Patient‐derived pancreatic tumour organoid implantation establishes novel pre‐cachexia mouse models

The poor survival of pancreatic cancer patients is largely attributable to cachexia, a syndrome of severe weight and muscle loss. To investigate the aetiology of cancer cachexia, preclinical models that closely recapitulate the human disease process are essential. Patient derived tumour organoids are promising novel cancer models, but their ability to induce cachexia in mice has not been investigated. We developed two pancreatic tumour organoid‐based mouse models and demonstrate their potential for cancer cachexia research.


Introduction
Cachexia is estimated to be the direct cause of death in 20%-30% of all cancer patients and is one of the most severe complications of pancreatic cancer. 1,2 It is a syndrome characterized by unintentional weight and muscle loss 3 that affects more than 80% of pancreatic cancer patients. 4 Cachectic pancreatic cancer patients experience more post-operative complications, 5 more severe chemo-and radiotoxicity, 6 and have a lower overall survival. 7 As the 5 year survival of pancreatic cancer is as low as 9%, 8 it is crucial to increase our understanding of the underlying mechanisms of this deadly syndrome to improve treatment options and quality of life.
The main manifestations of cancer cachexia include severe muscle and fat wasting, systemic inflammation, and impaired physical function. 3,4 Much is still unknown about the cause of cachexia, but it is thought to be a multifactorial process where breakdown of fat and muscle are provoked by systemic inflammation, catabolic factors, reduced food intake, and abnormal metabolism. 4,9,10 Cachexia-instigating factors may be secreted by tumour cells and by immune cells. 4 Some of these factors have been identified in preclinical models such as tumour cell-line based (implantation) models, genetically engineered mouse models (GEMMs), and patient derived xenografts. 11 Although these models have contributed greatly to our understanding of cancer cachexia, they cannot fully explain the heterogeneous symptoms of pancreatic cancer cachexia. 4,12,13 Moreover, in tumour cell-implantation and xenograft models, cachexia occurs quickly, usually around 4-14 days after initial implantation, and tumour burden often approaches 10%-20% of body weight within a couple of weeks. 14,15 This is not consistent with the human disease as pancreatic cancers grow notoriously slow. 11 Furthermore, the excessive weight loss and ulceration of tumours seen in these models 16 are considered humane endpoints and often cause premature termination of animal experiments.
Talbert and colleagues generated a GEMM harbouring Kras +/G12D , Ptf1 +/ER-CRE , and Pten f/f (KPP) mutations, which produces pancreatic tumours upon injection of tamoxifen and imitates the slower pancreatic cancer progression seen in humans. 17 However, these tumours are of murine origin and may not precisely recapitulate human pancreatic cancer. 11,18 The absence of adequate cancer cachexia models may explain the lack of translation of preclinical results into approved anti-cachexia drugs. 4 Consequently, there is a need for new preclinical models that can be used to investigate both the underlying pathophysiologic mechanisms of cachexia that are common to patients as well as the aetiology of the interindividual differences in clinical presentation seen in this disease.
In recent years, a new cell culturing method has been developed which enables a highly efficient and fast growth of neoplastic and metastasized cells from pancreatic cancer into three-dimensional 'mini-organs', called organoids. 19 Organoids have been shown to recapitulate many relevant pathophysiological aspects of pancreatic cancer such as its metastatic potential, genetic driver mutations, and transcriptomic features. 19,20 Moreover, they form ductal structures when transplanted into immunocompromised mice, analogous to the parental tumour, in contrast to conventional two-dimensional pancreatic cancer cell cultures. 20 Organoid models are therefore increasingly used to investigate the pathological processes of pancreatic cancer and have been instrumental in identifying novel genes associated with pancreatic cancer progression. 19,21 Combining detailed cachexia phenotyping of patients with generation of organoids from their pancreatic tumours creates unique opportunities to identify fundamental molecular mechanisms underlying clinical cachexia phenotypes. Therefore, our group has recently generated a human pancreatic cancer organoid biobank that encompasses the whole cachexia spectrum as observed in clinical practice 22 and has shown that pancreatic cancer organoid cultures produce known cachexia-related factors that affect skeletal muscle cell differentiation. 23 This study aimed to investigate the impact of implantation of pancreatic tumour organoids from patients with and without cachexia on body weight and skeletal muscle characteristics in mice.

Ethical approval
This study was conducted according to the institutional guidelines for the care and use of laboratory animals established by the Ethics Committee for Animal Experimentation of the Maastricht University in consent with the European Directive 2010/63/EU for the Protection of Vertebrates Used for Scientific Purposes. Study approval was given by the Central Authority for Scientific Procedures on Animals (2013-053). The collection of data and tissue for the generation of organoids was done in accordance with the guidelines of the European Network of Research Ethics Committees following European, national, and local law (METC 13-4-107).
Tumour organoid-based mouse models of pre-cachexia

Pancreatic cancer organoid culture
We have recently generated a biobank of pancreatic tumour organoids of patients with varying degrees of cachexia. 22 In short, we performed extensive physical screening of patients before surgery, and established organoid cultures from resected tumour tissue according to the protocol of Boj and colleagues. 19 Two organoid cultures were selected based on the cachexia profile of the donor (PANCO-09b organoid from 'cachectic' patient 09 and PANCO-12a organoid from 'non-cachectic' patient 12). These organoids were cultured in basement membrane matrix (BME, Geltrex ™ LDEV-Free Reduced Growth Factor Basement Membrane Matrix, Gibco, Cat. No. A1413202) supplemented with either 'Tumour a'-or 'Tumour b'-organoid medium (see Vaes et al. 22 ), in a humidified 37°C/5% CO 2 incubator. Medium was changed every 2-3 days. Organoids were passaged every 7-10 days. The experiments were performed with passage number below 15.

Animals and organoid xenograft implantation
Twenty female, 9 weeks old, Crl:NMRI-Foxn1 nu mice were purchased from Charles River (strain 639, Wilmington, Den Bosch, the Netherlands) and maintained in SPF housing with environmental enrichment at 26°C with a 12-h light/dark cycle. All animals were born on the same day. Only female mice were used to avoid a potential impact of sex differences on cachexia development. 24 Animals were provided with ad libitum access to water and food (ssniff, cat# v1124-703) and co-housed with five animals per cage. Mice were randomized into three groups to receive PANCO-09b organoids (n = 8), PANCO-12b organoids (n = 8), or BME injection (control) (n = 4), using a random number generator. The experimental groups were divided equally over five cages. Organoids were dissociated into single cells and 20 000 cells were dissolved in 100% Geltrex ™ BME. Fifty microliters of organoid suspension were injected into the flank. Mice in the control group were injected with 50 μL of 100% Geltrex ™ BME from the same batch as the experimental groups without organoid cells. The researchers performing the surgery and weight measurements were blinded for group allocation.

Data and sample collection
Body weight of the mice was monitored every 2-3 days immediately after lights-on by researchers blinded to group allocation. After 38 days, the mice were euthanized using an intraperitoneal injection of ketamine/xylazine. This endpoint was pre-determined according to the expected growth rate of the organoids and the expected timeframe of presentation of (pre-)cachexia. 14 No mice were prematurely removed from the study due to reaching humane endpoints. At the end of the experiment, tumours were resected for histological analysis. Animals without tumour engraftment were excluded from all analyses. Blood was collected via heart puncture in BD microtainer ™ serum tubes (Cat. No. 365957). Blood was processed by centrifugation at 1000× g, 20°C for 10 min and serum was stored at À80. The spleen, stomach, liver, kidney, lungs, retroperitoneum, diaphragm, mesentery, bowel loops, and abdominal wall were investigated for formation of distant metastases. White adipose tissue was resected from the left and right posterior subcutaneous deposit and was flash frozen and stored at À80°C for further analyses. The soleus, gastrocnemius, tibialis anterior (TA), plantaris, and extensor digitorum longus (EDL) muscles were collected from both hind limbs using standardized procedures and weighed in pairs using an analytical balance. 25 After weighing (wet weight), these muscles and the livers were flash frozen and stored at À80°C for further analyses.

Dry weight muscle measurement
Tibialis anterior muscles were frozen in liquid nitrogen immediately after dissection and wet weight measurement. Subsequently, they were subjected to lyophilization in a vacuum chamber for 72 h and weighed before thawing to measure dry muscle weight. Dry weight was analysed and wet/dry weight ratio was calculated.

Histological examination of white adipocyte size
For histological examination, flash frozen white adipose tissue was fixed in 4% formaldehyde at 4°C for 24 h. No thawing was performed prior to incubation in formaldehyde. Subsequently, white adipose tissue was paraffin embedded and five sections of 5 at 100 μm intervals were made of each sample. These five sections were stained using haematoxylin and eosin, and whole sections were scanned using the Aperio ImageScope CS2 (Leica, Amsterdam) at 200× magnification. Five images per section were analysed using Fiji according to Parlee and colleagues and the Adiposoft Fiji plugin. [26][27][28] Immunohistochemistry Tumour sections were stained for human cytokeratin 7 and 8 and alpha-smooth muscle actin (αSMA). Sections (5 μm thickness) were cut, deparaffinized in xylene, and rehydrated in graded alcohols. Endogenous peroxidases were blocked using H 2 O 2 in methanol followed by blocking with bovine-serum albumin (cytokeratin 7/8) or normal goat serum (αSMA). The sections were incubated overnight at 4°C with mouse monoclonal anti-cytokeratin CAM5.

Statistical analysis
Results are expressed as mean ± SEM or median (min to max). Raw data were entered in IBM SPSS 24 for Microsoft Windows® and R (R Core Team). 30 Statistical analyses were performed in SPSS using Spearman's correlation, the nonparametric Mann-Whitney U-test to compare differences between two groups, and the nonparametric Kruskal-Wallis test for three groups, followed by Dunn's post hoc testing with Bonferroni correction. To compare the frequency distribution of adipocyte sizes between groups, a two-way ANOVA followed by Bonferroni post hoc analysis was used, as described in Parlee and colleagues. 26 Runs test for deviation of linearity was used to assess curvilinear trends in weight development (change from baseline) where the intercept was constrained to 0. A P-value of <0.05 was considered statistically significant. Figures were produced using Prism Graphpad (version 6.00 for Windows, La Jolla, CA, USA, www.graphpad.com), R using the package ggplot2, 31 and Adobe Illustrator CC 2017 Tumour organoid-based mouse models of pre-cachexia (Adobe Creative Suite for Windows, San Jose, CA, USA, www.adobe.com).

Organoid engraftments grow in ductal structures and induce a stromal reaction
We selected two organoid lines at opposite ends of the cachexia spectrum from our biobank. 22 PANCO-09b was derived from patient 09 and PANCO-12a from patient 12. Complete patient characteristics, including nutritional and body composition parameters, have been previously reported. 22 According to the international consensus definition of cancer cachexia, 3 patient 09 was considered cachectic, whereas patient 12 was not. Patient 09 had lost 13.4% of body weight in the last 6 months, whereas patient 12 had lost only 1.2%. Analysis of body composition using computed tomograpy scans at the third lumbar level showed that both patients were sarcopenic according to previously published sex-specific cut-off values for skeletal muscle index (i.e., skeletal muscle area normalized for height). 5 Moreover, following published cut-off values for visceral adipose tissue, 5 patient 09 and patient 12 had low visceral adipose tissue (VAT) volumes.
Evaluation of the tumours at the end of the experiment showed that organoids had engrafted successfully in 87.5% of mice implanted with PANCO-09b and in half of the mice implanted with PANCO-12a. Histological examination between the respective patient tumour, organoids, and xenografts showed clear similarities in growth pattern ( Figure 1A). PANCO-12a was a rare adeno-squamous pancreatic cancer and showed a mixture of glandular and squamous differentiation, resembling the parent tumour. Although the squamous component of PANCO-12a was not clearly visible in the xenografts, the adenocarcinomatous component of both organoid cultures and respective mouse xenografts closely resembled those of the parent tumour. This was supported by human cytokeratin 7 and 8 staining, confirming the human epithelial origin of the tumours. Moreover, αSMA staining showed an extensive desmoplastic stromal reaction characteristic of pancreatic cancer ( Figure 1A). There were no signs of macroscopic metastases in lungs, liver, peritoneum, or intestine at the end of the experiment in any of the mice. Average tumour weight at the end of the experiment was low, and although median tumour weight for PANCO-12a tumours was lower, it did not differ significantly between the groups [34.4 ± 25.1 mg for PANCO-09b vs. 32.8 ± 40.2 mg for PANCO-12a ( Figure 1B), P = 0.450].

PANCO-12a organoid engraftments negatively affect body weight
Because the main manifestation of cancer cachexia is weight loss, we weighed the mice every 2-3 days until the end of the experiment. At the day of implantation, day 0, there were no significant differences in body weight between the groups (25.9 ± 1.3 g for controls, 26.1 ± 0.8 g for PANCO-09b, and 26.7 ± 0.9 g for PANCO-12a, P = 0.44) (Figure 2A). The first 4 weeks after implantation, we observed a steady body weight increase to a peak weight of 28.8 ± 1.1 g for controls, 28.1 ± 1.2 g for PANCO-09b, and 27.4 ± 1.2 g for PANCO-12a, P = 0.75. However, from day 28 onwards, mice implanted with PANCO-12a organoids progressively lost body weight with an average  Figure 2B). At the end of the experiment (day 38), mice implanted with PANCO-12a organoids had gained significantly less weight from baseline compared to control mice (0.7 ± 0.6 g or 2.5 ± 1.1% vs. 2.9 ± 1.6 g or 10.1 ± 2.5%, respectively, P = 0.027) ( Figure 2C,D). Linear regression analysis showed that the weights of PANCO-12a deviated significantly from linearity (P = 0.048), but not for PANCO-09b mice (P = 0.167) or SHAM mice (P = 0.9603), indicating that although body weight increased linearly for SHAM and PANCO-09b mice, it did not for PANCO-12a mice. Nevertheless, mice implanted with PANCO-09b organoids consistently gained less weight than controls, although the difference was not statistically significant at any time point (tumour free body weight difference between baseline and end of the experiment: 2.0 ± 1.2 g, P = 0.961 versus controls) (Figure 2C,D). Moreover, tibia length between the three groups did not differ (mean 17.5 ± 0.3 vs. 17.4 ± 0.3 vs. 17.2 ± 0.5 mm, P = 0.84 for controls vs. PANCO-09b vs. PANCO-12a, respectively), illustrating that the lower body Tumour organoid-based mouse models of pre-cachexia weight was not caused by a general growth deficit. Taken together, these data show that implantation of pancreatic tumour organoids can result in body weight loss, the key characteristic of cancer cachexia.

Association between tumour weight and hind leg muscle weight
Since tumour organoid implantation had a negative impact on body weight, we next investigated skeletal muscle mass loss, a central aspect of cancer cachexia. To this end, we collected and weighed the soleus, plantaris, gastrocnemius, TA, and EDL muscles at the end of the experiment. There were no significant differences in wet weights ( Figure 3A), dry weights, or wet/dry weight ratios (Table S2) of these muscles between controls, PANCO-09b, and PANCO-12a implanted mice. Moreover, we did not detect an increase in immune infiltration markers EMR1, encoding the macrophage marker F4/80, and monocyte marker CD68 in gastrocnemius, soleus, or EDL muscle ( Figure S1). However, when we analysed the association between tumour weight and muscle weight to investigate subtle effects of implantation of tumour organoids on muscle ( Figure 3B), soleus muscle weight was significantly negatively correlated with tumour weight (r = À0.62, P = 0.03). Similarly, wet weights of plantaris (r = À0.58, P = 0.062), gastrocnemius (r = À0.41, P = 0.2), TA (r = À0.52, P = 0.10), and EDL muscles were negatively correlated with tumour weight, although the correlations for these muscles were not significant (see Table S3). This suggests that muscle weight might be affected by tumour growth.

Skeletal muscle atrophy related genes are not affected by implantation of pancreatic tumour organoids
Given the observed association between tumour burden and muscle mass in our models, we next investigated key ubiquitin proteasome-related regulators of muscle protein breakdown during atrophy responses, including Atrogin-1, Murf-1, SMART, and MUSA. 9 Expression analysis of Atrogin-1, MuRF-1, MUSA, and SMART showed no significant differences in mice implanted with PANCO-09b or PANCO12a compared with control mice ( Figure 4A). Furthermore, expression of REDD1, an mTORC1 inhibitor that negatively regulates skeletal muscle protein synthesis, was not different between the groups ( Figure 4A).
Additionally, we investigated whether the expression of myosin heavy chain isoforms in the muscles was altered, as changes in functional muscle properties in cachexia are often associated with a shift in myosin isoform expression. 32 Our results showed the expected expression patterns for the three muscle groups analysed ( Figure 4B). The gastrocnemius and EDL are fast-twitch muscles that mainly express myosin heavy chain IIb (MHCIIb), whereas the soleus muscle is a mixed-type muscle expressing mainly myosin heavy chain I (MHCI) and myosin heavy chain IIa (MHCIIa) isoforms, and almost no myosin heavy chain IIb (MHCIIb). 32,33 Except for a down-regulation of myosin heavy chain type IIx (MHCIIx, gene MyH1) in EDL muscle for mice implanted with PANCO-09b organoids, we observed no differences in the expression of myosin isoforms between the three groups ( Figure 4B).

White adipose tissue wasting in PANCO-12a implanted mice
As adipose tissue wasting is one of the primary manifestations of cancer cachexia, we next investigated the effect of tumour organoid implantation on adipocyte size in white adipose tissue. The relative frequency distribution of adipocyte cross sectional areas clearly showed a leftward shift towards smaller adipocytes at the expense of larger adipocytes in PANCO-12a implanted mice compared to both PANCO-09b and SHAM mice (P < 0.0001, see Figure 5A). Moreover, the mean diameter of white adipocytes of PANCO-12a implanted mice was significantly smaller than those of PANCO-09b and SHAM mice (876.2 ± 14.3 vs. 1159 ± 14.0 vs. 1003 ± 13.3 μm, respectively, P < 0.0001, see Figure 5B). Interestingly, PANCO-09b mice had significantly larger adipocyte diameters compared to both PANCO-12a and SHAM mice (P < 0.0001). The smaller size of the adipocytes in PANCO-12a implanted mice relative to SHAM and PANCO-09b mice was also clearly visible in sections of white adipose tissue (see Figure 5C).

No evidence of systemic inflammation or liver inflammation in mice implanted with tumour organoids
Chronic systemic inflammation is an important factor in the aetiology of cancer cachexia in pancreatic cancer. 35 Therefore, we next assessed circulating cytokine concentrations in mouse serum. Overall, systemic levels of the cytokines TNFα, IL-10, and IL-6 were low or undetectable in all mice (see Figure S2). Moreover, expression levels of acute phase proteins pentraxin 2 (APCS) and Alpha-1-acid glycoprotein (Orm1), and inflammatory cytokine TNFα in the liver did not differ between the three groups (see Figure S2). Thus, systemic inflammation is not likely to the cachexia related changes induced by tumour organoid implantation. Tumour organoid-based mouse models of pre-cachexia

Discussion
There is a need for better mouse models of pancreatic cancer cachexia that recapitulate the whole-body encompassing symptoms and natural development seen in pancreatic cancer patients. Here, we present novel models using patient derived organoids of pancreatic cancer that may more accurately represent the underlying pathophysiologic mechanisms of pancreatic cancer cachexia than current pancreatic cancer cachexia models. We show that pancreatic tumour organoids engraft in mice and induce a marked local stromal reaction with clear similarities in growth patterns between the parent tumours, organoids, and organoid engraftments, as previously seen by others. 19,36 Importantly, tumour organoid implantation had a negative effect on body weight development, although wet muscle weight was not affected. Nevertheless, the negative correlation between tumour size and muscle weight indicates that tumour derived factors may affect muscle homeostasis in these organoid models. Moreover, adipocyte size was significantly smaller in organoid implanted mice with weight loss, consistent with fat wasting. Overall, the phenotype of these mice is in line with a state of precachexia.  (Atrogin-1, Murf-1, SMART, and MUSA) and mTORC1 inhibitor REDD1. Data represent mean fold change ± SEM compared with controls. (B) Genes encoding myosin heavy chain isoforms. Data represent mean relative mRNA level normalized for reference genes ± SEM. N = 4-7 mice per group. MHC, myosin heavy chain. *P < 0.05 using Kruskal-Wallis test.
Surprisingly, the phenotypic changes in the mice were not entirely in line with the patient phenotypes, as PANCO-12aimplanted mice lost weight, whereas the corresponding donor patient reported only 1.2% weight loss over the last 6 months and accordingly was not considered cachectic according to the international consensus definition for cancer cachexia. This definition identifies weight loss and skeletal muscle loss as key factors in the cancer cachexia syndrome 3 and uses >2% weight loss as a cut-off for sarcopenic patients. Nevertheless, the 1.2% weight loss combined with the presence of sarcopenia may indicate that this patient suffered from pre-cachexia. In addition, several longitudinal studies found that loss of adipose tissue predicted poorer survival, 6,37 and a definition focused on weight and muscle loss may be too restrictive to accurately represent the cachexia syndrome. Indeed, both patient 09 and patient 12 had low visceral adipose tissue volumes 22 according to the cut-off values described by van Dijk and colleagues. 5 Taken together, this underscores the difficulty of diagnosing cancer cachexia, and the importance of accurately phenotyping donor patients to aid in translating experimental findings to clinically relevant cachexia parameters. It has been demonstrated that the loss of white adipose tissue is an early sign of cachexia or pre-cachexia. For instance, Hetzler and colleagues showed that in an Apc min/+ model of cancer cachexia, fat loss occurred at lower levels of body weight loss and was already apparent in mice with <5% body weight loss, whereas muscle wasting occurred over a longer time and became apparent at body weight losses larger than 5%. 38 Additionally, Tsoli and colleagues showed a reduction in adipocyte size accompanied by increased circulating free fatty acids in C26 tumour bearing mice. 39 Moreover, there is evidence that the initiation of cachexia in female mice may be characterized mainly by a rapid loss of fat tissue, 38 whereas cachexia in male mice is accompanied by a loss of both muscle and fat. 40 It cannot be excluded that the presentation of cachexia in our models is influenced by sex of the mice. In order to investigate the effect of biological sex differences on cachexia development, simultaneous experiments in male and female mice should be performed.
Our data indicate that organoid engraftment models of pancreatic cancer can be of great value for cancer cachexia research. First, tumour organoids can be frozen and reused, potentially indefinitely. Moreover, in comparison with traditional xenograft models, where the success rate of establishing a xenograft directly from a tumour specimen is 40%-60% (initial engraftment rate), we had a 100% initial engraftment rate, as both of our organoids engrafted successfully, with a tumour take rate of 50% and 87.5%, respectively. The combination of the ease of expansion of these organoids to adequate cell numbers for experiments and the high chance of generating xenografts could result in a great increase in the number of phenotypically different pancreatic cancer models available for disease modelling. Secondly, the comparatively slow growth rate of tumours in our organoid engraftment models allows for a larger time window during which cachexia progression can be investigated or during which treatment re-sponses to anti-cachexia compounds can be monitored. There may be both a longer pre-cachectic state during which compounds preventing cachexia can be tested, and, potentially, a longer disease stage as weight loss in PANCO-12a-implanted mice already occurred at a low tumour burden. To increase tumour burden, the number of implanted organoids could be increased to accelerate the induction of cachectic symptoms. Finally, GEMM and cell line implantation models of cancer cachexia have the limitation that experimental results may not be reflective of the human disease: GEMMs are of murine origin and may not recapitulate human pancreatic cancer, and in cell culture models, mutational drift and/or expansion may occur with time in culture. 18 Our laboratory has currently established >25 different organoid cultures from patients with pancreatic cancer with varying degrees of cachexia as assessed by extensive phenotyping. These organoid cultures and corresponding organoid engraftment models can be applied to identify biological processes and gene signatures that are commonly changed in cachexia-inducing tumours such as pancreatic cancer.
Despite showing the potential of tumour organoids as models of pre-cachexia, as described above, our study has several limitations. The impact of tumour organoid implantation on body weight and muscle loss was not as pronounced as expected. This may be explained in several ways. First of all, tumour burden was low at the end of the experiment, with a largest tumour of 93 mg. This is small compared to other mouse models using implantation of cell lines or patient derived xenografts. 14,15 Considering the time frame and the initial tumour cell load injected into the mice, it was expected that tumour burden would be relatively low at the end of the experiment. We aimed for a lower tumour burden since this more accurately mimics human disease: pancreatic cancers are usually comparatively small and notably slow growing. Nevertheless, within this timeframe of 38 days of follow-up, none of the mice reached humane end- points, and follow-up could have been prolonged. The first signs of body weight loss occurred around 31 days, which is comparable to what is seen in genetically engineered cachexia models. For instance, in their GEMM harbouring Kras +/G12D , Ptf1 +/ER-CRE , and Pten f/f (KPP) mutations, Talbert and colleagues observed weight loss approximately 30 days after induction, and did not see muscle loss until 47 days after induction. Furthermore, significant up-regulation of atrogenes did not occur until 60 days after induction. 17 Compared to implantation models using conventional 2D tumour cell lines, this timeframe is much more consistent with human disease. We expect that a similar time frame would be appropriate for organoid implantation models as well.
Second, the location of the engraftment may play an important role in the development of cachexia. Michaelis and colleagues showed that although orthotopic implantation of KPC cells induced both weight loss and muscle wasting, subcutaneous implantation did not result in weight loss or sarcopenia. 41 Another study by Delitto and colleagues showed that implantation of PANC-1 cells to the flanks of mice did not produce significant weight loss or wasting of the TA muscle, whereas orthotopic implantation resulted in 15% weight loss and a 21% reduction in weight of TA muscle compared to controls. 14 This might be explained by the fact that subcutaneous tumours do not accurately recapitulate the tumour microenvironment. Cells in the stroma surrounding the tumour, such as cancer associated fibroblasts, are known to secrete pro-inflammatory cytokines that may promote pancreatic cancer-induced cachexia. 11,42 Furthermore, subcutaneous tumours have a lower potential to metastasize and appear to resemble more benign tumours rather than infiltrative malignancies. 43 In addition, pancreatic exocrine insufficiency may play a role. Because of the subcutaneous location of the tumours in our model, any contribution of pancreatic exocrine insufficiency to the development of cachexia can be excluded. Therefore, the effects on weight loss in the current study must be caused by direct action of tumour-derived factors and/or by indirect mechanisms initiated by these tumour-derived factors. In future studies, the effects of orthotopic implantation of pancreatic tumour organoids on cachexia development should be investigated.
Third, the lack of T-lymphocytes in the NMRI-Nude mice may have important implications for the development of cachexia since cytokines secreted or stimulated by T-cells are thought to be important drivers of cancer induced body weight loss and muscle loss in cancer cachexia. 35 As these cells are not present in our organoid implantation models, the changes in body weight and muscle of the organoid-implanted mice are unlikely to be caused by direct or indirect immune effector functions of T-lymphocytes. Other important immune factors implicated in cachexia are cytokines such as IL-6 released by cells of the innate immune system, tumour cells, or cells in the tumour microenvironment. 44,45 In previous research, we observed that IL-6 was detectable in serum from all pancreatic cancer patients, but only a small subset of patients showed secretion of IL-6 by their respective tumour organoids. 22 Notably, we did not observe any secretion of IL-6 from both PANCO-09b and PANCO-12a organoids. 22 The high levels of IL-6 in patient serum observed in the literature are therefore most likely not caused by direct secretion of IL-6 by pancreatic cancer cells but rather result from other (tumour adjacent) cells such as human cancer associated fibroblasts or mononuclear cells. As NMRI-nude mice do possess a functioning innate immune system, and we observed a strong stromal reaction in our models, cells of these compartments may also play a role in the development of cachexia in the current study. However, the levels of circulating cytokines IL-6, IL-10, and TNFα in tumour organoid implanted mice were too low to be detected and there was no induction of inflammatory gene expression in the livers of these mice. Despite the lack of an increase in inflammatory cytokines, we did observe weight loss, indicating that other factors than these immune factors may play an important role in weight loss in our models.
Finally, the impact of PANCO-09b versus PANCO-12a on body weight was not in line with the cachexia status of patients 09 and 12 who contributed the tumour organoids. Whereas patient 12 was not cachectic according to the consensus definition of cancer cachexia, implantation of PANCO-12a did have a significant impact on body weight development of the mice. Conversely, patient 09 was cachectic yet implantation of PANCO-09b organoids did not majorly affect body weight of the animals. As the criteria used to categorize patients into cachectic versus non-cachectic groups are predominantly based on body weight loss, other factors, such as decreased muscle strength and increased inflammatory factors in blood, that are important aspects in the cancer cachexia syndrome, 3 are undervalued. Indeed, both patients were sarcopenic based on analysis of their skeletal muscle mass using CT scans. This may indicate that although patient PANCO-12 was not cachectic based on a lack of weight loss, important processes of cancer cachexia could still be active, and this patient might have pre-cachexia. In line, there was a negative correlation between tumour size and muscle weight in these models combined. However, we could not detect an up-regulation of ubiquitin proteasome-related regulators of muscle protein breakdown. The ubiquitin proteasome is thought to play an important role in cancer cachexia, and both MuRF-1 and Atrogin-1 have been shown to be up-regulated in animal models of cancer cachexia. 14,17,41 Nonetheless, there is limited evidence for the involvement of these proteasomerelated regulators in human skeletal muscle. Studies in patients with pancreatic cancer have shown contradictory results, where one study showed an increase of proteasome subunits C2 and C5 only in patients with more than 10% weight loss, 46 whereas another study even showed a decrease in expression of MurF in muscle of cachectic patients. 47 The lack of activation of regulators of the ubiquitin proteasome in our models may therefore not exclude the presence of sarcopenia-associated mechanisms in skeletal muscle.
In conclusion, implantation of human pancreatic tumour organoids into mice negatively affected body weight and adipocyte size. Although there was no evidence for an up-regulation of genes associated with the ubiquitin proteasome, the correlation between tumour weight and muscle weight suggests that tumour-derived factors may directly or indirectly promote sarcopenia in both models. In future experiments, we suggest to increase the follow-up time, include more organoid models from patients with varying degrees of cachexia, and investigate the effect of orthotopic implantation on cachexia development, as the tumour microenvironment may play a pivotal role in cancer cachexia. Although the donor patient's cachexia phenotype did not correspond directly with weight loss in these mouse models, the thorough analysis of cachexia parameters of donor patients is essential to aid in the understanding and translation of preclinical research. Implantation of tumour organoids into mice provides a valuable model to investigate processes underlying the heterogeneous presentation of cancer cachexia, and can contribute to the identification of biological processes and gene signatures that are commonly changed in cachexia inducing tumours.  Non-significant using Kruskal-Wallis test. Table S1. primer sequences Table S2. Dry and wet/dry weight ratio of tibialis anterior muscles. Table S3. Correlations between tumour weight and muscle weights.