RKRaphael Koch

Raphael Koch

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Raphael Koch has a total of 563 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Top Cited Papers
    Open Access

    Prediction of antimicrobial resistance from MALDI-TOF mass spectra using machine learning: a validation study

    by Niklas Wiesmann, Dominic Enders, Antje Westendorf, Raphael Koch, Frieder Schaumburg

    Matrix-assisted laser desorption-ionization-time of flight (MALDI-TOF) mass spectra can be used to predict antimicrobial resistance (AMR) using machine learning (ML). This study aimed to validate the performance of ML models for AMR prediction using own and publicly available MALDI-TOF data and to test how these models perform over time. Mass spectra of Escherichia coli ( n = 7,897), Klebsiella pneumoniae ( n = 2,444), and Staphylococcus aureus ( n = 4,664) from routine diagnostics (Germany) and the DRIAMS-A database (Switzerland) were used. Six classification models were benchmarked for AMR prediction using cross-validation (regularized logistic regressions [LR], multilayer perceptrons [MLP], support vector machines [SVM], random forests [RF], gradient boosting machines [LGBM, XGB]). Performance was prospectively observed for 18 months after training. The performance of AMR prediction evaluated by the mean area under the receiver operating characteristic curve (AUROC) was comparable between the DRIAMS-A data set and own data. The best predictive performance (classifier, AUROC) on own data was achieved for oxacillin resistance in S. aureus (RF, 0.85), ciprofloxacin resistance in E. coli (XGB, 0.83), and piperacillin-tazobactam resistance in K. pneumoniae (XGB, 0.81). ML performance was poor if training and test data were unrelated in terms of location and time. Performance (change in AUROC) decreased within 18 months after training for S. aureus (oxacillin resistance, RF: −0.10), E. coli (ciprofloxacin, XGB: −0.19), and K. pneumoniae (piperacillin-tazobactam, XGB: −0.25). The performance of ML for the prediction of AMR based on MALDI-TOF data is good (AUROC ≥ 0.8) but classifiers need to be trained on local data and retrained regularly to maintain the performance level. IMPORTANCE: MALDI-TOF mass spectrometry can be used not only for bacterial species identification but also for the prediction of antimicrobial resistance (AMR) using machine learning (ML). Such an approach would provide antimicrobial susceptibility test results one day earlier than conventional routine diagnostics. This is essential for an early targeted treatment to reduce mortality of severe infections. We show that the performance of ML for the prediction of AMR based on MALDI-TOF data is good (AUROC ≥ 0.8). However, the ML models need to be trained on local data and retrained regularly to maintain a good performance.


  • Positive impact of standardized perioperative management on spinal cord ischemia rate during endovascular repair of thoracoabdominal aneurysms

    by Stefan Walczak, Raphael Koch, Marco Virgilio Usai, Yousef Shehada, Bärbel Berekhoven, Martin Austermann


  • Open Access

    Retrospective Evaluation of GEC-ESTRO Constraints for Definitive Radiochemotherapy with Brachytherapy and Correlation with Oncologic Outcome in Cervical Cancer: A Monocenter Study

    by Tom Schönicke, Raphael Koch, Isabel Vogt, Isabel Falke, Hans Theodor Eich, Gabriele Reinartz

    Background: This study aims to evaluate patients with locally advanced cervical cancer who underwent definitive radiochemotherapy, including brachytherapy, at the University Hospital of Muenster (UKM), focusing on target volume coverage, oncologic outcome parameters, and organs at risk (OAR) toxicities. Results are compared with the Gyn GEC-ESTRO (GGE) recommendations. Methods: Of a cohort of 48 patients, treated between 2013 and 2023, the physical radiation treatment planning with application of CT and MRI and oncologic follow-up data was analyzed. Target volume structures, comprising the high-risk clinical target volume (HR-CTV), intermediate-risk clinical target volume (IR-CTV), Point A, and corresponding EQD2(α/β=10) doses were determined. Endpoints included local tumor control, overall survival (OS), recurrence-free survival (RFS), and progression-free survival (PFS). Total OAR (D2cc) EQD2(α/β=3) doses were correlated with adverse events defined by CTCAE v5.0 and LENT-SOMA criteria. Results: Median follow-up was 58.0 months (95% CI [27.6, 88.4]). FIGO stage I was present in 7 (15%) patients, II in 13 (27%), and III in 28 (58%) patients. A total of 38 (79%) patients showed a complete remission 3 months after treatment. The 5-year event-free rate was 67.4% (95% CI [49.3, 80.3]) for OS, 77.0% (95% CI [56.7, 88.6]) for RFS and 68.1% (95% CI [49.7, 80.9]) for PFS. Incomplete radiation treatment and advanced tumor stages led to worse outcomes. Meeting Point A GGE recommendations increased chances for complete remission and could decrease chances of an event occurring for OS, PFS, and RFS. Compliance with GGE recommendations lowered the chances of OAR toxicity occurring. Conclusions: MRI-based target volume definition for brachytherapy in cervical cancer may improve patients’ OS, PFS, and RFS. Time-to-event endpoints are consistent with comparable studies, and adherence to current ESGO/ESTRO/ESP guidelines is endorsed.

  • Open Access

    Microvascular Density Analysis of Patients with Inactive Systemic Lupus Erythematosus—A Two-Year Follow-Up Optical Coherence Tomography Angiography Study

    by Martin Dominik Leclaire, Eliane Luisa Esser, Sebastian Dierse, Raphael Koch, Julian Alexander Zimmermann, Jens Julian Storp, Marie-Louise Gunnemann, Larissa Lahme, Nicole Eter, Nataša Mihailovic

    Background/Objectives: This study aims to investigate the long-term effect of inactive systemic lupus erythematosus (SLE) on the retinal microcirculation measured via optical coherence tomography angiography (OCT-A). Methods: Twenty-four eyes of 24 patients with inactive SLE under hydroxychloroquine (HCQ) therapy were included. The OCT-A data (mainly vessel density (VD) and foveal avascular zone (FAZ) data of the superficial and of the deep capillary plexus (SCP, DCP) and the choriocapillaris (CC)) were analyzed and compared between the baseline examination (t0) and 2 years later (t1). Results: At t1, VD in the whole en face SCP and in the CC was notably reduced compared to t0 (SCP: p = 0.001, CC: p = 0.013). VD in the DCP, CRT and FAZ area showed no difference at t1 compared to t0 (DCP: p = 0.128, FAZ: p = 0.332, CRT fovea: p = 0.296). Correlation analysis between the increase in cumulative doses of HCQ between t0 and t1 and the VD of the whole en face SCP did not show any correlation (Spearman r = 0.062 (95% CI −0.367; 0.477). Conclusions: SLE patients demonstrated a decrease in the retinal VD of the SCP and CC over a 2-year period. There was no correlation with the change in cumulative doses of HCQ. These results suggest an ongoing effect of the disease on the retinal and choriocapillary microcirculation.


  • Zoledronic Acid Add-on Therapy for Standard-Risk Ewing Sarcoma Patients in the Ewing 2008R1 Trial

    by Raphael Koch, Lianne Haveman, Ruth Ladenstein, Benedicte Brichard, Heribert Juergens, Sona Cyprova, Henk van Den Berg, Wolf Hassenpflug, Anna Raciborska, Torben Ek, et al.

    Purpose: The phase III, open-label, prospective, multicenter, randomized Ewing 2008R1 trial (EudraCT2008-003658-13) was conducted in 12 countries to evaluate the effect of zoledronic acid (ZOL) maintenance therapy compared with no add-on regarding event-free survival (EFS, primary endpoint) and overall survival (OS) in standard-risk Ewing sarcoma (EWS).Patients and Methods: Eligible patients had localized EWS with either good histologic response to induction chemotherapy and/or small tumors (<200 mL). Patients received six cycles of VIDE induction and eight cycles of VAI (male) or eight cycles of VAC (female) consolidation. ZOL treatment started parallel to the sixth consolidation cycle. Randomization was stratified by tumor site (pelvis/other). The two-sided adaptive inverse–normal four-stage design (planned sample size 448 patients, significance level 5%, power 80%) was changed after the first interim analysis using the Müller–Schäfer method.Results: Between April 2010 and November 2018, 284 patients were randomized (142 ZOL/142 no add-on). With a median follow-up of 3.9 years, EFS was not significantly different between ZOL and no add-on group in the adaptive design (HR, 0.74; 95% CI, 0.43–1.28, P = 0.27, intention-to-treat). Three-year EFS rates were 84.0% (95% CI, 77.7%–90.8%) for ZOL vs. 81.7% (95% CI, 75.2%–88.8%) for no add-on. Results were similar in the per-protocol collective. OS was not different between groups. The 3-year OS was 92.8% (95% CI, 88.4%–97.5%) for ZOL and 94.6% (95% CI, 90.9%–98.6%) for no add-on. Noticeable more renal, neurologic, and gastrointestinal toxicities were observed for ZOL (P < 0.05). Severe renal toxicities occurred more often in the ZOL arm (P = 0.003).Conclusions: In patients with standard-risk localized EWS, there is no additional benefit from maintenance treatment with ZOL.


  • Fetal Thymus Size at 19–22 Weeks of Gestation: A Possible Marker for the Prediction of Low Birth Weight?

    by Julia Maria Kim, Kathrin Oelmeier, Janina Braun, Kerstin Hammer, Johannes Steinhard, Helen Ann Köster, Raphael Koch, Walter Klockenbusch, Ralf Schmitz, Mareike Möllers

    Introduction: The purpose was to compare thymus size measured during second trimester screening of fetuses who were subsequently small for gestational age at birth (weight below 10th percentile, SGA group) with fetuses with normal birth weight (control group). We hypothesized that measuring the fetal thymic-thoracic ratio (TT-ratio) might help predict low birth weight. Methods: Using three-vessel view echocardiograms from our archives, we measured the anteroposterior thymus size and the intrathoracic mediastinal diameter to derive TT-ratios in the SGA (n = 105) and control groups (n = 533) between 19+0 and 21+6 weeks of gestation. We analyzed the association between TT-ratio and SGA adjusted to the week of gestation using logistic regression. Finally, we determined the possible TT-ratio cut-off point for discrimination between SGA and control groups by means of receiver operating characteristics (ROC) curve analysis. Results: The TT-ratio was significantly higher in the SGA group than in the control group (p < 0.001). An increase of the TT-ratio by 0.1 was associated with a 3.1-fold increase in the odds of diagnosing SGA. We determined that a possible discrimination cut-off point between SGA and healthy controls was achieved using a TT-ratio of 0.390 (area under the ROC curve 0.695). Conclusion: An increased TT-ratio may represent an additional prenatal screening parameter that improves the prediction of birth weight below the 10th percentile. Prospective studies are now needed to evaluate the use of fetal thymus size as predictive parameter for adverse fetal outcome.


  • Open Access

    Evaluation of Proenkephalin A 119–159 for liberation from renal replacement therapy: an external, multicenter pilot study in critically ill patients with acute kidney injury

    by Thilo von Groote, Felix Albert, Melanie Meersch, Raphael Koch, Joachim Gerss, Birte Arlt, Mahan Sadjadi, Christian Porschen, Peter Pickkers, Alexander Zarbock, et al.

    Introduction: Recent evidence suggests an association of plasma Proenkephalin A 119–159 (penKid) with early and successful liberation from continuous renal replacement therapy (CRRT) in critically ill patients with acute kidney injury. However, these exploratory results are derived from a monocentric trial and therefore require external validation in a multicenter cohort. Methods: Data and plasma samples from the “Effect of Regional Citrate Anticoagulation versus Systemic Heparin Anticoagulation During Continuous Kidney Replacement Therapy on Dialysis Filter Life Span and Mortality Among Critically Ill Patients With Acute Kidney Injury—A Randomized Clinical Trial” (RICH Trial) were used for this validation study. PenKid was measured in all plasma samples available at CRRT initiation and at day 3 of CRRT. Patients were categorized into low and high penKid groups with a cutoff at 100 pmol/l. Competing-risk time-to-event analyses were performed. Competing risk endpoints were successful and unsuccessful liberation from CRRT, the latter meaning death or initiation of a new RRT within one week of discontinuation of primary CRRT. Then penKid was compared to urinary output. Results: Low pre-CRRT penKid levels at CRRT initiation were not associated with early and successful liberation from CRRT compared to patients with high pre-CRRT penKid levels [subdistribution hazard ratio (sHR) 1.01, 95% CI 0.73–1.40, p = 0.945]. However, the landmark analysis on day 3 of ongoing CRRT demonstrated an association between low penKid levels and successful liberation from CRRT (sHR 2.35, 95% CI 1.45–3.81, p < 0.001) and an association between high penKid levels and unsuccessful liberation (sHR 0.46, 95% CI 0.26–0.80, p = 0.007). High daily urinary output (> 436 ml/d) was even stronger associated with successful liberation (sHR 2.91, 95% CI 1.80–4.73, p < 0.001) compared to penKid. Discussion: This study suggests that penKid may be a competent biomarker to monitor the recovery of kidney function during CRRT. This is in line with previous findings and investigated this concept in a multicenter cohort. Again, low penKid was associated with early and successful CRRT liberation, but was outperformed by high daily urinary output. The findings of this study now warrant further evaluation in prospective studies or a randomized controlled trial. Trial registration The RICH Trial was registered at clinicaltrials.gov: NCT02669589. Registered 01 February 2016.


  • Open Access

    Long-Term Outcomes of the PRESERFLO MicroShunt Implant in a Heterogeneous Glaucoma Cohort

    by Jens Julian Storp, Friederike Elisabeth Vietmeier, Ralph-Laurent Merté, Raphael Koch, Julian Alexander Zimmermann, Nicole Eter, Viktoria Constanze Brücher

    The Preserflo MicroShunt represents a novel glaucoma treatment device, necessitating long-term follow-up data to accurately assess its efficacy. The aim of this study is to report real-world data of a heterogenous glaucoma cohort who received Preserflo implantation at a specialized glaucoma clinic. A total of 160 eyes of 160 patients who underwent Preserflo MicroShunt implantation were retrospectively enrolled in this study. Patient characteristics, as well as success and failure rates, were assessed. The numbers of adverse events and revision procedures were recorded, along with any reduction in supplementary medication. The progression of intraocular pressure (IOP) was assessed over the course of 12 months, and fluctuations were analyzed. The overall success rate was 61.9% (complete success: 51.3%, qualified success: 10.6%). Revision surgery was performed in 25% of cases. Excessive hypotony occurred postoperatively in 54.4% of patients and regressed after 7 days in 88.8% of all cases. Median IOP decreased from 22 (interquartile range (IQR): 17–27) mmHg preoperatively to 14 (IQR 12–16) mmHg at 12 months postoperatively (p < 0.01). The median number of antiglaucomatous agents decreased from three to zero at latest follow-up. The Preserflo MicroShunt achieved a noticeable reduction in IOP over the course of 12 months in glaucoma patients, irrespective of disease severity or disease subtype. The frequency of postoperative adverse events and number for revision surgeries over the course of the follow-up period were low.


  • Open Access

    Clinical benefits of systemic amoxicillin/metronidazole may depend on periodontitis stage and grade: An exploratory sub‐analysis of the ABPARO trial

    by Peter Eickholz, Raphael Koch, Moritz Göde, Katrin Nickles, Thomas Kocher, Katrin Lorenz, Ti‐Sun Kim, Jörg Meyle, Doğan Kaner, Ulrich Schlagenhauf, et al.

    Aim: Assessment of treatment response after systemic amoxicillin/metronidazole adjunctive to subgingival instrumentation (SI) according to stages and grades of the 2018 classification of periodontal diseases.Materials and Methods: We carried out exploratory re‐analysis of the placebo‐controlled, multi‐centre ABPARO trial (52; 45/60 years of age; 205 males, 114 active smokers). Patients were randomized to SI with systemic amoxicillin 500 mg/metronidazole 400 mg (three times a day for 7 days, n = 205; ANTI) or placebo (n = 200; PLAC) and maintenance therapy every 3 months. Patients were reclassified according to the 2018 classification (stage/extent/grade). Treatment effect was the percentage of sites per patient with new attachment loss ≥1.3 mm (PSAL ≥ 1.3 mm) at 27.5 months post‐baseline/randomization.Results: All patients were assigned according to the stage (n = 49 localized stage III, n = 206 generalized stage III, n = 150 stage IV). Because of missing radiographs, only 222 patients were assigned to grades (n = 73 B, n = 149 C). Treatment (PLAC/ANTI) resulted in PSAL ≥ 1.3 mm (median; lower/upper quartile) in localized stage III (PLAC: 5.7; 3.3/8.4% vs. ANTI: 4.9; 3.0/8.3%; p = .749), generalized stage III (8.0; 4.5/14.3% vs. 4.7; 2.4/9.0%; p < .001), stage IV (8.5; 5.1/14.4% vs. 5.7; 3.3/10.6%; p = .008), grade B (4.4; 2.4/6.7% vs. 3.6; 1.9/4.7%; p = .151) and grade C (9.4; 5.3/14.3% vs. 4.8; 2.5/9.4%; p < .001).Conclusions: In generalized periodontitis stage III/grade C, a clinically relevant lower percentage of disease progression after adjunctive systemic amoxicillin/metronidazole was observed compared to placebo (PLAC: 9.7; 5.8/14.3% vs. ANTI: 4.7; 2.4/9.0%; p < .001).