iTReX: Interactive exploration of mono- and combination therapy dose response profiling data

ElHarouni, Dina; Berker, Yannick; Peterziel, Heike; Gopisetty, Apurva; Turunen, Laura; Kreth, Sina; Stainczyk, Sabine; Oehme, Ina; Pietiäinen, Vilja; Jäger, Natalie; Witt, Olaf; Schlesner, Matthias; Oppermann, Sina

doi:10.1016/j.phrs.2021.105996

Cited by 13 publications

(24 citation statements)

References 66 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Drug sensitivity of the FTC samples was tested for at least 76 clinically relevant drugs (mostly approved or late-phase clinical trials). The drug response was evaluated via the drug sensitivity score (DSS) as proposed by Yadav et al for high-throughput compound testing studies [28] with minor modifications (DSS asym ) [25]. We first confirmed that the different culture models reflected the molecular profile of the original tumor via DNA methylation analysis.…”

Section: Characterization and Comparison Of The Three Patient-derived...mentioning

confidence: 73%

“…Bioluminescence was read on a TECAN Spark (TECAN, Männedorf, Switzerland) or PheraStar (BMG Labtech, Ortenberg, Germany) plate reader. Results were analyzed for dose-response curves and drug sensitivity with the iTReX Shiny app [25] (LTCs and PDX-Cs; DSS asym computed with A min set to 10% from I max ) or an in-house developed automated drug screen analysis pipeline (FTCs).…”

Section: Drug Screening and Metabolic Activity Assaysmentioning

confidence: 99%

See 1 more Smart Citation

Functional Therapeutic Target Validation Using Pediatric Zebrafish Xenograft Models

Gatzweiler

Ridinger

Herter

et al. 2022

Cancers

Self Cite

View full text Add to dashboard Cite

The survival rate among children with relapsed tumors remains poor, due to tumor heterogeneity, lack of directly actionable tumor drivers and multidrug resistance. Novel personalized medicine approaches tailored to each tumor are urgently needed to improve cancer treatment. Current pediatric precision oncology platforms, such as the INFORM (INdividualized Therapy FOr Relapsed Malignancies in Childhood) study, reveal that molecular profiling of tumor tissue identifies targets associated with clinical benefit in a subgroup of patients only and should be complemented with functional drug testing. In such an approach, patient-derived tumor cells are exposed to a library of approved oncological drugs in a physiological setting, e.g., in the form of animal avatars injected with patient tumor cells. We used molecularly fully characterized tumor samples from the INFORM study to compare drug screen results of individual patient-derived cell models in functional assays: (i) patient-derived spheroid cultures within a few days after tumor dissociation; (ii) tumor cells reisolated from the corresponding mouse PDX; (iii) corresponding long-term organoid-like cultures and (iv) drug evaluation with the corresponding zebrafish PDX (zPDX) model. Each model had its advantage and complemented the others for drug hit and drug combination selection. Our results provide evidence that in vivo zPDX drug screening is a promising add-on to current functional drug screening in precision medicine platforms.

show abstract

Section: Characterization and Comparison Of The Three Patient-derived...mentioning

confidence: 73%

Section: Drug Screening and Metabolic Activity Assaysmentioning

confidence: 99%

Functional Therapeutic Target Validation Using Pediatric Zebrafish Xenograft Models

Gatzweiler

Ridinger

Herter

et al. 2022

Cancers

Self Cite

View full text Add to dashboard Cite

show abstract

“…The drug library consisted of 76 drugs of which most were approved and some were investigative compounds and the plate production was performed as previously published. 35 All concentrations were tested in duplicates, readout (after 72 hrs) was done on a PHERAstar FS microplate reader (BMG Labtech) using CellTiterGlo 2.0 (Promega) according to the manufacturer’s protocol. Drug sensitivity analyses were performed using the web-based drug analysis pipeline BREEZE ( https://breeze.fimm.fi ), developed at the Institute for Molecular Medicine Finnland (FIMM).…”

Section: Methodsmentioning

confidence: 99%

Predictive modeling of resistance to SMO inhibition in a patient-derived orthotopic xenograft model of SHH medulloblastoma

Krausert

Brabetz

Mack

et al. 2022

Neuro-Oncology Advances

Self Cite

View full text Add to dashboard Cite

Background Inhibition of the sonic hedgehog (SHH) pathway with SMO-inhibitors is a promising treatment strategy in SHH-activated medulloblastoma, especially in adult patients. However, the problem is that tumors frequently acquire resistance to the treatment. To understand the underlying resistance mechanisms and to find ways to overcome the resistance, preclinical models that became resistant to SMO-inhibition are needed. Methods To induce SMO-inhibitor resistant tumors, we have treated a patient-derived xenograft (PDX) model of SHH medulloblastoma, sensitive to SMO-inhibition, with 20 mg/kg Sonidegib using an intermitted treatment schedule. Vehicle-treated and resistant models were subjected to whole genome and RNA sequencing for molecular characterization and target engagement. In vitro drug screens (76 drugs) were performed using Sonidegib-sensitive and -resistant lines to find other drugs to target the resistant lines. One of the top hits was then validated in vivo. Results Nine independent Sonidegib-resistant PDX lines were generated. Molecular characterization of the resistant models showed that eight models developed missense mutations in SMO and one gained an inactivating point mutation in MEGF8, which acts downstream of SMO as a repressor in the SHH pathway. The in vitro drug screen with Sonidegib-sensitive and -resistant lines identified good efficacy for Selinexor in the resistant line. Indeed, in vivo treatment with Selinexor revealed that it is more effective in resistant than in sensitive models. Conclusions We report the first human SMO-inhibitor resistant medulloblastoma PDX models, which can be used for further preclinical experiments to develop the best strategies to overcome the resistance to SMO-inhibitors in patients.

show abstract

“…The CNNs fine-tuned using screen-specific controls (phases II and III, respectively) were subsequently applied to images acquired from all wells of that screen. Drug-sensitivity scoring used a prerelease version of the iTReX web application for drug-sensitivity scoring ( [40], v1.1.1.9000, functionally identical to release v1.2.0), run in R v4.1. Due to label smoothing (α = 0.4), raw image-based cell viabilities generally range around the interval [0.2, 0.8] and were therefore normalized using positive (STS, set to 0%) and negative (DMSO, 100%) controls to yield percentage image-based cell viabilities.…”

Section: Inference and Drug Scoringmentioning

confidence: 99%

“…For each drug, 10 percentage image-based cell viabilities (5 concentrations, duplicates) were converted to percentage inhibition values and used to fit a curve model based on the Hill equation (namely, a generalized 5-parameter logistic curve model in the log-concentration domain [42] having an asymmetry coefficient). The curve model was then integrated in the log-concentration domain to yield the which was then converted to an in-house-developed DSS (DSS asym , [40]) inspired by [43,Fig. 1], using an automated drug-scoring pipeline such as previously shown by others [43], [44], [45], [46].…”

Section: Inference and Drug Scoringmentioning

confidence: 99%

Patient-by-Patient Deep Transfer Learning for Drug-Response Profiling Using Confocal Fluorescence Microscopy of Pediatric Patient-Derived Tumor-Cell Spheroids

Berker

ElHarouni

Peterziel

et al. 2022

IEEE Trans. Med. Imaging

Self Cite

View full text Add to dashboard Cite

Image-based phenotypic drug profiling is receiving increasing attention in drug discovery and precision medicine. Compared to classical end-point measurements quantifying drug response, image-based profiling enables both the quantification of drug response and characterization of disease entities and drug-induced cell-death phenotypes. Here, we aim to quantify imagebased drug responses in patient-derived 3D spheroid tumor cell cultures, tackling the challenges of a lack of single-cell-segmentation methods and limited patientderived material. Therefore, we investigate deep transfer learning with patient-by-patient fine-tuning for cell-viability quantification. We fine-tune a convolutional neural network (pre-trained on ImageNet) with 210 control images specific to a single training cell line and 54 additional screenspecific assay control images. This method of image-based drug profiling is validated on 6 cell lines with known drug sensitivities, and further tested with primary patientderived samples in a medium-throughput setting. Network outputs at different drug concentrations are used for drugsensitivity scoring, and dense-layer activations are used in t-distributed stochastic neighbor embeddings of drugs to visualize groups of drugs with similar cell-death phenotypes. Image-based cell-line experiments show strong correlation to metabolic results (R ≈ 0.7) and confirm expected hits, indicating the predictive power of deep learning to identify drug-hit candidates for individual patients. In patient-derived samples, combining drug sensitivity scoring with phenotypic analysis may provide opportunities for complementary combination treatments. Deep transfer Manuscript

show abstract

iTReX: Interactive exploration of mono- and combination therapy dose response profiling data

Cited by 13 publications

References 66 publications

Functional Therapeutic Target Validation Using Pediatric Zebrafish Xenograft Models

Functional Therapeutic Target Validation Using Pediatric Zebrafish Xenograft Models

Predictive modeling of resistance to SMO inhibition in a patient-derived orthotopic xenograft model of SHH medulloblastoma

Patient-by-Patient Deep Transfer Learning for Drug-Response Profiling Using Confocal Fluorescence Microscopy of Pediatric Patient-Derived Tumor-Cell Spheroids

Contact Info

Product

Resources

About