A model of Zebrafish Avatar for co-clinical trials

Animal ‘‘Avatars’’ and co-clinical trials represent an emerging concept for implementing schemes of personalized medicine in oncology. In a co-clinical trial, the cancer cells of the patient tumor are xenotransplanted in the animal Avatar for drug efficacy studies and data collected in the animal trial are used to plan the best drug treatment in the patient trial. Recently, zebrafish has been proposed for implementing Avatar models but the lack of a general criterion for chemotherapy dose conversion from humans to fishes represents a limitation for conducting co-clinical trials. Here, we validate a simple, reliant and cost-effective Avatar model based on the use of zebrafish larvae; by crossing data from safety and efficacy studies, we found a basic formula for the estimation of the dose to be used for running co-clinical trials and we validate it in a clinical study enrolling 24 adult patients with solid cancers (XenoZ, NCT03668418).


INTRODUCTION 48
Precision medicine refers to the approaches for tailoring a medical treatment 49 to the individual characteristics of each patient (1). In particular, the "Mouse 50 Avatar" is an emerging approach of precision medicine in oncology that has 51 recently grown in importance (2); it implicates the xenotransplantation of 52 cancer cells from patient tumor sample in mouse models to use them in drug 53 efficacy studies. Mouse Avatars can be used to run "co-clinical trials" (3). In a 54 co-clinical trial, the patient and murine trials are concurrently conducted and 55 the drug efficacy response of the mouse study provides data to plan the best 56 drug treatment of the patient tumor (4). The advantage of this approach is that 57 each patient has his/her own tumor growing in an in vivo system, thereby 58 allowing the identification of a personalized therapeutic approach. Nowadays, 59 there are companies providing mouse Avatar generation and drug testing 60 services to patients at a cost of tens thousands of dollars (5). The high cost is 61 directly associated to the time-consuming process and the requirement of 62 immunosuppressed strains (6). Unfortunately, this makes Avatars a cutting-  However, in order to move forward in new paradigm of co-clinical trial using 85 zebrafish Avatars, some critical aspects need to be solved. The biggest issue 86 is related to the lack of the "equivalent dose" for translating the chemotherapy 87 dosage used in humans to zebrafish larvae because one cannot apply the 88 interspecies allometric approach for dose conversion from human to animal.

89
The caveat is that chemotherapy drugs have to be administered in the fish 90 water rather than injected as parenteral formulations. Cisplatin), Gem/nab-P (Gemcitabine + nab-Paclitaxel), GEMOX (Gemcitabine 101 + Oxaliplatin), Gemcitabine, 5-Fluorouracil. We found a general criterion for 102 dose equivalence that has been validated on zebrafish Avatar receiving fresh 103 tissue fragments taken from surgical specimens of patients underwent 104 surgical operation for hepato-biliary-pancreatic cancer and gastro-intestinal 105 cancer.

107
Zebrafish safety study 108 Dose-response analysis for the determination of the effects of chemotherapy 109 treatment on larvae was based on the evaluation of the phenotype resulting 110 from the exposure (i.e. normal, aberrant and dead). In particular, we exposed 111 larvae to 10 different chemotherapy treatments (GEM, GEMOX, GEM/nab-P, 112 GEMCIS, 5-FU, FOLFOX, FOLFIRI, FLOT, FOLFOXIRI, ECF, see 113 supplementary tables S1 and S2) for 72 hours, from 48 to 120 hpf ( Figure 1).

114
Chemotherapy treatments induced death and a variety of malformations in 115 larvae, including yolk sac edema, pericardial edema and spine deformation.

116
For all regimens, deviation from phenotype without defect (normal phenotype) 117 increased with the increase of drug concentration. Linear regression analysis 118 showed an excellent relationship between the linear or logarithmic 119 concentration of the chemotherapy drug and the incidence of normal 120 phenotype (R 2 >0.95; p<0.05 for any protocol tested) or the incidence of 121 mortality (R 2 >0.87; p<0.05 for any protocol tested), (Figure 1). For any 122 chemotherapy treatment, the dose that is lethal to 25% of the population 123 (LD25) and the concentration at which 50% of the normal phenotype is 124 inhibited (IC50) was determined ( Figure 2A). Such data were also expressed With EPC defined as human Equivalent Plasma Concentration, given by:

128
Eq.2 = 129 M being the total amount (mg) of chemotherapy administered to humans by 130 the clinicians involved in the present study, V (ml) being the mean volume of 131 human plasma (the EPC value for each regimen is given in table S2).

132
In the present study, we fixed the 75 percentiles of the box plots ( Figure 2B  larvae. According to the toxicity study, we used conversion factors CF>4.6.

171
First, a CF=8 was tested but data showed a statistically significant increase of 172 the DiI-stained area at 1 dpi and 2 dpi for all the regimens, suggesting the 173 inefficacy of chemotherapy treatment at the CF used ( Figure 3A). Therefore, 174 we tested chemotherapy protocols at a higher concentration, corresponding to 175 CF=5. Interestingly, FOLFOXIRI were found to inhibit the increase of the versus time in all cases, which was statistically significant at 2 dpi with respect 229 to the time point 2 hpi for 5 patients of 6 (83%, Figure 5). According to this 230 finding, the measure of the size of the relative stained area at 2 dpi has been 231 fixed as primary measure of the study. Sporadically we also detected cancer 232 cell migration ( Figure S3).  Figure S4). The chemotherapy protocols tested were 5-FU, FOLFOX, 257 FOLFIRI and FOLFOXIRI for colon cancer; GEM, GEMOX, GEM/nab-P and 258 FOLFOXIRI for pancreatic case and FOLFOX, FOLFORI, FLOT and ECF for 259 gastric cancer. We adapted the "Response evaluation criteria in solid tumors 260 (RECIST)" to the fish trial by defining the partial response (PR, at least a 30% 261 decrease in the relative stained area at 2 dpi / 2 hpi, taking as reference the 262 relative stained area at 2 dpi / 2 hpi of the control group) and complete 263 response (CR, at least a 90% decrease in the relative stained area at 2 dpi / 2 264 hpi, taking as reference the relative stained area at 2 dpi / 2 hpi of the control 265 group) ( Figure 6). For patients affected by colon cancer, we observed a PR in 266 62.5% of patients to FOLFOX, FOLFIRI and FOLFOXIRI but a less frequent 267 response (37.5% of patients) to 5-FU; CR was observed only in a limited number of patients (12.5%) and only to FOLFIRI chemotherapy. For patients 269 affected by pancreatic cancer, we observed a PR to GEM/nab-P (58.33 % of 270 patients), GEM (50%), GEMOX (50%), a limited PR to FOLFOXIRI (33.33 %) 271 but we never observed CR for any chemotherapy treatment. For patients 272 operated for gastric cancer, we observed high incidence of PR to FOLFIRI 273 (100% of patients) but low incidence of PR to FOLFOX, FLOT and ECF (25% 274 of patients); we also observed CR to FOLFIRI in one patient of four.

315
In this work, we found a general dose conversion criterion based on the 316 following formula: Where c fish (mg/ml) is the chemotherapy concentration in fish water, M is the 319 total amount (mg) of chemotherapy administered to humans, V (ml) is the 320 volume of human plasma and CF is the conversion factor that we estimated to 321 be CF=5. We estimated this value by matching data collected from the safety 322 and efficacy studies performed in zebrafish. The safety study was performed 323 on WT larvae. The efficacy study was performed on larvae xenotransplanted 324 with human cancer cell line whose response to chemotherapy has been 325 already characterized. Specifically, we found that HCT 116 responded to 326 FOLFOXIRI treatment with higher sensitivity, but not to 5-FU, FOLFOX and  Figure S1B). The survival 363 rate of the xenografted host was acceptable, at both 1 dpi (81%, n=101) and 364 at 2 dpi (68%, n=101). We also detected the capacity of cancer cell 365 extravasion and dissemination in distal tissues ( Figure S3). As the relative 366 area at 2 dpi/2 hpi has been fixed as primary measure of the study, we Groups of 30 embryos (2 dpf) arrayed in multiwell plates were exposed to E3 453 supplemented with 10000 U/ml penicillin and 100 μg/ml streptomycin 454 unmodified (control) and modified with the chemotherapy drug at 35°C for 24h 455 added with increasing concentrations (Tables S1, S2).

456
The drugs were refreshed each day for the three days of treatment plan. 3 457 days after treatment (3 dpt) zebrafish larvae were fixed in 4% 458 paraformaldehyde in PBS at 4°C over night. After that, they were dehydrated 459 with increasing concentration of ethanol, and analyzed by stereo microscope 460 to evaluate the phenotype (normal, death, aberrant).  Statistical analysis 484 We used GraphPad Prism 7 as statistical analysis software. Data analysis 485 was performed by ANOVA, followed by Bonferroni correction or Dunnett's 486 post-hoc test or t-test. Statistical significance was set to 5%.

488
The authors thank Ida Montesanti and Noemi Nardillo for supporting zebrafish