Organoids as a biomarker for personalized treatment in metastatic colorectal cancer: drug screen optimization and correlation with patient response

doi:10.21203/rs.3.rs-3794312/v1

Download PDF

Research Article

Organoids as a biomarker for personalized treatment in metastatic colorectal cancer: drug screen optimization and correlation with patient response

https://doi.org/10.21203/rs.3.rs-3794312/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 27 Feb, 2024

Read the published version in Journal of Experimental & Clinical Cancer Research →

You are reading this latest preprint version

Background

The inability to predict treatment response of colorectal cancer patients results in unnecessary toxicity, decreased efficacy and survival. Response testing on patient-derived organoids (PDOs) is a promising biomarker for treatment efficacy. The aim of this study is to optimize PDO drug screening methods for correlation with patient response and explore the potential to predict responses to standard chemotherapies.

Methods

We optimized drug screen methods on 5–11 PDOs per condition of the complete set of 23 PDOs from patients treated for metastatic colorectal cancer (mCRC). PDOs were exposed to 5-fluorouracil (5-FU), irinotecan- and oxaliplatin-based chemotherapy. We compared medium with and without N-acetylcysteine (NAC), different read-outs and different combination treatment layouts to capture the strongest association with patient response. We expanded the screens using the optimized methods for all PDOs. Organoid sensitivity was correlated to the patient’s response, determined by % change in the size of target lesions. We assessed organoid sensitivity in relation to prior exposure to chemotherapy, mutational status and sidedness.

Results

Drug screen optimization involved excluding N-acetylcysteine from the medium and biphasic curve fitting for 5-FU & oxaliplatin combination screens. CellTiter-Glo measurements were comparable with CyQUANT and did not affect the correlation with patient response. Furthermore, the correlation improved with application of growth rate metrics, when 5-FU & oxaliplatin was screened in a ratio, and 5-FU & SN-38 using a fixed dose of SN-38. Area under the curve was the most robust drug response curve metric. After optimization, organoid and patient response showed a correlation coefficient of 0.58 for 5-FU (n = 6, 95% CI -0,44-0.95), 0.61 for irinotecan- (n = 10, 95% CI − .03-0.90) and 0.60 for oxaliplatin-based chemotherapy (n = 11, 95% CI -0.01-0.88). Median progression-free survival of patients with resistant PDOs to oxaliplatin-based chemotherapy was significantly shorter than sensitive PDOs (3.3 vs 10.9 months, p = 0.007). Increased resistance to 5-FU in patients with prior exposure to 5-FU/capecitabine was adequately reflected in PDOs (p = 0.005).

Conclusions

Our study emphasizes the critical impact of the screening methods for determining correlation between PDO drug screens and mCRC patient outcomes. Our 5-step optimization strategy provides a basis for future research on the clinical utility of PDO screens.

Organoids

Oncology

Cancer

Precision medicine

Colorectal cancer

Chemotherapy

Drug screening

For metastatic colorectal cancer (mCRC) patients, treatment options have expanded in recent years. Chemotherapy remains the backbone of mCRC treatment. The chance of response to treatment for mCRC patients varies from 10%-90%, depending on patient and tumour characteristics, treatment type and treatment line(1–4). One of the holy grails in cancer research is being able to predict to which treatment a patient will respond, enabling personalized treatment and improving the prospects for prolonged patient survival. Considerable research has been done in the field of predictive biomarkers, with the main focus on genetic biomarkers. Genetic biomarkers are, however, not available for chemotherapy and do not robustly predict treatment response for targeted treatments(5). As a result, oncologists are currently unable to predict treatment response or estimate the contribution of the different agents in combination therapy for individual patients.

A promising predictive biomarker is in vitro response testing using patient-derived organoids (PDOs)(6). PDOs are (cancer) stem-cell derived, 3D self-organizing and proliferating structures comprised of epithelial cells representing their corresponding tumour(7–10). PDOs are well suited for drug screening due to their long-term stability, applicability for high throughput screens and inherent heterogeneity(10, 11). This makes PDOs a novel promising biomarker to predict treatment response for various cancer and treatment types, and a valuable screening platform to identify new treatment types(9). Previous studies involving colorectal cancer (CRC) patients have demonstrated a correlation between organoid response and patient response to chemotherapy(9, 12–20). PDOs have, however, not consistently predicted the clinical outcome for oxaliplatin(18, 21), resulting in considerable controversy surrounding the predictive value of PDOs for this standard mCRC treatment. N-acetyl cysteine (NAC) is commonly used in organoid culture media(11), but is also known to interfere with platinum-based chemotherapy(22)^,(23). Prior studies that used NAC in oxaliplatin screening found no correlation between organoid sensitivity and patient response(18, 21), raising questions about effective drug screening methodologies. Likewise, variations in drug sensitivity measurements in prior studies underscore the need for standardized drug screening protocols to improve translatability to patient responses.

In this study, we first aim to optimize the PDO drug screening methods. Subsequently, we explore if these optimized PDO screens can predict treatment response in patients for standard-of-care systemic chemotherapies 5-FU/capecitabine, irinotecan- and oxaliplatin-based combination treatment. In addition, we investigate whether the effect on drug sensitivity of prior chemotherapy, mutational status, and primary tumor location (sidedness) is recapitulated in organoids.

Study design

We included organoids of 23 patients based on the following criteria: diagnosed with mCRC, tissue for organoid culture was obtained prior to receiving a new line of standard-of-care systemic therapy for metastatic disease (either from primary tumour or metastatic disease) and available clinical data (including treatments given, response scans and vital status). Tissue samples were collected during surgery or fine-needle biopsies within the Biobanking protocol HUB-Cancer (TCBIO #12–093, n = 15) or within the prospective clinical trial Organoids to Predict Therapy Response In Colorectal Cancer (OPTIC(24) #17–356, n = 8). Written informed consent was obtained prior to study inclusion. We first optimized drug screen methods, by examining the optimal screening medium, readout type, curve fitting, drug response curve (DRC) metrics and combination screen set-up, and we evaluated whether the optimization steps affect the correlation of the PDO screens with patient response (Fig. 1). After drug screen optimization on 5–11 PDOs per condition, we expanded the screens for the complete set of 23 PDOs and assessed the correlation with clinical response per treatment category. Additionally we evaluated organoid sensitivity in relation to prior exposure to chemotherapy, mutational status and sidedness.

Clinical data collection

Clinical data was collected from the electronic patient database by data managers who were blinded for the drug screen response data. The following variables were collected: sex, date of primary and metastatic disease diagnosis, number and location of metastases upon diagnosis of metastatic disease, TNM status, mismatch repair status, mutational status of the tumour (including RAS and BRAF status), and sidedness of the primary tumour (defined as right-sided (coecum-transverse colon), left-sided (splenic flexure-sigmoid) and rectosigmoid/rectal. The mutation status has been determined by NGS panels Ion Torrent PGM Cancer Hotspot panel v2Plus/Plus2/Plus3, mMIPs PATH gene panel combined with SeqNext or AvL Panel V1.2. Class I and II BRAF mutations were considered pathogenic, but not Class III.(25) RAS mutations included KRAS amplifications. Treatment information was collected, including primary tumour resection, local treatment of metastases (including metastasectomy or ablation), adjuvant chemotherapy (type treatment received) and the type and duration of treatment for each systemic palliative therapy line given. Chemotherapy exposure naïve prior to organoid establishment was defined as any given systemic treatment including adjuvant XXXreatment. A radiologist, who was blinded for the organoid screen data, reassessed all CT scans, for each treatment line according to RECIST 1.1. Clinical response on treatment was defined as the percentage size change of all target lesions during treatment, compared to the baseline scan. Progression-free survival for each treatment line was determined as the days between the start of a given treatment and the date of progressive disease on the response scan. If a patient received local treatment for metastatic target lesions, the progression-free survival date was censored for the date of the local treatment.

Organoid culture isolation and establishment

PDO isolation was performed as van de Wetering et al. (2015) described(10). PDOs were passaged using mechanical dissociation by pipetting and enzymatic dissociation with TrypLE Express (Gibco, Breda, the Netherlands, #12604021) for 5–10 minutes at 37^oC, washed using organoid culturing medium and re-plated by embedding in a solution of ice-cold Matrigel (Corning, Amsterdam, the Netherlands, #356231) (70%) and organoid culturing medium (Supplementary Table 1, Additional File 1) and plated as droplets on a pre-warmed 6-well plate. After solidification of the Matrigel, organoid culturing medium, containing Rho-kinase inhibitor (10 µM, Abmole Bioscience, Brussels, Belgium, #M1817), was added to the plates. The organoid culture medium was refreshed three times per week. Organoid identity was confirmed using a single-nucleotide polymorphism (SNP) array targeting 64 SNPs using TagMan OpenArray technology in combination with the QuantStudio 12K Flex Real-Time PCR System at the Utrecht Sequencing Facility (USEQ). We calculated genetic distances of PDO DNA versus blood DNA, assigning values based on SNP comparisons (1 for XX vs XY and YY vs XY, 2 for XX vs YY, -2 for no call, and ± 0.333 for invalid calls). PDOs with a genetic distance < 5 were included in the drug screens.

Organoid in vitro drug screens

The timeline of the drug screen was as follows: On day − 1, PDOs were sheared and re-plated. Day 0, PDOs were harvested by incubating with 1 mg/mL Dispase II (Life Technologies Europe B.V., Zuid-Holland, the Netherlands, #17105041) for 30 minutes at 37^oC, washed twice using organoid culturing medium with Rho-kinase inhibitor, filtered using a100 and 20 µm mesh filter to collect the PDOs which have passed through the 100 µm filter but not through the 20 µm filter, to remove debris and single cells. PDOs were resuspended in organoid screening medium with Rho-kinase inhibitor (5 µM, Abmole Bioscience, Brussels, Belgium, #M1817) with 5% Matrigel® for a final 125,000 PDOs/20 mL concentration. The organoid screening medium was composed of the organoid culture medium (Supplementary Table 1, Additional File 1) with the following adjustments: 1 ng/mL human recombinant EGF (Sigma-Aldrich, #A9165) and 10 ng/mL heregulin (Peprotech, London, UK, #100-03) to avoid interference by epidermal growth factor (EGF) with panitumumab sensitivity. Using an automated Multidrop™ Combi Reagent Dispenser, 40 µL of PDO suspension was dispensed in clear-bottomed, black-walled 384-well plates with ultra low-attachment coating (Corning, Zuid-Holland, the Netherlands, #4588). A ten-point concentration range of the treatments (Supplementary Table 2, Additional File 1), the positive control (Staurosporine) and negative control 1% (Dimethyl sulfoxide, Phosphate Buffered Saline or combination depending on the solvent used), were dispensed in technical quadruplicates using a Tecan Fluent liquid handler or Tecan D300E dispenser. The following commonly used CRC treatments were screened: 5-fluorouracil (5-FU), SN-38 (active metabolite of irinotecan), oxaliplatin, panitumumab, combination treatment 5-FU & oxaliplatin, and combination treatment 5-FU & SN-38. On day 5, readouts were obtained by quantifying cell viability using CellTiter-Glo 3D (Promega, #G9681, 40 µL/well) with a Tecan Spark plate reader and CyQUANT Direct proliferation assay (Invitrogen, C7026, 20 µL/well) with Perkin Elmer Operetta® CLS™. A baseline readout was measured on day 0 in a separate plate with PDOs without treatment. Drug screens were performed in at least two duplicate experiments on different days. Percentage viability and growth rate inhibition (GR) metrics values were calculated(26). GR values range from 1 to -1, with 0 to 1 for partial growth inhibition, 0 for complete cytostasis, and 0 to -1 for cell death. Four parameter log-logistic curves and biphasic sigmoid curves were fit using the DRC package in R(27) and the biphasic dose-response model in GraphPad, respectively. We calculated the Akaike information criterion (AIC) values of both models for 5-FU & oxaliplatin, to objectify goodness-of-fit. As AIC values of the biphasic model were lower (median decrease 9.5, range 2.5–26.7), response curves for 5-FU & oxaliplatin were fit using a biphasic model instead of a log-logistic model, resulting in GR₅₀1 and GR₅₀2 parameters. The GR_AUC (area under the non-fitted ‘curve’ of the raw GR values), GR₅₀ (concentration that gives half-maximal growth rate inhibition), GR₅₀1 and GR₅₀2 (concentration of 50% growth rate inhibition of the upper and lower part of the biphasic curve, respectively) were calculated. Normalized values for GR_AUC and GR₅₀(2) were calculated using the maximum and minimum measured parameter per treatment to compare different treatment types with different concentration ranges.

Statistical methods

Drug screens were optimized by comparing correlation coefficients of different PDO screening methods with clinical response. Where applicable, two-sided Pearson correlation tests were utilized to examine linear associations between continuous variables. Spearman correlation tests were employed when the data did not meet the assumptions of normality. Associations between clinical and organoid response were examined by scatterplots. Univariate cox proportional hazard regression analysis was performed to estimate the hazard ratio (HR) with a 95% CI. AUC values were transformed to a standard normal distribution to estimate the HR. Kaplan-Meier survival curves were used to plot progression-free survival, and a two-sided Wald test was used to compare survival between patients with different organoid sensitivity. PDOs were classified as sensitive if the normalized GR_AUC of the 5-FU & oxaliplatin drug screen was below the upper tertile. This cut-off was based on the response rate of approximately 60% for patients treated with FOLFOX in the first line in order to evaluate whether the upper third most resistant PDOs indeed predict failure of FOLFOX (1, 2). We evaluated organoid sensitivity in relation to prior exposure to chemotherapy, mutational status and sidedness using boxplots. A two-sided Mann-Whitney U-test were applied for comparing 2 groups. Analysis was performed in R (version 4.0.3). DRCs and derived metrics were fit and calculated with the GRmetrics (version 1.16.0) and DRC (version 3.0–1) packages. Heatmaps were plotted using the Pheatmap (version 1.0.12) package.

Patient characteristics

The study cohort consists of 23 mCRC patients who were treated with standard-of-care systemic therapy for metastatic disease. Four PDOs were derived from primary tumours and nineteen from liver metastases (Table 1). Most patients were chemotherapy naïve when the PDO was established (14/23). The clinical characteristics of the included patients reflect the heterogeneous mCRC population (Supplementary Table 3, Additional File 1). The cohort includes eleven RAS-mutants, three BRAF-mutants and ten RAS/BRAF-wildtype tumours. The cohort comprises five patients with rectal tumours, eleven patients with left-sided tumours, and seven patients with right-sided tumours (Supplementary Table 3, Additional File 1). Evaluable treatments in the first line of treatment after PDO establishment include capecitabine monotherapy (prodrug of 5-FU, n = 6), 5-FU & oxaliplatin combination treatment (n = 13) and irinotecan-based treatment (5-FU & irinotecan combination treatment or irinotecan monotherapy, n = 4). Late-line treatments include 5-FU & oxaliplatin combination treatment (n = 1), irinotecan-based treatment (n = 6) and panitumumab (n = 6). Bevacizumab treatment was given to the vast majority of the patients (n = 19). However, this treatment is not evaluable in vitro, since its main effect is on the tumor vasculature(28).

Table 1

**Overview of evaluable treatments per patient**.
Patient ID	Evaluable treatment 1	Response 1	Evaluable treatment 2	Response 2	Evaluable treatment 3	Response 3	Mutational status*	PDO origin	Stage	Sidedness
04	5-FU	SD	5-FU & irinotecan	SD	5-FU & oxaliplatin	PR	RAS mutant	Metastatic	IV	Left
06	5-FU	SD					RAS mutant	Metastatic	IV	Left
07	5-FU	SD					RAS mutant	Metastatic	IV	Left
11	5-FU	PD					Wildtype	Metastatic	IV	Right
16	5-FU	SD	Panitumumab	PD			BRAF mutant	Primary	IV	Right
18	5-FU	PD					Wildtype	Metastatic	III	Rectum
01	Irinotecan	SD					RAS mutant	Metastatic	II	Rectum
02	5-FU & irinotecan	SD					RAS mutant	Metastatic	III	Right
23	5-FU & irinotecan	SD	Panitumumab	SD			BRAF mutant	Metastatic	III	Left
20	5-FU & irinotecan	PD
20	5-FU & oxaliplatin	PD					RAS mutant	Primary	IV	Left
03	5-FU & oxaliplatin	PD					RAS mutant	Primary	IV	Right
05	5-FU & oxaliplatin	PR					RAS mutant	Primary	IV	Right
08	5-FU & oxaliplatin	PR			Panitumumab	SD	Wildtype	Metastatic	IV	Rectum
09	5-FU & oxaliplatin	SD					BRAF mutant	Metastatic	IV	Left
10	5-FU & oxaliplatin	PR	Panitumumab	SD			Wildtype	Metastatic	IV	Left
12	5-FU & oxaliplatin	PR	5-FU & irinotecan	SD	Panitumumab	SD	Wildtype	Metastatic	IV	Left
13	5-FU & oxaliplatin	SD	Irinotecan	SD			RAS mutant	Metastatic	IV	Left
14	5-FU & oxaliplatin	PR	5-FU & irinotecan	PR			Wildtype	Metastatic	III	Rectum
15	5-FU & oxaliplatin	PR	5-FU & irinotecan	SD	Panitumumab	SD	Wildtype	Metastatic	III	Rectum
17	5-FU & oxaliplatin	PR	5-FU & irinotecan	SD			RAS mutant	Metastatic	IV	Right
19	5-FU & oxaliplatin	PD					Wildtype	Metastatic	III	Left
21	5-FU & oxaliplatin	PD					Wildtype	Metastatic	II	Right
22	Panitumumab	PR					Wildtype	Metastatic	IV	Left

Table 1. Table describing the evaluable treatments, ranked per primary evaluable treatment. The first, second and third lines after organoid establishment are shown for each patient, along with the patient’s best RECIST response during treatment. The RAS/BRAF mutational status, of the original tumour, PDO origin, tumour stage at presentation and primary tumour location (sidedness) are presented. All patients had a proficient mismatch repair (pMMR) tumour. Patient 20 was resistant to FOLFOXIRI triplet chemotherapy and is therefore evaluable for 5-FU & oxaliplatin and 5-FU & irinotecan. Clinically capecitabine (prodrug 5-FU) is used often, while in in vitro screens 5-FU is evaluated. Abbreviations: 5-FU (5-fluorouracil), CAP (capecitabine), ID (identification), PD (progressive disease), PDO (patient-derived organoid), PR (partial response), SD (stable disease).

Organoid drug screen quality

The drug screen quality was analyzed by examining the Z’-factor(29), correlation in DRC metrics among the biological replicates and Bland-Altman plot for the difference in GR_AUC between replicates. The mean Z’-factor for all PDOs and treatments was 0.64 with only 4% below 0.5, indicating a good quality drug screen (Supplementary Table 4, Additional File 1). Biological replicates were not included in the analysis if the runs had a Z’-factor < 0.3 or if technical errors had occurred (e.g. dispensing error). When drug suspension errors occurred or replicate curves deviated, the respective biological replicates were excluded (9% of the drug screens) and a third and/or fourth run was performed. The calculated DRC metrics GR_AUC GR₅₀ and GR_max of different included biological replicates were significantly correlated (Spearman correlation > 0.84, p < 0.05, (Supplementary Fig. 1A, Additional File 1). The mean difference in GR_AUC between included replicates was − 0.09, with limits of agreement ranging from − 1.17 to 0.98 (Supplementary Fig. 1B, Additional File 1). The DRC per PDO for each treatment screened are displayed in Supplementary Fig. 2, Additional File 1.

Drug screen optimization

Removal of N-acetyl cysteine from screening medium

We compared screening medium with and without NAC for oxaliplatin-based growth inhibition assays. We confirmed that NAC increased resistance to oxaliplatin-based treatment, with a 2.2-fold increase in the median GR₅₀ and a 24% increase in the median GR_AUC (Supplementary Fig. 3A, Additional File 1). This limits the evaluability of the results as it leads to high doses oxaliplatin in a clinically irrelevant range. Withdrawal of NAC from the PDO drug screening medium had no disadvantageous effect on organoid growth of the untreated control. The growth rate from day 0 to day 5 with NAC was 5.5 (95% CI 4.2, 6.7) and 6.8 with (95% CI 5.5, 8.1) without NAC. Moreover, the correlation with clinical response improved after removal of NAC (Pearson correlation 0.6 without NAC vs 0.3 with NAC, Table 2). We, therefore, continued all drug screens for oxaliplatin-based treatment without NAC.

Drug screen readout (CellTiter-Glo versus CyQUANT)

Various approaches exist for quantifying organoid viability as a proxy for drug response. We used ATP-based CellTiter-Glo, as this luminescence-based read-out is most commonly used in literature as it is a rapid, high-throughput assay(30). In addition, we explored the nuclear fluorescent-based CyQUANT readout. CyQUANT offers the benefit of directly measuring number of viable cells instead of relying on surrogate measures(31) and enables PDO screening with < 10 organoids per well. We found that the Pearson correlation between drug sensitivity measurements (GR_AUC) utilizing CellTiter-Glo and measurements utilizing CyQUANT was 0.78 for 5-FU monotherapy (95% CI 0.50, 0.92), 0.69 for 5-FU & SN-38 combination treatment (95% CI -0.04, 0.94) and 0.57 for 5-FU & oxaliplatin combination treatment (95% CI -0.23, 0.91 Supplementary Fig. 3B-D, Additional File 1). As drug sensitivity measurement for 5-FU & oxaliplatin combination obtained from the QyQUANT readout showed limited concordance with the CellTiter-Glo readout, we further explored the CyQUANT readout for this drug combination in 6 PDOs. We found that QyQUANT and CellTiter-Glo measurements showed similar Pearson correlations of 0.77 and 0.81, respectively (Table 2).

Viability, growth rate metrics and curve fitting

For all evaluated treatments, we compared GR metrics to percentage viability measurements (supplementary Fig. 3E, Additional File 1). Using growth rate inhibition metrics instead of percentage viability corrects for confounders in organoid drug sensitivity, related to differences in cell division rate in untreated controls. Applying GR metrics improved correlations compared to percentage viability for 5-FU & oxaliplatin (Pearson correlation 0.6 vs 0.46), 5-FU & SN-38 combination treatment (Pearson correlation 0.66 vs 0.2) and 5-FU monotherapy (Pearson correlation 0.58 vs 0.30, Table 2). Response curves of most drugs exhibit a sigmoidal shape. Interestingly, we observed a biphasic drug response with a static plateau for 5-FU & oxaliplatin combination screens (Fig. 2A).

Table 2

Comparison of organoid drug screening conditions.
Comparison	Condition 1	Condition 2
Screening medium 5-FU & oxaliplatin	With NAC (CQ ratio) r = 0.32 [-0.78, 0.94]	Without NAC (CQ ratio) r = 0.77 [-0.35, 0.98]
Readout 5-FU & oxaliplatin	CQ (ratio without NAC) r = 0.77 [-0.35, 0.98]	CTG (ratio without NAC) r = 0.81 [-0.24, 0.99]
Viability vs GR 5-FU & oxaliplatin	Viability (ratio without NAC) r = 0.46 [-0.19, 0.83]	GR (ratio without NAC) r = 0.6 [-0.01, 0.88]
Viability vs GR 5-FU & SN-38	Viability (CTG fixed) r = 0.26 [-0.54, 0.82]	GR (CTG fixed) r = 0.66 [-0.08, 0.93]
Viability vs GR 5-FU	Viability (CTG) r = 0.3 [-0.68, 0.89]	GR (CTG) r = 0.58 [-0.44, 0.95]
Combination screening 5-FU & oxaliplatin	Fixed (CTG with NAC) r= -0.25 [-0.76, 0.45]	Ratio (CQ without NAC) r = 0.6 [-0.01, 0.88]
Combination screening 5-FU & SN-38	Fixed (CTG) r = 0.66 [-0.08, 0.93]	Ratio (CTG) r = 0.3 [-0.79, 0.94]

Table 2. Pearson correlation coefficients with 95% confidence interval for different organoid drug screening conditions. Abbreviations: 5-FU (5-fluorouracil), CTG (CellTiter-Glo), CQ (CyQUANT), GR (growth rate), NAC (N-acetyl cysteine), SN-38 (active metabolite of irinotecan).

Correlation of organoid response with patient response

Next, the optimized screening methods were used to examine the association between organoid and patient response on the complete set of PDOs per treatment type, for 5-FU & oxaliplatin, irinotecan-based treatment and 5-FU monotherapy. We were able to evaluate the most optimal strategy for combination screens in either a fixed or a ratio of the two compounds. Our primary outcome measure for patient response is RECIST-based percentage size change of all metastatic tumour lesions, since this is considered the most direct outcome measure.

5-FU & oxaliplatin

We expanded the CyQUANT screens for evaluating organoid sensitivity and correlation with patient response to 5-FU & oxaliplatin on fourteen PDOs, of which eleven PDOs were derived from metastatic lesions. 5-FU & oxaliplatin comprises the largest treatment cohort and the majority of patients was treated in the first line, directly after the organoid biopsy was obtained. Sensitivity of PDOs derived from metastatic lesions to 5-FU & oxaliplatin without NAC, screened in a 1.8:1 ratio and measured by GR_AUC, showed a Pearson correlation of 0.6 (p = 0.053) with change in tumour size during 5-FU & oxaliplatin treatment. The median GR_AUC was 0.57 for PDOs derived from patients with a decrease in size of metastatic lesions and 0.87 for PDOs derived from patients with stable or an increase in size of metastatic lesions (p = 0.012). There was discordance between patient response to 5-FU & oxaliplatin and PDO response in two of the three PDOs derived from primary tumors. The Pearson correlation for PDOs derived from metastases and primary tumours was limited to 0.29 (Supplementary Fig. 4A, Additional File 1). Next to GR_AUC, other DRC metrics were evaluated, including the GR₅₀2 parameter for biphasic curves. GR₅₀2 showed a Pearson correlation of 0.7 (p = 0.02) with patient response to 5-FU & oxaliplatin (Supplementary Fig. 4B, Additional File 1). We explored PFS as a secondary outcome measure for patient response. First, in a Cox proportional hazards model, the hazard ratio was estimated to be 2.30 (95% CI 0.68, 7.71, p = 0.2). Secondly, PDOs were classified as sensitive or resistant based on the GR_AUC. PFS of patients with PDOs classified as sensitive was substantially longer compared to patients with resistant PDOs (median PFS 3.3 vs 10.9 months, log rank p = 0.007, Fig. 2C). For further optimization, we compared combination drug screens using a fixed ratio with increasing dosage of both 5-FU & oxaliplatin (‘ratio’) versus using a fixed concentration of oxaliplatin and increasing only 5-FU (‘fixed’). How drugs were combined in the screen influenced organoid sensitivity to the drug combination. Organoid response to oxaliplatin in a fixed concentration and increasing 5-FU did not show a positive correlation with patient response to 5-FU & oxaliplatin (Pearson correlation − 0.25, Table 2).

Irinotecan-based treatment

Ten patients received irinotecan-based treatment in 1st or 2nd treatment line, either monotherapy (n = 2) or in combination with 5-FU (n = 8). For irinotecan-based treatment, organoid sensitivity to a fixed dose SN-38 and increasing concentrations 5-FU (with NAC, using CellTiter-Glo) measured by GR_AUC showed a Pearson correlation of 0.61 (p = 0.059) with patient response (Fig. 3A, B). For 5-FU & SN-38 no other DRC parameters then GR_AUC were suitable for measuring drug sensitivity. GR₅₀ could not reliably be measured due the absence of an evident lower plateau in the DRC of these screens (Fig. 3A). As patients in this cohort were treated in different lines, PFS could not reliably be compared. For further optimization, we compared combination drug screens using a fixed ratio of 1500:1 with increasing dosage of both 5-FU & SN-38 (‘ratio’) versus using a fixed concentration of SN-38 and increasing only 5-FU (‘fixed’). When PDOs were screened using a ratio, the sensitivity of the PDOs did not result in improved correlations with patient response to the combination treatment (Pearson correlation 0.3, Table 2), in contrast to 5-FU & oxaliplatin.

5-FU

Six patients were treated with capecitabine monotherapy (prodrug of 5-FU) which allows a direct single agent comparison of patient response with organoid response to 5-FU. All six patients received capecitabine as 1st treatment line and directly after the organoid biopsy was obtained. Organoid sensitivity to 5-FU measured by GR_AUC showed a Pearson correlation of 0.58 (p = 0.23) with % change in tumor size during capecitabine treatment (Fig. 3C, D).

PDOs reflect clinical heterogeneity in drug sensitivity, effect of prior treatment, primary tumor sidedness and mutational status

All PDOs in the cohort were screened for commonly used treatments in CRC, allowing comparison of drug sensitivity for patients with different tumour characteristics. We investigated the impact of mutational status, primary tumour location, and prior chemotherapy exposure on organoid drug sensitivity. When we examined organoid sensitivity for different standard-of-care drugs, a wide range of responses was seen (Fig. 4A). Some PDOs display a generally resistant phenotype (e.g. PDO 21) to most treatments, while other PDOs display a differential treatment response (e.g. PDO 19). We found that PDOs obtained from patients with a left-sided RAS/BRAF-wildtype tumour were more sensitive to EFGR-inhibitor panitumumab, compared to PDOs obtained from patients with right-sided or rectal RAS/BRAF-wildtype tumours, RAS-mutant tumours and BRAF-mutant tumours (median normalized GR_AUC 0.17 vs 0.74, p = 1.38). Interestingly, we observed varying organoid sensitivity to panitumumab within the group of left-sided RAS/BRAF-wildtype tumours (Fig. 4B). Organoid sensitivity to chemotherapeutic treatment with 5-FU was increased in organoids derived from RAS mutant tumours (median normalized GR_AUC 0.42 vs 0.64, p = 0.057, Fig. 4C).

PDOs from tumours exposed to 5-FU or capecitabine (in combination with oxaliplatin) before establishment (n = 9), mainly in the form of adjuvant treatment (n = 6), are more resistant to 5-FU than untreated PDOs (median normalized GR_AUC of 0.92 versus 0.43, p = 0.005). However, this resistance does not extend to other chemotherapy types (Fig. 5).

Our study shows that CRC PDO drug screens need optimization and standardization before being reliably utilized as biomarker for patient response in clinical practice. Key optimization steps include NAC removal from the screening medium, using biphasic or logistic curve fitting based on the type of treatment, applying growth rate metrics and selecting ratio or fixed concentrations depending on the combination treatments used. Our data supports the evidence that organoid response can be associated with patient response to 5-FU/capecitabine, irinotecan- and oxaliplatin-based chemotherapy, but only after drug screen optimization. In addition, we show that PDOs adequately reflect key clinical aspects and capture heterogeneity in treatment response depending on prior chemotherapy exposure, mutational status and tumour sidedness. Our results provide several methods to improve PDO drug screening for CRC. Furthermore, we offer insight into clinical factors that affect organoid sensitivity.

We aimed to optimize CRC PDO drug screen methods given the significant variation in drug screen medium composition, read-outs and combination screen layout used in different studies(9, 32). A standardized screening protocol could enhance cross-study comparisons and improve reproducibility. Our study's strength lies in comparing various PDO drug screening methods for diverse treatment types on a set of heterogenous PDOs. We examined aspects such as medium composition, screening readout, response curve fitting, response curve metrics, and combination screen layout, to capture the strongest association (best correlation coefficient) with patient response. Firstly, regarding medium composition, it was previously reported that NAC affects sensitivity to oxaliplatin in CRC organoids from peritoneal metastasis(23). Our screens confirmed increased resistance to oxaliplatin-based treatment with NAC-containing screening medium. This led to the necessity of using high, clinically irrelevant oxaliplatin doses and poor correlation with patient response. Therefore, we recommend removing NAC from the medium in oxaliplatin-based drug screens. Secondly, we compared two types of drug screen readouts to define the most optimal readout. ATP-based readout is the most commonly used for PDO drug screening. Nevertheless, based on the compound mechanism of action, ATP measurements can lead to misleading results as metabolic activity does not always correlate with cell viability(30). Therefore, we explore the use of a well-established readout based on DNA-content for 2D cell lines (CyQUANT) to measure cell viability in PDO-based screens. Nuclear fluorescent-based assays such as CyQUANT or propidium iodide dye combined with Hoechst are less affected by cell changes unrelated to viability, such as senescence. We found that drug sensitivity measurements with the CyQUANT readout instead of the ATP-based CellTiter-Glo readout was possible in 3D-based screens and did not affect the correlation with patient response. Thirdly, bias caused by the proliferation rate of PDOs was avoided by using growth rate metrics for organoid response analysis (26) and improved the correlation with patient response. Finally, we optimized the drug screen layout for combination screens and found that a fixed concentration of SN-38 is recommended for SN-38 based combination screening. This recommendation is based on the finding that SN-38 & 5-FU in a fixed screen correlated best with patient response to 5-FU & irinotecan. This might be explained by the fact that 5-FU can inhibit the in vitro efficacy of SN-38 at high concentrations(33). For oxaliplatin-based treatments, we recommend a ratio screen to capture the additive effect of both compounds. A summary of all recommendations is provided in Table 3.

Table 3

**Recommendations for CRC organoid drug screening**.
Topic	Evidence	Recommendations
Medium composition	Resistance to oxaliplatin-based treatment increases with NAC in screening medium	Remove NAC from screening medium in oxaliplatin-based drug screens
DRC fitting	PDOs exhibit a biphasic drug response to 5-FU & oxaliplatin	Apply a biphasic model for DRC fitting instead of a log-logistic model
Readouts	CellTiter-Glo measurements are in agreement with CyQUANT measurements.	Both readouts can be used. QyQUANT provides the advantage of performing drug screens with 5–10 PDOs per well.
Plate layout	SN-38 & 5-FU in a ratio combination screen did not reflect patient response to 5-FU & irinotecan. Oxaliplatin & 5-FU with a fixed oxaliplatin concentration did not reflect patient response to 5-FU & oxaliplatin.	Use a fixed concentration of SN-38 and increase the 5-FU concentration for 5-FU & SN-38 combination screens. Use a concentration ratio 5-FU:oxaliplatin for oxaliplatin-based combination screens.
DRC metrics	Growth rate inhibition metrics showed better correlation with patient response than percentage viability metrics, for 5-FU (& SN-38). GR_AUC is the most robust DRC metric and best reflects PDO and patient response.	Apply growth rate metrics to correct for confounders in organoid drug sensitivity, related to differences in natural cell division rate. Employ GR_AUC for comparison with patient response, or GR₅₀ if a clear lower curve plateau is present.

Table 3. Abbreviations: 5-FU (5-fluorouracil), CTG (CellTiter-Glo), CQ (CyQUANT), DRC (drug response curve), GR_AUC (area under the growth rate inhibition curve), NAC (N-acetylcysteine), PDO (patient-derived organoid), SN-38 (active metabolite of irinotecan).

After PDO drug screen optimization, we assessed the correlation with patient response on mCRC treatment. Our findings confirm the reported results from previous studies in which correlations were found for organoid response and patient response to chemotherapy. In the literature, consistent results are seen for 5-FU and irinotecan-based treatment, showing good correlations of the AUC with both RECIST response and PFS in several, relatively small, studies which aligns with our results(9, 12–14, 21). However, there is still considerable controversy surrounding the predictive potential for oxaliplatin-based treatments, one of the most prescribed treatments. PDOs failed to predict clinical outcome in two previous studies(18, 21). In contrast, others showed that PDOs could predict response to oxaliplatin-based chemotherapy(12–14, 19). In our study, organoid response (GR_AUC) positively correlated with patient response (% size change in metastatic lesions) for 5-FU & oxaliplatin, despite the small sample size. Furthermore, classifying PDOs as sensitive and resistant based on GR_AUC, resulted in a clinically relevant difference in median PFS in the sensitive and resistant group. This cut-off should be validated in separate future studies. The agreement between organoid and patient response to oxaliplatin-based treatment could result from removing of NAC from the screening medium and using the 5-FU:oxaliplatin ratio combination layout. A good association with patient response was found in a study that used a comparable ratio(14, 20), while no association was found in studies that used different methods for combination screens(18, 21). Previous studies used DRC parameters AUC, IC₅₀, and GR₅₀ to evaluate organoid response(12–14, 21). In our research, GR_AUC proved robust and patient-response reflective. For full sigmoidal/biphasic curves, GR₅₀(2) could also serve as reliable drug sensitivity measures.

In addition to the correlation with patient response, we show that CRC PDOs adequately reflect key clinical aspects regarding drug sensitivity and capture heterogeneity in treatment response. Patient response to EGFR inhibitors is influenced by factors such as tumour sidedness and RAS/BRAF mutational status. In line with the findings of large clinical trials(34, 35), PDOs from patients with a left-sided colon RAS/BRAF-wildtype tumour were most sensitive to panitumumab in our cohort. In addition to mutational status, we showed that the resistance after 5-FU or capecitabine containing treatment is well captured in PDOs. This implies that PDOs provide a representative model for the 5-FU resistant state after prior chemotherapy exposure. As recognized in daily clinical practice, adjuvant chemotherapy might also affect response to palliative treatment. In this context, PDOs could be applied to study resistance mechanisms and evasion strategies. It also underlines the importance of deriving organoids directly before a new treatment starts, as prior treatments affect sensitivity to subsequent therapies.

In this small retrospective study utilizing biobank samples, most samples were not initially collected for direct comparison with patient response. Consequently, these samples were not acquired immediately before initiating the evaluated treatment. The cohort’s diversity further stems from the fact that the collected tissue does not exclusively mirror the metastatic lesions under evaluation in patient responses; it might encompass primary tissue or other resected metastases. While this aspect presents challenges for direct comparison with patient response, the cohort is suitable for refining methodologies and demonstrates the predictive potential of PDOs. The limited number of PDOs in the treatment subgroups, however, constrains the demonstration of significant correlations with patient response for all treatments. The predictive potential of PDOs using the optimized drug screening method will be validated in the ongoing OPTIC trial(24), where we establish PDOs for mCRC patients from newly obtained biopsies immediately prior to the start of treatment, and directly compare patient response with organoid response. These results are key to enhance screening methods and advance towards the clinical application of PDOs for personalized treatment. Our data support the importance of PDOs as a model system as it adequately reflects the effect of prior treatment, primary tumour sidedness and mutational status.

Our study emphasizes the critical impact of the screening methods for determining correlation between PDO drug screens and mCRC patient outcomes. We used a 5-step optimization strategy including NAC removal from the screening medium, using biphasic or logistic curve fitting based on the type of treatment, applying growth rate metrics and selecting ratio or fixed concentrations depending on the combination treatments used. By using the optimized methods, PDO response correlated with patient response for oxaliplatin-based treatment. This optimization provides a basis for future research on the clinical utility of PDO screens. Furthermore, PDOs adequately reflect the effect of prior treatment, primary tumor sidedness and mutational status, supporting their use for modeling colorectal cancer in vitro.

5-FU 5-fluorouracil

ATP Adenosine triphosphate

AIC Akaike information criterion
AUC Area under the curve

CRC Colorectal cancer

CI Confidence interval

CT Computed tomography

DMSO Dimethylsulfoxide

DNA Desoxyribonucleïnezuur

DRC Drug response curve

EFG® Epidermal growth factor (receptor)

FOLFOX 5-FU & oxaliplatin

FOLFIRI 5-FU & irinotecan

GR Concentration that gives half-maximal growth rate inhibition

GR₅₀ Concentration of 50% growth inhibition

GR₅₀1/2 Concentration of 50% growth rate inhibition of the upper and lower part of
the biphasic curve

GR_AUCArea under the non-fitted ‘curve’ of the raw GR values

GR_maxConcentration that gives maximal growth rate inhibition

HR Hazard ratio

mCRC Metastatic colorectal cancer

NAC N-acetyl cysteine
NGS Next generation sequencing

pMMR Proficient mismatch repair

PDO Patient-derived organoid

PFS Progression-free survival

RECIST Response evaluation criteria in solid tumours
SNP Single-nucleotide polymorphism

Ethics approval and consent to participate

The collection of patient data and tissue for the generation of organoids has

been performed according to the guidelines of the European Network of Research Ethics

Committees (EUREC) following European, national, and local law. This research was performed under the HUB-Cancer biobanking protocol and OPTIC protocol. The Biobank Research Ethics Committee of the UMC Utrecht (TCBio) approved the biobanking protocol: 12-093 HUB-Cancer. The medical ethical committee of the University Medical Center Utrecht (UMCU) approved the prospective clinical trial Organoids to Predict Therapy Response In Colorectal Cancer (OPTIC #17-356(24)). All donors participating in this study signed informed consent. The study was performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable

Availability of data and materials

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request after ethical approval. The underlying code for this study is not publicly available but may be made available to qualified researchers on reasonable request from the corresponding author. The organoid models are available and can be requested via Foundation Hubrecht Organoid Biobank (https://www.hubrechtorganoidbiobank.org/)

Competing interests

Miriam Koopman reports institutional financial interests with Amgen, Bayer, Bristol-Myers Squibb, Merck-Serono, Nordic Pharma, Pierre Fabre, Servier, Sirtex, Roche, Sanofi, and Personal Genome Diagnostics. Miriam Koopman reports the following non-financial interests: an advisory role for ZON-MW, membership of the scientific board of the Dutch Cancer Society (KWF), chairmanship of the Dutch Colorectal Cancer Group (DCCG), and principal investigator (PI) of the Prospective Dutch Colorectal Cancer (PLCRC) cohort. Jeanine Roodhart reports institutional financial interests with BMS, Pierre Fabre, Servier, Cleara, Xilis, DoMore diagnostics and HUB organoids B.V. Jeanine Roodhart reports an advisory role for Bayer, BMS, Merck-Serono, Pierre Fabre, Servier, AMGEN and board member of Foundation Hubrecht Organoid Biobank. Liselot Valkenburg-van Iersel reports financial interests with and an advisory role for Servier and Pierre Fabre. All other authors declare that they have no competing interests.

Funding

This study was funded by HUB Organoids B.V. and Cancer Genomics Center (NWO Gravitation Program).

Authors' contributions

OK, JR, SB, SE, RV and MK initiated the project. AB, LV and GC contributed to the successful enrollment of a diverse and representative study population. LS collected the clinical data. MB revised all CT scan reports according to RECIST 1.1. EW, LS, EK and CV designed the experiments. KCP and CHB performed the experiments and contributed to the analyses. EK and MD analyzed the PDO drug screen data. RO and MH were major contributors to interpreting the drug screen data. LS and EW analyzed the PDO drug screen and clinical data and wrote the manuscript. JR was a major contributor in writing the manuscript. SE and BV advised in the statistical analysis and BV provided the script for the biphasic curve fitting. SB, CV, SE, OK and MK reviewed the manuscript. All authors read and approved the final manuscript.

Acknowledgements

We extend our gratitude to Utrecht Platform for Organoid Technology (UPORT) for their pivotal role in patient enrollment and tissue acquisition. We acknowledge the commitment of all participating hospitals in the recruitment and enrollment of eligible patients. We thank the patients for providing consent to the HUB-Cancer Biobank and OPTIC study.

Authors' information (optional)
Not applicable.

Yamazaki K, Nagase M, Tamagawa H, Ueda S, Tamura T, Murata K, et al. Randomized phase III study of bevacizumab plus FOLFIRI and bevacizumab plus mFOLFOX6 as first-line treatment for patients with metastatic colorectal cancer (WJOG4407G). Ann Oncol. 2016;27(8):1539–46.
Venook AP, Niedzwiecki D, Lenz HJ, Innocenti F, Fruth B, Meyerhardt JA, et al. Effect of First-Line Chemotherapy Combined With Cetuximab or Bevacizumab on Overall Survival in Patients With KRAS Wild-Type Advanced or Metastatic Colorectal Cancer: A Randomized Clinical Trial. JAMA. 2017;317(23):2392–401.
Kwakman JJM, Simkens LHJ, van Rooijen JM, van de Wouw AJ, Ten Tije AJ, Creemers GJM, et al. Randomized phase III trial of S-1 versus capecitabine in the first-line treatment of metastatic colorectal cancer: SALTO study by the Dutch Colorectal Cancer Group. Ann Oncol. 2017;28(6):1288–93.
Overman MJ, Lonardi S, Wong KYM, Lenz HJ, Gelsomino F, Aglietta M, et al. Durable Clinical Benefit With Nivolumab Plus Ipilimumab in DNA Mismatch Repair-Deficient/Microsatellite Instability-High Metastatic Colorectal Cancer. J Clin Oncol. 2018;36(8):773–9.
Sawyers CL. The cancer biomarker problem. Nature. 2008;452(7187):548–52.
Sakshaug BC, Folkesson E, Haukaas TH, Visnes T, Flobak A. Systematic review: predictive value of organoids in colorectal cancer. Sci Rep. 2023;13(1):18124.
de Souza N, Organoids. Nat Methods. 2018;15(1):23.
Li M, Izpisua Belmonte JC. Organoids - Preclinical Models of Human Disease. Reply. N Engl J Med. 2019;380(20):1982.
Wensink GE, Elias SG, Mullenders J, Koopman M, Boj SF, Kranenburg OW, Roodhart JML. Patient-derived organoids as a predictive biomarker for treatment response in cancer patients. npj Precision Oncology. 2021;5(1):30.
van de Wetering M, Francies HE, Francis JM, Bounova G, Iorio F, Pronk A, et al. Prospective derivation of a living organoid biobank of colorectal cancer patients. Cell. 2015;161(4):933–45.
Sato T, Stange DE, Ferrante M, Vries RGJ, van Es JH, van den Brink S, et al. Long-term Expansion of Epithelial Organoids From Human Colon, Adenoma, Adenocarcinoma, and Barrett's Epithelium. Gastroenterology. 2011;141(5):1762–72.
Ganesh K, Wu C, O’Rourke KP, Szeglin BC, Zheng Y, Sauvé C-EG, et al. A rectal cancer organoid platform to study individual responses to chemoradiation. Nat Med. 2019;25(10):1607–14.
Cho YW, Min DW, Kim HP, An Y, Kim S, Youk J, et al. Patient-derived organoids as a preclinical platform for precision medicine in colorectal cancer. Mol Oncol. 2022;16(12):2396–412.
Wang T, Pan W, Zheng H, Zheng H, Wang Z, Li JJ, et al. Accuracy of Using a Patient-Derived Tumor Organoid Culture Model to Predict the Response to Chemotherapy Regimens In Stage IV Colorectal Cancer: A Blinded Study. Dis Colon Rectum. 2021;64(7):833–50.
Vlachogiannis G, Hedayat S, Vatsiou A, Jamin Y, Fernández-Mateos J, Khan K, et al. Patient-derived organoids model treatment response of metastatic gastrointestinal cancers. Science. 2018;359(6378):920–6.
Yao L, Zao XL, Pan XF, Zhang HG, Wang FJ, Qiao PF. Application of tumoroids derived from advanced colorectal cancer patients to predict individual response to chemotherapy. J Chemother. 2022:1–13.
Yao Y, Xu X, Yang L, Zhu J, Wan J, Shen L, et al. Patient-Derived Organoids Predict Chemoradiation Responses of Locally Advanced Rectal Cancer. Cell Stem Cell. 2020;26(1):17–26e6.
Narasimhan V, Wright JA, Churchill M, Wang T, Rosati R, Lannagan TRM, et al. Medium-throughput Drug Screening of Patient-derived Organoids from Colorectal Peritoneal Metastases to Direct Personalized Therapy. Clin Cancer Res. 2020;26(14):3662–70.
Mo S, Tang P, Luo W, Zhang L, Li Y, Hu X, et al. Patient-Derived Organoids from Colorectal Cancer with Paired Liver Metastasis Reveal Tumor Heterogeneity and Predict Response to Chemotherapy. Adv Sci (Weinh). 2022;9(31):e2204097.
Tang Y, Wang T, Hu Y, Ji H, Yan B, Hu X, et al. Cutoff value of IC(50) for drug sensitivity in patient-derived tumor organoids in colorectal cancer. iScience. 2023;26(7):107116.
Ooft SN, Weeber F, Dijkstra KK, McLean CM, Kaing S, van Werkhoven E, et al. Patient-derived organoids can predict response to chemotherapy in metastatic colorectal cancer patients. Sci Transl Med. 2019;11:513.
Mohn C, Hacker HG, Hilger RA, Gutschow M, Jaehde U. Defining the role of MRP-mediated efflux and glutathione in detoxification of oxaliplatin. Pharmazie. 2013;68(7):622–7.
Laoukili J, Constantinides A, Wassenaar ECE, Elias SG, Raats DAE, van Schelven SJ, et al. Peritoneal metastases from colorectal cancer belong to Consensus Molecular Subtype 4 and are sensitised to oxaliplatin by inhibiting reducing capacity. Br J Cancer. 2022;126(12):1824–33.
Netherlands Trial Register. [Available from: https://www.trialregister.nl/trial/7305.
Sahin IH, Klostergaard J. BRAF Mutations as Actionable Targets: A Paradigm Shift in the Management of Colorectal Cancer and Novel Avenues. JCO Oncol Pract. 2021;17(12):723–30.
Hafner M, Niepel M, Chung M, Sorger PK. Growth rate inhibition metrics correct for confounders in measuring sensitivity to cancer drugs. Nat Methods. 2016;13(6):521–7.
Ritz C, Baty F, Streibig JC, Gerhard D. Dose-Response Analysis Using R. PLoS ONE. 2015;10(12):e0146021.
Song SL, Li B, Carvalho MR, Wang HJ, Mao DL, Wei JT, et al. Complex in vitro 3D models of digestive system tumors to advance precision medicine and drug testing: Progress, challenges, and trends. Pharmacol Ther. 2022;239:108276.
Sui Y, Wu Z. Alternative statistical parameter for high-throughput screening assay quality assessment. J Biomol Screen. 2007;12(2):229–34.
Chan GK, Kleinheinz TL, Peterson D, Moffat JG. A simple high-content cell cycle assay reveals frequent discrepancies between cell number and ATP and MTS proliferation assays. PLoS ONE. 2013;8(5):e63583.
Jones LJ, Gray M, Yue ST, Haugland RP, Singer VL. Sensitive determination of cell number using the CyQUANT cell proliferation assay. J Immunol Methods. 2001;254(1–2):85–98.
Driehuis E, Kretzschmar K, Clevers H. Establishment of patient-derived cancer organoids for drug-screening applications. Nat Protoc. 2020;15(10):3380–409.
Zoetemelk M, Ramzy GM, Rausch M, Nowak-Sliwinska P. Drug-Drug Interactions of Irinotecan, 5-Fluorouracil, Folinic Acid and Oxaliplatin and Its Activity in Colorectal Carcinoma Treatment. Molecules. 2020;25(11).
Tejpar S, Stintzing S, Ciardiello F, Tabernero J, Van Cutsem E, Beier F, et al. Prognostic and Predictive Relevance of Primary Tumor Location in Patients With RAS Wild-Type Metastatic Colorectal Cancer: Retrospective Analyses of the CRYSTAL and FIRE-3 Trials. JAMA Oncol. 2017;3(2):194–201.
Holch JW, Ricard I, Stintzing S, Modest DP, Heinemann V. The relevance of primary tumour location in patients with metastatic colorectal cancer: A meta-analysis of first-line clinical trials. Eur J Cancer. 2017;70:87–98.

Additionalfile1.docx

Download PDF

Journal Publication

published 27 Feb, 2024

Read the published version in Journal of Experimental & Clinical Cancer Research →

Editorial decision: Major revision
04 Jan, 2024
Reviewers agreed at journal
23 Dec, 2023
Reviewers invited by journal
22 Dec, 2023
Editor assigned by journal
21 Dec, 2023
First submitted to journal
20 Dec, 2023

You are reading this latest preprint version

Organoids as a biomarker for personalized treatment in metastatic colorectal cancer: drug screen optimization and correlation with patient response

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Methods

Study design

Clinical data collection

Organoid culture isolation and establishment

Organoid in vitro drug screens

Statistical methods

Results

Patient characteristics

Organoid drug screen quality

Drug screen optimization

Removal of N-acetyl cysteine from screening medium

Drug screen readout (CellTiter-Glo versus CyQUANT)

Viability, growth rate metrics and curve fitting

Correlation of organoid response with patient response

5-FU & oxaliplatin

Irinotecan-based treatment

5-FU

Discussion

Conclusions

Abbreviations

Declarations

Ethics approval and consent to participate

Consent for publication

Availability of data and materials

Competing interests

Funding

Authors' contributions

Acknowledgements

Authors' information (optional) Not applicable.

References

Supplementary Files

Status:

Journal Publication

Version 1

Authors' information (optional)
Not applicable.