Deep Learning and Registration-Based Mapping for Analyzing the Distribution of Nodal Metastases in Head and Neck Cancer Cohorts: Informing Optimal Radiotherapy Target Volume Design

We introduce a deep-learning- and a registration-based method for automatically analyzing the spatial distribution of nodal metastases (LNs) in head and neck (H/N) cancer cohorts to inform radiotherapy (RT) target volume design. The two methods are evaluated in a cohort of 193 H/N patients/planning CTs with a total of 449 LNs. In the deep learning method, a previously developed nnU-Net 3D/2D ensemble model is used to autosegment 20 H/N levels, with each LN subsequently being algorithmically assigned to the closest-level autosegmentation. In the nonrigid-registration-based mapping method, LNs are mapped into a calculated template CT representing the cohort-average patient anatomy, and kernel density estimation is employed to estimate the underlying average 3D-LN probability distribution allowing for analysis and visualization without prespecified level definitions. Multireader assessment by three radio-oncologists with majority voting was used to evaluate the deep learning method and obtain the ground-truth distribution. For the mapping technique, the proportion of LNs predicted by the 3D probability distribution for each level was calculated and compared to the deep learning and ground-truth distributions. As determined by a multireader review with majority voting, the deep learning method correctly categorized all 449 LNs to their respective levels. Level 2 showed the highest LN involvement (59.0%). The level involvement predicted by the mapping technique was consistent with the ground-truth distribution (p for difference 0.915). Application of the proposed methods to multicenter cohorts with selected H/N tumor subtypes for informing optimal RT target volume design is promising.

Non-Professional Phagocytosis Increases in Melanoma Cells and Tissues with Increasing E-Cadherin Expression

Non-professional phagocytosis in cancer has been increasingly studied in recent decades. In malignant melanoma metastasis, cell-in-cell structures have been described as a sign of cell cannibalism. To date, only low rates of cell-in-cell structures have been described in patients with malignant melanoma. To investigate these findings further, we examined twelve primary melanoma cell lines in both adherent and suspended co-incubation for evidence of engulfment. In addition, 88 malignant melanoma biopsies and 16 healthy tissue samples were evaluated. E-cadherin levels were determined in the cell lines and tissues. All primary melanoma cell lines were capable of phagocytosis, and phagocytosis increased when cells were in suspension during co-incubation. Cell-in-cell structures were also detected in most of the tissue samples. Early T stages and increasingly advanced N and M stages have correspondingly lower rates of cell-in-cell structures. Non-professional phagocytosis was also present in normal skin tissue. Non-professional phagocytosis appears to be a ubiquitous mechanism in malignant melanoma. The absence of phagocytosis in metastases may be one reason for the high rate of metastasis in malignant melanoma.

Impact of COVID-19 on Quality of Life in Long-Term Advanced Rectal Cancer Survivors

Colorectal cancer remains one of the most commonly diagnosed cancers. Advanced rectal cancer patients receive neoadjuvant radiochemotherapy as well as surgery and suffer from reduced health-related quality of life due to various side effects. We were interested in the role of the COVID-19 pandemic and how it affected those patients' quality of life. A total of 489 advanced rectal cancer patients from the University Hospital Erlangen in Germany were surveyed between May 2010 and March 2022 and asked to fill out the EORTC QLQ-C30 and QLQ-CR38 questionnaires over eight different time points: at the beginning, during and after radiochemotherapy, right before surgery, and in yearly intervals after surgery for up to four years. Answers were converted to scores to compare the COVID-19 period to the time before March 2020, focusing on the follow-ups, the developments over time-including by sex and age-and the influence of the TNM cT-stage. Overall, a trend of impaired functional and symptom scores was found across all surveys with few significances (body image -10.6 percentage points (pp) after one year; defecation problems +13.5 pp, insomnia +10.2 pp and weight loss +9.8 pp after three years; defecation problems +11.3 pp after four years). cT4-stage patients lost significantly more weight than their cT1-3-stage counterparts (+10.7 to 13.7 pp). Further studies should be conducted to find possible causes and develop countermeasures for future major infectious diseases.

Deep learning for automatic head and neck lymph node level delineation provides expert-level accuracy

Background

Deep learning-based head and neck lymph node level (HN_LNL) autodelineation is of high relevance to radiotherapy research and clinical treatment planning but still underinvestigated in academic literature. In particular, there is no publicly available open-source solution for large-scale autosegmentation of HN_LNL in the research setting.

Methods

An expert-delineated cohort of 35 planning CTs was used for training of an nnU-net 3D-fullres/2D-ensemble model for autosegmentation of 20 different HN_LNL. A second cohort acquired at the same institution later in time served as the test set (n = 20). In a completely blinded evaluation, 3 clinical experts rated the quality of deep learning autosegmentations in a head-to-head comparison with expert-created contours. For a subgroup of 10 cases, intraobserver variability was compared to the average deep learning autosegmentation accuracy on the original and recontoured set of expert segmentations. A postprocessing step to adjust craniocaudal boundaries of level autosegmentations to the CT slice plane was introduced and the effect of autocontour consistency with CT slice plane orientation on geometric accuracy and expert rating was investigated.

Results

Blinded expert ratings for deep learning segmentations and expert-created contours were not significantly different. Deep learning segmentations with slice plane adjustment were rated numerically higher (mean, 81.0 vs. 79.6, p = 0.185) and deep learning segmentations without slice plane adjustment were rated numerically lower (77.2 vs. 79.6, p = 0.167) than manually drawn contours. In a head-to-head comparison, deep learning segmentations with CT slice plane adjustment were rated significantly better than deep learning contours without slice plane adjustment (81.0 vs. 77.2, p = 0.004). Geometric accuracy of deep learning segmentations was not different from intraobserver variability (mean Dice per level, 0.76 vs. 0.77, p = 0.307). Clinical significance of contour consistency with CT slice plane orientation was not represented by geometric accuracy metrics (volumetric Dice, 0.78 vs. 0.78, p = 0.703).

Conclusions

We show that a nnU-net 3D-fullres/2D-ensemble model can be used for highly accurate autodelineation of HN_LNL using only a limited training dataset that is ideally suited for large-scale standardized autodelineation of HN_LNL in the research setting. Geometric accuracy metrics are only an imperfect surrogate for blinded expert rating.

The Distribution of Pelvic Nodal Metastases in Prostate Cancer Reveals Potential to Advance and Personalize Pelvic Radiotherapy

Background

Traditional clinical target volume (CTV) definition for pelvic radiotherapy in prostate cancer consists of large volumes being treated with homogeneous doses without fully utilizing information on the probability of microscopic involvement to guide target volume design and prescription dose distribution.

Methods

We analyzed patterns of nodal involvement in 75 patients that received RT for pelvic and paraaortic lymph node metastases (LNs) from prostate cancer in regard to the new NRG-CTV recommendation. Non-rigid registration-based LN mapping and weighted three-dimensional kernel density estimation were used to visualize the average probability distribution for nodal metastases. As independent approach, the mean relative proportion of LNs observed for each level was determined manually and NRG and non-NRG levels were evaluated for frequency of involvement. Computer-automated distance measurements were used to compare LN distances in individual patients to the spatial proximity of nodal metastases at a cohort level.

Results

34.7% of patients had pelvic LNs outside NRG-consensus, of which perirectal was most common (25.3% of all patients) followed by left common iliac nodes near the left psoas major (6.7%). A substantial portion of patients (13.3%) had nodes at the posterior edge of the NRG obturator level. Observer-independent mapping consistently visualized high-probability hotspots outside NRG-consensus in the perirectal and left common iliac regions. Affected nodes in individual patients occurred in highly significantly closer proximity than at cohort-level (mean distance, 6.6 cm vs. 8.7 cm, p < 0.001).

Conclusions

Based on this analysis, the common iliac level should extend to the left psoas major and obturator levels should extend posteriorly 5 mm beyond the obturator internus. Incomplete coverage by the NRG-consensus was mostly because of perirectal involvement. We introduce three-dimensional kernel density estimation after non-rigid registration-based mapping for the analysis of recurrence data in radiotherapy. This technique provides an estimate of the underlying probability distribution of nodal involvement and may help in addressing institution- or subgroup-specific differences. Nodal metastases in individual patients occurred in highly significantly closer proximity than at a cohort-level, which supports that personalized target volumes could be reduced in size compared to a “one-size-fits-all” approach and is an important basis for further investigation into individualized field designs.

Volumetric Regression in Brain Metastases After Stereotactic Radiotherapy: Time Course, Predictors, and Significance

Background

There is insufficient understanding of the natural course of volumetric regression in brain metastases after stereotactic radiotherapy (SRT) and optimal volumetric criteria for the assessment of response and progression in radiotherapy clinical trials for brain metastases are currently unknown.

Methods

Volumetric analysis via whole-tumor segmentation in contrast-enhanced 1 mm³-isotropic T1-Mprage sequences before SRT and during follow-up. A total of 3,145 MRI studies of 419 brain metastases from 189 patients were segmented. Progression was defined using a volumetric extension of the RANO-BM criteria. A subset of 205 metastases without progression/radionecrosis during their entire follow-up of at least 3 months was used to study the natural course of volumetric regression after SRT. Predictors for volumetric regression were investigated. A second subset of 179 metastases was used to investigate the prognostic significance of volumetric response at 3 months (defined as ≥20% and ≥65% volume reduction, respectively) for subsequent local control.

Results

Median relative metastasis volume post-SRT was 66.9% at 6 weeks, 38.6% at 3 months, 17.7% at 6 months, 2.7% at 12 months and 0.0% at 24 months. Radioresistant histology and FSRT vs. SRS were associated with reduced tumor regression for all time points. In multivariate linear regression, radiosensitive histology (p=0.006) was the only significant predictor for metastasis regression at 3 months. Volumetric regression ≥20% at 3 months post-SRT was the only significant prognostic factor for subsequent control in multivariate analysis (HR 0.63, p=0.023), whereas regression ≥65% was no significant predictor.

Conclusions

Volumetric regression post-SRT does not occur at a constant rate but is most pronounced in the first 6 weeks to 3 months. Despite decreasing over time, volumetric regression continues beyond 6 months post-radiotherapy and may lead to complete resolution of controlled lesions by 24 months. Radioresistant histology is associated with slower regression. We found that a cutoff of ≥20% regression for the volumetric definition of response at 3 months post-SRT was predictive for subsequent control whereas the currently proposed definition of ≥65% was not. These results have implications for standardized volumetric criteria in future radiotherapy trials for brain metastases.

FSRT vs. SRS in Brain Metastases-Differences in Local Control and Radiation Necrosis-A Volumetric Study

Background: While the role of stereotactic radiotherapy for brain metastases is increasing, evidence on the comparative efficacy and safety of fractionated stereotactic radiotherapy (FSRT) and single-session radiosurgery (SRS) is scarce.

Methods: Longitudinal volumetric analysis was performed in a consecutive cohort of 120 patients and 190 brain metastases (>0.065 cm³ in volume / > ~5 mm in diameter) treated exclusively with FSRT (n = 98) and SRS (n = 92), respectively. A total of 972 tumor segmentations was used, averaging 5.1 time points per metastasis. Progression was defined using a volumetric extension of the RANO-BM criteria. Local control and radionecrosis were compared for lesions treated with FSRT and SRS, respectively.

Results: Metastases treated with FSRT were significantly larger at baseline (mean, 4.66 vs. 0.40 cm³, p < 0.001). Biologically effective dose (BED) for metastases (α/β = 12, linear-quadratic-cubic model) was significantly associated with local control, whereas BED for normal brain (α/β = 2, linear-quadratic model) was significantly associated with radionecrosis. Median time to local progression was 22.9 months in the FSRT group compared to 14.5 months in the SRS group (p = 0.022). Overall radionecrosis rate at 12 months was 3.4% for FSRT and 14.8% for SRS (p = 0.010). Radionecrosis °IV requiring resection with histologic proof of radiation necrosis also was significantly reduced in the FSRT group (FSRT 0.0% vs. SRS 3.9%, p = 0.041). In multivariate analysis, FSRT was associated with reduced risk of progression (HR 0.47, p = 0.015) and reduced risk of radionecrosis (HR 0.18, p = 0.045).

Conclusions: This volumetric study provides initial evidence that the improvements in therapeutic ratio expected for FSRT in larger brain metastases, might equally extend into the domain of smaller metastases, traditionally less considered for fractionated treatment. FSRT might constitute an important tool to further increase local control and reduce radionecrosis risk in stereotactic radiotherapy for brain metastases, that should be assessed in randomized intervention trials.

Cell-to-cell distances between tumor-infiltrating inflammatory cells have the potential to distinguish functionally active from suppressed inflammatory cells

Beyond their mere presence, the distribution pattern of inflammatory cells is of special interest. Our hypothesis was that random distribution may be a clear indicator of being non-functional as a consequence of lack of interaction. Here, we have assessed the implication of cell-to-cell distances among inflammatory cells in anal squamous cell carcinoma and a possible association with survival data. Thirty-eight patients suffering from anal carcinoma were studied using tissue microarrays, double staining immunohistochemistry, whole slide scanning and image analysis software. Therapy consisted of concurrent radiochemotherapy. Numbers of stromal and intraepithelial tumor-infiltrating inflammatory cells (TIC) and the distances between cells were quantified. Double-staining of FoxP3(+) cells with either CD8(+), CD1a(+) or CD20(+) cells was performed. Measured cell-to-cell distances were compared to computer simulated cell-to-cell distances leading to the assumption of non-randomly distributed and therefore functional immune cells. Intraepithelial CD1a(+) and CD20(+) cells were randomly distributed and therefore regarded as non-functional. In contrary, stromal CD20(+) cells had a non-random distribution pattern. A non-random distance between CD20(+) and FoxP3(+) cells was associated with a clearly unfavorable outcome. Measured distances between FoxP3(+) cells were distinctly shorter than expected and indicate a functional active state of the regulatory T cells (Treg). Analysis of cell-to-cell distances between TIC has the potential to distinguish between suppressed non-functional and functionally active inflammatory cells. We conclude that in this tumor model most of the CD1a(+) cells are non-functional as are the intraepithelial CD20(+) cells, while stromal CD20(+) cells and FoxP3(+) cells are functional cells.

Radiosensitization by BRAF inhibitor therapy-mechanism and frequency of toxicity in melanoma patients

BACKGROUND

Recent evidence suggests that ionizing radiation may be associated with unexpected side-effects in melanoma patients treated with concomitant BRAF inhibitors. A large multicenter analysis was carried out to generate reliable safety data and elucidate the mechanism.

METHODS

A total of 161 melanoma patients from 11 European skin cancer centers were evaluated for acute and late toxicity, of whom 70 consecutive patients received 86 series of radiotherapy with concomitant BRAF inhibitor therapy. To further characterize and quantify a possible radiosensitization by BRAF inhibitors, blood samples of 35 melanoma patients were used for individual radiosensitivity testing by fluorescence in situ hybridization of chromosomal breaks after ex vivo irradiation.

RESULTS

With radiotherapy and concomitant BRAF inhibitor therapy the rate of acute radiodermatitis ≥2° was 36% and follicular cystic proliferation was seen in 13% of all radiotherapies. Non-skin toxicities included hearing disorders (4%) and dysphagia (2%). Following whole-brain radiotherapy, rates of radiodermatitis ≥2° were 44% and 8% (P < 0.001) for patients with and without BRAF inhibitor therapy, respectively. Concomitant treatment with vemurafenib induced acute radiodermatitis ≥2° more frequently than treatment with dabrafenib (40% versus 26%, P = 0.07). In line with these findings, analysis of chromosomal breaks ex vivo indicated significantly increased radiosensitivity for patients under vemurafenib (P = 0.004) and for patients switched from vemurafenib to dabrafenib (P = 0.002), but not for patients on dabrafenib only. No toxicities were reported after stereotactic treatment.

CONCLUSION

Radiotherapy with concomitant BRAF inhibitor therapy is feasible with an acceptable increase in toxicity. Vemurafenib is a more potent radiosensitizer than dabrafenib.

Analysis of radiation- and 5-FU-induced inhibition of cell proliferation by an automatic colony analyser

Proliferation of Chinese hamster cells B-14 was inhibited by irradiation, by incubation with 5-fluorouracil (5-FU) and by a combination of both treatments. The reduction in proliferation was assayed by the colony formation test, which was evaluated by an automatic colony analyser according to the number and volume of the colonies. It was demonstrated that the number of colonies multiplied by the volume was equivalent to the number of cells in a Petri dish and is called total colony volume. Since this quantity reflects the entire proliferation of cells, it is a more sensitive parameter for measuring cell viability than the clonogenicity of cells. The drug-radiation interaction showed a supra-additive effect, if total colony volume is taken into account, while the traditional scoring of colonies yielded only an additive effect.

The effect of serum albumin on the radiolysis of DNA studied by constant field electrophoresis and compared to alterations caused by low molecular weight OH. scavengers

After radiolysis of calf thymus DNA in 10(-2) mol dm-3 phosphate buffer at pH7 under N2, N2O and air the yields of double-strand breaks (dsb) have been determined by constant field electrophoresis. Double-strand (dsb) breaks were formed according to a linear-quadratic relationship with dose showing a lower G-value under aerobic than under anaerobic conditions (G (air) = 1.4 nmolJ-1; G (N2) = 2.1 nmolJ-1; G (N2O) = 4.9 nmolJ-1). To test the reliability of this system the effect of low molecular weight OH. scavengers which were already used in comparable work with plasmid DNA were studied. The results with plasmid DNA and calf thymus DNA obtained by different techniques of electrophoresis agreed quite well. Under N2 more protection was obtained with ethanol than with DMSO or with t-butanol. Under air, double-strand breakage was further decreased and reached the same level with all of these scavengers. Furthermore the constant field electrophoresis gives similar results as the low-angle light scattering technique for radiation induced double strand breakage of calf thymus DNA. When BSA was used at the same scavenger capacity as the low molecular weight scavengers, the protection against double strand breakage was less if radiolysis was carried out in the presence of proteins. Under anaerobic conditions the protection factor was 13 in the presence of BSA, while with DMSO or t-butanol this factor was about 100 and with ethanol 300. In contrast to the low molecular weight OH. scavengers oxygen enhanced radiation-induced double-strand breakage with BSA. It is assumed that protein peroxyl radicals may cause strand breakage.