@article{HechtMeierZimmeretal.2018, author = {Hecht, Markus and Meier, Friedegund and Zimmer, Lisa and Polat, B{\"u}lent and Loquai, Carmen and Weishaupt, Carsten and Forschner, Andrea and Gutzmer, Ralf and Utikal, Jochen S. and Goldinger, Simone M. and Geier, Michael and Hassel, Jessica C. and Balermpas, Panagiotis and Kiecker, Felix and Rauschenberg, Ricarda and Dietrich, Ursula and Clemens, Patrick and Berking, Carola and Grabenbauer, Gerhard and Schadendorf, Dirk and Grabbe, Stephan and Schuler, Gerold and Fietkau, Rainer and Distel, Luitpold V. and Heinzerling, Lucie}, title = {Clinical outcome of concomitant vs interrupted BRAF inhibitor therapy during radiotherapy in melanoma patients}, series = {British Journal of Cancer}, volume = {118}, journal = {British Journal of Cancer}, doi = {10.1038/bjc.2017.489}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-227970}, pages = {785-792}, year = {2018}, abstract = {Background: Concomitant radiation with BRAF inhibitor (BRAFi) therapy may increase radiation-induced side effects but also potentially improve tumour control in melanoma patients. Methods: A total of 155 patients with BRAF-mutated melanoma from 17 European skin cancer centres were retrospectively analysed. Out of these, 87 patients received concomitant radiotherapy and BRAFi (59 vemurafenib, 28 dabrafenib), while in 68 patients BRAFi therapy was interrupted during radiation (51 vemurafenib, 17 dabrafenib). Overall survival was calculated from the first radiation (OSRT) and from start of BRAFi therapy (OSBRAFi). Results: The median duration of BRAFi treatment interruption prior to radiotherapy was 4 days and lasted for 17 days. Median OSRT and OSBRAFi in the entire cohort were 9.8 and 12.6 months in the interrupted group and 7.3 and 11.5 months in the concomitant group (P=0.075/P=0.217), respectively. Interrupted vemurafenib treatment with a median OSRT and OSBRAFi of 10.1 and 13.1 months, respectively, was superior to concomitant vemurafenib treatment with a median OSRT and OSBRAFi of 6.6 and 10.9 months (P=0.004/P=0.067). Interrupted dabrafenib treatment with a median OSRT and OSBRAFi of 7.7 and 9.8 months, respectively, did not differ from concomitant dabrafenib treatment with a median OSRT and OSBRAFi of 9.9 and 11.6 months (P=0.132/P=0.404). Median local control of the irradiated area did not differ in the interrupted and concomitant BRAFi treatment groups (P=0.619). Skin toxicity of grade ≥2 (CTCAE) was significantly increased in patients with concomitant vemurafenib compared to the group with treatment interruption (P=0.002). Conclusions: Interruption of vemurafenib treatment during radiation was associated with better survival and less toxicity compared to concomitant treatment. Due to lower number of patients, the relevance of treatment interruption in dabrafenib treated patients should be further investigated. The results of this analysis indicate that treatment with the BRAFi vemurafenib should be interrupted during radiotherapy. Prospective studies are desperately needed.}, language = {en} } @article{BrinkerHeklerHauschildetal.2019, author = {Brinker, Titus J. and Hekler, Achim and Hauschild, Axel and Berking, Carola and Schilling, Bastian and Enk, Alexander H. and Haferkamp, Sebastian and Karoglan, Ante and von Kalle, Christof and Weichenthal, Michael and Sattler, Elke and Schadendorf, Dirk and Gaiser, Maria R. and Klode, Joachim and Utikal, Jochen S.}, title = {Comparing artificial intelligence algorithms to 157 German dermatologists: the melanoma classification benchmark}, series = {European Journal of Cancer}, volume = {111}, journal = {European Journal of Cancer}, doi = {10.1016/j.ejca.2018.12.016}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-220569}, pages = {30-37}, year = {2019}, abstract = {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.}, language = {en} } @article{BrinkerHeklerEnketal.2019, author = {Brinker, Titus J. and Hekler, Achim and Enk, Alexander H. and Berking, Carola and Haferkamp, Sebastian and Hauschild, Axel and Weichenthal, Michael and Klode, Joachim and Schadendorf, Dirk and Holland-Letz, Tim and von Kalle, Christof and Fr{\"o}hling, Stefan and Schilling, Bastian and Utikal, Jochen S.}, title = {Deep neural networks are superior to dermatologists in melanoma image classification}, series = {European Journal of Cancer}, volume = {119}, journal = {European Journal of Cancer}, doi = {10.1016/j.ejca.2019.05.023}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-220539}, pages = {11-17}, year = {2019}, abstract = {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).}, language = {en} }