Refine
Has Fulltext
- yes (35)
Is part of the Bibliography
- yes (35)
Year of publication
Document Type
- Journal article (35) (remove)
Language
- English (35) (remove)
Keywords
- melanoma (35) (remove)
Institute
- Klinik und Poliklinik für Dermatologie, Venerologie und Allergologie (22)
- Theodor-Boveri-Institut für Biowissenschaften (10)
- Medizinische Klinik und Poliklinik I (3)
- Pathologisches Institut (3)
- Comprehensive Cancer Center Mainfranken (2)
- Rudolf-Virchow-Zentrum (2)
- Abteilung für Funktionswerkstoffe der Medizin und der Zahnheilkunde (1)
- Frauenklinik und Poliklinik (1)
- Institut für Klinische Transfusionsmedizin und Hämotherapie (1)
- Institut für Pharmazie und Lebensmittelchemie (1)
- Klinik und Poliklinik für Unfall-, Hand-, Plastische und Wiederherstellungschirurgie (Chirurgische Klinik II) (1)
- Lehrstuhl für Biochemie (1)
- Medizinische Klinik und Poliklinik II (1)
Sonstige beteiligte Institutionen
Background
A basic requirement for artificial intelligence (AI)–based image analysis systems, which are to be integrated into clinical practice, is a high robustness. Minor changes in how those images are acquired, for example, during routine skin cancer screening, should not change the diagnosis of such assistance systems.
Objective
To quantify to what extent minor image perturbations affect the convolutional neural network (CNN)–mediated skin lesion classification and to evaluate three possible solutions for this problem (additional data augmentation, test-time augmentation, anti-aliasing).
Methods
We trained three commonly used CNN architectures to differentiate between dermoscopic melanoma and nevus images. Subsequently, their performance and susceptibility to minor changes (‘brittleness’) was tested on two distinct test sets with multiple images per lesion. For the first set, image changes, such as rotations or zooms, were generated artificially. The second set contained natural changes that stemmed from multiple photographs taken of the same lesions.
Results
All architectures exhibited brittleness on the artificial and natural test set. The three reviewed methods were able to decrease brittleness to varying degrees while still maintaining performance. The observed improvement was greater for the artificial than for the natural test set, where enhancements were minor.
Conclusions
Minor image changes, relatively inconspicuous for humans, can have an effect on the robustness of CNNs differentiating skin lesions. By the methods tested here, this effect can be reduced, but not fully eliminated. Thus, further research to sustain the performance of AI classifiers is needed to facilitate the translation of such systems into the clinic.
Recent research revealed the superiority of artificial intelligence over dermatologists to diagnose melanoma from images. However, 30–50% of all melanomas and more than half of those in young patients evolve from initially benign lesions. Despite its high relevance for melanoma screening, neither clinicians nor computers are yet able to reliably predict a nevus’ oncologic transformation. The cause of this lies in the static nature of lesion presentation in the current standard of care, both for clinicians and algorithms. The status quo makes it difficult to train algorithms (and clinicians) to precisely assess the likelihood of a benign skin lesion to transform into melanoma. In addition, it inhibits the precision of current algorithms since ‘evolution’ image features may not be part of their decision. The current literature reveals certain types of melanocytic nevi (i.e. ‘spitzoid’ or ‘dysplastic’ nevi) and criteria (i.e. visible vasculature) that, in general, appear to have a higher chance to transform into melanoma. However, owing to the cumulative nature of oncogenic mutations in melanoma, a more fine-grained early morphologic footprint is likely to be detectable by an algorithm. In this perspective article, the concept of melanoma prediction is further explored by the discussion of the evolution of melanoma, the concept for training of such a nevi classifier and the implications of early melanoma prediction for clinical practice. In conclusion, the authors believe that artificial intelligence trained on prospective image data could be transformative for skin cancer diagnostics by (a) predicting melanoma before it occurs (i.e. pre-in situ) and (b) further enhancing the accuracy of current melanoma classifiers. Necessary prospective images for this research are obtained via free mole-monitoring mobile apps.
Background
Melanoma is the most dangerous type of skin cancer but is curable if detected early. Recent publications demonstrated that artificial intelligence is capable in classifying images of benign nevi and melanoma with dermatologist-level precision. However, a statistically significant improvement compared with dermatologist classification has not been reported to date.
Methods
For this comparative study, 4204 biopsy-proven images of melanoma and nevi (1:1) were used for the training of a convolutional neural network (CNN). New techniques of deep learning were integrated. For the experiment, an additional 804 biopsy-proven dermoscopic images of melanoma and nevi (1:1) were randomly presented to dermatologists of nine German university hospitals, who evaluated the quality of each image and stated their recommended treatment (19,296 recommendations in total). Three McNemar's tests comparing the results of the CNN's test runs in terms of sensitivity, specificity and overall correctness were predefined as the main outcomes.
Findings
The respective sensitivity and specificity of lesion classification by the dermatologists were 67.2% (95% confidence interval [CI]: 62.6%–71.7%) and 62.2% (95% CI: 57.6%–66.9%). In comparison, the trained CNN achieved a higher sensitivity of 82.3% (95% CI: 78.3%–85.7%) and a higher specificity of 77.9% (95% CI: 73.8%–81.8%). The three McNemar's tests in 2 × 2 tables all reached a significance level of p < 0.001. This significance level was sustained for both subgroups.
Interpretation
For the first time, automated dermoscopic melanoma image classification was shown to be significantly superior to both junior and board-certified dermatologists (p < 0.001).
Background
Several recent publications have demonstrated the use of convolutional neural networks to classify images of melanoma at par with board-certified dermatologists. However, the non-availability of a public human benchmark restricts the comparability of the performance of these algorithms and thereby the technical progress in this field.
Methods
An electronic questionnaire was sent to dermatologists at 12 German university hospitals. Each questionnaire comprised 100 dermoscopic and 100 clinical images (80 nevi images and 20 biopsy-verified melanoma images, each), all open-source. The questionnaire recorded factors such as the years of experience in dermatology, performed skin checks, age, sex and the rank within the university hospital or the status as resident physician. For each image, the dermatologists were asked to provide a management decision (treat/biopsy lesion or reassure the patient). Main outcome measures were sensitivity, specificity and the receiver operating characteristics (ROC).
Results
Total 157 dermatologists assessed all 100 dermoscopic images with an overall sensitivity of 74.1%, specificity of 60.0% and an ROC of 0.67 (range = 0.538–0.769); 145 dermatologists assessed all 100 clinical images with an overall sensitivity of 89.4%, specificity of 64.4% and an ROC of 0.769 (range = 0.613–0.9). Results between test-sets were significantly different (P < 0.05) confirming the need for a standardised benchmark.
Conclusions
We present the first public melanoma classification benchmark for both non-dermoscopic and dermoscopic images for comparing artificial intelligence algorithms with diagnostic performance of 145 or 157 dermatologists. Melanoma Classification Benchmark should be considered as a reference standard for white-skinned Western populations in the field of binary algorithmic melanoma classification.
The cystine/glutamate antiporter xCT is an important source of cysteine for cancer cells. Once taken up, cystine is reduced to cysteine and serves as a building block for the synthesis of glutathione, which efficiently protects cells from oxidative damage and prevents ferroptosis. As melanomas are particularly exposed to several sources of oxidative stress, we investigated the biological role of cysteine and glutathione supply by xCT in melanoma. xCT activity was abolished by genetic depletion in the Tyr::CreER; Braf\(^{CA}\); Pten\(^{lox/+}\) melanoma model and by acute cystine withdrawal in melanoma cell lines. Both interventions profoundly impacted melanoma glutathione levels, but they were surprisingly well tolerated by murine melanomas in vivo and by most human melanoma cell lines in vitro. RNA sequencing of human melanoma cells revealed a strong adaptive upregulation of NRF2 and ATF4 pathways, which orchestrated the compensatory upregulation of genes involved in antioxidant defence and de novo cysteine biosynthesis. In addition, the joint activation of ATF4 and NRF2 triggered a phenotypic switch characterized by a reduction of differentiation genes and induction of pro-invasive features, which was also observed after erastin treatment or the inhibition of glutathione synthesis. NRF2 alone was capable of inducing the phenotypic switch in a transient manner. Together, our data show that cystine or glutathione levels regulate the phenotypic plasticity of melanoma cells by elevating ATF4 and NRF2.
Highlights
• The integrated stress response leads to a general ATF4-dependent activation of NRF2
• ATF4 causes a CHAC1-dependent GSH depletion, resulting in NRF2 stabilization
• An elevation of NRF2 transcript levels fosters this effect
• NRF2 supports the ISR/ATF4 pathway by improving cystine and antioxidant supply
Summary
The redox regulator NRF2 becomes activated upon oxidative and electrophilic stress and orchestrates a response program associated with redox regulation, metabolism, tumor therapy resistance, and immune suppression. Here, we describe an unrecognized link between the integrated stress response (ISR) and NRF2 mediated by the ISR effector ATF4. The ISR is commonly activated after starvation or ER stress and plays a central role in tissue homeostasis and cancer plasticity. ATF4 increases NRF2 transcription and induces the glutathione-degrading enzyme CHAC1, which we now show to be critically important for maintaining NRF2 activation. In-depth analyses reveal that NRF2 supports ATF4-induced cells by increasing cystine uptake via the glutamate-cystine antiporter xCT. In addition, NRF2 upregulates genes mediating thioredoxin usage and regeneration, thus balancing the glutathione decrease. In conclusion, we demonstrate that the NRF2 response serves as second layer of the ISR, an observation highly relevant for the understanding of cellular resilience in health and disease.
Defects in DNA repair pathways have been associated with an improved response to immune checkpoint inhibition (ICI). In particular, patients with the nucleotide excision repair (NER) defect disease Xeroderma pigmentosum (XP) responded impressively well to ICI treatment. Recently, in melanoma patients, pretherapeutic XP gene expression was predictive for anti-programmed cell death-1 (PD-1) ICI response. The underlying mechanisms of this finding are still to be revealed. Therefore, we used CRISPR/Cas9 to disrupt XPA in A375 melanoma cells. The resulting subclonal cell lines were investigated by Sanger sequencing. Based on their genetic sequence, candidates from XPA exon 1 and 2 were selected and further analyzed by immunoblotting, immunofluorescence, HCR and MTT assays. In XPA exon 1, we established a homozygous (c.19delG; p.A7Lfs*8) and a compound heterozygous (c.19delG/c.19_20insG; p.A7Lfs*8/p.A7Gfs*55) cell line. In XPA exon 2, we generated a compound heterozygous mutated cell line (c.206_208delTTG/c.208_209delGA; p.I69_D70delinsN/p.D70Hfs*31). The better performance of the homozygous than the heterozygous mutated exon 1 cells in DNA damage repair (HCR) and post-UV-C cell survival (MTT), was associated with the expression of a novel XPA protein variant. The results of our study serve as the fundamental basis for the investigation of the immunological consequences of XPA disruption in melanoma.
Tumour progression stage-dependent secretion of YB-1 stimulates melanoma cell migration and invasion
(2020)
Secreted factors play an important role in intercellular communication. Therefore, they are not only indispensable for the regulation of various physiological processes but can also decisively advance the development and progression of tumours. In the context of inflammatory disease, Y-box binding protein 1 (YB-1) is actively secreted and the extracellular protein promotes cell proliferation and migration. In malignant melanoma, intracellular YB-1 expression increases during melanoma progression and represents an unfavourable prognostic marker. Here, we show active secretion of YB-1 from melanoma cells as opposed to benign cells of the skin. Intriguingly, YB-1 secretion correlates with the stage of melanoma progression and depends on a calcium- and ATP-dependent non-classical secretory pathway leading to the occurrence of YB-1 in the extracellular space as a free protein. Along with an elevated YB-1 secretion of melanoma cells in the metastatic growth phase, extracellular YB-1 exerts a stimulating effect on melanoma cell migration, invasion, and tumourigenicity. Collectively, these data suggest that secreted YB-1 plays a functional role in melanoma cell biology, stimulating metastasis, and may serve as a novel biomarker in malignant melanoma that reflects tumour aggressiveness.
Bioprinting offers the opportunity to fabricate precise 3D tumor models to study tumor pathophysiology and progression. However, the choice of the bioink used is important. In this study, cell behavior was studied in three mechanically and biologically different hydrogels (alginate, alginate dialdehyde crosslinked with gelatin (ADA–GEL), and thiol-modified hyaluronan (HA-SH crosslinked with PEGDA)) with cells from breast cancer (MDA-MB-231 and MCF-7) and melanoma (Mel Im and MV3), by analyzing survival, growth, and the amount of metabolically active, living cells via WST-8 labeling. Material characteristics were analyzed by dynamic mechanical analysis. Cell lines revealed significantly increased cell numbers in low-percentage alginate and HA-SH from day 1 to 14, while only Mel Im also revealed an increase in ADA–GEL. MCF-7 showed a preference for 1% alginate. Melanoma cells tended to proliferate better in ADA–GEL and HA-SH than mammary carcinoma cells. In 1% alginate, breast cancer cells showed equally good proliferation compared to melanoma cell lines. A smaller area was colonized in high-percentage alginate-based hydrogels. Moreover, 3% alginate was the stiffest material, and 2.5% ADA–GEL was the softest material. The other hydrogels were in the same range in between. Therefore, cellular responses were not only stiffness-dependent. With 1% alginate and HA-SH, we identified matrices that enable proliferation of all tested tumor cell lines while maintaining expected tumor heterogeneity. By adapting hydrogels, differences could be accentuated. This opens up the possibility of understanding and analyzing tumor heterogeneity by biofabrication.