High rates of recurrence and distant metastasis are a foremost challenge in the management of adenoid cystic carcinoma (ACC), occurring in approximately 40% of all ACC patients. Despite the morbidity and mortality resulting from recurrent/metastatic (R/M) disease, there are no FDA-approved systemic agents for these patients. In this review, we summarize pertinent ACC pathophysiology and its implications for different systemic treatment regimens in R/M ACC. We review the evidence for the most widely used systemic agents - cytotoxic chemotherapy and tyrosine kinase inhibitors (TKIs) targeting VEGFR - in addition to immune checkpoint inhibitors and non-TKI biologic agents. Exciting emerging targets for R/M ACC, including inhibitors of Notch signaling, stemness, PRMT5, and Axl, are also discussed. Lastly, we review local therapies for small-volume lung disease in patients with oligometastatic ACC, specifically pulmonary metastasectomy and stereotactic body radiation therapy (SBRT). Future development of targeted molecular agents which exploit the underlying biology of this disease may yield novel therapeutic options to improve clinical outcomes in patients with R/M ACC.

Purpose

Urgent indications for palliative radiation therapy (RT) include malignant spinal cord compression, symptomatic brain metastases, pain, airway obstruction, and bleeding. Data on the timing of palliative RT in the inpatient setting are limited. We report our experience with inpatient palliative RT at a tertiary academic center and evaluate the effect of a dedicated inpatient palliative RT nurse practitioner (NP) on treatment timelines.

Methods and materials

We performed a retrospective, single-institution review of 219 inpatients consulted for RT to sites of metastatic disease between May 2012 and May 2018. We compared time-to-treatment intervals before and after integrating an NP for palliative RT in August 2017.

Results

The median age of the 219 patients receiving RT was 61 years (interquartile range [IQR], 51-69 years). The most frequent indications were symptomatic brain metastases (73 patients [33%]), pain (61 patients [28%]), and cord/cauda compression (48 patients [22%]). The median time from consultation request to consult was 1 day (IQR, 0-2 days), and the median time from consultation request to first RT fraction was 3 days (IQR, 2-6 days). The median time from consultation request to RT was shorter for cord compression (2 [IQR, 1-4] days) than for pain (5 [IQR, 2-7] days) (P = .001) or symptomatic brain metastases (3 [IQR, 1-6] days; P = .037). With an NP, patients were more likely to undergo same-day consultation and simulation (75% vs 60%; P = .045), which was associated with shorter median duration from consultation to initiation of RT (1 [IQR, 0-3] days vs 4 [IQR, 2-7] days; P <.001). After the integration of an NP for palliative RT, patients had a higher median Karnofsky Performance Score (70 [IQR, 60-80] vs 50 [IQR, 40-60]; P < .001) and were more likely to complete their prescribed RT course (93% vs 82%; P = .05).

Conclusions

Time from consultation request to RT is necessarily short for urgent inpatient palliative RT. Advanced practice providers may facilitate and potentially expedite treatment, with significantly shorter times to treatment among patients who undergo same-day consultation and simulation.

Objective Fast and accurate deformable image registration (DIR) that include DIR uncertainty estimation are essential for safe and reliable clinical deployment. While recent deep learning models have shown promise in predicting DIR with its uncertainty, challenges persist in proper uncertainty evaluation and hyperparameter optimization for these methods. This work aims to develop and evaluate a model that can perform fast DIR and predict its uncertainty in seconds. Approach In this study, we introduce a novel probabilistic multi-resolution image registration model utilizing convolutional neural networks (CNNs) to estimate a multivariate normal distributed dense displacement field (DDF) in a multimodal image registration problem. To assess the quality of the DDF distribution predicted by the model, we propose a new metric based on the Kullback-Leibler divergence. The performance of our approach was evaluated against three other DIR algorithms (VoxelMorph, Monte Carlo Drop-Out, and Monte Carlo B-splines) capable of predicting uncertainty. The evaluation of the models included not only the quality of the deformation but also the reliability of the estimated uncertainty. Our application investigated registration of a treatment planning computed tomography (CT) to follow-up cone beam CT for daily adaptive radiotherapy. Main results The hyperparameter tuning of the models showed that there is a trade-off between the reliability of the estimated uncertainty and the accuracy of the deformation. In the optimal trade-off our model excelled in contour propagation and uncertainty estimation (p < 0.01) compared to existing uncertainty estimation models. We obtained an average dice similarity coefficient of 0.89 and a KL-divergence of 0.15. Significance By addressing challenges in DIR uncertainty estimation and evaluation, our work showed that both the DIR and its uncertainty can be reliably predicted paving the way for safe deployment in clinical environment. .

This systematic review summarizes the current applications of ¹⁸F-FDG PET imaging in the diagnosis, staging, radiation treatment response assessment, and outcome prognostication of head and neck cancers. For head and neck cancers of unknown primary origin, ¹⁸F-FDG PET/CT increases the likelihood of identifying the primary tumor and establishing the diagnosis. ¹⁸F-FDG PET/CT is important in the accurate staging of locoregionally advanced cases of HNSCC, which can greatly affect recommendations for treatment. Following definitive chemoradiation, ¹⁸F-FDG PET/CT is validated as a means of treatment response assessment. Emerging PET tracers of hypoxia and their potential applications are reviewed.

Deep learning algorithms have recently been developed that utilize patient anatomy and raw imaging information to predict radiation dose, as a means to increase treatment planning efficiency and improve radiotherapy plan quality. Current state-of-the-art techniques rely on convolutional neural networks (CNNs) that use pixel-to-pixel loss to update network parameters. However, stereotactic body radiotherapy (SBRT) dose is often heterogeneous, making it difficult to model using pixel-level loss. Generative adversarial networks (GANs) utilize adversarial learning that incorporates image-level loss and is better suited to learn from heterogeneous labels. However, GANs are difficult to train and rely on compromised architectures to facilitate convergence. This study suggests an attention-gated generative adversarial network (DoseGAN) to improve learning, increase model complexity, and reduce network redundancy by focusing on relevant anatomy. DoseGAN was compared to alternative state-of-the-art dose prediction algorithms using heterogeneity index, conformity index, and various dosimetric parameters. All algorithms were trained, validated, and tested using 141 prostate SBRT patients. DoseGAN was able to predict more realistic volumetric dosimetry compared to all other algorithms and achieved statistically significant improvement compared to all alternative algorithms for the V₁₀₀ and V₁₂₀ of the PTV, V₆₀ of the rectum, and heterogeneity index.

Background

To evaluate acute and late genitourinary and gastrointestinal toxicities and patient reported urinary and sexual function following accelerated, hypofractionated external beam radiotherapy to the prostate, seminal vesicles and pelvic lymph nodes and high dose rate (HDR) brachytherapy or stereotactic body radiation therapy (SBRT) prostate boost.

Methods

Patients at a single institution with NCCN intermediate- and high-risk localized prostate cancer with logistical barriers to completing five weeks of whole pelvic radiotherapy (WPRT) were retrospectively reviewed for toxicity following accelerated, hypofractionated WPRT (41.25 Gy in 15 fractions of 2.75 Gy). Patients also received prostate boost radiotherapy with either HDR brachytherapy (1 fraction of 15 Gy) or SBRT (19 Gy in 2 fractions of 9.5 Gy). The duration of androgen deprivation therapy was at the discretion of the treating radiation oncologist. Toxicity was evaluated by NCI CTCAE v 5.0.

Results

Between 2015 and 2017, 22 patients with a median age of 71 years completed accelerated, hypofractionated WPRT. Median follow-up from the end of radiotherapy was 32 months (range 2-57). 5%, 73%, and 23% of patients had clinical T1, T2, and T3 disease, respectively. 86% of tumors were Gleason grade 7 and 14% were Gleason grade 9. 68% and 32% of patients had NCCN intermediate- and high-risk disease, respectively. 91% and 9% of patients received HDR brachytherapy and SBRT prostate boost following WPRT, respectively. Crude rates of grade 2 or higher GI and GU toxicities were 23% and 23%, respectively. 3 patients (14%) had late or persistent grade 2 toxicities of urinary frequency and 1 patient (5%) had late or persistent GI toxicity of diarrhea. No patient experienced grade 3 or higher toxicity at any time. No difference in patient-reported urinary or sexual function was noted at 12 months.

Conclusions

Accelerated, hypofractionated whole pelvis radiotherapy was associated with acceptable GU and GI toxicities and should be further validated for those at risk for harboring occult nodal disease.

Purpose

The aim was to develop a novel artificial intelligence (AI)-guided clinical decision support system, to predict radiation doses to subsites of the mandible using diagnostic computed tomography scans acquired before any planning of head and neck radiation therapy (RT).

Methods and materials

A dose classifier was trained using RT plans from 86 patients with oropharyngeal cancer; the test set consisted of an additional 20 plans. The classifier was trained to predict whether mandible subsites would receive a mean dose >50 Gy. The AI predictions were prospectively evaluated and compared with those of a specialist head and neck radiation oncologist for 9 patients. Positive predictive value (PPV), negative predictive value (NPV), Pearson correlation coefficient, and Lin concordance correlation coefficient were calculated to compare the AI predictions to those of the physician.

Results

In the test data set, the AI predictions had a PPV of 0.95 and NPV of 0.88. For 9 patients evaluated prospectively, there was a strong correlation between the predictions of the AI algorithm and physician (P = .72, P < .001). Comparing the AI algorithm versus the physician, the PPVs were 0.82 versus 0.25, and the NPVs were 0.94 versus 1.0, respectively. Concordance between physician estimates and final planned doses was 0.62; this was 0.71 between AI-based estimates and final planned doses.

Conclusion

AI-guided decision support increased precision and accuracy of pre-RT dental dose estimates.

Purpose

Radiation of the draining lymph node basin remains controversial for Merkel cell carcinoma, particularly in the era of sentinel lymph node biopsy (SLNB).

Methods and materials

Based on a 20-year experience using SLNB-guided adjuvant radiation therapy (RT), we conducted a retrospective review of clinically node-negative patients testing 2 hypotheses: (1) whether nodal RT could be safely omitted in SLNB-negative Merkel cell carcinoma and (2) whether the excised primary site should always be radiated. Clinically node-positive patients were excluded.

Results

Among 57 clinically node-negative patients who underwent SLNB and wide local excision (WLE), 42 (74%) had a negative SLNB, and 15 (26%) had a positive SLNB. At a median follow-up of 43 months (range, 5-182), SLNB-negative patients irradiated to the primary site had improved 4-year disease-specific survival (100% vs 65%, P = .008), local recurrence-free survival (100% vs 76%, P = .009), and distant recurrence-free survival (100% vs 75%, P = .008), but not overall survival (87.5% vs 57.7%, P = .164) compared with SLNB-positive patients receiving comprehensive RT. Among SLNB-negative patients treated with WLE only, 67% (6/9) had a disease relapse, half of which were local relapses (33%).

Conclusions

In this single-institution retrospective review, after negative SLNB and WLE, RT given only to the primary site provided 100% disease control without a need for nodal RT. Among SLNB-negative patients who had WLE, omission of postoperative primary-site RT was associated with 67% cancer relapse, of which half was local.

Purpose

To suggest an attention-aware, cycle-consistent generative adversarial network (A-CycleGAN) enhanced with variational autoencoding (VAE) as a superior alternative to current state-of-the-art MR-to-CT image translation methods.

Materials and methods

An attention-gating mechanism is incorporated into a discriminator network to encourage a more parsimonious use of network parameters, whereas VAE enhancement enables deeper discrimination architectures without inhibiting model convergence. Findings from 60 patients with head, neck, and brain cancer were used to train and validate A-CycleGAN, and findings from 30 patients were used for the holdout test set and were used to report final evaluation metric results using mean absolute error (MAE) and peak signal-to-noise ratio (PSNR).

Results

A-CycleGAN achieved superior results compared with U-Net, a generative adversarial network (GAN), and a cycle-consistent GAN. The A-CycleGAN averages, 95% confidence intervals (CIs), and Wilcoxon signed-rank two-sided test statistics are shown for MAE (19.61 [95% CI: 18.83, 20.39], P = .0104), structure similarity index metric (0.778 [95% CI: 0.758, 0.798], P = .0495), and PSNR (62.35 [95% CI: 61.80, 62.90], P = .0571).

Conclusion

A-CycleGANs were a superior alternative to state-of-the-art MR-to-CT image translation methods.© RSNA, 2020.