Sensitivity profiling reveals consistent drug responses across preclinical neuroblastoma models

This study demonstrates that rapid ex vivo short-term drug screening offers a clinically viable alternative to slower organoid and PDX models for functional precision oncology in neuroblastoma, as it achieves higher success rates, faster turnaround times, and consistently recapitulates drug sensitivities observed in long-term preclinical models.

The Gist

Imagine you are a chef trying to save a patient who has a very stubborn, dangerous tumor called neuroblastoma. This tumor is like a fortress that is hard to break down. For a long time, doctors have tried to guess which "weapon" (drug) will work best by looking at the tumor's blueprints (DNA). But often, the blueprints don't tell the whole story, and the wrong weapons are used, wasting precious time.

This paper is about a new way to test weapons before they are used on the patient. The researchers compared three different "testing grounds" to see which one is the fastest, most reliable, and most accurate.

Here is the breakdown of their experiment using simple analogies:

1. The Three Testing Grounds

The researchers tried three different ways to test drugs on tumor cells:

• The "Slow-Motion" Garden (Organoids):
  • What it is: Taking tumor cells and growing them into tiny, 3D mini-tumors in a lab dish.
  • The Problem: It's like trying to grow a giant oak tree to test if a specific fertilizer works. It takes 3 to 12 months to get a big enough tree. By the time the tree is ready, the patient might be too sick to wait. Also, many seeds (samples) just refuse to grow (low success rate).
• The "Mouse Lab" (PDX Models):
  • What it is: Putting human tumor cells into mice to see how they grow and react to drugs.
  • The Problem: This is accurate but very expensive, slow, and ethically complicated. It's like sending a test pilot into a real storm to see if their plane holds up. It works, but it takes a long time and uses a lot of resources.
• The "Speed-Run" Kitchen (Ex Vivo Short-Term Screens):
  • What it is: Taking fresh tumor cells and testing them immediately in a lab dish for just one week.
  • The Benefit: This is like tasting a soup while it's cooking. You get the result in 14 days. The researchers found this method worked 65% of the time, compared to only 23% for the slow gardens.

2. The Big Discovery: "Does the Taste Match?"

The researchers wanted to know: If we taste the soup now (Short-Term), will it taste the same as the finished meal (Organoids) or the meal cooked in the oven (Mouse)?

• The Result: Yes! The taste was almost identical.
• The Analogy: Imagine you have a recipe. Whether you taste the sauce after 10 minutes of cooking or after 10 hours, the flavor profile is the same. The "Short-Term" test predicted exactly which drugs would kill the tumor, just as well as the slow, expensive methods.
• The Correlation: They found a 0.87 match (on a scale of 0 to 1) between the fast test and the slow test. This means the fast test is a very reliable crystal ball.

3. The "Good Guy vs. Bad Guy" Test (Selectivity)

A major worry in cancer treatment is that drugs might kill the tumor but also hurt the healthy body parts (like the heart or nerves).

• The Experiment: They tested the drugs on the "Bad Guys" (tumor cells) and the "Good Guys" (healthy cells from children, like skin or brain cells).
• The Findings:
  • Some drugs (like Cisplatin) were like snipers: They hit the tumor hard but barely touched the healthy cells.
  • Other drugs were like bombs: They hurt both the tumor and the healthy cells equally.
  • This helps doctors pick the "snipers" to save the patient's healthy organs.

4. The "Real World" Check (Mouse vs. Human Cells)

Finally, they checked if the results from the lab dish matched what happened when they actually treated the mice (the "Mouse Lab").

• The Result: For 7 out of 10 drugs, the lab dish predictions were spot on. If the dish said "This drug works," the mouse tumor shrank. If the dish said "This drug fails," the mouse tumor kept growing.

The Bottom Line

This paper is a game-changer for Precision Medicine.

• Before: Doctors had to wait months for a "slow-motion garden" to grow, hoping it would tell them what to do. Often, the patient got too sick to wait.
• Now: We have a "Speed-Run Kitchen." We can test a patient's tumor against dozens of drugs in just two weeks.
• Why it matters: It's faster, cheaper, works more often, and gives the same accurate results as the slow methods. It allows doctors to pick the right weapon for the patient while they are still fighting the battle, rather than waiting until the battle is almost lost.

In short: We found a way to test cancer drugs quickly and accurately, giving patients a much better chance of survival without waiting months for the answer.

Technical Summary

1. Problem Statement

High-risk and relapsed neuroblastoma remains a critical challenge in pediatric oncology, with overall survival rates stagnating below 50% despite intensive treatment. While molecular profiling (genomic/transcriptomic) is used to stratify patients, it fails to identify druggable targets for a significant subset of patients. Consequently, there is an urgent need for functional precision oncology approaches to guide treatment selection.

Current functional models, such as Patient-Derived Xenografts (PDXs) and Patient-Derived Organoids (PDOs), have shown promise but suffer from significant clinical limitations:

• Long Turnaround Times: Establishing organoids or PDXs often takes 3–12 months, exceeding the median time to progression for relapsed patients (4 months).
• Low Success Rates: Many models fail to establish due to growth arrest or insufficient biomass.
• Cost and Ethics: PDX models require animal use and are expensive to maintain.

The study aims to evaluate ex vivo short-term drug screening as a rapid, high-success-rate alternative to determine if it can reliably recapitulate drug sensitivities found in long-term organoids and in vivo PDX models.

2. Methodology

The study utilized a multi-center approach involving the Princess Máxima Center (Máxima) and the Hopp Children's Cancer Center (KiTZ) in Germany.

• Cohort: 55 drug screens were performed on samples from 38 neuroblastoma patients and 5 pediatric non-malignant samples (fibroblasts, astrocytes, PBMCs).
• Model Systems Compared:
  1. Ex vivo Short-term Screens: Freshly dissociated tumor cells cultured for 2–7 days (KiTZ) or 5–8 days (Máxima) before screening.
  2. Organoid Screens: Long-term cultures established over 3–12 months.
  3. PDX Models: Tumor tissue engrafted into immunodeficient mice, followed by short-term or organoid derivation.
  4. In vivo PDX Trials: Conducted within the ITCC-P4 consortium, testing 10 compounds in 8 matched PDX models.
• Drug Libraries:
  • Máxima: Libraries of 224–226 drugs (FDA-approved and clinical trial compounds).
  • KiTZ: Library of 78 clinically relevant anticancer drugs.
  • Concentrations: Tested across a range of 0.1 nM to 100 µM (or up to 250 µM).
• Assay Metrics:
  • Drug response quantified via Area Under the Curve (AUC) (0 = sensitive, 100 = resistant).
  • Viability assessed via Cell Titer Glo 3D (CTG3D).
  • Incubation times: 72 hours for short-term screens; 120 hours for organoids.
• Molecular Profiling: Whole Exome Sequencing (WES), RNA-seq, and methylation arrays were performed to correlate genomic alterations (e.g., MYCN, ALK, TP53) and transcriptomic pathways with drug sensitivity.
• Statistical Analysis: Pearson correlation for model consistency, Spearman rank correlation for ex vivo vs. in vivo concordance, and Wilcoxon rank-sum tests for differential sensitivity.

3. Key Contributions

• Comparative Feasibility Analysis: A direct head-to-head comparison of turnaround times and success rates between short-term screens, organoids, and PDXs.
• Cross-Model Validation: Demonstration that drug sensitivity profiles are highly consistent across:
  • Sample origin (Patient tissue vs. PDX-derived).
  • Culture duration (Short-term vs. Long-term organoids).
  • Assay types (Ex vivo vs. In vivo).
• Tumor Selectivity Assessment: Integration of pediatric non-malignant tissues to identify drugs with favorable therapeutic windows (tumor-selective vs. toxic to normal cells).
• Biomarker Discovery: Correlation of specific genomic (TP53, ALK) and transcriptomic (PI3K-AKT pathway) features with drug response, highlighting limitations of using mutational status alone as a predictor.

4. Key Results

A. Feasibility and Turnaround

• Success Rates: Short-term screens achieved a 65% success rate (13/20 in Máxima PDX cohort), whereas organoid screens succeeded in only 23% (5/22). Organoid failures were primarily due to growth arrest during long-term culture.
• Turnaround Time: Short-term screens provided results in 14 days, compared to 3–12 months for organoids. This makes short-term screening viable for real-time clinical decision-making.

B. Consistency of Drug Sensitivity

• Patient vs. PDX: Matched samples showed high correlation in drug response (Mean Pearson r = 0.84), indicating PDX models faithfully recapitulate patient drug sensitivity.
• Short-term vs. Organoid: Matched samples showed strong correlation between short-term and organoid screens (Mean Pearson r = 0.87). No significant differences in drug efficacy were observed between the two methods, suggesting short-term screens capture the essential drug response phenotype without the need for long-term culture.

C. Genomic and Transcriptomic Correlations

• TP53 Status: TP53 wild-type samples showed significantly higher sensitivity to MDM2 inhibitors (idasanutlin) and XPO1 inhibitors (selinexor) compared to TP53-altered samples.
• ALK Status: While ALK mutations predicted sensitivity to lorlatinib, they did not consistently predict sensitivity to other ALK inhibitors.
• Pathway Analysis: Reduced activity of the PI3K-AKT pathway correlated with increased sensitivity to ALK inhibitors (ceritinib, alectinib, lorlatinib), suggesting pathway activity is a better predictor than mutational status alone.
• Transcriptomics: Sensitivity to futibatinib correlated with FGFR4 expression; ribociclib sensitivity correlated with cell cycle gene expression.

D. Tumor Selectivity (Safety)

• Selective Drugs: BCL2 inhibitors (navitoclax, venetoclax), MDM2 inhibitor (idasanutlin), and topoisomerase inhibitor (mitoxantrone) were significantly more effective in tumor cells than in non-malignant pediatric tissues.
• Toxic Drugs: MEK inhibitors (cobimetinib, trametinib), mTOR inhibitors (everolimus, temsirolimus), and HDAC inhibitors (panobinostat) showed higher sensitivity in non-malignant cells, flagging potential toxicity risks.
• Standard of Care: Doxorubicin and vincristine showed similar efficacy in tumor and non-malignant cells, consistent with their known toxicity profiles (cardiotoxicity and neuropathy).

E. Ex Vivo vs. In Vivo Concordance

• In the ITCC-P4 cohort, ex vivo short-term screens showed concordant response patterns with in vivo PDX trials for 7 out of 10 compounds.
• High Concordance: Strong Spearman correlations were observed for alectinib (ρ = 1.0), cobimetinib (ρ = 0.71), and JQ-1 (ρ = 0.61).
• Discrepancies: Some compounds (e.g., copanlisib) showed weak concordance, likely due to pharmacokinetic differences (drug metabolism, dosing schedules) between in vitro and in vivo settings.

5. Significance and Conclusion

This study establishes ex vivo short-term drug screening as a superior alternative for functional precision oncology in neuroblastoma compared to traditional organoid and PDX models.

• Clinical Impact: The rapid turnaround (14 days) and high success rate (65%) allow for the integration of drug sensitivity data into multidisciplinary tumor boards for patients with relapsed or high-risk disease, where time is critical.
• Reliability: The high correlation between short-term screens, organoids, and in vivo PDX models validates that short-term cultures retain the biological fidelity necessary for predicting therapeutic response.
• Safety Profiling: The inclusion of non-malignant pediatric tissues provides a crucial layer of safety assessment, helping to distinguish between tumor-selective drugs and those with high toxicity risks.
• Future Direction: While organoids and PDXs remain essential for drug discovery and mechanistic studies, short-term screening offers a practical, scalable, and ethically favorable solution for guiding personalized treatment in the clinic. The authors recommend further optimization to handle small biopsy samples and the development of larger non-malignant reference panels.