Abstract
Background: We aimed to investigate the difference in engraftment rates depending on the transplant site for a patient-derived xenograft (PDX) of pancreatic ductal adenocarcinoma (PDAC) and the effects of the microenvironment on engraftment. Materials and Methods: Frozen cancer tissues from PDAC tumors were used, and tumor fragments were directly implanted into the subcutaneous, orthotopic pancreas, peritoneum, and liver of X-linked severe combined immunodeficiency (XSCID) rats. We assessed the success of engraftment in each organ. Additionally, to evaluate the effect of the microenvironment in each organ, we performed immunohistochemical analysis. Results: Subcutaneous transplantation was successful in 8 of 10 PDAC cases (16 of 30 rats). This was a higher rate than for other organ transplants. The vascular endothelial cells in the stroma were replaced with those from rats instead of humans. Vascular endothelial growth factor-A (VEGF-A) and cluster of differentiation-31 (CD31) was significantly more strongly expressed in the subcutaneous transplantation model (VEGF-A: p<0.001, CD31: p=0.0036). Conclusion: The engraftment rate was significantly higher for the subcutaneous PDX model than for the orthotopic pancreatic, peritoneal, and liver PDX models. Blood vessels of the PDX stroma had been replaced by rat-derived vessels instead of the original human vessels, suggesting that angiogenesis in the PDX microenvironment may be a major factor in engraftment.
- Patient-derived xenograft
- pancreatic ductal adenocarcinoma
- microenvironment
- vascular endothelial cells
- engraftment rate
Various animal tumor models and tumor cell lines have been used to analyze the pathogenesis of malignant tumors and develop new therapeutic methods, but these models do not necessarily reflect the clinical pathology of the tumor and are not directly applicable to clinical use. In recent years, a variety of highly immunodeficient animals have been developed, increasing the efficiency of human tumor cell transplantation and facilitating the creation of patient-derived xenograft (PDX) (1, 2).
Since PDXs retain characteristics, such as heterogeneity and gene expression profiles of patient-derived tumors, it is hoped that the advantages of PDXs will promote drug development efficiency and help select drugs based on individual molecular characteristics, paving the way towards precision medicine (3). Various types of PDX models have been established in many institutions in the past. In particular, patient-derived orthotopic xenograft (PDOX) models have been shown to preserve the genetic alterations and histopathological features of primary tumors (4). In recent years, some clinical and preclinical studies have reported anticancer drug sensitivity testing of PDX models of malignancy. However, limitations, such as low engraftment rate, time to establish the model, and high cost, have been a major hindrance.
Subcutaneous and orthotopic grafts are often used to create PDX models, but the engraftment rate of xenografts varies depending on the type of cancer, and the characteristics of the primary tumor, such as tumor grade, histological type, and percentage of tumor cells in the tissue, are important (5, 6). The subcutaneous model is easy to create and has the great advantage of allowing visualization of the formed tumor size measurement. However, when conducting drug sensitivity tests in PDX models of pancreatic cancer, it is uncertain if the subcutaneous model can even mimic the microenvironment of the primary tumor. Since the liver and peritoneum are the preferred metastatic organs for pancreatic cancer in clinical practice, we hypothesized that there is a niche in their microenvironment where pancreatic cancer grafts tend to be accepted by the host. We decided to compare PDX engraftment rates and investigate differences in cancer microenvironments in the subcutaneous, orthotopic pancreas, liver and peritoneal transplantation sites.
Materials and Methods
Patients. Informed written consent was obtained from all patients for this research according to an Institutional Review Board-approved protocol (approval number: 4376). The animal study was approved by the Institutional Animal Care and Use Committee of Osaka City University (approval number: 17027).
Patient-derived primary pancreatic ductal adenocarcinoma (PDAC) samples. Tumor samples (X0) were obtained from PDAC patients who had undergone surgery at our facility between August 2019 and April 2020. The clinical characteristics of the 10 PDAC cases are shown in Table I.
Clinicopathological characteristics of 10 patients with patient-derived xenograft for PDACs.
Cryopreservation and thawing procedure. The resected specimens were washed with sterile phosphate-buffered saline (PBS) twice and cut into 5-mm cubes for implantation. The 5-mm cube samples were transferred into sterile cryotubes containing Cellbanker® 1 plus (Zenoaq, Fukushima, Japan) and subsequently placed in a –80°C freezer. The median cryopreservation duration was 38.7 (range=10-100) days (Table I). For thawing, the cryotubes were placed in a 37°C water bath until they were no longer frozen. Tumor samples were held on ice until implantation.
Surgical transplantation. PDAC tumors were implanted in the subcutaneous, pancreas, peritoneum, and liver of 6- to 8-week-old interleukin 2 receptor γ chain (IL2Rg) knockout XSCID rats (Osaka University) (7). The rats were kept in a sterile environment.
Transplantation of cryopreserved PDAC tumors into 12 rats was performed into 3 rats for each organ (subcutaneous, pancreatic, peritoneal and liver). A total of 120 rats were used for each of the 10 PDAC cases.
The rats were initially anesthetized with 3% isoflurane for induction and then 1%-2% for maintenance. After placing the rats in the supine position, the abdomen was extensively disinfected with 70% ethyl alcohol, and a 2-cm lateral incision was made in the epigastric region. Concerning liver transplantation, using a cotton swab, the left lobe of the liver was maneuvered outside the body and then placed and fixed on gauze. The surface of the liver was incised horizontally using a No. 11 scalpel (AD Surgical, Sunnyvale, CA, USA) blade to form a pocket in the liver parenchyma without cutting any major vessels (5, 8). A 5-mm cube tumor sample was implanted into the liver pocket. The incision site was sealed using an absorbable hemostatic material (SURGICEL®, Johnson and Johnson, New Brunswick, NJ, USA) to stop the bleeding. The left liver lobe was returned to the abdominal cavity. For pancreas transplantation, a 5-mm cube tumor sample was fixed in the pancreatic tail with 5-0 absorbable sutures (PDSPLUS®, Johnson & Johnson). The pancreas was returned to the abdominal cavity. For the peritoneal transplantation, a 2-cm midline incision was made in the lower abdomen and a 5-mm cube tumor sample was fixed to the peritoneum with 5-0 absorbable sutures (PDSPLUS®, Johnson & Johnson). For the subcutaneous transplantation, a 1-cm skin incision was placed on the back of the rat approximately 2 cm from the base of the tail, and a 5-mm cube tumor sample was inserted subcutaneously. The incision was closed in layers with 3-0 absorbable sutures (PDSPLUS®, Johnson & Johnson).
Follow-up schedule and collection of the engrafted tumor. The general condition and survival of rats were observed twice a week, and weight was measured once a week. The rats were sacrificed at 12-13 weeks after implantation. Resected tumors were washed with sterile PBS, and a portion of the tumor was fixed with 4% formalin. Paraffin blocks and 4-μm slide sections were prepared for analyses of the tumor characteristics. The remaining tumors were cryopreserved for analyses of tumor characteristics and for reimplantation.
Histology and immunohistochemistry. For histopathological evaluation, hematoxylin and eosin (H&E) staining of the primary PDAC and engrafted tumor was performed on paraffin sections with a thickness of 4 μm. In the engraftment cases, immunohistochemical staining was performed using the following antibodies: i) Rabbit VEGF-A (1:100, polyclonal antibody, bs-1957R; Beijing Biosynthesis Biotechnology, ii) Rabbit p53 (1:100, polyclonal antibody, bs-0033R; Beijing Biosynthesis Biotechnology, iii) the rCD31 antibody, which is cross reactive only in rat blood vessels: Mouse CD31 (1:100, monoclonal antibody, ab64543; abcam), iv) reactive hCD31 antibody, which is cross reactive only in human blood vessels: Rabbit CD31 (1:100, monoclonal Antibody, ab76533; abcam, v) mouseαSMA (1:100, monoclonal antibody, M0851; Dako).
All antibodies except CD31 are cross reactive to both humans and rats. Each expression level of histological image was evaluated by quantitative analysis using Image J Java 1.8.0_172 (9).
Passage. After successful engraftment of first-generation rats (X1), xenograft tumors were removed 12-13 weeks later. The resected specimens from X1 rats were washed with sterile PBS twice and cut into 5-mm cubes. Approximately 10 pieces were cryopreserved in a –80°C freezer using Cellbanker® 1 plus. Twelve pieces were transplanted in the subcutaneous, pancreas, peritoneum, and liver of second-generation rats (X2) within 2 h of obtaining the tumor, without cryopreservation. The rats were sacrificed at 12-13 weeks after implantation similar to X1 rats. Resected tumors were evaluated using immunohistochemistry.
Statistical analysis. The associations of clinicopathological characteristics with the tumor engraftment success were determined by performing Fisher’s exact test. Continuous data, such as the Immunostaining area and cryopreservation duration, were analyzed by performing the nonparametric Mann–Whitney U-test. Groups were considered to be significantly different when p<0.05. JMP13 (Statistical Discovery, SAS Institute, NC, USA) was used for statistical analyses.
Results
Comparison of engraftment rates by transplant site. Engraftment images of each organ are shown in Figure 1. No metastasis to organs other than the transplanted organs was seen in any of the rats, and there was no perioperative death of rats in this study. Table II shows the clinicopathological characteristics of each PDAC case and the number of subcutaneous, orthotopic pancreatic, peritoneal and liver engraftment cases. Subcutaneous PDX was successfully engrafted into 16 of 30 rats (53.3%) in 8 PDAC cases (80%); pancreatic PDX was successfully engrafted into 10 rats (33.3%) in 5 PDAC cases (50%), and peritoneal PDX was successfully engrafted into 9 rats (30%) in 4 PDAC cases (40%) (Table III). PDX to the liver was engrafted into only one rat (0.033%) in one PDAC case (10%). Subcutaneous transplantation showed significantly higher engraftment compared with liver transplantation (p<0.001). The engraftment rate of X2 was 75% and that of third-generation rats (X3) was 100% (Table IV). The liver transplantation model, which had a low engraftment rate in X1, showed a high engraftment rate by passaging.
Macroscopic findings of each PDX model. (a) Subcutaneous engraftment, (b) pancreatic engraftment, (c) peritoneal engraftment, (d) liver engraftment.
Histology features of the primary PDAC and number of cases xenograft tumors.
Engraftment rate of PDX by transplantation site.
Engraftment after passage.
Comparison of cancer phenotype and microenvironment in different transplantation models. The degree of differentiation of the cancers in each of the X1 and X2 transplantation models was similar to that of the primary tumors, and it was determined that the degree of differentiation was maintained. Furthermore, the positive staining for p53 seen in X0 was also observed in X1 (Figure 2). The morphology of the stroma in each transplant model of X1 and X2 was similar to that of primary tumors (Figure 2). When αSMA staining was performed to compare the expression of cancer-associated fibroblast (CAFs) among the different tumor organs of X1, no significant differences were found among the models (Figure 2). However, the expression level of CAF decreased with passaging, and it was significantly decreased in the passaging from X2 to X3 (p=0.003) (Figure 2 and Figure 3). Then, we examined angiogenesis and angiogenic factors in PDX. The results showed that hCD31, which is cross reactive only in humans, stained X0 stromal vessels but not X1 stromal vessels. In contrast, X1 stromal vessels were stained with the rCD31, which is cross reactive only in the rat (Figure 4). Furthermore, a comparative study of the number of blood vessels in the stroma was conducted. As shown in Figure 4, the subcutaneous PDX model had significantly more blood vessels (16.9%) than the pancreatic orthotopic (5.83%) and peritoneal models (8.4%). Further, the number of blood vessels tended to increase with passaging. In the liver transplantation model, the number of blood vessels increased as X2 and X3 were passed on, and there was a significant difference between X2 and X3 (Liver X2: 15%; Liver X3: 15.2%). Vascular endothelial growth factor-A -A (VEGF-A) staining was 30.6% in the subcutaneous transplantation model, 18.5% in the orthotopic pancreas model, 21% in the peritoneal transplantation model, and 13% in the liver transplantation model, which was significantly higher in the subcutaneous transplantation model than in other organs (p<0.001).
Hematoxylin and eosin and immunohistochemical staining of primary (X0) and xenograft tumors (X1, X2, X3). Histological differentiation of cancer cells in X0 was maintained in X1 and X2. Furthermore, the positive staining for p53 observed in X0 was also observed in X1. αSMA staining showed no change in fiber formation in each organ. However, when X2 was passed on to X3, the amount of fiber formation was significantly decreased (p<0.01). Scale bar=200 μm.
Immunostaining area in PDX stroma (%). (a) CD31, (b) VEGF-A, (c) aSMA. *p<0.05.
Immunohistochemical staining of primary tumor and liver transplantation models. (a) X0 (hCD31; reactive human), (b) liver engraftment PDX (hCD31: reactive human), (c) liver engraftment PDX (rCD31: reactive rat), (d) liver engraftment PDX (VEGF-A).CD31, which is cross reactive in humans in which X0 stromal vessels are stained, did not stain in X1 stromal vessels (b: negative staining). In contrast, X1 stromal vessels are stained with CD31, which is cross reactive in rats (c: positive staining). (arrow: stained vascular endothelial cells). Scale bar=200 μm.
Discussion
In this study, we focused not only on the commonly used subcutaneous PDX models, but also on the orthotopic pancreatic models, liver and peritoneal PDX models as commonly used metastatic organs in pancreatic cancer. We hypothesized that the liver and peritoneum, where most metastases occur in pancreatic cancer, would have a better microenvironment for metastasis. Consequently, it was assumed that a high engraftment rate would be observed in these organs. However, the results were different. The results of the model study can be summarized as follows: 1) the subcutaneous transplantation had a higher engraftment rate when compared with the pancreatic orthotopic, liver, and peritoneal models; 2) the vascular endothelial cells in the stroma were replaced with those from rats instead of humans; 3) the number of blood vessels in the stroma was significantly higher in the subcutaneous model with a higher engraftment rate, and VEGF-A was a factor affecting angiogenesis; 4) the microenvironment changed with the passage, although the character of the tumor did not change.
Additionally, there have been many reports of PDX model experiments using severely immunodeficient mice, but only a few reports using rats (10). This is the first report of a PDX rat model using pancreatic cancer tissue implanted in the four organs. In the past, several reports have described the comparative tumor properties of subcutaneous and orthotopic transplantation models, their engraftment rates, and differences from primary tumors (4, 11, 12). In a comparative study of subcutaneous, peritoneal, and pancreatic models of PDX in pancreatic cancer, Mersedes et al. reported that the engraftment rates of the subcutaneous, peritoneal, and pancreatic models were 69.9%, 57.6%, and 55%, respectively. They reported that the subcutaneous model tended to have a higher survival rate (but no significant difference was found) and that the three PDX models retained similar characteristics in terms of differentiation, fibrogenesis and angiogenesis (13). There are no reports of comparative studies of the engraftment rates in multiple organs, including liver transplantation, or the pursuit of engraftment factors, including the cancer microenvironment.
In the current study, we focused on angiogenesis to investigate the effect of engraftment in each PDX microenvironment. Previous reports have shown that PDX angiogenesis maintained the same characteristics as human tumors, with lower microvessel density compared with transplantation models derived from cultured cell lines. However, whether the angiogenesis is of human or host origin has not been reported, and there are few reports examining the amount of angiogenesis by the site of implantation (14-16). In this study, we investigated the origin of angiogenesis by immunostaining. Unexpectedly, when using hCD31, which stains vessels in the primary tumor, none of the X1 tumor stroma was stained. Further, rCD31, which does not stain primary tumor vessels, did stain the vascular endothelial cells in the X1 tumor. These findings suggest that it is essential that human blood vessels be replaced by rat blood vessels for engraftment to succeed, and angiogenesis is a very important factor for engraftment. In line with this, the subcutaneous PDX model, which had the most angiogenesis, had the highest engraftment rate. Furthermore, each tumor was stained with VEGF-A, an important growth factor that specifically affects vascular endothelial cells in the process of angiogenesis. VEGF signaling is critical for proliferation in many preclinical cancer models, including angiogenesis and tumorigenesis, and the VEGF pathway has been reported to promote PDX formation (17-19). In this study, staining for VEGF-A was significantly higher in the subcutaneous models than in the other models. As shown in Table III and Table IV, the liver transplantation model showed a very low engraftment rate in X1 and a high engraftment rate in X2 and X3. Further, in the liver transplantation model, the expression of VEGF-A and CD31 increased with the increasing passage. This result suggests that there is a close relationship between engraftment rate and angiogenesis.
As one of the phenotypic changes in the PDX microenvironment, CAF expression was altered by passage. H&E staining showed that the histopathological differentiation and characteristics of tumors themselves did not change by passage. Although there were no significant differences in the expression of CAFs at implantation sites, the volume of CAFs was decreased by passage. It has been reported that excessive passaging significantly reduces fibrogenesis, and the rate of tumor formation had increased by reimplantation (13, 20-22). In our study, the engraftment rate was higher for rat-to-rat transplantation than for human-to-rat transplantation, and the time to engraft was shorter in X2 (1.5 months) vs X1 (3 months). We speculated that this finding was a result of decreased tumor and fibrogenesis with progressing passage, which facilitates tumor development (23). However, this also means that creating a PDX model from the primary tumor will change the characteristics of the tumor. PDX is associated with rapid loss of stromal cells and immune cells upon passaging, which affects drug sensitivity (24). Because the characteristics of xenograft tumors differ from those of the primary tumor due to excessive passages, the PDX model requires fewer passages to maintain the same characteristics as the xenograft tumor (25). According to previous reports, the tumor microenvironment plays an important role in the responsiveness of tumors to anticancer drugs (26). However, it would have been desirable to maintain the stromal component and tumor characteristics in X3 and subsequent passages, but as mentioned earlier, the original characteristics of the primary tumor may have been lost through passages, and it may be difficult to test the sensitivity of the primary tumor to anticancer agents.
The limitation of this study is that only 10 PDACs were used. Increasing the number of PDAC cases and using a large number of immunodeficient rats remains a major problem because of the very high cost. Additionally, the cryopreservation period varies, and it is desirable to transplant the tumor within a short period of time after removal (27).
In conclusion, the subcutaneous PDX model had a significantly higher engraftment rate than the orthotopic pancreatic, peritoneal and liver PDX models. Subcutaneous tumors are a useful PDX model for pancreatic cancer. Blood vessels of the PDX stroma are replaced by rat-derived vessels from the original human vessels, suggesting that angiogenesis in the PDX microenvironment may be a major factor in engraftment. Excessive passage altered the tumor microenvironment by reducing PDX stroma, including CAFs.
Acknowledgements
We would like to give special thanks Mr. Nishino and Ms. Sagaki at the laboratory animal center of Osaka City University Graduate School of Medicine for taking care of rats.
Footnotes
Authors’ Contributions
KK, KK and AY conceived this study. KK, GO and RA performed the surgeries. SE performed the animal experiments and analyzed the data. SE, KK, KK and SK were involved in the drafting of the manuscript. KK, KK and SK supervised this study. NT, RT, KN, HS, GO, RA, ST, AY, ST and MY were involved in the critical revision of the manuscript.
Conflicts of Interest
The Authors declare no conflicts of interest for this article.
- Received February 20, 2022.
- Revision received March 19, 2022.
- Accepted March 21, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
