*Article* **A Patient-Derived Organoid-Based Radiosensitivity Model for the Prediction of Radiation Responses in Patients with Rectal Cancer**

**Misun Park <sup>1</sup> , Junhye Kwon <sup>1</sup> , Joonseog Kong <sup>2</sup> , Sun Mi Moon <sup>3</sup> , Sangsik Cho <sup>3</sup> , Ki Young Yang <sup>4</sup> , Won Il Jang <sup>5</sup> , Mi Sook Kim <sup>5</sup> , Younjoo Kim 1,4,\* ,† and Ui Sup Shin 1,3,\* ,†**


**Simple Summary:** Predicting the tumor regression grade of locally advanced rectal cancer after neoadjuvant chemoradiation is important for customized treatment strategies; however, there are no reliable prediction tools. A novel preclinical model based on patient-derived tumor organoids has shown promising features of the recapitulation of real tumors and their treatment response. We conducted a small co-clinical trial to determine the correlation between the irradiation response of individual patient-derived rectal cancer organoids and the results of actual radiotherapy. Among the quantitative experimental data, the survival fraction was best matched and correlated with the patients' real treatment outcome. In the machine learning-based prediction model for radiotherapy results using the survival fraction data, the prediction accuracy was excellent at more than 89%. Enhanced machine learning with the accumulation of further new experimental data would help in creating a more reliable prediction model, and this new preclinical model can lead to more advanced precision medicine.

**Abstract:** Patient-derived tumor organoids closely resemble original patient tumors. We conducted this co-clinical trial with treatment-naive rectal cancer patients and matched patient-derived tumor organoids to determine whether a correlation exists between experimental results obtained after irradiation in patients and organoids. Between November 2017 and March 2020, we prospectively enrolled 33 patients who were diagnosed with mid-to-lower rectal adenocarcinoma based on endoscopic biopsy findings. We constructed a prediction model through a machine learning algorithm using clinical and experimental radioresponse data. Our data confirmed that patient-derived tumor organoids closely recapitulated original tumors, both pathophysiologically and genetically. Radiation responses in patients were positively correlated with those in patient-derived tumor organoids. Our machine learning-based prediction model showed excellent performance. In the prediction model for good responders trained using the random forest algorithm, the area under the curve, accuracy, and kappa value were 0.918, 81.5%, and 0.51, respectively. In the prediction model for poor responders, the area under the curve, accuracy, and kappa value were 0.971, 92.1%, and 0.75, respectively. Our patient-derived tumor organoid-based radiosensitivity model could lead to more advanced precision medicine for treating patients with rectal cancer.

**Citation:** Park, M.; Kwon, J.; Kong, J.; Moon, S.M.; Cho, S.; Yang, K.Y.; Jang, W.I.; Kim, M.S.; Kim, Y.; Shin, U.S. A Patient-Derived Organoid-Based Radiosensitivity Model for the Prediction of Radiation Responses in Patients with Rectal Cancer. *Cancers* **2021**, *13*, 3760. https://doi.org/ 10.3390/cancers13153760

Academic Editor: Samuel C. Mok

Received: 21 June 2021 Accepted: 19 July 2021 Published: 27 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Keywords:** machine learning; patient-derived tumor organoid; precision medicine; radiation response; rectal cancer

#### **1. Introduction**

Since the German trial of 2004, neoadjuvant chemoradiation therapy (NCRT), followed by radical surgery with total mesorectal excision, has been a standard treatment for locally advanced rectal cancer without metastasis [1,2]. With NCRT, the rate of local recurrence is significantly reduced, and the survival rate of cancer patients is significantly increased among good radiation responders [3–5]. Tumor response is evaluated based on pathologic findings of tumor regression, or the amount of TNM downstaging in postoperative surgical specimens compared with the clinical TNM staging [6]. The downstaging rate is 60–80%, of which 15–20% show a pathological complete response. However, approximately 20–40% of patients do not benefit from NCRT.

Currently, even if a complete response is clinically observed after NCRT, radical resection is recommended, which can be accompanied by serious surgical morbidity or impaired quality of life. However, it has been suggested that radical surgery is unnecessary if NCRT eradicates all tumor cells. Beets et al. suggested the 'wait and see' approach for rectal cancer patients [7]. According to these authors, if rectal cancer patients have a clinical complete response, as determined based on strict preoperative endoscopic criteria, after NCRT, undertaking nonoperative management or delayed surgery does not compromise long-term oncologic results [8]. In contrast, to improve the radiation response rate, many studies have been conducted by adding more intensive drug therapies during the periradiation period. The single-agent 5-fluorouracil (5-FU) or its derivatives have been used as a radiosensitizer. However, more intensive chemotherapeutic drugs (oxaliplatin or irinotecan) or biologics (cetuximab, bevacizumab, or panitumumab) have been added to enhance the radiation response [9–16]. However, administering these intensive treatments to all patients with rectal cancer is not cost-effective and is associated with increased toxicity. Moreover, the issue of overtreatment cannot be avoided.

In terms of precision medicine, rectal cancer is an ideal candidate, as treatment strategies can be tailored according to the expected radioresponsiveness. If a pathological complete response is expected, patients could avoid radical surgery, or if the expected radioresponsiveness is poor, more intensive preoperative chemotherapy could be administered. Therefore, the development of reliable prediction tools for radioresponsiveness is important.

As a preclinical model for precision medicine, patient-derived tumor organoids (PDTOs) have shown advantages over patient-derived tumor xenograft models, but have many limitations in clinical usage owing to their high cost and time taken to establish individual patient-derived models [17,18]. For pancreatic cancer and metastatic gastrointestinal cancer, the PDTO models showed a high correlation with clinical outcomes in terms of drug response [19,20]. Regarding radiation response, Ganesh et al. [21] and Yao et al. [22] recently generated PDTOs from patients with rectal cancer, and reported that PDTOs mirrored individual radiotherapy outcomes. Their results suggest that PDTOs can be used to predict individual responses to chemoradiation. However, prior studies have not identified the method that can best determine the correlation between PDTO response and patient outcome.

In this co-clinical trial, we attempted to reproduce previous study results to determine whether there is a correlation between experimental results obtained after irradiation in PDTOs and actual individual NCRT results of patients. In addition, we constructed a simple machine learning model that predicts patients' actual NCRT results based on the experimental data.

#### **2. Materials and Methods**

#### *2.1. Patient Enrolment and Treatment*

Between November 2017 and March 2020, we prospectively enrolled 33 patients diagnosed with mid-to-lower rectal adenocarcinoma pathologically confirmed by endoscopic biopsy. All patients underwent a staging workup using pelvic MRI; chest, abdominal, and pelvic computed tomography (CT); and 18-fluoro-2-deoxy-glucose positron emission tomography/CT. For patients diagnosed with locally advanced rectal cancer, NCRT was performed over a long course with a dose of 50.4 Gy in 28 fractions administered during weekdays. Chemotherapy was administered with a single-agent infusional 5-FU (425 mg per body square meter) for 5 days every 4 weeks before surgery. Radical surgeries were performed 6–8 weeks after completing radiotherapy with the aim of total mesorectal excision. Adjuvant chemotherapy was recommended for all medically fit patients after radical resection. For one patient who was diagnosed with a small resectable liver metastasis during staging workup, short-course radiotherapy with 25 Gy was administered in 5 Gy fractions over 5 days, followed by three cycles of neoadjuvant therapy: FOLFOX (5-FU, leucovorin, and oxaliplatin) with bevacizumab (the first cycle of FOLFOX only) every 2 weeks. Radical surgery, including liver metastasectomy, was performed 8 weeks after completing radiotherapy.

#### *2.2. Pathologic Examination of Surgical Specimens*

Standard pathologic tumor staging of the surgical specimen was performed and recorded according to the 8th edition of the TNM classification of the American Joint Committee on Cancer by dedicated gastrointestinal pathologists [23]. Pathologic response after NCRT was evaluated using the tumor regression grade (TRG) system suggested by the Gastrointestinal Pathology Study Group of the Korean Society of Pathologists [24]. The definitions of the TRG system are as follows: (A) TRG 0, complete response (no residual tumor cells were identified); (B) TRG 1, near complete response (only a few scattered tumor cells were present); (C) TRG 2, partial response (residual tumor glands with predominant fibrosis were easily identified); and (D) TRG 3, poor or no response (tumor cells did not demonstrate any response to chemoradiotherapy).

#### *2.3. Tissue Acquisition*

Pre-NCRT rectal cancer samples were obtained from enrolled patients at the endoscopic evaluation stage. Four or five rectal cancer biopsy samples were collected. A pathologist verified that the collected samples were histologically adenocarcinoma or normal crypts using hematoxylin and eosin (H&E) staining. The biopsy samples were pooled and immediately placed in cold phosphate-buffered saline with 50 µg/mL gentamicin (Gibco, Grand Island, NY, USA).

#### *2.4. Organoid Cultures*

Tumor organoids were isolated and cultured as previously described [25]. The composition of the PDTO culture medium is described in Supplementary Table S1. To prevent anoikis, 10 µM of Y-27632 was added to the culture medium for the first 2–3 days. When organoids were >200 µm, they were passaged by pipetting using Gentle Cell Dissociation Reagent (STEMCELL Technologies, Vancouver, BC, Canada) according to the manufacturer's instructions. Most of PDTO used in experiments were cultured more than 14 days.

#### *2.5. Immunocytochemistry and Immunohistochemistry*

For immunocytochemistry, PDTOs were fixed in 4% paraformaldehyde at 25 ◦C for 24 h, embedded in paraffin, and then dissected into 3-µm-thick sections. After treatment with Smartblock solution (CANDOR Bioscience GmbH, Wangen im Allgäu, Germany) for 30 min at 25 ◦C, the slides were incubated with primary antibodies at 4 ◦C overnight and then incubated with secondary antibodies for 1 h at 25 ◦C. Images were acquired using the EVOS FL Cell Imaging System (Thermo Fisher Scientific, Carlsbad, CA, USA).

Immunohistochemistry was performed to characterize organoids and their tissues of origin with the colorectal markers caudal type homeobox 2 transcription factor, cytokeratin 7, and cytokeratin 20 on 3-µm-thick formalin-fixed paraffin-embedded tissue and organoid sections. Sections were incubated for 1 h at 37 ◦C with primary antibodies. Detection was performed using an Envision/Horseradish Peroxidase system (Dako; Agilent Technologies, Inc., Santa Clara, CA, USA) and counterstained with hematoxylin for 10 min at 25 ◦C. Finally, the sections were dehydrated through a graded series of alcohol, cleared in xylene, and mounted. Images were acquired using an IX73 inverted microscope (Olympus Corporation, Tokyo, Japan). The antibodies and dilutions used are described in Supplementary Table S2.

#### *2.6. Survival Fraction Analysis*

For survival fraction analysis, organoids were resuspended in TrypLE Express (Thermo Fisher Scientific, Carlsbad, CA, USA) via pipetting with a p200 pipette and incubated at 37 ◦C for 10 min. Cells were centrifuged at 600× *g* for 5 min, and the supernatant was discarded. The pellet was resuspended in Matrigel and distributed into a 48-well plate (500–1000 cells/20 µL of Matrigel per well). After the Matrigel had polymerized, 100 µL of culture medium was added. Over the following days, organoids were treated with 0 Gy, 2 Gy, 4 Gy, and 6 Gy using a <sup>137</sup>Cs γ-ray source (Atomic Energy of Canada, Ltd., Renfrew County, ON, Canada) at a dose rate of 3.81 Gy/min. After 14 days, viable organoids were counted using Cell3 iMager Neo cc-3000 (Screen Holdings Co., Ltd., Kyoto, Japan). Analysis recipes were as follows: organoid diameter min 93, max 2907; organoid area min 6833, max 6,640,106; and circularity min 0.24, max 1. The plating efficiency was defined as the number of formed organoids/seeded cells × 100%. Survival fraction was calculated as follows: number of formed organoids/number of seeded cells in plate × (plating efficiency/100)]. A single-hit multitarget model was used to fit the survival curves, and D0, called the 'mean lethal dose', was the dose required to reduce the fraction of surviving organoids to 37% [26], calculated using GraphPad Prism software (version 8.0; GraphPad Software, Inc., La Jolla, CA, USA). For each PTDO, experimental replication of 4 wells was used. We obtained a total of 76 sets of survival fraction data.

#### *2.7. Viability Assay*

For the viability assay, organoids were resuspended in TrypLE. Cells were resuspended in Matrigel and distributed into a 96-well plate (5000 cells/10 µL of Matrigel per well). After the Matrigel had polymerized, 100 µL of culture medium was added. Over the following days, organoids were treated with 0 Gy, 2 Gy, 4 Gy, and 6 Gy. After 7 days, organoid viability was evaluated using CellTiter 96 AQUEOUS One Solution contains a tetrazolium compound [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4 sulfophenyl)-2H-tetrazolium, inner salt; MTS] (Promega, Madison, WI, USA) according to the manufacturer's instructions. Optical density was measured using a BioTek Eon microplate absorbance reader (BioTek Instruments Inc., Winooski, VT, USA) at 490 nm. Matrigel without organoids (10 µL) was used as a control.

#### *2.8. Second Passage*

For analysis at the second passage, organoids were treated with 5 Gy. After 72 h, organoids were passaged by pipetting using Gentle Cell Dissociation Reagent with a 1:2–1:4 split ratio. After 72 h, viable organoids were counted using the EVOS FL Cell Imaging System (Thermo Fisher Scientific, Carlsbad, CA, USA).

#### *2.9. EdU Staining*

Organoids were incubated with 10 µM EdU for 2 h and evaluated using Click-iT Plus EdU Imaging Kits (Thermo Fisher Scientific, Carlsbad, CA, USA) according to the manufacturer's instructions. Images were acquired using the EVOS FL Cell Imaging System (Thermo Fisher Scientific, Carlsbad, CA, USA).

#### *2.10. Western Blot Analysis*

For Western blot analysis, organoids were washed with cold phosphate-buffered saline and lysed in radioimmunoprecipitation assay buffer (Thermo Fisher Scientific, Carlsbad, CA, USA). Proteins were quantified using the Bradford method, and 20–40 µg of protein was resolved using SDS-PAGE. The membranes were incubated with primary antibodies overnight at 4 ◦C, followed by incubation with a secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 1 h at 25 ◦C. Proteins were visualized using enhanced chemiluminescence (Thermo Fisher Scientific, Grand Island, NE, USA). Western blot images were analyzed using the Bio-Rad ChemiDoc (Bio-Rad, Richmond, CA, USA).

#### *2.11. Targeted Next-Generation Sequencing Analysis*

To analyze the mutational status of tissues and organoids, they were harvested using a cell recovery solution (Corning, Inc., Corning, NY, USA). DNA extraction and library construction were performed using the Gentra Puregene kit (Qiagen, Hilden, Germany) and SureSelect XT library prep kit (Agilent Technologies, Santa Clara, CA, USA). Deep targeted sequencing using Axen Cancer Panel 2 (170 cancer-related genes; Macrogen, Seoul, Korea) and the NextSeq 500 mid-output system platform (Illumina, San Diego, CA, USA) was conducted on 19 PDTOs. Libraries comprising 150-bp end reads were sequenced via high-throughput sequencing using synthesis technology to a depth coverage of approximately 2000×.

#### *2.12. Statistical Analysis*

Data obtained from a minimum of three independent experiments are expressed as mean ± standard deviation. Unpaired two-tailed Student's *t*-tests were used to determine significant differences between the two groups. One-way analysis of variance followed by Tukey's and Bonferroni tests was performed to compare the means between multiple groups, and *p* values < 0.05 were considered significant. Statistical and receiver operating characteristic (ROC) curve analyses were performed using R 4.0.2 (https://www.r-project. org; accessed on 15 May 2020). Analysis of the mutation-annotated files was conducted using the R package 'maftools' (version 3.12), which included the generation of figure oncoplots [27]. Comparison of linear-quadratic (LQ) cell survival curves was performed using analysis of variance calculated with the R package 'CFAssay' (version 1.22.0) [28].

#### *2.13. Development of Predictive Models Using Machine Learning*

To build the prediction model for TRG, we used survival fraction data. A total of 76 experimental data points were randomly split in a 1:1 ratio into training and testing datasets. The machine learning model was built using the training data and subsequently tested on the remaining 50% of the data comprising the testing set. The supervised machine learning classification algorithm performed binary logistic regression and random forest classification with the R package, 'randomForest' version 4.6-14. For model training with a random forest method, we used 200 trees and two variables as training hyperparameters. We calculated the area under the ROC curve (AUC), accuracy, and kappa value of the testing dataset to evaluate the model performance.

#### **3. Results**

#### *3.1. Patient Characteristics and Treatment Outcomes*

Tumor tissues were collected by endoscopic biopsy from 33 patients with rectal cancer. Among 33 tumor tissues, 10 PDTOs could not be established due to bacterial contamination in one case and no expansion in the culture medium in nine cases (70% success rate). In addition, two patients were excluded as they were diagnosed with unresectable metastatic rectal cancer; radical surgeries were not planned for these patients, and it would not have been possible to evaluate their TRG. Two patients refused radical surgeries and were also excluded. Finally, 19 patients and their PDTOs were analyzed in this study (Figure 1A). Representative images of the 19 PDTOs are displayed in Supplementary

Figure S1. Individual patient characteristics and clinical treatment results are summarized in Table 1. The median age of patients was 59 (interquartile range, 53.0–70.5) years. The male-to-female ratio was 14:5. Eighteen patients had stage III disease, and one patient had stage IV disease with resectable liver metastasis. After R0 surgery following NCRT, TRGs were as follows: five patients achieved TRG 0 (26.3%), and one patient had TRG 1. Three patients had TRG 3, and the other 11 patients had TRG 2 (Figure 1B). During a median of 19.0 (interquartile range, 12.5–26.5) months of follow-up, six patients developed tumor recurrence (five distant, one local), and one patient died due to recurrence (Table 1). tary Figure S1. Individual patient characteristics and clinical treatment results are summarized in Table 1. The median age of patients was 59 (interquartile range, 53.0–70.5) years. The male-to-female ratio was 14:5. Eighteen patients had stage III disease, and one patient had stage IV disease with resectable liver metastasis. After R0 surgery following NCRT, TRGs were as follows: five patients achieved TRG 0 (26.3%), and one patient had TRG 1. Three patients had TRG 3, and the other 11 patients had TRG 2 (Figure 1B). During a median of 19.0 (interquartile range, 12.5–26.5) months of follow-up, six patients developed tumor recurrence (five distant, one local), and one patient died due to recurrence (Table 1).

Tumor tissues were collected by endoscopic biopsy from 33 patients with rectal cancer. Among 33 tumor tissues, 10 PDTOs could not be established due to bacterial contamination in one case and no expansion in the culture medium in nine cases (70% success rate). In addition, two patients were excluded as they were diagnosed with unresectable metastatic rectal cancer; radical surgeries were not planned for these patients, and it would not have been possible to evaluate their TRG. Two patients refused radical surgeries and were also excluded. Finally, 19 patients and their PDTOs were analyzed in this study (Figure 1A). Representative images of the 19 PDTOs are displayed in Supplemen-

*Cancers* **2021**, *13*, x FOR PEER REVIEW 6 of 16

*3.1. Patient Characteristics and Treatment Outcomes* 

**3. Results** 

**Figure 1.** Patient characteristics and treatment outcomes. (**A**) Flow chart indicating the number of patients with rectal cancer, including reasons for non-evaluability, and the success rate of establishing cultures from patients. (**B**) Pre- and post-RT endoscopic clinical responses, magnetic resonance images. and H&E staining images are shown for TRG 0, TRG 2, and TRG 3 patients. Magnification, ×4. Scale bars, 200 µm Abbreviations: H&E, hematoxylin and eosin; PDTO, patientderived tumor organoid; RT, radiotherapy; and TRG, tumor regression grade. **Figure 1.** Patient characteristics and treatment outcomes. (**A**) Flow chart indicating the number of patients with rectal cancer, including reasons for non-evaluability, and the success rate of establishing cultures from patients. (**B**) Pre- and post-RT endoscopic clinical responses, magnetic resonance images. and H&E staining images are shown for TRG 0, TRG 2, and TRG 3 patients. Magnification, ×4. Scale bars, 200 µm Abbreviations: H&E, hematoxylin and eosin; PDTO, patient-derived tumor organoid; RT, radiotherapy; and TRG, tumor regression grade.

*Cancers* **2021**, *13*, 3760


103

BMI, body mass index; RT, radiotherapy; MRI, magnetic resonance imaging; TRG0, (complete response) TRG1, (Nearly complete) TRG2, (Moderate) TRG3, (Minimal).

#### *3.2. Histological and Genomic Characterization of PDTOs*

*Cancers* **2021**, *13*, x FOR PEER REVIEW 8 of 16

To verify PDTOs, immunostaining was performed using paraffin-embedded organoid sections and tissues. Our PDTOs differentiated into enterocytes (villin 1), goblet cells (mucin 2), and enterochromaffin cells (chromogranin A) and contained amplifying cells (Ki-67; Figure 2A). To analyze the mutational status of the 19 PDTOs, we performed targeted next-generation sequencing analysis using Axen Cancer Panel 2. Variants were filtered based on a multivariate alteration detection of <2%, type of alteration (multi-hit, missense, nonsense, splicing site, in-frame del, and frame-shift), fusion gene, copy number alterations, and functional consequence (pathogenic, likely pathogenic, benign, and likely benign). Genes of the WNT signaling pathway (*APC* and *FBXW7*) were mutated in 68.4% (13/19) of all PDTOs. *APC* and *FBXW7* mutations were identified in 13 of 19 PDTOs (68.4%) and 6 of 19 PDTOs (31.5%), respectively (Figure 2B). All mutation alterations are displayed in Supplementary Figure S2. We performed H&E staining and immunostaining of the proteins cytokeratin 7, cytokeratin 20, and caudal type homeobox 2 transcription factor to confirm that our PDTOs originated from rectal cancer tissue and not from normal rectal mucosa. Our PDTOs showed similar histological morphologies and CK protein expression patterns to those of original tumor tissues (Figure 2C). Overall, these data demonstrated that PDTOs recapitulated the histological morphologies and marker expression of the paired patient tissues, as previously reported [21,29]. To define the capacity of colorectal cancer organoids to mirror the genome heterogeneity of the corresponding patient tumor, we compared the mutational status of three genes (*KRAS*, *NRAS,* and *BRAF)* in 19 PDTOs and corresponding tumor tissues. *KRAS*, *NRAS,* and *BRAF* mutations in PDTOs were matched to 86.6%, 100%, and 100% of those in corresponding tumor tissues, respectively (Figure 2D). *3.2. Histological and Genomic Characterization of PDTOs*  To verify PDTOs, immunostaining was performed using paraffin-embedded organoid sections and tissues. Our PDTOs differentiated into enterocytes (villin 1), goblet cells (mucin 2), and enterochromaffin cells (chromogranin A) and contained amplifying cells (Ki-67; Figure 2A). To analyze the mutational status of the 19 PDTOs, we performed targeted next-generation sequencing analysis using Axen Cancer Panel 2. Variants were filtered based on a multivariate alteration detection of <2%, type of alteration (multi-hit, missense, nonsense, splicing site, in-frame del, and frame-shift), fusion gene, copy number alterations, and functional consequence (pathogenic, likely pathogenic, benign, and likely benign). Genes of the WNT signaling pathway (*APC* and *FBXW7*) were mutated in 68.4% (13/19) of all PDTOs. *APC* and *FBXW7* mutations were identified in 13 of 19 PDTOs (68.4%) and 6 of 19 PDTOs (31.5%), respectively (Figure 2B). All mutation alterations are displayed in Supplementary Figure S2. We performed H&E staining and immunostaining of the proteins cytokeratin 7, cytokeratin 20, and caudal type homeobox 2 transcription factor to confirm that our PDTOs originated from rectal cancer tissue and not from normal rectal mucosa. Our PDTOs showed similar histological morphologies and CK protein expression patterns to those of original tumor tissues (Figure 2C). Overall, these data demonstrated that PDTOs recapitulated the histological morphologies and marker expression of the paired patient tissues, as previously reported [21,29]. To define the capacity of colorectal cancer organoids to mirror the genome heterogeneity of the corresponding patient tumor, we compared the mutational status of three genes (*KRAS*, *NRAS,* and *BRAF)* in 19 PDTOs and corresponding tumor tissues. *KRAS*, *NRAS,* and *BRAF* mutations in PDTOs were matched to 86.6%, 100%, and 100% of those in corresponding tumor tissues, respectively (Figure 2D).

**Figure 2.** Histological and genomic characterization of PDTOs. (**A**) Fluorescence microscopy images of FFPE sections of organoids and corresponding tissues for goblet cells (mucin 2 [Muc2]+, red), entero-endocrine cells (CgA+, red), enterocytes (villin 1 [VL1]+, red), and proliferating cells (Ki-67, red). Counterstain, DAPI (blue) and epithelial, E-cadherin (green). Scale bars, 100 µm. (**B**) The mutation landscape of 19 PDTOs. The frequency of alterations in PDTO is noted with the type of genetic alteration (indicated by color code). The top 21 mutated genes observed in PDTOs, including the most known significant cancer driver genes, are shown. (**C**) Immunohistochemical profile of FFPE sections of organoids and corresponding tissues for cytokeratin 7, cytokeratin 20, and caudal type homeobox 2 transcription factor along with corresponding H&E staining. Magnification, ×40. Scale bars, 50 µm. (**D**) *KRAS*, *NRAS,* and *BRAF* mutation status of PDTOs and paired tumor tissues. Abbreviations: CDX, caudal type homeobox; CK, cytokeratin; DAPI, 4′,6-diamidino-2-phenylindole; FFPE, formalin fixed paraffin-embedded; H&E, hematoxylin and eosin; and PDTO, patient-derived tumor organoid. **Figure 2.** Histological and genomic characterization of PDTOs. (**A**) Fluorescence microscopy images of FFPE sections of organoids and corresponding tissues for goblet cells (mucin 2 [Muc2]+, red), entero-endocrine cells (CgA+, red), enterocytes (villin 1 [VL1]+, red), and proliferating cells (Ki-67, red). Counterstain, DAPI (blue) and epithelial, E-cadherin (green). Scale bars, 100 µm. (**B**) The mutation landscape of 19 PDTOs. The frequency of alterations in PDTO is noted with the type of genetic alteration (indicated by color code). The top 21 mutated genes observed in PDTOs, including the most known significant cancer driver genes, are shown. (**C**) Immunohistochemical profile of FFPE sections of organoids and corresponding tissues for cytokeratin 7, cytokeratin 20, and caudal type homeobox 2 transcription factor along with corresponding H&E staining. Magnification, ×40. Scale bars, 50 µm. (**D**) *KRAS*, *NRAS,* and *BRAF* mutation status of PDTOs and paired tumor tissues. Abbreviations: CDX, caudal type homeobox; CK, cytokeratin; DAPI, 40 ,6-diamidino-2 phenylindole; FFPE, formalin fixed paraffin-embedded; H&E, hematoxylin and eosin; and PDTO, patient-derived tumor organoid.

#### *3.3. PDTOs Response to Irradiation*

*Cancers* **2021**, *13*, x FOR PEER REVIEW 9 of 16

To validate the response of PDTOs to irradiation in vitro, we performed a radiation dose-dependent (0 Gy, 2 Gy, 4 Gy, and 6 Gy) survival analysis of 19 PDTOs. Supplementary Figure S3 displays representative images of irradiated organoids, and we counted the number of viable organoids after irradiation to measure the survival fraction (Figure 3A and Supplementary Figure S4). We analyzed the D<sup>0</sup> value (the dose required to reduce the fraction of surviving organoids to 37%); a higher D<sup>0</sup> value indicates greater radioresistance [26]. Therefore, we defined radioresistant PDTOs and radiosensitive PDTOs according to the D<sup>0</sup> value (Figure 3B). These survival fraction data were validated by direct comparison using the MTS cell viability assay (Figure 3C and Supplementary Figure S5). The results demonstrated the heterogeneity of the radioresponse in 19 PDTOs. According to our data, PDTO-22 and PTDO-19 showed radioresistant and radiosensitive characteristics, respectively (Figure 3D,E). To confirm these different radioresponses, we tested this result using several in vitro analyses. The organoid viability of PDTO-19 cells was significantly reduced compared with that of PDTO-22 at 2 Gy, 4 Gy, and 6 Gy (*p* < 0.0001; Figure 3F). To directly assess the regenerative ability of organoids, we counted organoids at the second passage after splitting the irradiated organoids. Seventy-two hours after splitting, the relative number of PDTO-19 organoids was significantly lower than that of PDTO-22 after irradiation (*p* = 0.034; Figure 3G). To determine whether the change in cell viability was accompanied by cell proliferation, we performed EdU staining in PTDOs after irradiation and showed that 13% of the cells in the S phase decreased after irradiation in PDTO-22. In contrast, 30% of S phase cells were reduced after irradiation in PDTO-19 (*p* = 0.029; Figure 3H). To evaluate the apoptotic cellular response to radiation, apoptosis-related protein levels were analyzed. Cleaved-PARP and -caspase-3 levels, which are considered hallmarks of apoptosis, were increased in PDTO-19 after irradiation compared to those in PDTO-22 (Figure 3I). *3.3. PDTOs Response to Irradiation*  To validate the response of PDTOs to irradiation in vitro, we performed a radiation dose-dependent (0 Gy, 2 Gy, 4 Gy, and 6 Gy) survival analysis of 19 PDTOs. Supplementary Figure S3 displays representative images of irradiated organoids, and we counted the number of viable organoids after irradiation to measure the survival fraction (Figure 3A and Supplementary Figure S4). We analyzed the D0 value (the dose required to reduce the fraction of surviving organoids to 37%); a higher D0 value indicates greater radioresistance [26]. Therefore, we defined radioresistant PDTOs and radiosensitive PDTOs according to the D0 value (Figure 3B). These survival fraction data were validated by direct comparison using the MTS cell viability assay (Figure 3C and Supplementary Figure S5). The results demonstrated the heterogeneity of the radioresponse in 19 PDTOs. According to our data, PDTO-22 and PTDO-19 showed radioresistant and radiosensitive characteristics, respectively (Figure 3D,E). To confirm these different radioresponses, we tested this result using several in vitro analyses. The organoid viability of PDTO-19 cells was significantly reduced compared with that of PDTO-22 at 2 Gy, 4 Gy, and 6 Gy (*p* < 0.0001; Figure 3F). To directly assess the regenerative ability of organoids, we counted organoids at the second passage after splitting the irradiated organoids. Seventy-two hours after splitting, the relative number of PDTO-19 organoids was significantly lower than that of PDTO-22 after irradiation (*p* = 0.034; Figure 3G). To determine whether the change in cell viability was accompanied by cell proliferation, we performed EdU staining in PTDOs after irradiation and showed that 13% of the cells in the S phase decreased after irradiation in PDTO-22. In contrast, 30% of S phase cells were reduced after irradiation in PDTO-19 (*p* = 0.029; Figure 3H). To evaluate the apoptotic cellular response to radiation, apoptosis-related protein levels were analyzed. Cleaved-PARP and -caspase-3 levels, which are considered hallmarks of apoptosis, were increased in PDTO-19 after irradiation compared to those in PDTO-22 (Figure 3I).

**Figure 3.** The response of PDTOs to radiation. (**A**) Dose–response of survival fraction in 19 PDTOs (*n* = 4, independent experiments for each PDTO) is shown. Data are presented as mean ± standard deviation. The red line represents a survival fraction of 0.37. (**B**) D0 values were calculated according to the multitarget single-hit model. (**C**) MTS cell viability assay of **Figure 3.** The response of PDTOs to radiation. (**A**) Dose–response of survival fraction in 19 PDTOs (*n* = 4, independent experiments for each PDTO) is shown. Data are presented as mean ± standard deviation. The red line represents a survival fraction of 0.37. (**B**) D<sup>0</sup> values were calculated according to the multitarget single-hit model. (**C**) MTS cell viability assay of 19 PDTOs after 0 Gy, 2 Gy, 4 Gy, and 6 Gy irradiation (*n* = 6, independent experiments for each PDTO). Data are normalized to those of control cells. Data are presented as mean ± standard deviation. (**D**) Dose–response of survival fraction in PDTO-19

and PDTO-22 (*n* = 4 independent experiments for each PDTO) is shown. \*\* *p* < 0.01. (**E**) Morphology of PDTO-19 and PDTO-22 after irradiation with 5 Gy after 5 days. Scale bars, 1000 µm. (**F**) MTS cell viability assay of PDTO-19 and PDTO-22 after treatment with 0 Gy, 2 Gy, 4 Gy, and 6 Gy. Data are normalized to those of the control cells and presented as mean ± standard deviation. \*\* *p* < 0.01. (**G**) (**left**) Image of organoids after the second passage. Scale bars, 1000 µm. (**right**) The relative organoid number after the second passage. Data are presented as mean ± standard deviation. \* *p* < 0.05. (**H**) (**left**) Fluorescence microscopy images of EdU incorporation in PDTO-19 and PDTO-22 after irradiation. Scale bars, 400 µm. Blue, DAPI; red, EdU. (**right**) Statistical analysis representing EdU-positive cells per DAPI-stained cell (*n* = 3). \* *p* < 0.05. (**I**) Expression levels of c-PARP and c-caspase-3 in PDTOs. β-actin was the loading control. Abbreviations: c-PARP, cleaved poly-ADP-ribose polymerase; DAPI, 40 ,6-diamidino-2-phenylindole; IR, ionizing radiation; PDTO, patient-derived tumor organoid; RR, radioresistant; and RS, radiosensitive.

#### *3.4. Correlation of Experimental Data with Actual TRG Outcomes*

To compare the experimental results of the survival fraction, D<sup>0</sup> value, and cell viability, we regrouped TRGs into three categories: TRG 0, TRG 1/2, and TRG 3 (Figure 1B). The results of comparisons according to the three TRG groups and according to whether TRGs were at their two extreme categories, good responders (TRG 0 or not) and poor responders (TRG 3 or not), are shown in Figure 4A. Generally, *p* values obtained by comparing the mean (SD) values among the three TRG groups were more significant in the survival fraction and D<sup>0</sup> data than in cell viability. Furthermore, comparing after actual TRGs were regrouped according to whether TRGs were in the two extreme categories or not, the *p* values were more significant for comparisons of survival fraction and D<sup>0</sup> data (Figure 4A). Next, we performed ROC analyses to determine whether our experimental data could classify TRGs and which experimental data would be more appropriate to use for classifying TRGs. While D<sup>0</sup> data had a single value, D<sup>0</sup> only, the survival fraction data and cell viability data had multiple values at each radiation dose (2 Gy, 4 Gy, and 6 Gy). Therefore, we used a multiple logistic regression model to analyze survival fraction and cell viability data in this ROC analysis. In the ROC analysis of good responders (TRG 0), AUCs matched to D0, survival fraction, and cell viability tests were 0.753 (95% confidence interval (CI), 0.644–0.863), 0.897 (95% CI, 0.83–0.965), and 0.631 (95% CI, 0.525–0.737), respectively (Figure 4B). When analyzing poor responders (TRG 3), the AUCs of the respective experimental data were as follows: D0, 0.966 (95% CI, 0.926–1); the survival fraction model, 0.974 (95% CI, 0.941–1); and the cell viability model, 0.898 (95% CI, 0.827–0.968; Figure 4C). Sensitivity, specificity, positive predictive value, and negative predictive value were highest in the survival fraction model (Supplementary Table S3). We reconstructed the LQ curve according to the regrouped TRG using the 19 individual PDTO survival fraction curves (Figure 3A). When comparing the curves of the three groups, the LQ curves were clearly divided with statistical significance (*p* < 0.0001). In addition, TRG 0 or not (*p* < 0.0001) and TRG 3 or not (*p* < 0.0001) of the LQ curve were still significantly divided with respect to the TRG groups (Figure 4D).

*Cancers* **2021**, *13*, x FOR PEER REVIEW 11 of 16

#### *3.5. Machine Learning-Assisted Prediction Model*

As shown in Figure 4, the AUC, sensitivity, specificity, positive predictive value, and negative predictive value of the survival fraction model were highest among the values from the three experimental datasets. Therefore, we developed machine learning-based classification models using the survival fraction data. After building a prediction model using a training dataset, we evaluated the model performance using the testing dataset. In the prediction model for good responders (TRG 0) trained using logistic regression, the AUC was 0.916 (Figure 5A), the accuracy was 78.9%, and the kappa value was 0.38. The AUC, accuracy, and kappa value of the model trained using the random forest were 0.918, 81.5%, and 0.51, respectively. In the prediction model for poor responders (TRG 3) trained using logistic regression, the AUC, accuracy, and kappa value were 0.927, 89.5%, and 0.65, respectively (Figure 5B); those of the model trained using the random forest were 0.971, 92.1%, and 0.75, respectively. *3.5. Machine Learning-Assisted Prediction Model*  As shown in Figure 4, the AUC, sensitivity, specificity, positive predictive value, and negative predictive value of the survival fraction model were highest among the values from the three experimental datasets. Therefore, we developed machine learning-based classification models using the survival fraction data. After building a prediction model using a training dataset, we evaluated the model performance using the testing dataset. In the prediction model for good responders (TRG 0) trained using logistic regression, the AUC was 0.916 (Figure 5A), the accuracy was 78.9%, and the kappa value was 0.38. The AUC, accuracy, and kappa value of the model trained using the random forest were 0.918, 81.5%, and 0.51, respectively. In the prediction model for poor responders (TRG 3) trained using logistic regression, the AUC, accuracy, and kappa value were 0.927, 89.5%, and 0.65, respectively (Figure 5B); those of the model trained using the random forest were 0.971, 92.1%, and 0.75, respectively.

**Figure 5.** Machine learning-assisted prediction model. (**A**,**B**) Performance of machine learningbased prediction models for TRG 0 (**A**) and TRG 3 (**B**) on the testing dataset presented using receiver operating characteristic curves and their area under the curves. Red line: prediction model trained using the random forest algorithm, blue line: prediction model trained using binary logistic regression. Abbreviations: AUC, area under the ROC curve; ROC, receiver operating characteristic; and TRG, tumor regression grade. **Figure 5.** Machine learning-assisted prediction model. (**A**,**B**) Performance of machine learning-based prediction models for TRG 0 (**A**) and TRG 3 (**B**) on the testing dataset presented using receiver operating characteristic curves and their area under the curves. Red line: prediction model trained using the random forest algorithm, blue line: prediction model trained using binary logistic regression. Abbreviations: AUC, area under the ROC curve; ROC, receiver operating characteristic; and TRG, tumor regression grade.

#### **4. Discussion 4. Discussion**

In this co-clinical trial, we reproduced the results of previous studies [21,22]. The histology, genetic features, and irradiation response of PDTOs mirrored real treatment outcomes of original tumors and patients. Furthermore, our quantitative experimental data correlated well with actual TRG results. With these results, we built a machine learningbased prediction model by inputting the survival fraction values of PTDOs. At the beginning of this study, we did not know which experimental indicator would best match the patient's actual TRG results; thus, we conducted various experiments regarding organoid irradiation responses. Among them, we selected D0, survival fraction, and cell viability data, which were easily measurable and reproducible by repeated tests. We found that survival fraction data were the best-matched experimental results to the patient's TRG results in statistical analyses. The machine learning-based prediction model using the survival fraction data showed an excellent performance. In this co-clinical trial, we reproduced the results of previous studies [21,22]. The histology, genetic features, and irradiation response of PDTOs mirrored real treatment outcomes of original tumors and patients. Furthermore, our quantitative experimental data correlated well with actual TRG results. With these results, we built a machine learning-based prediction model by inputting the survival fraction values of PTDOs. At the beginning of this study, we did not know which experimental indicator would best match the patient's actual TRG results; thus, we conducted various experiments regarding organoid irradiation responses. Among them, we selected D0, survival fraction, and cell viability data, which were easily measurable and reproducible by repeated tests. We found that survival fraction data were the best-matched experimental results to the patient's TRG results in statistical analyses. The machine learning-based prediction model using the survival fraction data showed an excellent performance.

Organoid technology has been a highlight for cancer research due to the close resemblance of organoids to original tumors [29–35]. Due to its rapid establishment with a high Organoid technology has been a highlight for cancer research due to the close resemblance of organoids to original tumors [29–35]. Due to its rapid establishment with a high success rate, the organoid model is noted as a pre- or co-clinical model for pre-

cision medicine. Although not commented on in this study, the growth rate of PDTOs was heterogeneous, but could acquire enough volume to assess the irradiation response within 1–2 weeks in most cases. It added a testing time of approximately 4 weeks to obtain irradiation response data that can predict the TRG results of real patients. It is a clinically significant period to produce treatment recommendations, as stated by Yao et al. [22].

This study has some limitations. First, the study sample size was small. The goal of a machine learning model is to generalize patterns using training data to correctly predict new data that have never been presented to the model. Overfitting occurs when a model adjusts excessively to the training data, sees patterns that do not exist, and consequently performs poorly in predicting new data. The fewer samples for training, the more models that can fit. Our treatment-naive sample number was not smaller than that of previous studies [21,22]; however, it was not sufficient to obtain a reproducible prediction model, although we used the random forest method for model training and obtained acceptable model performance results. Random forest is an ensemble machine learning model that increases the model performance, but is not a solution for small sample size issues. To develop a reliable predictive model using organoids, a reliable volume of training samples is required [36]. Given that it is difficult to generate sufficient data in a single laboratory, it is necessary to collect and share data produced under consented standard experimental conditions among clinical organoid researchers.

Second, in the current organoid model itself, one can only observe the irradiation response of cancer cells themselves. For the actual therapeutic response of tumor cells to radiation, the role of the microenvironment is very important. Although organoid cultures provide more favorable conditions than traditional cell line models for tissue physiology and structure, which are close to in vivo situations, the model does not robustly retain the complexity and diversity of the tumor microenvironment (T-ME). The T-ME has been gradually recognized as a key contributor to cancer progression and a determinant of treatment outcomes [37,38]. In radiotherapy, vascular, stromal, and immunological changes in T-ME induced by radiation promote radioresistance and tumor recurrence; furthermore, radiotherapy has recently been proposed to target the T-ME to overcome radioresistance [39]. However, organoid cultures typically contain epithelium. Thus, to overcome these limitations, models for co-culture of tumor organoids and T-ME have recently been introduced. Öhlund et al. developed a co-culture system of pancreatic cancer organoids and cancer-associated fibroblasts that can recapitulate some of the features observed in patients [40]. Dijkstra et al. described a patient-personalized in vitro model that enabled the induction and analysis of tumor-specific T-cell responses using colorectal cancer organoids and T lymphocytes isolated from patients' peripheral blood [41]. In addition, a unique co-culture method based on an air-liquid interface system permitted the propagation of PDO and tumor-infiltrating lymphocytes [42]. Organoid culture methods that partially retain the patient's T-ME might overcome the hurdles of organoid culture and offer reliable results.

Finally, in this study, we only evaluated the response against irradiation. In a real situation, various chemotherapeutic agents are combined to obtain improved NCRT results [9–14]. However, we did not perform a drug sensitivity test as our study population comprised a homogenous patient group that used a single agent, 5-FU, as a concurrent treatment for all patients except one, and the difference in radioresponse affected by the combination of various drugs could not be observed. Based on this study result of radiosensitivity, and through further validations, we believe that we will be able to identify which element or combinations of current multimodal treatments would be most helpful and identify a more advanced tailored treatment via current ex vivo tests with PDTOs.

#### **5. Conclusions**

As revealed by previous studies, individual PDTOs recapitulated responses of original tumors to irradiation. The radiation response of PDTOs could predict the patient's TRG with statistical significance. The PDTO-based radiosensitivity model could be a reliable diagnostic tool for the tailored treatment of rectal cancer.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/cancers13153760/s1, Figure S1: Observed morphologies of 19 PDTOs. Figure S2: The mutation of 19 PDTOs for all alterations is displayed. Figure S3: Morphologies of PDTOs after irradiation at 2 Gy, 4 Gy, and 6 Gy. Figure S4: Dose–response of survival fraction in 19 PDTOs. Figure S5: Dose–response of cell viability in 19 PDTOs. Figure S6: Whole blot showing all the bands with molecular weight marker. Table S1: List of chemical and reagents used for studies. Table S2. List of antibodies used for studies. Table S3: Results of ROC about two extreme categories.

**Author Contributions:** Conceptualization: M.P., Y.K. and U.S.S.; Methodology: M.P., J.K. (Junhye Kwon), J.K. (Joonseog Kong), Y.K. and U.S.S.; Software: W.I.J., M.S.K. and U.S.S.; Validation: M.P., J.K. (Junhye Kwon), J.K. (Joonseog Kong), Y.K. and U.S.S.; Formal analysis: M.P., J.K. (Junhye Kwon), J.K. (Joonseog Kong), Y.K. and U.S.S.; Resources: J.K. (Joonseog Kong), S.M.M., S.C., K.Y.Y., W.I.J., M.S.K., Y.K. and U.S.S.; Writing—Original Draft: M.P., J.K. (Junhye Kwon), J.K. (Joonseog Kong), Y.K. and U.S.S.; Visualization: M.P., J.K. (Junhye Kwon), J.K. (Joonseog Kong), Y.K. and U.S.S.; and Supervision: Y.K. and U.S.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by a grant from the Korea Institute of Radiological and Medical Sciences, funded by the Ministry of Science and ICT, Republic of Korea (grant no. 50542-2020).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki. The Ethics Committee of Korea Cancer Center Hospital approved this study (approval no. KIRAMS-2017-07-001). All research was performed according to the approved guidelines and regulations of the institution.

**Informed Consent Statement:** All samples were obtained from patients who provided written informed consent.

**Data Availability Statement:** All data in this study are included as Supplementary Tables and Figures. Any additional information is available upon request.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Review* **The Role of Cancer Stem Cells in Colorectal Cancer: From the Basics to Novel Clinical Trials**

**Céline Hervieu <sup>1</sup> , Niki Christou 1,2 , Serge Battu <sup>1</sup> and Muriel Mathonnet 1,2,\***


**Simple Summary:** Cancer stem cells (CSCs) fuel tumor growth, metastasis and resistance to therapy in colorectal cancer (CRC). These cells therefore represent a promising target for the treatment of CRC but are difficult to study because of the complexity of their isolation. This review presents the methods currently used to isolate colorectal CSCs as well as the techniques for characterizing these cells with their advantages and limitations. The aim of this review is to provide a state-of-the-art on the clinical relevance of CSCs in CRC by outlining current treatments for CRC, the resistance mechanisms developed by CSCs to overcome them, and ongoing clinical trials of drugs targeting CSCs in CRC. Overall, this review addresses the complexity of studying CSCs in CRC research and developing clinically effective treatments to enable CRC patients to achieve a short and long-term therapeutic response.

**Abstract:** The treatment options available for colorectal cancer (CRC) have increased over the years and have significantly improved the overall survival of CRC patients. However, the response rate for CRC patients with metastatic disease remains low and decreases with subsequent lines of therapy. The clinical management of patients with metastatic CRC (mCRC) presents a unique challenge in balancing the benefits and harms while considering disease progression, treatment-related toxicities, drug resistance and the patient's overall quality of life. Despite the initial success of therapy, the development of drug resistance can lead to therapy failure and relapse in cancer patients, which can be attributed to the cancer stem cells (CSCs). Thus, colorectal CSCs (CCSCs) contribute to therapy resistance but also to tumor initiation and metastasis development, making them attractive potential targets for the treatment of CRC. This review presents the available CCSC isolation methods, the clinical relevance of these CCSCs, the mechanisms of drug resistance associated with CCSCs and the ongoing clinical trials targeting these CCSCs. Novel therapeutic strategies are needed to effectively eradicate both tumor growth and metastasis, while taking into account the tumor microenvironment (TME) which plays a key role in tumor cell plasticity.

**Keywords:** colorectal cancer; cancer stem cells; drug resistance; clinical trials

#### **1. Introduction**

Colorectal cancer (CRC) is the fourth leading cause of cancer-related death worldwide [1]. While the occurrence and mortality rates of CRC is declining in the European countries, these rates are increasing in rapidly transitioning countries, such as many African and South Asian countries [2]. The tumor–node–metastases (TNM) classification allows the stratification of patient groups according to the stage of the disease, based on anatomical information [3,4]. The location and stage of the tumor enable both the assessment of the patient's prognosis and the determination of the therapeutic approach, depending on the

**Citation:** Hervieu, C.; Christou, N.; Battu, S.; Mathonnet, M. The Role of Cancer Stem Cells in Colorectal Cancer: From the Basics to Novel Clinical Trials. *Cancers* **2021**, *13*, 1092. https://doi.org/10.3390/ cancers13051092

Academic Editors: Marta Baiocchi and Ann Zeuner

Received: 28 January 2021 Accepted: 27 February 2021 Published: 4 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

patient's overall health as well as the status of the tumor in terms of mutation and mismatch repair (MMR) [1,5]. Therapeutic options for the treatment of CRC are surgical resection, systemic therapy including chemotherapy, targeted therapy and immunotherapy, local therapy for metastases and palliative therapy [1,6]. Importantly, surgical resection is the only curative treatment, if all macroscopic and microscopic tumor foci can be removed [1,6]. Unfortunately, even after well directed curative treatment, some patients experience treatment failure that may be associated with the development of multidrug resistance (MDR) during or after treatment. In addition, despite initially successful therapy, the development of drug resistance often leads to relapse in cancer patients, known as minimal residual disease (MRD) [7]. Both MDR and MRD can be attributed to a subpopulation of tumor cells with self-renewal and multi-lineage differentiation capabilities, the cancer stem cells (CSCs), known as colorectal cancer stem cells (CCSCs) for CRC [8]. CSCs contribute to tumor initiation and dissemination, treatment resistance and metastasis development. Tumor microenvironment (TME) and metabolic plasticity may also be involved in therapeutic failure by imposing selective pressures on cancer cells that lead to chemoresistance and cancer progression [9,10]. Therefore, the development of new therapies targeting CSCs, taking into account the TME and tumor metabolism, represents an interesting approach to overcome resistance to therapies [11]. In this review, we will present the origin of CCSCs and provide an overview of the techniques currently used to isolate them. Then, we will review current knowledge on the clinical relevance of CCSCs, through the clinical management of CRC and the mechanisms of resistance to therapies associated with CCSCs. Finally, we will introduce some clinical trials based on drugs targeting CCSCs.

#### **2. Colorectal Cancer Stem Cells**

The CSC theory suggests that tumor growth is driven by a small number of dedicated stem cells (SCs), the CSCs [8]. By definition, a CSC has the ability to self-renew in order to expand its pool and to generate all the differentiated cells that comprise the tumor (multipotency). The transformation of a colorectal stem cell into CCSC requires the acquisition of tumor-related features.

#### *2.1. Colorectal Cancer Stem Cell Origin*

The history of CSCs began two decades ago with the discovery of CSCs in human acute myeloid leukemia (AML) by Dick and colleagues [12]. For the first time, a cell capable of initiating human AML in immunodeficient mice and possessing differentiation, proliferation and self-renewal capabilities was described. A few years later, using similar experimental approaches, the presence of CSC was demonstrated in solid cancers such as colorectal cancer. The origin of CSCs in CRC is controversial, and several hypotheses have been proposed. CCSCs are associated with the acquisition of malignant molecular and cellular changes either due to the accumulation of genetic and epigenetic alterations in restricted stem/progenitor cells and normal tumor cells, or to the dedifferentiation of somatic cells caused by various genetic and environmental factors [13–15]. CSCCs exhibit tumor-related characteristics such as uncontrolled growth, tumorigenicity and therapy resistance, and may constitute the small reservoir of drug-resistant cells that are responsible for relapses after chemotherapy-induced remission, known as MRD, and distant metastasis [7,11]. Thus, CCSCs play a key role in the initiation, invasion and progression of CRC as well as resistance to therapy. These CCSCs give rise to heterogeneous tumors that can be serially transplanted into immunodeficient mice that resemble the original tumor [16]. In addition, CCSCs have the ability to form disseminated metastatic tumors due to their extensive proliferative potential [15]. One of the main challenges in the study of CCSCs is their isolation, due to their low percentage within the tumor [16]. However, the CCSC population appears to be phenotypically and functionally heterogeneous and dynamic, which is another barrier to their isolation [17]. Therefore, the development of therapies that selectively eradicate CCSCs offers promising opportunities for a sustainable clinical response but requires effective technologies to detect and isolate them [11].

#### *2.2. Colorectal Cancer Stem Cell Isolation Methods*

Different methods are used to isolate CCSCs, based either on the expression pattern of CCSC markers, the functional aspect of CCSCs, or their biophysical features [18]. The objective of this chapter is to present the techniques currently in use with the advantages and disadvantages of each approach.

#### 2.2.1. CCSC Isolation Based on Phenotypic Features

Many stem cells markers were found to be associated with CCSC features. However, the heterogeneous and dynamic nature of CCSCs challenges their isolation and enrichment. The first publications from the literature identifying subpopulations of CSCs in CRC are summarized in Table 1. Experimental models, CCSC isolation methods and characterization techniques used by the authors are detailed in this table. Studies conducted by O'Brien et al. and Ricci-Vitiani et al. identified the first CCSC marker: the five-transmembrane glycoprotein CD133 [19,20]. However, its use has become controversial as the tumorigenic and clonogenic potential of CD133<sup>+</sup> -CSCs depends on the positivity for a specific glycosylated epitope of the CD133 protein [21].

**Table 1.** Experimental models, markers and CCSC isolation and characterization methods used in the first publications identifying CSCs in CRC.


CRC: colorectal cancer; CCSC: colorectal cancer stem cells; CD: cluster of differentiation; MACS: magnetic-activated cell sorting; FACS: fluorescence-activated cell sorting; ALDH: aldehyde dehydrogenase.

> Then, Clarke's group showed that EpCAMhigh/CD44<sup>+</sup> cells isolated from human CRC could establish a tumor in mice with morphological and phenotypic heterogeneity of the original tumor and concluded that CD44 and EPCAM markers could be considered robust CCSC markers [22]. In addition, the study by Dalerba et al. highlights an additional differentially expressed marker, CD166, which could be used to further enrich CCSCs in the EpCAhigh/CD44<sup>+</sup> population [22]. Using lineage-tracing experiments in mice, Clevers and coworkers identified stem cells in the small intestine and colon using the marker gene *Lgr5* [28] and proposed them as the cells-of-origin of intestinal cancer [23]. At the same time, Sangiorgi and Capecchi's study found another intestinal stem cell marker in vivo,

Bmi1 [24]. Importantly, Bmi-1 and Lgr5 markers define two types of SCs, quiescent and rapidly cycling SCs, respectively [23,24], and may identify CCSCs. Vermeulen et al. showed that spheroid cultures from primary CRC have a tumor-initiating capacity and that a cell subpopulation expresses CD24, CD29, CD44 and CD166 markers, suggested as CCSC markers [25]. The study by Pang et al. identifies a subpopulation of CD26<sup>+</sup> cells capable of developing distant metastases when injected into the mouse cecal wall and associated with increased invasiveness and chemoresistance, whereas CD26− cells cannot [26]. Interestingly, the presence of CD26<sup>+</sup> cells in the primary tumor of patients without distant metastases at that time may predict future distant metastases, highlighting a critical role of CSCs in the progression of metastatic cancer and important clinical implications [26]. The transmembrane glycoprotein CD44 has several splicing variants, including CD44v6, which appears to negatively impact the prognosis of CRC patients [29,30]. Todaro et al. demonstrated that all identified CCSCs express the CD44v6 marker, which supports their migration and promotes metastasis [27]. Each of these markers has its own function and role in the prognosis of CRC, as shown in Table 2.


**Table 2.** Functions and roles in CRC prognosis of CCSC markers.

CCSC: colorectal cancer stem cells; CD: cluster of differentiation; ECM: extracellular matrix; CRC: colorectal cancer.

All these markers can be expressed by CCSCs, but they do not all have the same capacity. Some, such as CD133, Lgr5, Bmi-1, CD26 and CD44v6 alone identify CCSCs, while the other presented markers allow the identification of CCSCs only in combination with one or more of the aforementioned markers. In conclusion, these markers play a key role in the identification of CCSCs and can be used alone or in combination to sort CCSCs by magnetic-activated cell sorting (MACS) or fluorescence-activated cell sorting (FACS) techniques.

MACS is a magnetic-based cell isolation technique, using a positive selection strategy, presented in Figure 1 panel 1 [44]. Magnetic beads are conjugated to highly specific monoclonal antibodies that recognize CCSC marker on the surface of cells of interest. Then, the heterogeneous suspension of cells is passed through a separation column, in a magnetic field, to retain the cells labeled with magnetic beads and antibodies [45]. By switching off the magnetic field, target cells will be eluted. MACS is a fast and easy method of cell separation, especially for the isolation of CCSCs that represent a small cell population in the tumor mass. However, MACS is only a mono-parameter separation method that requires cell labelling and is unable to separate cells based on the variable expression of markers [44,45].

**Figure 1.** Phenotypic sorting of CSCs through the expression of CSC markers recognized by antibodies coupled to either magnetic beads, MACS (1), or fluorochromes, FACS (2). Once the antibodies are added, the cell suspension is passed through either a MACS column in a magnetic field that retains the antibody-labeled cells (1) or through a flow cytometer that distinguishes and isolates labeled cells from unlabeled cells (2). CSC: cancer stem cell; MACS: magnetic-activated cell sorting; FACS: fluorescence-activated cell sorting.

FACS uses fluorescently labeled antibodies that target the cell surface or intracellular markers to isolate CCSCs [44]. Antibodies are conjugated to fluorochromes and recognize the marker of interest within a cell suspension, as shown in Figure 1 panel 2 [44]. The cell suspension is then hydrodynamically focused into a stream of individual cells by the flow cytometer and passed through a laser which provides information on the size, granularity and fluorescent properties of single cells [18]. Fluorochromes with different emission wavelengths can be used simultaneously to allow multiparameter separations [44]. Both technologies allow the sorting of CCSCs with high purity but require the availability of antibodies and cell labeling, which can modify their properties and induce cell differentiation [16,44,46]. In addition, phenotypic characterization is insufficient to define a CCSC because these markers are also expressed by normal SCs.

Therefore, in order to confirm the detection and isolation of CCSCs, their functional capabilities need to be evaluated by in vitro and in vivo assays [18].

#### 2.2.2. CCSC Isolation Based on Functional Features

CCSCs have many intrinsic properties that can be used to identify them, such as their capacity for self-renewal, multi-lineage differentiation, detoxification due to aldehyde dehydrogenase 1 (ALDH1) activity and dye exclusion ability, colony/sphere formation and tumorigenicity, which are illustrated in Figure 2. These functional characteristics have been used to develop effective methods for isolating CCSCs. The ALDH activity assay is based on the use of a fluorescent and non-toxic ALDH substrate that freely diffuses into intact and viable cells [47]. Then, in the presence of the detoxifying enzyme ALDH, the substrate is converted into a negatively charged fluorescent product that is retained inside the cells. Thus, cells with high ALDH activity become brightly fluorescent and can be measured by flow cytometry as presented in Figure 2 panel 1a [47,48]. CCSCs increase their ALDH1 activity to resist to chemotherapeutic agents and prevent apoptosis by maintaining low levels of reactive oxygen species [47]. The advantage of the ALDH assay is high stability compared to the use of surface markers, but its specificity is low due to its expression in both normal SCs and CSCs [48].

**Figure 2.** Functional sorting of CSCs due to their specific properties such as enhanced detoxification (1), ALDH (1a) and SP (1b), in vitro self-renewal and differentiation capacity, colony- (2) and sphere-forming (3) assays, and the ability to form tumors in vivo, tumorigenicity assay (4). CSC: cancer stem cell; ALDH: aldehyde dehydrogenase; SP: side population.

The side population (SP) assay relies on the differential ability of the cells to efflux dye via ATP-binding cassette (ABC) transporters [49]. Hoechst33342 is a fluorescent dye that binds all nucleic acids and has the particularity of passing through the plasma membrane of living cells. When excited by UV lights, Hoechst dye emits a fluorescence that can be detected by a flow cytometer [49]. SP cells are capable of actively removing the dye from the cell and have a unique low Hoechst fluorescence emission, as shown in Figure 2 panel 1b. CCSCs highly express efflux transporters, such as multidrug resistance protein 1 (ABCB1), multidrug resistance-associated proteins (ABCC1) and breast cancer resistance protein (ABCG2), to protect themselves against cytotoxic substances and therefore look like SP cells [18]. The SP assay is an easy and reliable method that does not require cell labeling, but due to its low purity and specificity, the SP assay is often combined with cell labeling to significantly increase the purity of sorted CSCCs [18,49].

Colony and sphere formation assays evaluate in vitro the self-renewal and differentiation capacities of individual cells in two (2D) and three (3D) dimensions, respectively, which are shown in Figure 2, panels 2 and 3 [50,51]. Both assays are based on non-adherent cultures using either a soft agar layer (2D) or low adherent plates (3D) [52,53]. In the soft agar method illustrated in Figure 2 panel 2, the suspension of individual cells is mixed with the soft agar which may, after several weeks of incubation, give colonies that can be stained with crystal violet to determine their number and size [50,52]. In comparison, in the 3D culture shown in Figure 2 panel 3, the individual cells in suspension are grown at very low cell density and in serum-free medium (DMEM/F12 medium) supplemented with growth factors (human recombinant basic fibroblast growth factor and human recombinant epidermal growth factor), N2 supplement, glucose, insulin and optionally antibiotics such as penicillin/streptomycin for several weeks to obtain spheroids [51,54]. The produced spheroids mimic various characteristics of solid tumors, such as growth kinetics, gene expression pattern and cellular organization with the outer layer containing highly proliferative cells, the middle layer with senescent or quiescent cells and the inner layer comprising necrotic cells due to a lack of oxygen and nutrients [53]. CCSCs can be identified in both techniques as they have the ability to form larger and more numerous colonies and are capable of giving rise to a tumor sphere (colonosphere) resembling the primary sphere when passed in series, due to their ability to grow and divide independently of their environment which normal cells are unable to do because of anoikis [18,52,55]. Thus, in vitro, 3D models appear to be a relevant preclinical model for testing new drugs, evaluating potential combinations and understanding drug resistance, by mimicking CSC-containing tumors in vitro, before testing them in vivo [18,53,55]. However, these models require well-established protocols and appropriate cell dilution to certify that each colony/sphere is derived from a single cell [18].

The tumorigenicity assay is considered the gold standard method for studying the CSC properties of human tumors in vivo [18,56]. This approach allows to determine the tumorinitiating ability of cancer cells in immunodeficient mice and their capacity for self-renewal in vivo after the dissociation of primary tumors and transplantation in secondary recipient mice, as illustrated in Figure 2 panel 4 [57]. In vivo limiting dilution is the best method for identifying the lowest concentration of cells capable of forming a tumor and determining the frequency of CSCs [18,58]. Importantly, only CSCs have the ability to generate a xenograft that is histologically similar to the parental tumor from which it originated, to be serially transplanted in a xenograft assay due to their long-term self-renewal capacity, and to generate daughter cells [56,58]. However, the use of mouse models requires ethical consideration and complicated laboratory equipment. In addition, the results of xenograft experiments are highly dependent on the number of cells, the implantation site and the incubation period, which leads to certain limitations [18]. Nevertheless, mouse models remain unique models for studying the biology of CSCs in vivo [57,58].

#### 2.2.3. CCSC Isolation Based on Biophysical Features

The development of enrichment and isolation methods for CCSCs without cell labeling offers new perspectives, such as sorting techniques based on biophysical characteristics. The sedimentation field-flow fractionation (SdFFF) is a gentle, non-invasive and label-free method that prevents interference for further cell use and the allows separation of cells according to their size, density, shape and rigidity [16,59]. Cell separation by SdFFF depends on the differential elution of cell subpopulations submitted both to the action of a parabolic profile generated by the mobile phase in the channel and to a multigravitational external

field generated by the rotation of the channel, as presented in Figure 3 [16,59]. In the past decade, SdFFF cell sorting has been adapted and applied in many fields such as neurology, oncology and stem cells [16,60–62]. The study by Mélin et al. describes a strategy, based on SdFFF elution, to obtain activated and quiescent CSC subpopulations from eight different human CRC cell lines [16]. The combination of cell sorting by SdFFF with the grafting of these CSC-enriched fractions into chick embryo chorioallantoic membrane (CAM) model demonstrates the potential of SdFFF to produce innovative matrices for the study of carcinogenesis and the analysis of treatment sensitivity [16,63]. The advantages of this isolation method are the use of biophysical characteristics for cell sorting without cell labeling; however, this technique requires a large number of cells and is time consuming [46].

**Figure 3.** Biophysical sorting of CSCs according to their size, density, shape and rigidity using the SdFFF technique, which does not require cell labelling or fixation. The SdFFF is composed of a pump (1) to transport the mobile phase (PBS) and the cells, an injector (2) to introduce the cell suspension, a motor (3) to rotate the separation channel (4) and a detector (5) coupled to a computer to obtain the elution profile of the cell suspension (6). Psi is a common unit of pressure. CSC: cancer stem cell; SdFFF: sedimentation field-flow fractionation; PBS: phosphate-buffered saline; Abs: absorbance.

#### 2.2.4. CCSC Isolation Methods: Discussion

Taken together, this chapter provides an overview of the techniques commonly used to identify and sort CCSCs, which are summarized in Table 3. The use of cell surface markers remains the most widely used in cancer research, however, it remains controversial due to the lack of a universal marker for CCSCs. Moreover, nowadays, none of the CSC isolation techniques are capable of 100% enrichment of CCSCs due to the shared properties between normal SCs, non-CCSCs and CCSCs [14,17]. As an example, Shmelkov and colleagues have shown that CD133 expression in the colon is not limited to SCs but is also expressed on differentiated tumors cells [64]. In addition, the authors found that both CD133<sup>+</sup> and CD133− isolated from metastatic colon tumors are capable of initiating tumors in a serial xenotransplantation model [64]. A few years later, the study by Kemper et al. demonstrated that CD133 is expressed on the cell surface of CSCs and differentiated tumor cells but is differentially glycosylated [21]. Similarly, using the ALDH activity assay, Huang et al. found that ALDH1 is a marker of both normal and malignant human colonic SCs [48]. Consequently, cell surface markers and ALDH activity cannot be used alone to sort and define CSCs. Thus, the SdFFF technique offers new perspectives for CSC sorting that does not require cell labeling or fixation and thereby allows the combination of this technique with other CSC characterization methods. Therefore, the combined use of CCSC isolation methods can provide a more powerful and efficient tool for identifying and sorting CCSCs. The advantages and weaknesses of each method must be known in order to select the best method based on the experimental question, as shown in Table 3.


**Table 3.** Advantages and disadvantages of CCSC isolation methods.

CCSC: colorectal cancer stem cell; MACS: magnetic-activated cell sorting; FACS: fluorescence-activated cell sorting; ALDH: aldehyde dehydrogenase; SdFFF: sedimentation field flow fractionation.

#### **3. Clinical Relevance of Colorectal Cancer Stem Cells**

Therapeutic advances made in recent decades now enable most cancer patients to achieve major clinical responses [6]. However, although therapeutic approaches are increasing, none of these treatment modalities is curative in most cases of advanced CRC [65]. Furthermore, despite initially successful treatment reflecting the therapeutic effect on the cells that form the tumor bulk, tumor recurrence is almost inevitable due to the development of drug resistance attributed to CCSCs [8].

#### *3.1. Clinical Management of Colorectal Cancer*

Treatment options and recommendations depend on several factors, including the patient's overall health, possible side effects, the type and stage of the tumor, and its mutational and MMR status [1,5]. Therapeutic approaches for the treatment of CRC include surgical resection, local therapies for metastatic disease, systemic therapy comprising chemotherapy, targeted therapy and immunotherapy as presented in Table 4, and palliative chemotherapy [6]. To ensure the optimal survival and quality of life for patients, personalized therapy is crucial to enable cancer patients to maximize the benefits while minimizing the harms [5].

Surgical resection is the mainstay of curative intent treatment for localized and advanced CRCs but needs to be complete to be considered curative when there is regional invasion or histological factors with a poor prognosis [66,67]. Surgery can be associated with neoadjuvant therapy in order to shrink tumor mass and facilitate medical operation and/or with adjuvant therapy to limit cancer recurrence [1]. Importantly, neoadjuvant chemotherapy, possibly coupled with radiotherapy, is mainly indicated for rectal cancers [68]. Treatment regimens for patients with localized CRC generally include chemotherapy such as 5-fluorouracil (5-FU) or capecitabine, oxaliplatin and irinotecan, alone or in combination [69–73]. Leucovorin is commonly administered with 5-FU to enhance its antitumor effect [74]. Despite many advances in CRC treatment, approximately 20% of new CRC cases are already metastatic [75]. The most common sites of metastatic colorectal cancer (mCRC) are the liver, lungs and peritoneum. Unfortunately, up to 50% of patients with early-stage disease at diagnosis will eventually develop metastatic disease, and 80–90% of them have unresectable metastatic disease because of the size, location, and/or extent of disease [76,77].

Local therapies are approved for mCRC with inoperable lesions. The choice of local therapies depends on the location and the extent of the metastases [78]. For patients with unresectable liver or lung metastases, radiofrequency ablation is recommended for the treatment of small and medium-sized lesions, but for larger lesions and those near vascular structures, microwave ablation or stereotactic body radiation therapy may be good alternatives [1,6]. Liver metastases can also be treated by administering a higher dose of chemotherapy directly into the hepatic artery compared to systemic therapy (hepatic arterial infusion) or by combining drug/radiation administration with blood vessel obstruction (chemo/radio-embolization) [79]. For patients with peritoneal metastases, cytoreductive surgery and hyperthermic intraperitoneal chemotherapy are recommended [6]. Local therapies can be administered with curative or palliative intent and are the most often used in combination with systemic therapy [6,79].

Systemic therapy for CRC aims to downsize the primary tumor or metastases in order to convert them to a resectable status and increase progression-free survival [6]. Patients with advanced CRC usually receive several lines of therapy, most often including a combination of chemotherapy with targeted therapy or immunotherapy, depending on tumor mutational and MMR status [5]. Targeted therapies are recommended for patients with *KRAS/NRAS/BRAF* mutated or wild-type tumors, *HER2*-amplified tumors and *NTRK* gene fusion-positive tumors, while immunotherapy is only offered for tumors with high microsatellite instability (MSI), as shown in Table 4. Thus, both statuses must be determined prior to the start of therapy [80]. Unfortunately, for advanced CRC patients whose overall health has deteriorated despite treatment, palliative treatments and the best supportive care are the only remaining options [5]. Therefore, the clinical management of patients with mCRC represents a unique challenge to balance benefits and harms, including the identification of strategies that improve disease response, limit treatment-associated toxicities, and improve the overall quality of life [81].


**Table 4.** Systemic therapies for localized and advanced colorectal cancer.


**Table 4.** *Cont.*

TP: thymidine phosphorylase; mAb: monoclonal antibody; VEGF: vascular endothelial growth factor; VEGFR: vascular endothelial growth factor receptor; EGFR: epidermal growth factor receptor; PD-1: programmed death cell receptor 1; CTLA4: cytotoxic T-lymphocyteassociated antigen 4; MEK: mitogen-activated kinases; TRK: tropomyosin receptor kinases; MSI: microsatellite instability; NTRK: neurotrophic receptor tyrosine kinase gene.

#### *3.2. Mechanisms of Drug Resistance Associated with Colorectal Cancer Stem Cells*

The effectiveness of current anticancer therapies is limited by the resistance of tumors to chemotherapy and targeted molecular therapies [99]. Resistance to anticancer drugs may be intrinsic, meaning that it occurs prior to treatment and involves pre-existing resistance factors in the mass of tumor cells, or it may be acquired during the treatment of tumors that were initially sensitive due to the induction of various adaptive responses [99]. Furthermore, due to the high degree of tumor heterogeneity, drug resistance may also result from the therapy-induced selection of a drug-resistant tumor subpopulation, such as CCSCs [99]. A wide range of molecular mechanisms are involved in drug resistance, as illustrated in Figure 4, and will be detailed in this chapter [74].

**Figure 4.** Major mechanisms of anticancer drug resistance attributed to CSCs such as changes in drug transport (1); impaired drug metabolism (2); alterations in drug targets (3); enhanced DNA damage repair (4); impaired balance between apoptosis and survival pathways (5); and the role of the tumor microenvironment comprising cellular and non-cellular components (6). CSC: cancer stem cell; DNA: deoxyribonucleic acid.

#### 3.2.1. Changes in Drug Transport

The anticancer activity of a drug can be limited by poor drug influx or excessive efflux, which alters the amount of drug reaching the tumor, as shown in Figure 4 panel 1 [99]. Several transporter proteins, belonging to the superfamilies ABC and solute carrier (SLC), have been linked to anticancer drug resistance by interfering with drug transport [74]. The ABC transporters ABCB1, ABCC1 and ABCG2 play a pivotal role in the efflux of anticancer drugs [100,101]. In colon cancer, ABCB1 may be overexpressed, leading to reduced cellular accumulation of chemotherapy and therefore therapeutic failure, or may be induced by chemotherapy resulting in the acquired development of multidrug resistance [99]. The impact of SLCs on cancer therapy has been less documented, however, some members of the SLC superfamily are also involved in the transport of anticancer drugs [100]. Changes in the expression of SLC transporters, such as the organic cation transporter OCT2 and the organic zwitterion/cation transporters OCTN1, may affect the ability of tumor cells to uptake anticancer drugs and lead to the development of chemoresistance [100]. The Zhang et al. study shows that the overexpression of human OCT2 transporters increases oxaliplatin accumulation and cytotoxicity in colon cancer cell lines [102]. Taken together, efflux and influx transporters may confer resistance to anticancer agents, and the intrinsic drug resistance of CCSCs may be explained by the higher expression of these transporters [99,100,102].

#### 3.2.2. Impaired Drug Metabolism

The efficacy of anticancer drugs may also be affected by changes in their metabolism, such as the production of an inactive metabolite, as highlighted in Figure 4 panel 2 [99]. The inactivation of anticancer drugs may be associated with the overexpression of drug-metabolizing enzymes, such as cytochrome P450-related enzymes (CYP), UDPglucuronosyltransferase (UGT) and glutathione S-transferase (GST) [74]. CYP enzymes play a crucial role in the metabolism of many therapeutic drugs, including SN-38, the active metabolite of irinotecan. Indeed, SN-38 can be inactivated by CYP3A4- and CYP3A5 dependent oxidations that form inactive metabolites [103]. The study by Buck et al. shows a significant correlation between CYP3A5 expression and tumor response to irinotecan therapy in CRC [103]. In addition, increased CYP expression in CSCs appears to be associated with chemoresistance [104]. SN-38 is predominantly eliminated by glucuronidation which is mainly mediated by the polypeptide A1 of the UGT1 family, encoded by the *UGT1A1* gene [105]. However, inter-individual variations in UGT1A1 activity exist and are related to the presence of genetic polymorphisms. For example, patients with UGT1A1\*28/\*28 genotype have a higher risk of developing irinotecan-induced hematological toxicity and require a reduction in irinotecan dose which may impact its anti-cancer effect [105]. The GSTP1 subclass of the GST superfamily is overexpressed in patients with colon cancer and is an important mediator of intrinsic and acquired platinum resistance [106]. Stoehlmacher et al. demonstrated that GSTP1 Ile105Val polymorphism is associated with increased survival in patients with advanced CRC receiving 5-FU/oxaliplatin chemotherapy [106]. Thus, the enhanced ability of tumor cells, particularly CCSCs, to inactivate anti-cancer drugs is mainly due to the overexpression of drug-metabolizing enzymes or polymorphisms [74].

#### 3.2.3. Alterations in Drug Targets

One of the most common mechanisms of resistance to targeted therapy is mediated by alterations in the target protein as suggested in Figure 4 panel 3 [107]. Somatic mutations have been identified in the *KRAS* gene as a biomarker of intrinsic resistance to EGFRtargeting agents in patients with CRC [108]. The Misale et al. study reports for the first time that a substantial fraction of CRC patients who exhibit an initial response to anti-EGFR therapies have, at the time of disease progression, tumors with focal amplification or somatic mutations in *KRAS* which were not detectable prior to therapy initiation [108]. Thus, drug resistance resulting from *KRAS* alterations can be attributed not only to the selection of pre-existing *KRAS* mutant and amplified clones, but also to new mutations resulting from ongoing mutagenesis [108]. The acquisition of mutations in target proteins

also contributes to chemotherapy drug resistance. Irinotecan exerts its cytotoxic activity by inhibiting topoisomerase 1 (TOP1). However, increased TOP1 gene copy number at 20q11.2-q13.1 or mutations in the gene that result in reduced affinity for its active metabolite may be involved in increased drug resistance in CCR [74]. Therefore, the alteration of drug targets primarily due to the acquisition of mutations may result in resistance to targeted therapy and chemotherapy.

#### 3.2.4. Enhanced DNA Damage Repair

Drug resistance can also be explained by an enhanced ability of tumor cells, especially CCSCs, to repair drug-induced DNA damage, as presented in Figure 4 panel 4. The repair of DNA adducts induced by platinum-based chemotherapy, such as oxaliplatin, is primarily mediated by the nucleotide excision repair (NER) pathway [74]. The upregulation of excision repair cross-complementing 1 (ERCC1), a key protein of the NER pathway, has been associated with oxaliplatin resistance in CRC [74]. In addition, the level of intratumoral ERCC1 mRNA expression is a predictive marker of survival in mCRC patients receiving combination chemotherapy with 5-FU and oxaliplatin [109]. Mismatched or wrongly matched nucleotides are corrected by the MMR system, which plays a crucial role in maintaining genome integrity [74]. DNA repair deficiency can be caused by mutations in *MMR* genes, such as *MLH1* and *MSH2,* and can lead to the MSI phenotype [99]. The study by Valeri et al. shows that the microRNA-21 (miR-21) downregulates hMSH2, and miR-21 overexpression reduces the therapeutic efficacy of 5-FU in a CRC xenograft model, suggesting that the downregulation of MSH2 with miR-21 overexpression may be an important indicator of therapeutic efficacy in CRC [110]. Consequently, the defects or upregulation of DNA repair pathways can serve as biomarkers of therapeutic response, and therapeutic effects can be enhanced by combining the inhibition of a DNA damage response pathway with DNA-damaging agents to eradicate CCSCs [111].

#### 3.2.5. Impaired Balance between Apoptosis and Survival Pathways

Resistance to cell death is one of the hallmarks of human cancers that contribute to tumor progression and drug resistance [101]. Cell death by apoptosis is a physiological program controlled by a tightly regulated balance between pro-apoptotic, anti-apoptotic and pro-survival mechanisms [112]. However, this balance is frequently altered in cancer cells and particularly in CCSCs, as shown in Figure 4 panel 5. The tumor suppressor p53, encoded by the *TP53* gene, is essential for the induction of apoptosis in response to chemotherapy [74]. Nevertheless, p53 is found mutated in approximately 85% of CRC cases, and *TP53*-mutated colon cancer cells tend to be more resistant to many anticancer drugs, including 5-FU and oxaliplatin, compared to *TP53* wild-type cells [74,101]. The BCL-2 family, which contains pro- and anti-apoptotic members, plays a crucial role in the regulation of apoptosis. The loss of expression and/or activity of the pro-apoptotic factor BAX can be explained by frameshift mutations in the *BAX* gene and may result in chemoresistance [74]. The study by Nehls et al. suggests a major prognostic impact of BAX, whose protein expression appears to be important for the clinical outcome of 5-FU-based adjuvant chemotherapy in stage III colon cancer [113]. The balance between apoptosis and survival may also be altered by aberrantly overexpressed or overactivated anti-apoptotic factors, such as Bcl-2, Bcl-XL, the inhibitor of apoptosis proteins and the caspase 8 inhibitor FLIP [74,99]. Importantly, alterations in the genes encoding these anti-apoptotic factors have been linked to resistance to chemotherapy and targeted therapy [99]. Finally, the overactivation of several pro-survival signaling pathways, including Notch, Hedgehog, Wnt, Bone morphogenetic proteins, Janus kinase/signal transducers and activators of transcription (JAK/STAT) and nuclear factor-κB pathways, may also be associated with drug resistance [112]. Taken together, the altered balance between apoptosis and survival in cancer cells, and especially in CCSCs, prevents apoptosis even when DNA repair fails, which is another mechanism of resistance to therapy [112].

#### 3.2.6. Role of the Tumor Microenvironment

In recent years, the TME has emerged as a key driver of tumor progression and drug resistance, challenging the development of new therapies in clinical oncology. TME contains both cellular components with cancerous and non-cancerous cells such as stromal myofibroblasts, endothelial cells, immune cells and cancer-associated fibroblasts (CAFs), and non-cellular components including extracellular matrix (ECM), cytokines, growth factors and extracellular vesicles, as illustrated in Figure 4 panel 6 [114]. In the tumor stroma, CAFs secrete the cytokines CXCL1 and CXCL2 as well as the interleukin-6, which promote angiogenesis and tumor progression [46,114]. Vermeulen et al. showed that myofibroblastsecreted factors, in particular hepatocyte growth factor (HGF), enhance Wnt signaling activity in colon cancer cells and can restore the CSC phenotype in more differentiated tumor cells, both in vitro and in vivo [115]. Furthermore, CSCs reside in anatomically specialized regions of the TME, known as the CSC niche, which retain their properties and protect them from anticancer drugs, contributing to their enhanced resistance to treatment [46,114,116]. Importantly, CSCs can also be maintained in a quiescent state with minimum energy consumption and a low proliferation rate to resist therapies [114]. In response to environmental signals such as hypoxia, the niche adapts to ensure optimal conditions for CSC proliferation and differentiation [46]. CSCs may contribute to vessel recruitment during tumorigenesis by secreting angiogenic factors, such as vascular endothelial growth factor (VEGF) and CXCL12, in order to accelerate angiogenesis in endothelial cells, which in turn secrete factors such as nitric oxide and osteopontin to maintain the stemness of CSCs [15]. Hypoxia is a key factor in cancer progression that regulates cell survival, angiogenesis, invasion and metastasis, via hypoxia-inducible factor (HIF) [14,116]. In addition, hypoxia can induce the epithelial-to-mesenchymal transition (EMT) that leads to the dissemination and invasion of tumor cells due to the loss of cell adhesion properties and the acquisition of a mesenchymal phenotype with motility and invasiveness [8,116]. The expression of SNAI1 protein, the main inducer of EMT, has been detected at the tumor–stromal interface in colon cancer [116] and elevated endogenous levels of SNAI1 in cancer cells have been shown to increase tumor initiation capacity and metastatic potential in mouse and human models [8].

#### 3.2.7. Mechanisms of Drug Resistance Associated with CCSCs: Discussion

Several publications point out that one of the main technical issues in the CSC field is the plasticity of CCSCs and tumor cells, which may be involved in drug resistance [117–120]. Using the CRISPR-Cas9 technology, Shimokawa et al. demonstrated that the selective ablation of LGR5<sup>+</sup> CCSCs in human CRC organoids leads to tumor regression in xenografts produced by these organoids [120]. However, after several weeks, tumor regrowth is observed and associated with differentiated tumor cells that dynamically replenish the pool of LGR5<sup>+</sup> CCSCs, indicative of cellular plasticity [120]. Another study confirmed these results using CRC organoids that express the diphtheria toxin receptor under the control of the LGR5 locus to selectively ablate LGR5<sup>+</sup> CCSCs [117]. Importantly, the removal of CCSCs limits primary tumor growth but does not prevent the regrowth of the tumor at the primary tumor site upon discontinuation of treatment due to proliferative LGR5− cells, whereas it does in metastatic lesions [117]. Thus, the authors demonstrated a protective role of selective CSC depletion in primary tumors on the appearance of distant metastases, suggesting an interesting therapeutic perspective for the management of metastatic diseases [117]. The process of cellular plasticity is crucial for the repopulation of impaired SC niches and tissue homeostasis, but its role in the formation of metastases is poorly studied [118]. Using a CRC mouse model and human tumor xenografts, Fumagalli et al. investigated the role of cellular plasticity in metastasis [118]. Surprisingly, the authors show that the majority of disseminated CRC cells in the circulation are LGR5− cancer cells and are capable of forming distant metastases, in which LGR5<sup>+</sup> CSCs subsequently emerge and contribute to long-term metastatic growth [118]. Importantly, microenvironmental factors may enhance cellular plasticity [118]. Thus, cellular plasticity complicates the

development of new therapeutic strategies and the eradication of CCSCs does not appear to be sufficient to completely cure cancer due to the impact of the microenvironment [8]. The heterogeneous and dynamic nature of SCCCs constitutes another obstacle to their targeting. Using a marker-free and quantitative analysis of colon cancer growth dynamics, Lenos et al. showed that cells with CSC functionality are not necessarily the same cells as those expressing CSC markers [121]. Interestingly, the authors demonstrated that all tumor cells have the capacity to fuel tumor growth when placed in an appropriate environment, preferentially at the edge of the tumor close to the CAFs [121,122]. Thus, from the authors' point of view, the stem cell function in established cancers is not intrinsically but entirely spatiotemporally orchestrated, suggesting a major role of the microenvironment [121]. Consequently, cellular plasticity and the microenvironment appear to allow tumors to easily adapt to the loss of key compartments, thereby compromising therapeutic efficacy [122]. Therefore, TME plays a crucial role in primary tumor growth and metastasis formation by protecting CSCs from therapeutic agents and appears to be an important target along with the other resistance mechanisms discussed in this chapter for the development of new therapies [116].

#### **4. Clinical Trials on Colorectal Cancer Stem Cells**

Clinical trials targeting CCSCs are rare. The complexity relies on the identification of molecular targets required to maintain cancer stemness in CSCs, but not or less by normal tissue SCs, to selectively target CSCs [123]. All clinical trials from the National Institute of Health are listed on the ClinicalTrials.gov website [124]. We narrowed our search by focusing on the terms "colorectal cancer" and "cancer stem cells", resulting in the identification of eight intervention studies as of September 30, 2020. Among them, we excluded all clinical trials whose status was withdrawn (N = 1) and terminated (N = 2) and focused on the remaining clinical trials (N = 5). Subsequently, from these five clinical trials, we selected and reviewed the clinical trials on pharmacological agents under investigation (N = 3), as presented in Table 5.


**Table 5.** Clinical trials on colorectal cancer and cancer stem cells from ClinicalTrials.gov.

Napabucasin (BBI608) is the first-in-class cancer stemness inhibitor that targets the STAT3 pathway [123,125]. In a preclinical study, Li et al. showed that BBI608 inhibits the expression of stemness genes and the self-renewal of CSCs and succeeds in depleting CSCs whereas standard chemotherapy leads to the enrichment of these cells [123]. In addition, the authors demonstrated the ability of BBI608 to block both cancer relapse and metastasis in vivo, using a mouse CRC model [123]. These preclinical results provide an interesting approach for the development of new anticancer therapies targeting cancer stemness [123,125]. Several clinical trials were designed prior to the ongoing phase III clinical trial, shown in Table 5. Firstly, a phase I dose-escalation study was conducted in adult patients with advanced malignancies who had failed standard therapies in order to investigate the safety and anti-tumor activity of BBI608 as monotherapy (NCT01775423) [126,127]. BBI608 showed encouraging signs of clinical activity with only mild adverse events observed and an unreached maximum tolerated dose (MTD), suggesting an excellent safety profile of BBI608 at 500 mg twice daily [126,127]. Subsequently, two additional phase Ib/II clinical trials were conducted to determine the safety and anti-tumor activity of BBI608 in combination with panitumumab in *KRAS* wild-type patients with mCRC (NCT01776307) or with FOLFIRI (5-FU/leucovorin/irinotecan) +/− bevacizumab in mCRC (NCT02024607). Both clinical trials showed a high disease control rate (DCR) including patients with partial response, stable disease or tumor regression, which confirms the safety of these combinations with encouraging anti-tumor activity [128–130]. Thereafter, a phase III study was designed to evaluate the efficacy and safety of BBI608 versus placebo with the best supportive care in patients with advanced CRC who had failed all available standard therapy (NCT01830621) [131]. In this trial, BBI608 did not improve overall survival (OS) or progression-free survival (PFS) in unselected patients with advanced CRC but did improve OS in pSTAT3-positive patients compared to the placebo group, suggesting that STAT3 may be an important target for CRC treatment [131,132]. Finally, the ongoing phase III clinical trial aims to assess the efficacy of BBI608 plus FOLFIRI versus FOLFIRI alone in previously treated mCRC patients (N = 1250) (NCT02753127) [133]. Patients are randomized 1:1 in each group and stratified by time to progression to first-line therapy, *RAS* mutation status and primary tumor location [133]. The endpoints of this clinical trial are OS, PFS, DCR and objective response rate in both the general population and p-STAT3-positive subpopulation [133].

Demcizumab (OMP-21M18) is a humanized anti-DLL4 (delta-like ligand 4) antibody that inhibits the Notch pathway and CSC activity through the inhibition of tumor growth and reduction in tumor-initiating cell frequency [134–136]. The study by Ridgway et al. shows that treatment with a DLL4-selective antibody disrupts tumor angiogenesis and inhibits tumor growth in several mouse tumor models [137], these results were confirmed by Hoey et al. using xenograft models of colon tumors [136]. In addition, treatment with anti-human DLL4, alone or in combination with irinotecan, delays tumor recurrence and reduces the frequency of CSCs, as demonstrated by the limiting dilution assay and in vivo tumorigenesis studies [136]. As a result of these preclinical results, several clinical trials were conducted with OMP-21M18. A phase I dose-escalation study was designed to determine the safety, MTD and pharmacokinetics of OMP-21M18 in patients with a previously treated solid tumor for which there is no remaining standard curative therapy (NCT00744562) [138]. In this trial, no more than one dose-limiting toxicity (DLT) was observed at each dose, corresponding to the appearance of side effects severe enough to prevent an increase in the dose of the drug, and the MTD was not reached [138]. OMP-21M18 was generally well tolerated by patients at doses below 5 mg per week and showed anti-tumor activity highlighted by the stabilization of the disease and decrease in tumor size. However, the prolonged administration of this drug was associated with an increased risk of congestive heart failure [138]. Subsequently, a phase Ib study failed to demonstrate enhanced anti-tumor activity of OMP-21M18 in combination with the anti-PD-1 pembrolizumab in patients with advanced or metastatic solid tumors, despite the fact that the combination therapy was well tolerated (NCT02722954) [139]. Finally, a phase I study was conducted to determine the safety and optimal dose of OMP-21M18 in combination

with FOLFIRI in patients with mCRC (N = 32) (NCT01189942). Safety was scheduled to be assessed in each patient group after 56 days of treatment and disease status every 8 weeks. The endpoints of this clinical trial were to determine the MTD of OMP-21M18 plus FOLFIRI and to evaluate the safety, pharmacokinetics and preliminary efficacy of this combination. Unfortunately, to date, no results from this clinical trial have been found, although the actual completion date of the study indicated on ClinicalTrials.gov is February 2011.

Mithramycin A (Mit-A) is an antineoplastic antibiotic agent and a potent inhibitor of specificity protein 1 (SP1) [140]. In various human malignancies, SP1 is overexpressed and contributes to the malignant phenotype by regulating genes involved in proliferation, invasion, metastasis, stemness and chemoresistance [141,142]. The study by Zhao et al. demonstrates that the inhibition of SP1 by Mit-A suppresses the growth of colon CSCs and attenuates their characteristics by significantly reducing the percentage of CD44+/CD166<sup>+</sup> cells in vitro and in vivo [142]. Another study shows that Mit-A inhibits tumor growth and significantly induces cell death and the PARP cleavage of CSC and non-CSC cells [140]. Thus, these preclinical results highlighted Mit-A as a potentially promising drug candidate for the treatment of CRC [140]. Several clinical trials have been conducted to investigate the safety and efficacy of Mit-A in chest cancers, solid tumors and Ewing sarcoma (NCT01624090 and NCT01610570) [143]. Despite the promising preclinical activity of Mit-A in various advanced malignancies, several patients developed severe hepatotoxicity due to the altered expression of hepatocellular bile transporters resulting in the early termination of the clinical trial [144]. The objective of the ongoing phase I/II clinical trial NCT02859415 was to determine the safest dose of Mit-A in patients with chest cancers, including CRC patients, by specifically selecting patients without these alterations. The endpoints of this clinical trial are to evaluate the DLT, MTD, and pharmacokinetics of Mit-A in patients with primary thoracic malignancies or carcinomas, sarcomas or germ cell neoplasms with pleuropulmonary metastases and to determine their overall response rates.

Clinical Trials on CCSCs: Discussion

Thus, the development of therapeutic agents specifically targeting CCSCs is complex, as outlined in this chapter. Unfortunately, despite encouraging preclinical results, the majority of ongoing clinical trials fail to demonstrate relevant results in phase I/II development, which examines the safety of the drug, and does not allow them to proceed to the next phase. In our search of ClinicalTrials.gov, we found only three clinical trials focusing on CSCs and recruiting patients with CRC, underscoring the rarity and complexity of clinical trial design [124]. Among these trials, no study results were found for one of the drugs tested, demcizumab, although the actual completion date of the study has passed [139]. In addition, of the three drugs in clinical trials, two drugs showed drug-related toxicities in the current or previous study. The prolonged administration of demcizumab was associated with an increased risk of congestive heart failure [138] and some patients treated with Mit-A developed severe hepatotoxicity [144]. However, one of these three drugs, napabucasin, has shown interesting results in previous clinical trials and is currently in a phase III study [128–130]. In conclusion, the development of clinical trials can encounter many problems related either to drugs, to patients with unexpected side effects or toxicities, or to the design of the study.

#### **5. Future Perspectives**

The main challenge in preclinical studies is to obtain relevant results that translate into meaningful clinical activity in patients with CRC [134]. Unfortunately, despite interesting preclinical results, many clinical trials fail to demonstrate the benefits of a new pharmacological agent due to the absence of anticancer activity in cancer patients or the presence of toxicities incompatible with the continuation of the trial. The development of new clinical trials must consider the intra- and inter-tumor heterogeneity of CRC patients, which influences their responses to therapies. Nowadays, targeted therapies and immunotherapy have significantly improved the survival of CRC patients, and the newly developed therapies are increasing the therapeutic options for patients with advanced CRC harboring specific

genetic abnormalities [5]. However, despite the initial success of commonly used therapies, most drugs fail to target the MRD associated with CSCs which often leads to relapse in cancer patients. Unfortunately, up to 50% of patients with early-stage CRC at diagnosis will eventually develop metastatic disease, and most of them have unresectable metastatic disease because of the size, location, and/or extent of the disease [76,77]. New clinical trials must therefore be designed to test drugs that could become relevant treatment options for patients with early-stage and advanced CRC. However, the lack of accurate preclinical models that take into account intrinsic and extrinsic characteristic of tumors, such as CSC subpopulation, tumor stroma and TME, is a major technical problem [134]. The CCSC isolation and characterization methods presented in this review highlighted the limitations of the methods currently in use, particularly those using CCSC markers. Cell sorting using phenotypic characteristics allows the sorting of only part of the CCSC population because they are heterogeneous, plastic, and subject to many signals from the TME. Thus, the use of new innovative techniques such as SdFFF which sorts cells according to cell characteristics other than marker expression or the combination of several isolation techniques is crucial. In conclusion, more accurate preclinical models are required because current approaches are not precise enough to identify therapies that may be clinically effective, particularly those targeting CCSCs [145].

#### **6. Conclusions**

Targeting CCSCs holds promise for preventing disease relapse and metastasis in CRC patients. In addition, as a major driving force of drug resistance, CCSCs are attractive potential targets for the treatment of CRC. However, the development of therapeutic agents specifically targeting CCSCs is complex, as highlighted by the clinical trial results presented in this review. Despite the increasing number of therapies, resistance mechanisms may emerge and thus complicate the therapeutic management of patients with CRC. In order to achieve a short- and long-term therapeutic response, the ideal therapeutic strategy should target both the cancer cells of the tumor mass to obtain tumor regression, CCSCs to prevent relapse and metastasis, and TME to limit cellular plasticity and the reappearance of CCSCs.

**Author Contributions:** Writing—original draft preparation, C.H.; writing—review and editing, N.C., S.B. and M.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

