might influence the result.Results:Presets 1 and 2 met a clinical requirement of 75% of structures to be rated “Clear” or above and 75% of matches to be rated “Acceptable” or above. Clarity is significantly affected by preset, patient, observer and struct...

Stereotactic radiotherapy (SBRT) and radiofrequency ablation (RFA) are common ablative techniques for lung metastases. A retrospective review of all patients treated with either modality at a single institution between 2011 and 2019 was conducted. Baseline characteristics and outcomes were compared. Local and distant progression, and overall survival were estimated using the Kaplan–Meier method. Univariable analysis was carried out using Cox regression; this was followed by multivariable modelling. In total, 106 patients treated with RFA and 70 treated with SBRT were identified. Tumours treated with SBRT were larger (median size 18 mm vs. 11 mm) and primarily oligometastatic (91.4% vs. 20%). Median progression-free survival (PFS) was 12.5 months for SBRT and 7.9 months for RFA (p = 0.009). Median OS was similar (p = 0.66). In multivariable analysis, lesion size > 20 mm was predictive of adverse local PFS (p = 0.001), PFS (p = 0.0034) and OS (p = 0.001). A statistically significant interaction effect suggested that RFA was associated with better local PFS within colorectal primary patients (p = 0.035). This study highlights differences in patient selection and outcomes for RFA or SBRT in the treatment of lung metastases at our institution. Future studies for SBRT should focus on the optimum dose schedules required for different histologies. For less-radiosensitive tumours, RFA may offer a superior alternative where dose-escalated SBRT is not possible.