MINI-CANCERS UTILIZATION FOR PERSONALIZED CANCER DRUG REGIMENS
20210333265 · 2021-10-28
Inventors
Cpc classification
G16H50/20
PHYSICS
G16B40/10
PHYSICS
G16H20/10
PHYSICS
A61B10/00
HUMAN NECESSITIES
G16H50/30
PHYSICS
International classification
Abstract
The present invention includes methods for identifying chemotherapy treatments specifically suited for each cancer patient using mini-cancers to develop personalized cancer drug regimes.
Claims
1. A method for identifying a chemotherapy treatment for a subject comprising: obtaining a biopsy from a subject; establishing a cancer culture from the biopsy; mixing the cancer culture with a matrix forming mini cancers; growing the mini cancers; dividing the mini cancers into one or more samples; adding one or more agents separately or combined to the one or more samples; analyzing the phenotypes of the mini cancers wherein the phenotypes are selected from the group of cell growth, invasion, or a combination thereof; and identifying a chemotherapy treatment for the subject by choosing the one or more agents determined to stop growth of the mini cancers, inhibit growth of the mini cancers, and/or inhibit invasion of the mini cancers when compared to a reference mini cancer without the addition of one or more agent.
2. The method of claim 1 wherein the biopsy is of liver cancer and the cancer culture is created by mixing the biopsy with an establishing growth media comprising: HGF, FSK, FSK-10, Gastrin 1, NAC B27, N2, Nicotinamide, Wnt3A, R-spondin1, EGF, Noggin, A83-01, FGF2, PGE2, and Y27632.
3. The method of claim 2 wherein the cancer culture is then grown on a second growth media comprising: HGF, FSK, FGF10, Gastrin I, NAC, B27 supplement, N2, Nicotinamide, Wnt3A-conditioned medium, R-spondin1, EGF, Noffin conditioned medium, A83-01, FGF2, and PGE2.
4. The method of claim 1, wherein the biopsy is of a desmoplastic cancer and cells selected from the group comprising fibroblasts, endothelial cells, immune cells, other tumor-specific cells, or combinations thereof, are added to the cancer culture.
5. The method of claim 1, wherein the matrix comprises a polymer selected from the group consisting of collagen, gelatin, chitosan, heparin, fibrinogen, hyaluronic acid, chondroitin sulfate, pullulan, xylan, dextran, polyethylene glycol, derivatives thereof, or a combination thereof.
6. The method of claim 1, wherein the biopsy is of liver cancer.
7. The method of claim 1, wherein the biopsy is of pancreatic cancer.
8. The method of claim 1, wherein the step of identifying a chemotherapy treatment is based on a percentage of mini cancers killed by the agent when compared to a reference mini cancer substantially free of the one or more agents.
9. The method of claim 1, wherein the step of identifying a chemotherapy treatment is based on a percentage of mini cancers having inhibited growth when compared to a reference mini cancer substantially free of the one or more agents.
10. The method of claim 1, wherein the step of identifying a chemotherapy treatment is based on a percentage of mini cancers that are inhibited from invasion when compared to a reference mini cancer substantially free of the one or more agents.
11. The method of claim 1, wherein the samples are in the range of 10 to 4,000 samples.
12. A cell culture comprising HGF, FSK, FSK-10, Gastrin 1, NAC B27, N2, Nicotinamide, Wnt3A, R-spondin1, EGF, Noggin, A83-01, FGF2, PGE2, and Y27632.
13. A cell culture comprising: HGF, FSK, FGF10, Gastrin I, NAC, B27 supplement, N2, Nicotinamide, Wnt3A-conditioned medium, R-spondin1, EGF, Noffin conditioned medium, A83-01, FGF2, and PGE2.
14. A method of treating a patient having cancer comprising: obtaining a biopsy from a subject with cancer; establishing a cancer culture from the biopsy; mixing the cell culture with a matrix forming mini cancers; growing the mini cancers; dividing the mini cancers into one or more samples; adding one or more agents separately or combined to the one or more samples; analyzing the phenotypes of the mini cancers wherein the phenotypes are selected from the group of cell growth, invasion, or a combination thereof; identifying an agent for the subject by choosing the one or more agents determined to stop growth of the mini cancers, inhibit growth of the mini cancers, or inhibit invasion of the mini cancers when compared to a reference mini cancer without the addition of one or more agents; and administering the one or more agents to the subject to treat the cancer.
15. The method of claim 14 wherein the process is repeated should the subject have a cancer recurrence after treatment.
16. The method of claim 15 wherein the subject is treated with a second one or more agent identified during the repeated process.
17. The method of claim 14 wherein the biopsy is of liver cancer and the cancer culture is created by mixing the biopsy with an establishing growth media comprising: HGF, FSK, FSK-10, Gastrin 1, NAC B27, N2, Nicotinamide, Wnt3A, R-spondin1, EGF, Noggin, A83-01, FGF2, PGE2, and Y27632.
18. The method of claim 17 wherein the cancer culture is then grown on a second growth media comprising: HGF, FSK, FGF10, Gastrin I, NAC, B27 supplement, N2, Nicotinamide, Wnt3A-conditioned medium, R-spondin1, EGF, Noffin conditioned medium, A83-01, FGF2, and PGE2.
19. The method of claim 15, wherein the cancer is a desmoplastic cancer and cells selected from the group comprising fibroblasts, endothelial cells, immune cells, other tumor-specific cells, or combinations thereof are added to the cancer culture.
20. The method of claim 15, wherein the matrigel comprises a polymer selected from the group consisting of collagen, gelatin, chitosan, heparin, fibrinogen, hyaluronic acid, chondroitin sulfate, pullulan, xylan, dextran, and polyethylene glycol as well as their derivatives or a combination thereof.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
[0043]
[0044]
[0045]
[0046]
[0047]
[0048]
[0049]
[0050]
[0051]
[0052]
[0053]
[0054]
[0055]
[0056]
[0057]
[0058]
[0059]
[0060]
[0061]
[0062]
[0063]
DETAILED DESCRIPTION OF THE INVENTION
[0064] Liver cancer has a global incidence of approximately 850,000 cases and represents the second leading cause of cancer-related mortality. Hepatocellular carcinoma (HCC) accounts for 80-85% of all cases of primary liver cancers. Cholangiocarcinoma (CCA), the second most prevalent primary liver cancer, has a median survival after diagnosis of less than 12 months. Next generation sequencing (NGS) approaches identified a significant number of somatic mutations in liver cancers. These strategies led to the postulate that the identification of actionable mutations will guide the selection of targeted drugs aimed at those specific mutations which in turn will result in patient- and cancer-specific therapeutics and better clinical outcomes. By inference, if each cancer had actionable mutations and if a targeted therapy existed for each, then cancer genomics could inform drug selection in all cases. However, only 3-6% of cancer patients who have their cancer exome sequenced are paired with a drug (http://ecog-acrin.org/nci-match-eay131/interim-analysis). Specific to liver cancers, multiple somatic alterations were identified in HCC, but the vast majority of molecular/targeted agents tested to date failed. Sorafenib, a multi-kinase inhibitor approved for HCC improves survival by a mere 3 months at the cost of severe side effects. Similar to clinical outcomes in HCC, in spite of NGS efforts to elucidate molecular pathways in CCA, the first line and mainstay of CCA chemotherapy continues to be Gemcitabine and Cisplatin (with a poor response rate).
[0065] The lack of effective therapeutics in liver cancers and the high failure rate of the vast majority of drugs to date led some to advance that HCC is not druggable. These efforts to identify actionable mutations or druggable pathways in liver cancers appeared to suffer an additional setback once intratumor heterogeneity was proven. First recognized in primary renal carcinomas, the phenomenon was later recognized in most solid tumors, including HCC and CCA. The implication of these studies is that intratumor genetic heterogeneity leads to a drug response/functional, intratumor heterogeneity that is clinically relevant. The importance of demonstrating this tenet cannot be overemphasized, since it may offer an explanation for at least some of the numerous therapeutic failures to date. In addition, if genetic heterogeneity is translated into drug response heterogeneity in liver cancers, then logical approaches can be devised to account for this phenomenon, perhaps through designing combinations of drugs that are effective on each clinically significant intra-tumor cancer cell population. To our knowledge, however, there is no data in liver cancers to demonstrate that genetically defined intra-tumor phylogenetic groups translate into identification of discreet drugs that are effective on each group.
[0066] Functional diagnostics describes a category of techniques to enable the experimental testing of a drug or set of drugs on live cancer cells obtained from a patient, with the goal of informing drug choice. While precision oncology is oftentimes equated to NGS of cancer exomes, the incomplete understanding of the genotype to phenotype relationship poses significant clinical translatability challenges. Carefully characterized functional diagnostics approaches bring the promise of addressing this limitation. The American Society of Clinical Oncology (ASCO), however, found that first generation functional diagnostics—Chemotherapy Sensitivity and Resistance Assays (CSRA)—have not been successful in surpassing published reports of clinical trials. These efforts, however, were restricted by a number of limitations, including lack of standardization in regards to the study design, type of assay, reproducibility and accuracy of such assays, number of drugs used (typically low number of drugs), or type of drugs used (typically standard chemotherapy).sup.13. Issues such as assay validation, reproducibility and more importantly clinical translatability were incompletely and heterogeneously described. However, next generation functional diagnostics—including patient derived organoids (PDO) could address these concerns.
[0067] Functional diagnostics describes a category of techniques to enable the experimental testing of a drug or set of drugs on liver cancer cells obtained from a patient, with the goal of informing the drug choice. While precision oncology is oftentimes equated to genomics, the incomplete understanding of the genotype to phenotype relationship poses significant clinical translatability challenges. Carefully characterized functional diagnostics approaches bring the promise of addressing this limitation. Of note, the idea of matching drugs to patients has been validated in other clinical fields, such as in infectious diseases where choice of best antibiotic is guided by functional testing and not by site of infection. In oncology, The American Society of Clinical Oncology (ASCO) found that first generation functional diagnostics—Chemotherapy Sensitivity and Resistance Assays—have not been successful in surpassing published reports of clinical trials. These efforts, however, were not standardized in regards to the study design, type of assay, reproducibility and accuracy of such assays, number of drugs used (typically low number of drugs), or type of drugs used (typically standard chemotherapy). Issues such as assay validation, reproducibility and more importantly clinical translatability were incompletely and heterogeneously described. However, next generation functional diagnostics—including patient derived organoids (PDO), have been recently developed and they could address these concerns. The inventors propose to implement state-of-the-art patient-derived cancer models with the purpose of validating our existing high throughput (HT) functional diagnostics pipeline in CCA. The deliverable is a comprehensively vetted algorithm for precision functional diagnostics that is accurate, reproducible, relatively inexpensive and clinically useful.
[0068] Cancer genomics postulates that the identification of actionable mutations will guide the selection of targeted drugs aimed at those specific mutations. By extension, this method is correlative, not based on direct, experimental verification of drug effects. By inferential reasoning, if each cancer had actionable mutations and if a targeted therapy existed for each, then cancer genomics could inform drug selection in all cases. However, only 3-6% of cancer patients who have their cancer exome sequenced are paired with a drug (http://ecog-acrin.org/nci-match-eay131/interim-analysis). The chief reasons are our incomplete knowledge of these “actionable mutations” as well as the paucity of targeted drugs for such mutations. Another concern is that outcomes for the few patients who are paired to a targeted drug are not dramatically better. Drug choice based on biomarkers improved progression-free survival by only 5.7 months. In conclusion, the low numbers of patients for whom a targeted drug is identified coupled with the relatively low response rate creates an imperative for direct, experimentally verifiable precision-oncology approaches. The relevant clinical question is how to utilize the information from genome-based diagnostics as a part of a larger approach, not as the sole guide. Experiments the inventors propose will correlate and integrate drug functional testing with genomic profiling, therefore elevating the significance of our application through its clinical impact.
[0069] Patient Derived Xenografts (PDX) represent the gold standard for recapitulating the biology of a patient's cancer, including stable genomic, transcriptomic and proteomic profiles across many passages. Of clinical interest, PDX also predict drug responses in patients from whom they were derived. According to some PDX experts, this model provide in excess of 80% positive predictive value for drug activity in patients. If there were no barriers to use, PDX would be the ideal assays to inform cancer drug selection. Unfortunately, PDX display a number of limitations to their widespread, systematic use in clinical oncology. For example, PDX are labor-intensive, expensive and slow to develop (up to 10 months to establish the first engraftment), with low and/or variable take rate. High costs limit the number of drugs that can realistically be tested. PDX also induce loss of clonal heterogenetity due to mouse selection pressure, replacement of the human stromal component with mouse cells, and the lack of an immune system. Given the high accuracy of PDX, the relevant question is how to develop surrogates to parallel PDX in an accurate, reproducible, rapid and cost conscious fashion. The experiments the inventors propose will benchmark patient-derived organoids (PDO) against PDX and will clarify precisely when and how to use PDO and at which step to rely on PDX.
[0070] First, the inventor's findings bring them closer to a functional drug testing trial in CCA patients, who desperately need effective therapies. Second, the inventors are develop principles widely applicable in designing similar trials in other cancers. Third, the inventor's are modeling the stromal contribution to drug responses and determining the optimal assay complexity, which is of high interest to other desmoplastic cancers, such as pancreatic, prostate and subsets of breast cancer. Fourth, the inventor's are delineating the merits of current precision modalities—genomic-based diagnostics and PDX. Finally, the inventor's are determining a clinically feasible strategy for initial drug screening (genome based diagnostics vs. PDO functional diagnostics) and validation (PDO models to include fibroblasts and endothelial cells vs. PDX, see
[0071] The inventors have established PDO cultures from 10/10 CCA patients seen at the Johns Hopkins Hospital (
[0072] The inventors are exploring three critically important features: (1) The ability of CCA PDO to maintain characteristics of original tumor. There is a pressing need to authenticate cancer PDO as key resources to be used in scientific and clinical applications; (2) The ability to assess drug efficacy through accurate, direct, assays as well as molecular profiling. The inventors will contrast several imaging modes to quantify cancer phenotypes (growth; invasion) at high accuracy and acceptable cost. In addition, the inventors will investigate the possibility that responses to specific drugs (such as mutation targeted therapies) can be predicted by genomic-based diagnostics without recourse to functional diagnostics; (3) The ability of various PDO constructs (simple and/or complex) to inform drug selection. The inventors are assessing if they can use simple PDO for initial drug screen, or do they have to use more resource-intensive cPDO (that include fibroblasts and endothelial cells. Furthermore, the inventors plan to clarify the best cPDO to parallel PDX in terms of drug selection positive predictive value.
[0073] In preliminary studies, the inventors have obtained resection cancer tissue from 10 CCA patients. The inventors successfully established all 10 in patient-derived organoids (PDO) by adapting culture conditions reported for normal liver (19). As shown in
[0074] There is mounting evidence that cell types outside of epithelial cancer cells are salient to cancer response to drugs (22). It is not feasible to recapitulate all cell types from a tumor in an ex-vivo model. Here, we will add 2 major types of stromal cells—fibroblasts and endothelial cells—and assess for changes in the PDO positive predictive value in regards to drug efficacy. Future studies can incorporate innate and adaptive immune cells to model response to immune-modulatory drugs. We have prototyped 2 methodologies to create complex PDO (cPDO). First, we directly mixed human CCA PDO (2×10{circumflex over ( )}4 cells) with liver stellate/fibroblast cells LX2 (ratio 1:1). LX2 are liver stellate cells immortalized spontaneously by low serum conditions (23). LX2 cells were transduced to express Red Fluorescence Protein (RFP). After 6 days of culturing of CCA PDO mixed with LX2 cells in Matrigel and fed with organoid media adapted by our lab, we noted that both cancer organoids and myofibroblasts appeared healthy as seen in
[0075] Simple PDOs (epithelial cells only) will be utilized to perpetuate patient-specific cancer cells. Ensuring that PDOs maintained in culture continue to reflect the biology of the original tumor is of paramount importance. These considerations will be evaluated (
[0076] CCA occurs in males and females. The inventors will enroll a population representing local demographics, including males and females of multiple ethnic and racial categories (see “Inclusion of Women and Minorities”). Based on demographics, the inventors anticipate equal male/female representation, 30-50% white European ancestry, and additional representation from Asian, Hispanic, and African-American ancestry. Sex, age, and ancestry (self-identified and estimated from DNA analysis using Eigenstrat principle components will be investigated as covariates for statistical tests. If necessary, we will analyze individual well-identified ethnic ancestries separately and merge results using standard meta-analysis approaches.
[0077] The ability to perpetuate cancer PDO virtually indefinitely creates the need to assess the number of passages up to which the PDO still reflects the drug response/biology of the original tumor. The inventors will perform this authentication, as well as seek DNA/RNA profiles to predict changes in drug responses. Such molecular markers can be used to predict a change in drug response profiles.
[0078] Human cancers accumulate mutations due to genomic instability and selection pressures. While the selection pressures in vitro are different vs. in vivo, they do exist and they can affect drug response profiles. Furthermore, the past utilization of poorly authenticated cancer cell lines has led to reports of unreproducible findings and billions of dollars wasted. The inventors propose to study the effects of passaging CCA PDO on drug response profiles (
[0079] There are 2 deliverables that the inventors will achieve. First, once the inventors assign a drug response profile to each passage point (4, 25, 50, 75 and 100) for each of the 10 CCA PDOs (Aim 1a), the inventors will correlate these profiles to DNA/RNA profiles at each of these passage points. The goal is to assess if molecular profiles are indicative, and therefore appropriate, as biomarkers of drug responses. Second, the inventors will compare the stability of DNA and RNA profiles at each of these passage points amongst themselves. The goal is to document the authentication of CCA PDOs as a key biologic resource. To reach these deliverables, for DNA analyses, we will utilize the Illumina HumanCytoSNP-12 v2.1 BeadChip kit for scanning the whole genome for genetic and structural variations (https://www.illumina.com/products/by-type/clinical-research-products/human-cytosnp-12.html). This assay is a panel of ˜300,000 Single Nucleotide Polymorphisms (SNPs) that will identify DNA CNVs (deletions, duplications, amplifications) and copy-neutral loss of heterozygosity. For RNA analyses, we will utilize the Agilent 8X60K SurePrint G3 Expression Microarray, which assays the expression of ˜28,000 unique Entrez genes and ˜30,000 lncRNAs (http://www.genomics.agilent.com/article.jsp?crumbAction=push&pageId=1516). DNA and RNA assays will be performed in the SKCCC Microarray Core.
[0080] The inventors will extract DNA and RNA from PDO following standard protocols, as the inventors and others have done. The recommended DNA quantity per sample (200 ng) will be provided to the core facility to run the 8-beadchip workflow. DNA processing will use the recommended workflow of the Illumina BlueFuse CytoChip module including QCs for overall call rate (>98%) and median Log R deviation (<0.2). CNV and LOH status will be extracted from the BlueFuse CGH/LOH reports. For RNA analysis with the Agilent platform, we will use the recommended 10 ng of RNA per sample for the 8X60K, one-color format. RNA data will be analyzed with limma, including methods for normalization, quality testing, and linear models.
[0081] The inventors will perform the following 5 types of analyses to integrate drug response/molecular profiles: (a) Trajectory of drug response. The inventors will perform tests of trend (Cochran-Armitage and linear model) for each PDO/drug combination against the null hypothesis that there is no change in IC50 due to number of passages. While mutations that accrue during passaging could conceivably affect IC50 in either direction, the inventors' hypothesis is that rates are low, making multiple changes in different directions less likely. Should the inventors observe many changes in drug response, the inventors can investigate more complex models. The inventors will perform a separate test for each PDO/drug combination. The effect size giving 80% power to detect is R.sup.2>0.61 for p<0.05 for a single PDO/drug and R.sup.2>0.83 for 0.05 FWER with 1290 PDO/drug combinations. (b) Trajectory of changes to DNA (CNVs) and RNA (gene expression). For each, the inventors will independently test the null hypothesis that the rate of change per passage is constant. For DNA, the inventors will calculate the Bayesian posterior probability for a change in the copy number of each SNP across the genome relative to (i) the baseline copy number; (ii) the copy number at the previous measured passage. The inventors will then sum these probabilities across the genome to calculate the number of CNVs. The inventors will also calculate the number of events using a hidden Markov model to group together CNVs that likely arose from a single amplification or deletion. Next, the inventors will identify time intervals where gene expression changes using a Bayesian change-point algorithm, then sum over the Bayesian posterior probabilities to obtain a global change-point statistic. The inventors will test for correlation between the rate of change of DNA CNV and RNA expression, both overall and locus-by-locus (i.e. CNV at a locus vs. expression levels of transcript from the same locus). The inventors anticipate that most PDO will show a constant, low rate of change of DNA CNVs and of RNA expression, with occasional large bursts corresponding to DNA instability and genomic aberrations. The inventors will determine the number of passages that is 95% likely to not have deviated from simple constant accumulation of changes at the DNA and RNA level for 95% of PDOs. (c) Integration of the drug response data and the DNA/RNA profiling. To test the hypothesis that changes in drug response correspond to changes in genome or gene expression. The inventors will calculate the number of passages 95% likely to avoid changes in drug response for 95% of PDOs. (d) Behavior of known drug targets: many drugs have known targets. For these, the inventors will perform opportunistic analysis to determine whether observed changes in drug response corresponding to genomic or gene expression changes at the target locus. Based on a 0.05 FWER for 129 possible tests, the effect size corresponding to 80% power is R.sup.2>0.71 for DNA-based features and responder/non-responder fold-ratio >4.3 or <0.23 for RNA expression assuming typical unit variance on a log 2-scale. (e) Perform the same type of test opportunistically for all loci vs. drugs that display a change in response for at least one PDO. These may identify additional targets or genes in compensating pathways with potential value as targets for combination therapies.
[0082] The expected outcomes include: (1) It is possible there will be a change in drug response and DNA/RNA profiles prior to passage 100. These analyses, averaged over 10 specimens, will inform the determination of an approximate/average cutoff for passaging cancer PDOs in cultures. (2) Although unlikely, it is possible that there will be a significant change in DNA/RNA profiles without a significant change in drug response profiles. In this case, the informative cutoffs will be reported in terms of both DNA/RNA and drug response changes. (3) It is possible that we will be able to ascribe a rate limiting step in regards to DNA changes (for example a new driver mutation occurring) that predicts a change in drug response behaviors. Similarly, it is possible to identify an RNA signature that predicts drug response changes (in particular for non-targeted therapies, such as classic chemotherapy agents and/or newer HDAC inhibitors and others). It is also possible that we will identify certain RNA profiles that might predict responses to a certain drug class but not to others (vinca alkaloids vs. targeted therapies).
[0083] In one case scenario, the inventors find that there is no drug response/molecular change up to 100 passages. These data will inform future protocols designed to keep the maximum numbers of passages under these numbers. Similarly, a lower passage number identified as a cutoff will similarly serve as guidance. In another scenario, the inventors identify DNA/RNA alterations predictive of drug response changes and these data may serve as dynamic markers of specimen reliability. There is also the possibility that genomic/phenotypic stability may be a function of the initial genomic profile. Last, it is possible that the inventors will not be able to maintain all 10 PDO to passage 100. In this case, the inventors will utilize additional PDOs (from the 30 different CCA PDOs available in the inventor's lab) or establish new PDOs from CCA tumors available through our clinical practice.
[0084] Standardized assays are mandatory for clinical-grade tests. The few studies to test drugs on PDO have generally used viability assays, such as the CellTiter-Glo (Promega). The inventors have used CellTiter Glo (
[0085] The inventors will implement image analytics for PDO growth as well as to quantify invasion since these cancer phenotypes are the most important clinical determinants of CCA patient survival. Chemotherapy can induce EMT responses and increase the frequency of dissemination and this concern further supports evaluating invasion. Assays that only monitor ATP turnover would fail to distinguish between cytostatic drugs that induce a stationary arrest and those that increase invasion. The inventors have developed imaging pipelines that enable the quantification of invasive and disseminative behavior of cancer cells using non-perturbing white light imaging (differential interference contrast). The inventors have extensive experience in the analysis of invasion and dissemination of organoids derived from murine and human tumors.
[0086] For growth assessment, the inventors will measure the area of the organoid in an image. This represents the projected area of organoid volume. In preliminary experiments, tumor organoid area increases linearly with time in Matrigel culture, for at least 96 hours. The inventors will culture and image for 48 hours with vehicle, treat with drug, and then culture and image for an additional 48-96 hours to determine response. Cytostatic drugs will block an increase in organoid area. Cytotoxic drugs will also block an increase in organoid area but may not reduce area. For invasion assessment, the inventors will use spectral analysis of organoid boundaries to convert organoid images into quantitative phenotypes for analysis. In brief, the inventors decompose the boundary into Fourier components, with lower frequency components relating to large-scale deformations from a spherical shape, medium frequency components relating to branching structures, and high frequency components relating to invasive structures. Organoid images will capture 2D cross-sections, which will be traced using an existing interface in ImageJ. Images will be processed through an automated pipeline to extract spectral features. First, the (x,y) coordinates defining the boundary will be interpolated to a parametric curve x(n),y(n), where the index n ranges from 1 to N, with N=1024 to 4096 providing stable results. The individual dimensions will then be Fourier transformed to yield spectral components X(k),Y(k), where the discrete index k ranges from −N/2 to N/2. The spectral power at wave vector k, defined as P(k)=|X(k)|.sup.2+|Y(k)|.sup.2, provides a rotationally invariant representation of the organoid boundary over length scales approximately 1/k relative to the circumference. The component at k=0 represents the overall center of the boundary and is discarded. The k=1 component represents the image size, determined by the number of cells, magnification, and the location of the imaging plane relative to the organoid center; it is used for an overall normalization. The remaining components serve as a signature of the invasiveness of organoid boundary. In addition to rotational invariance, this representation is useful because it is low dimensional, with most power limited to a small number of components, <10 for even the most invasive organoids. The inventors will use the total spectral power summed over modes k=2 and above as a quantitative phenotype for invasiveness. In addition to organoid shape, we will separately track and quantify the number of disseminating clusters.
[0087] Each of 30 CCA PDO included here will be screened at 10 uM with the NCI Drugs Set VII (129 FDA approved drugs). The inventors will assess the readout for drug efficacy based on the CellTiter-Glo assay, and the inventors will also perform image acquisition and analysis for growth and invasion. Note in particular that CellTiter Glo and PDO image area both report on cell growth, whereas boundary spectral power reports on invasiveness. These phenotypes may have distinct molecular biomarkers with distinct therapeutic interventions. Drugs that are active at 10 uM (>50% growth inhibition based on CellTiter Glo or PDO volume as estimated from PDO area; >50% reduction in boundary spectral power) will be retested at 1 uM, 100 nM, 10 nM, 1 nM to determine IC50.
[0088] The expected outcome is that growth of untreated and treated PDO will be highly correlated when measured by CellTiter Glo and by PDO. Given that CellTiter Glo is an accepted standard for growth assays, strong correlation of PDO area with CellTiter Glo (R.sup.2>0.80) will indicate that PDO area is a suitable proxy. For drug response, the inventors will perform two comparisons. First, the inventors will define drugs as active or inactive based on the threshold of 50% growth inhibition (CellTiter Glo) or volume reduction (estimated from PDO area), then determine the PPV of PDO vs. CellTiter Glo as known positives. The inventors will consider 80% PPV as a threshold for use of PPO as a proxy. For compounds active based on CellTiter Glo and PDO area, the inventors will calculate the R.sup.2 for IC50 values on a log-scale, with R.sup.2>0.8 indicating excellent concordance and R.sup.2>0.5 indicating a useful substitute assay.
[0089] The inventors will compare the list of compounds/IC50s identified through boundary spectral power with those identified through growth-based assays. Compounds that are highly cytotoxic may also reduce boundary spectral power because the PDO will collapse. The inventors anticipate, however, that many compounds that are active in reducing boundary spectral power will have little to no effect on growth. Such compounds may have high utility as anti-metastatic candidates with reduced side-effects from general toxicity.
[0090] Cancer genomics has been utilized for (i) the identification of actionable mutations and subsequent (ii) selection of drugs to target that mutation. Here the inventors assess whether DNA/RNA profiles from PDO can provide rigorous, reproducible, and clinically valuable protocols for drug selection. The inventors will directly quantify the effects of 129 drugs on 30 CCA PDO. In parallel, the inventors will seek DNA/RNA profiles to predict the experimentally validated drug efficacy readout. Of note, when compared to DNA/RNA analyses, the ones included here will aid in building predictors. The inventors therefore use more informative data generation. For DNA, the inventors will use exome sequencing of cancer (PDO) vs. normal adjacent tissue, as the inventors have done before for CCA with Agilent SureSelect exome capture, Illumina HiSeq sequencing, and Eland/CASAVA processing. Use of PDO rather than original cancer tissue removes stromal cells, eliminating the need for microdissection. DNA features will be restricted to somatic mutations with likely functional effects: (i) known driver mutations including inferred gain-of-function (GOF) from recurrent missense (oncogenes) and inferred loss-of-function (LOF) from nonsense, frameshift, and deletion mutations; (ii) newly observed candidate drivers defined as recurrent inferred GOF/LOF (>2 samples); (iii) CNVs involving any drivers or candidate drivers. Somatic mutations will be summarized at the gene level as presence/absence of GOF or LOF, and quantitative CNV. The inventors will investigate whether merging GOF and LOF categories will lead to greater power, particularly for LOF inferred from an early stop that may delete a regulatory domain. For RNA, the inventors will use published protocols for organoid-based RNA-seq with Illumina HiSeq followed by a standard BowTie/TopHat/HTSeq/DESeq pipeline.
[0091] DNA/RNA features will be tested for association with growth (CellTiter Glo, PDO area) and invasion (PDO boundary spectral power). The inventors will first determine whether sex and other covariates are associated with observed phenotypes (p<0.05); if so, they will be regressed out. DNA tests of GOF/LOF categories will use ANOVA; tests of CNV will use linear model relating phenotypes to CNV status. At genome-wide stringency of 0.05 family-wise error rate and with at most 50 DNA features anticipated based on previous CCA exome sequencing (2), the inventors anticipate that a p-value of 0.001 is appropriate, with 80% power to detect associations with R.sup.2>0.39 for 0.05 FWER with 129 drugs. RNA-based features will be analyzed by converting to log-scale and testing linear models for growth and invasion. Assuming unit standard deviation on a log 2-scale and significance threshold of p<1×10.sup.−6 typical for our previous gene expression studies, the inventors will have 80% power to detect transcripts with fold-ratio approximately >2.3 or <0.43 between most responsive and least responsive quartiles. Nominal parametric p-values will be assessed in two ways. First, the inventors will ensure that robust non-parametric equivalents of parametric tests are also significant at a less stringent threshold of 5% FDR. Second, the inventors will perform permutation tests that hold DNA/RNA features constant for each sample while shuffling phenotypes and covariates.
[0092] Similarly, the inventors will seek to identify DNA & RNA features that correlate with drug response. We will perform these tests in two ways: (1) significance in an ANOVA model (DNA GOF/LOF categories) or logistic regression model (DNA/CNV and log2-scale RNA) for activity at 10 uM; (2) significance in an ANOVA (GOF/LOF) or linear model (DNA/CNV and log2-scale RNA) for IC50, with inactive compounds assigned a nominal IC50 of 100 uM. Critical R.sup.2 values for 80% power will depend on the number of samples showing activity. As with pre-treatment phenotypes, we will use non-parametric tests and permutation tests to confirm that findings are robust. The inventors expect the following outcomes: (1) To find that images are superior to chemical-based methodologies. If so, in future studies, the inventors will refine these analyses for complex PDO. If they are not uniformly superior, the inventors will identify applications for which they are best suited. (2) To find molecular profiles (DNA and more so RNA) that are predictive of drug sensitivity of PDO. These signatures will likely be applicable for drug screening, but unlikely to be superior to complex PDO for validation purposes. It is also possible that such profiles can predict the effects of certain but not other drugs. In doing these experiments, the inventors can assign a preliminary role to these approaches as part of a larger strategy to select drugs in a personalized fashion.
[0093] The inventors believe some potential problems & possible solutions are (1) The number of 30 samples is limited by clinical intake and by analysis costs. The inventors will include non-parametric confirmation of robust effects and permutations to confirm p-values. Splitting our samples into discovery-vs-validation cohorts is not appropriate for this sample size, particularly for DNA features where only a small number will have any single somatic mutation. Nevertheless, rigorous thresholds of 0.05 FWER should yield findings that will reproduce in subsequent validation studies. (2) Should power be too low at 0.05 FWER, the inventors will consider relaxing our threshold to 5% FDR also commonly used in such studies. (3) The inventors will use gene-set enrichment analysis, implemented by our own software based on MSigDB categories, to identify possible themes in the candidate DNA and RNA features. The inventors will also use the DrugBank drug-target database to provide support for treatment-response biomarker candidates. All cancer models demonstrate advantages and disadvantages (
[0094] The inventors will utilize the NCI Approved Oncology Drugs Set VII (129 FDA approved drugs) to: 1) contrast PDO vs. cPDO in terms of drug selection sensitivity; 2) benchmark cPDO vs. PDX in terms of drug selection specificity; and 3) retrospectively compare patient response in clinic to PDO/cPDO/PDX predicted responses. The initial screen (relative low cost and good sensitivity) qualifies drug leads for validation. Ideally, due to low cost and ease of manipulation, epithelial-only PDOs will be sufficiently sensitive for screening. The inventors will compare PDOs vs. complex PDO to verify that PDOs are as sensitive as cPDOs for drug selection.
[0095] Based on our preliminary experiments, 6 days of co-culture are sufficient to produce matrix contraction and change the expression profiles of CCA cells. In preliminary experiments, the inventors mixed organoids with endothelial media (procured from Lonza) in 1:1 ratio and have shown good viability of all cells up to 10 days after plating. The best model of cPDO in CCA remains to be established (fibroblast source, ratio fibroblasts-cancer cell need to be determined). Fibroblast source may influence response to drugs. LX2 cells are partially activated liver stellate cells immortalized in culture and are the inexpensive, easily obtainable choice. In addition, the inventors will derive myofibroblasts from same human cancer tissue specimens as the inventors utilize to derive cancer organoids. In brief, as the inventors have done in preliminary experiments, 4/5 of the resection piece will be used for isolating cancer cells, allowing 1/5 to be processed separately (collagenase, filtration) and then grown in DMEM/10%FBS/Antibiotics with TGB-β1 supplementation (5 ng/mL), as described previously, to obtain liver myofibroblasts.
[0096] The ratio of fibroblast: cancer cells is another important determinant of the level of desmoplasia and consequently response to therapy. We will vary the ratio of fibroblasts to cancer cells from 1:1 (as in our preliminary experiments) to 8:1. In parallel experiments, it was demonstrated a more aggressive cancer behavior in co-cultures where the ratio was 8:1 when compared to 1:1. Absent any data to suggest similarly high impact of percent endothelial cells, as in preliminary experiments, the inventors will keep the ratio cancer cells: endothelial cells 1:1
[0097] For each CCA, the inventors will test 1 PDO (simple organoid culture) and 4 cPDOs in regards to their response to each of the 129 drugs in the NCI Approved Oncology Drugs Set VII. As in preliminary experiments, the inventors will use a cutoff of 50% metabolically active cells remaining at 48 hours after applying the drugs as reported by an ATP assay (CellTiter Glo) and create a list of drug leads for each of the 5 PDO models. For 10 CCAs, the inventors will contrast the sensitivity of PDO vs. cPDO in choosing drug leads at this screening stage.
[0098] PDX recapitulate drug response profiles seen in patients from whom they were derived. Therefore, the inventors will treat PDX as the best available surrogate of drug response in patients. Recent data suggest that simple PDOs established from breast cancer PDX displayed a positive predictive value of 82.5%. There is no data, however, in CCA or other desmoplastic cancers to report positive predictive value of simple PDO or cPDO. The inventors will test drug leads identified to calculate positive predictive value of both PDO and matched cPDO when compared to drug responses in PDX. First, the inventors will validate the screening results in cPDO-based validation experiments that involve calculation of IC50. The top 5 most effective drugs will be tested at decreasing concentrations (10 uM, 1 uM, 100 nM, 10 nM and 1 nM) in each of the 4 cPDOs (
Establishment of Multiple PDO Lines from Geographically Distinct Regions of Human Liver Cancers
[0104] Patient Derived Xenografts (PDX) models can be used to recapitulate cancer features and behaviors as well as for anti-cancer drug screening, either to characterize activity of new lead compounds, or in a highly personalized fashion to identify best chemotherapy for a specific patient. As described in, as well as in the inventor's experience, there is a wide variability of human tumor engraftment in immunocompromised mice. Although data regarding engraftment of CCA samples in immunocompromised mice is sparse, the inventor's experience (data unpublished) mirrors the poor rate (less than 6%) published recently in one study. Engraftment of HCC is also problematic. Recently, PDOs were successfully established from primary liver cancers and their histology, genetic and transcriptomic profiles were shown to mirror their corresponding primary human tumors. In a similar fashion, the inventors established PDO cultures from a human intrahepatic CCA. A surgical resection specimen from a CCA as well as matched histologically matched liver were obtained from a patient with an intrahepatic CCA. In line with the inventor's focus on biologic replicates obtained from same cancer tissue, the inventors cut the resection piece into 20 tissue slices as shown in
[0105] To confirm that the cystic PDO structure identified in white light images was indeed derived from a CCA, we embedded patient CCA tissue, as well as matched PDOs extracted from these CCA tissues, into optimum cutting temperature (OCT) compound, cut slides, and stained with representative antibodies.
[0106] A human primary HCC specimens was processed similarly to the CCA specimen. In brief, 7 geographically distinct pieces from the same surgical specimen were utilized to establish 7 corresponding sister HCC PDO lines. The inventors confirmed that PDO cultures are similar to primary cancer.
Drug Testing and Effect Quantification on a PDO Line
[0107] Initial drug testing experiments were performed on one PDO line, with a 9-drug library, at a standard HT screening drug concentration of 10 uM. For these experiments, we utilized a CCA PDO line. For positive control, we used 10% Triton X, while for negative control we used DMSO a25 same concentration used to dissolve drugs. For these experiments, we utilized white light images and equated the killing effect of drugs into shrinking of PDO cystic structures. As shown in
The Identification of “Silver Bullets” by High Throughput Drug Screening in a Large Cohort of Liver Cancer PDO Lines
[0108] To investigate if liver cancer sister PDO lines can be utilized for high-throughput drug screening, we collected two additional human primary CCA surgical specimens and an additional human primary HCC surgical specimen. From each cancer specimen, we established between 3 and 7 PDO lines (Table 1).
TABLE-US-00001 TABLE 1 Summary of human primary liver cancer specimens, demographics and laboratory derivation of sister PDO lines. Lab PDO ID Disease Age Gender Race Predisposing Lines CCA8 iCCA 63 female hispanic None 6 CCA23 iCCA 60 female white None 5 CCA28 iCCA 57 female white None 6 HCC25 HCC 71 male white HCV 3 HCC26 HCC 69 female white HCV 7 Total 27
[0109] Next, each PDO line from each cancer was subjected to treatment with the NCI VII 129-drug panel and the killing effect was measured with CellTiter-Glo (measuring percent viability 4 days after treatment was applied). The experiment produced 27 readings (from 0% to 100%, one for each liver cancer PDO line) for each of the 129 drugs, for a total of 3,483 data cells. These data is shown in a color coded fashion in
Inter Patient Functional Heterogeneity—The Identification of Inter-Patient “Divergents”
[0110] The drug testing results indicates that a number of drugs might be efficacious in a subgroup of patients. For example, dasatinib, previously reported as efficacious in a liver cancer organoid drug testing study, appears to work well in HCC26, but not at all in HCC25 or in CCA8. To pick out inter patient divergent drugs (i.e., drugs with discordant effect across the 5 patients), we calculated the mean cell viability per patient cancer specimen by averaging the effect of the drugs across all PDO lines established for the same patient. This mean cell viability per patient is a measure of how the tumor, as a whole, responds to a certain drug. Next, we calculated the standard deviation of the 5 means (one per patient) and ordered the data in decreasing order of the standard deviation. We excluded drugs that did not induce a significant cell killing in at least one of the 5 cancers (arbitrarily defined as 30% viability or less in at least one specimen). The top interpatient divergent drugs are shown in
Intra Tumor Functional Heterogeneity—The Identification of Intra-Tumor “Divergents”
[0111] The results allows the visual identification of drugs that have a divergent effect among PDO lines established from a single patient tumor. This intra tumor functional heterogeneity was investigated by calculating the standard deviation of a drug response across all PDO lines established from a patient tumor.
Neither Exome nor Transcriptome Sequencing Analyses can Accurately Predict Response to Therapy
[0112] A principal component analysis that evaluated inter-PDO line similarity as a function of each line response to treatment demonstrated that, based on their functional heterogeneity (i.e., response to a panel of 129 cancer drugs), the 6 sister PDO lines appear to cluster in 3 distinct groups (
[0113] The American Society of Clinical Oncology (ASCO) affirms the need to develop standardized assays for direct drug testing in patient-derived cancer samples. The promise of precision oncology has been, to date, centered mainly on genomics. The paradigm is attractive: identify an actionable mutation, then choose the drug that targets that mutation. One of the caveats, however, is the fact that, employing existing drugs and protocols, only 2-6% of cancer patients are successfully paired with a drug. Therefore, the need to develop direct, accurate, replicable, quick and inexpensive functional drug testing on patient-derived cancer samples is urgent.
[0114] Drug testing on patient-derived specimens requires cancer models that recapitulate the biology and predict the response of the cancer of origin. To date, several systems had been employed, including cell lines (in vitro and xenotransplanted in immunocompromised animals), genetic animal models, patient derived xenografts (PDX) and recently patient derived organoids (PDO). Arguably, PDX models represent the gold standard in recapitulating cancer biology. Importantly, PDX was found to have in excess of 80% drug response positive predictive value in patients. There are, unfortunately, practical limitations to using PDX routinely to inform drug choice in oncology practice. PDX models are expensive. They are also slow (it may take up to 10 months to 1 year to establish some of these models). In this study, the inventors benchmark and standardize PDOs with the purpose of achieving high accuracy (similar or better than PDX) in a reproducible, inexpensive and quick fashion. Data we accrued to date shows that we are able to (i) successfully establish PDO from all CCA samples collected to date; (ii) develop and adapt protocols to keep these patient-specific cancer samples in culture virtually indefinitely; (iii) develop and adapt protocols to test up to 1,500 drugs in each patient derived culture (high-throughput); (iv) incorporate stromal cells to better recapitulate the biology and drug response of CCA; (v) perform initial drug screening as well as validation testing and inhibitory concentration 50 (IC50) calculations; (vi) provide with a drug sensitivity report in 6 weeks or less since specimen collection from patients.
[0115] The inventors identified patient-specific patterns of intra-tumor heterogeneity, implying that lessons about variable drug response for one individual's cancer cannot be applied to another individual's clinical scenario without additional patient-specific experimentation.
[0116] To the best of the inventors' knowledge, few prior studies have directly and simultaneously evaluated intra-tumor heterogeneity at the level of somatic mutation and that of variable gene expression. The present invention is the first to demonstrate that, on a global scale, a genetic clonal relationship across tumor tissue samples does not necessarily imply increased similarity in their gene expression profiles. Multiple possible mechanisms that affect differential gene expression in cancer, independent of DNA mutation effects, have been proposed and investigated, including DNA methylation [PMID 28423542], epigenetic histone modification [PMID 16369569], and micro RNA expression [PMID 26681654, PMID 48922647].
[0117] As the inventors evaluated the relationship between somatic, exomic mutations and gene expression, they also sought to analogously interrogate the level of similarity between sequencing-based heterogeneity and variable drug response of PDO's. The inventors were unable to relate, on a global level, either exomic genetic variation or gene expression to overall drug response profile. This was not especially surprising, as studies to date have not identified widespread patterns in the relationship between such pharmacogenomics biomarkers and clinically applicable drug efficacy [PMID 25338550]. Numerous factors on the tumor, individual, and population levels contribute to complex biology underlying a cancer's susceptibility to various anti-cancer agents. Our efforts to identify specific associations between particular mutations or genes and drug response profiles were limited by the small number of PDO's used in this study. However, in the clinical setting, for technical and cost reasons, only a limited number of PDO's can be generated to inform an individual patient's treatment; therefore, the constraint imposed by limited sample size is unlikely to be overcome. Therefore, performing assays of indirect biomarkers, such as DNA or RNA profiling, would seem to have much lower yield in clinical application, compared to direct interrogation of a tumor's heterogeneous response profile to a panel of potential drugs.
[0118] The patterns that emerged when we assessed intra-tumor and inter-tumor heterogeneity of drug response profile hold implications for how the results of PDO drug screening can be applied to individualized treatment in the clinical setting. The relatively similar drug response profiles for each cancer, when averaged across PDO's to eliminate intra-tumor heterogeneity, imply that individualized therapy is unlikely to result from assessing for inter-individual variability alone, as would be conducted using single biopsy specimens. While the presence of high cancer cell killing for some drugs across all cancer samples would seem to indicate that those drugs stand to be used universally to treat cholangiocarcinoma, this in fact more likely reflects in vitro conditions that favor the cytotoxicity of those specific medications.
[0119] In contrast to inter-tumor relative uniformity, the presence of some intra-tumor heterogeneous drug responses indicates that the use of PDO's in an in vitro assay provides biological, and also potentially clinical, value for at least a subset of drugs in the screening panel. Specifically, resistance to a drug that is localized to only a portion of a tumor implies a resistance factor that has emerged during the development of that tumor, and that needs to be overcome by treatment with another drug. The underlying reason for the difference between intra- and inter-tumor drug profile comparisons may lie in the accumulated genetic and non-genetic factors that confer drug resistance to cancer cells and lead to intra-tumor variability, in contrast to the root, or shared, somatic mutations that cause cancer to arise. Therefore, targeting a drug (or likely a combination of drugs) to the localized susceptibility patterns of a particular tumor may lead to improved efficacy of cancer cell killing.
[0120] Current personalized clinical oncology approaches include biopsying of the tumor followed by exome analyses to identify “actionable” mutation(s). The NCI-Match trial showed, however, that only XX % of cancer patients who underwent this approach to date as part of the trial were successfully matched with a targetable drug (CITATION). This fact suggests that the frequency of actionable mutations may not be sufficiently high to allow this exome-sequencing that informs drug choice approach to significantly impact clinical cancer patient care. Therefore, it appears that exome sequencing as a patient individualized data interrogation methodology to inform drug treatment may not be clinically impactful. In addition, the NCI-Match trial has shown that even those patients for whom exome sequencing did identify an actionable mutation did not survive significantly longer (CITATION). Additional discreet reports exist of targeted therapy that initially demonstrates good response, only to fail later due to either selection or emergence of a resistant cancer clone (CITATION NEJM melanoma in young person case). The reasons for which patients who do harbor actionable mutation(s) and who are successfully paired with a targeted therapy fail to have a significantly better clinical outcome are not clear. Data presented here argues that one potential explanation lies in the intra-tumor functional heterogeneity, which may impact clinical outcomes more than the presence of an actionable mutation in one of the clonal branches of the tumor. In addition, data presented here further argues that neither genetic heterogeneity, nor expression heterogeneity are able to accurately predict functional, drug-response intra-tumor heterogeneity.
[0121] Patient derived organoid (PDO) models of human cancers are utilized increasingly for research purposes. In addition, PDOs may appear as good candidates to augment or replace exome NGS (next generation sequencing) methodologies to inform best drug choice in a patient-specific fashion. Our data suggests that intra-tumor genetic heterogeneity does not mirror the expression heterogeneity and neither of them predict the drug response, functional, heterogeneity. However, since the drug response heterogeneity is the clinically important variable, these data cast a shadow of doubt onto clinical utility of genetic profiling of tumors for informing clinical drug choice. In addition, these data bring center stage the concept of multiple cancer biopsying protocols to ensure that the functional intra-tumor heterogeneity is well represented a PDO-driven methodology to inform patient-specific drug choice.
EXAMPLES/METHODS
[0122] The following Examples have been included to provide guidance to one of ordinary skill in the art for practicing representative embodiments of the presently disclosed subject matter. In light of the present disclosure and the general level of skill in the art, those of skill can appreciate that the following Examples are intended to be exemplary only and that numerous changes, modifications, and alterations can be employed without departing from the scope of the presently disclosed subject matter. The following Examples are offered by way of illustration and not by way of limitation.
Example 1
Media Compositions for Growing Liver Cancer Organoids/Mini-Cancers
[0123] A) Basic Media For Making Mini-Cancers: [0124] Advanced DMEM/F12 (Gibco—Ref: 12634-010). Add 1M HEPES buffer (Quality Biological) AND GlutaMAX (Gibco Ref: 35050). For 500 mL DMEM/F12 you add 5 mL HEPES and 5 mL GlutaMAX. Add Puromocin 1 mL Basic media can be used for up to 1 month
[0125] B) Growth Media for Establishing Mini-Cancers and for Passaging Mini-Cancers:
TABLE-US-00002 Cat Basic Media 7 mL As above HGF 10 uL (F Conc: 25 ng/uL) Peprotech (cat: 100-39H) FSK 10 uL (10 uM) TOCRIS(cat: 1099) FGF10 10 uL (100 ng/mL) Peprotech(cat: 100-26) Gastrin I 2 uL (10 nM) Sigma (10047-33-3) NAC 40 uL (1.25 mM) Sigma (A7250) B27 supplement 400 uL Gibco Ref: 12587-010 (50X) N2 800 uL Gibco Ref: 17502-048 (100X) Nicotinamide 400 uL (10 mM) Sigma (N3376) Wnt3A-conditioned medium 6 mL (30%) Made in the lab - 293T cells (purchased from ATCC) R-spondin1-conditioned 4 mL (20%) Made in the lab - 293T medium cells (purchased from Trevigen) - HA-R- Spondin1-Fc293T) EGF 2 uL (50 ng ml.sup.−1) Sigma(62253-63-8) Noggin 100 ng/mL (2 uL) Peprotech A83-01 20 μl (500 nM) TOCRIS(cat: 2939) FGF2 20 nG/mL Peprotech(cat: 100-18B) PGE2 10 nM TOCRIS (cat: 2296) Y27632 10 uM Sigma (129830-38-2) Total 20 mL
[0126] C) Growth Media for Growing Mini-Cancers:
TABLE-US-00003 Cat Basic Media 7 mL As above HGF 10 uL (F Conc: 25 ng/uL) Peprotech (cat: 100-39H) FSK 10 uL (10 uM) TOCRIS(cat: 1099) FGF10 10 uL (100 ng/mL) Peprotech(cat: 100-26) Gastrin I 2 uL (10 nM) Sigma (10047-33-3) NAC 40 uL (1.25 mM) Sigma (A7250) B27 supplement 400 uL Gibco Ref: 12587-010 (50X) N2 800 uL Gibco Ref: 17502-048 (100X) Nicotinamide 400 uL (10 mM) Sigma (N3376) Wnt3A-conditioned medium 6 mL (30%) Made in the lab - 293T cells (purchased from ATCC) R-spondin1-conditioned 4 mL (20%) Made in the lab - 293T medium cells (purchased from Trevigen) - HA-R- Spondin1-Fc293T) EGF 2 uL (50 ng ml.sup.−1) Sigma(62253-63-8) Noggin conditioned medium 100 ng/mL (2 uL) Peprotech A83-01 20 μl (500 nM) TOCRIS(cat: 2939) FGF2 20 nG/mL Peprotech(cat: 100-18B) PGE2 10 nM TOCRIS (cat: 2296) Total 20 mL
Example 2
Steps to Plating Mini-Cancers for High Throughput Drug Testing
[0127] Take out Matrigel from −20 C and keep at 4 C overnight. Also keep pipette tips at 4 C overnight. Count cells. Mix cells with Matrigel so that you have 200 cells per uL of Matrigel. Take a Falcon—Ref: 35321996 96-well plate and place 2.5 uL of Matrigel+cells per well. Plate 2 rows, then stop. Wait till the Matrigel is solid. Then add 100 uL of growth passage media. Then continue till all plate is filled up with cells. Place plate into incubator. Take picture with white light. After 4 days replace the media with media containing drugs (90 uL media+10 uL drug work solution). Use a multi-pipette.
[0128] All references, including publications, patent applications, and patents, cited herein are hereby incorporated by reference to the same extent as if each reference were individually and specifically indicated to be incorporated by reference and were set forth in its entirety herein.
Human Tissue
[0129] Fresh human CCA and matched histologically normal liver tissue were obtained from the Johns Hopkins Hospital (JHH), under an Institutional Review Board (IRB) approved informed consent. A diagnosis of CCA was performed by JHH pathologists, in accord with current standards in the field.
Cell Lines
[0130] L Wnt-3A cells (Cat: CRL-2647™, ATCC, Manassas, Va.), as well as HA-R-Spondin1-Fc293T (Trevigen, Gaithersburg, Md.) cells were grown in DMEM medium supplemented with 10% fetal bovine serum (FBS) and antibiotics, and maintained in 37° C. humidified incubators, supplied with 5% CO2. Mycoplasma tests were performed weekly and only mycoplasma-free cells were utilized for experiments.
Conditioned Medium
[0131] Conditioned media from L Wnt-3A and HA-R-Spondin1-Fc293T was obtained as previously described.
PDO Medium
[0132] Media reported previously for normal liver organoids was adapted for use in the current study.
[0133] Establishment of PDOs from Human CCA and Matched Histologically Normal Liver
[0134] Resection liver tissue (CCA, HCC or normal liver) obtained from patients was minced in small pieces measuring approximately 9 mm.sup.3. Tissue was then rinsed 3 times with DMEM supplemented with 1% FBS at 4° C. in 50 mL Falcon tubes. The tissue was then dissociated with collagenase (2 mG/mL, Sigma) supplemented with DNAse 1 (0.1 mG/mL in DMEM, Sigma) at 37° C. Digestion was stopped by adding cold DMEM supplemented with 1% FBS. Cells were filtered, washed by centrifugation and then counted. After mixing with growth factor reduced Matrigel (Corning, Cat#: 356231, Tewksbury, Mass.), cells were seeded into 24-well plates. After the matrigel solidified within 15 min, warm organoid culture medium was added.
PDOs Xenograft Establishment and Character
[0135] Animal experiments were performed under Johns Hopkins IACUC approved protocol, 500,000 PDO cells were mixed at 1:1 with Matrige and were subcutaneous injected into the flank of NSG mouse.
Immunohistochemistry and Immunofluorescence
[0136] Fresh tissues as well as PDOs in culture were embedded into optimal cutting temperature compound (OCT) at −80° C. Tissue and PDO sections (8 uM) were prepared and stained. We utilized the following primary antibodies: LGRS (stem cell marker Cat#: AP2745f, Abgent, San Diego, Calif.), SOX9 (stem cell marker—Cat#: 82630, Cell signaling, Danvers, Mass.), CK7 (biliary epithelium marker—Cat#: ab9021, Abcam, Cambridge, Mass.) and CK19 (biliary epithelium markers—Cat#: MAB3238 Millipore, Danvers, Mass.), EPCAM (epithelial marker—Cat#: 2929, Cell Signaling, Cell signaling, Danvers, Mass.), Ki67 (cell proliferation—Cat#: 550609, BD San Jose, Calif.)). Alexa Fluor dye-conjugated secondary antibodies (Life Technologies, Carlsbad, Calif.). Microphotographs were obtained with a Zeiss laser scanning microscope (LSM 510). Hematoxylin and Eosin (H&E) staining (Sigma Aldrich, St. Louis, Mo.), mucicarmine staining for mucin (Cat#: ab150667, Abcam, Cambridge, Mass.) were performed following standard protocols. Pictures were obtained with a Zeiss Axio Observer Inverted Microscope.
Whole-Exome Sequencing (WES) and Analysis
[0137] Total DNA was extracted from PDOs with Qiagen DNA kits. We used 1.5 uG DNA for WES with Agilent V5 capture probe set and sequenced by Illumina Hiseq2000 platform (BGI Hong Kong, China). The sequences of each sample were generated as 90/100 bp paired-end reads, and or each samples, around 10 Gb of unique sequences was generated. Standard quality control procedures were applied. Tumor-normal pair read alignment and variant identification, using SAMtools and GATK, were performed using standard parameters. A somatic variant was considered “present” in a tissue sample if it was present in at least 10% of reads. Using somatic coding (missense, nonsense, and canonical splice site) variations only, we performed phylogenetic tree estimation, using the maximum parsimony method with default settings in MEGA7.
High-Throughput RNA Sequencing (RNA-Seq) and Analysis
[0138] Total RNA was isolated from PDOs with passage 4 in culture using Trizol following the manufacturer's instructions. 100 nG of total RNA were utilized for RNA-Seq using HIseq4000 SE50 (BGI Hong Kong, China) with 50 or 75 bp paired-end reads. An average of 43 million reads were generated for each sample. The gene expression level was quantified using the RSEM software package by BGI (Hong Kong, China).
High-Throughput Drug Screening on CCA PDOs
[0139] A panel of 129 anti-cancer compounds (the NCI Approved Oncology Drugs Set VII) was utilized. PDOs were plated in 96-well or 384-well plates and cultured for 3 days. Next, media was changed with drug-containing media. The initial screen was performed at a concentration of 10 uM. After 4 days, viability was determined in each well with Cell TiterGlo (Cat#: G7572, Promega, Wis.) following the manufacturer's instruction. Fluorescent intensity was measured with a plate reader (Perkin-Elmer EnVision Plate Reader). Drugs that induced less than 50% viability were chosen for future studies. Inhibitory concentration 50 (IC50) experiments were performed at the following concentrations: 10 uM,1 uM, 100 nM, 10 nM and 1 nM.
[0140] The use of the terms “a” and “an” and “the” and similar referents in the context of describing the invention (especially in the context of the following claims) are to be construed to cover both the singular and the plural, unless otherwise indicated herein or clearly contradicted by context. The terms “comprising,” “having,” “including,” and “containing” are to be construed as open-ended terms (i.e., meaning “including, but not limited to,”) unless otherwise noted. Recitation of ranges of values herein are merely intended to serve as a shorthand method of referring individually to each separate value falling within the range, unless otherwise indicated herein, and each separate value is incorporated into the specification as if it were individually recited herein. All methods described herein can be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. The use of any and all examples, or exemplary language (e.g., “such as”) provided herein, is intended merely to better illuminate the invention and does not pose a limitation on the scope of the invention unless otherwise claimed. No language in the specification should be construed as indicating any non-claimed element as essential to the practice of the invention.
[0141] Preferred embodiments of this invention are described herein, including the best mode known to the inventors for carrying out the invention. Variations of those preferred embodiments may become apparent to those of ordinary skill in the art upon reading the foregoing description. The inventors expect skilled artisans to employ such variations as appropriate, and the inventors intend for the invention to be practiced otherwise than as specifically described herein. Accordingly, this invention includes all modifications and equivalents of the subject matter recited in the claims appended hereto as permitted by applicable law. Moreover, any combination of the above-described elements in all possible variations thereof is encompassed by the invention unless otherwise indicated herein or otherwise clearly contradicted by context.