A systematic investigation of the maximum tolerated dose of cytotoxic chemotherapy with and without supportive care in mice

Background Cytotoxic chemotherapeutics form the cornerstone of systemic treatment of many cancers. Patients are dosed at maximum tolerated dose (MTD), which is carefully determined in phase I studies. In contrast, in murine studies, dosages are often based on customary practice or small pilot studies, which often are not well documented. Consequently, research groups need to replicate experiments, resulting in an excess use of animals and highly variable dosages across the literature. In addition, while patients often receive supportive treatments in order to allow dose escalation, mice do not. These issues could affect experimental results and hence clinical translation. Methods To address this, we determined the single-dose MTD in BALB/c and C57BL/6 mice for a range of chemotherapeutics covering the canonical classes, with clinical score and weight as endpoints. Results We found that there was some variation in MTDs between strains and the tolerability of repeated cycles of chemotherapy at MTD was drug-dependent. We also demonstrate that dexamethasone reduces chemotherapy-induced weight loss in mice. Conclusion These data form a resource for future studies using chemotherapy in mice, increasing comparability between studies, reducing the number of mice needed for dose optimisation experiments and potentially improving translation to the clinic. Electronic supplementary material The online version of this article (10.1186/s12885-017-3677-7) contains supplementary material, which is available to authorized users.


Background
Cytotoxic chemotherapy still forms the basis of systemic therapy for many cancers. Treatment plans typically consist of repeated cycles of chemotherapy at as high a dose as possible, without causing unacceptable toxicity, the maximum tolerated dose (MTD). Maximum drug doses are determined in dose escalating phase I clinical trials until reaching an appropriate balance between efficacy and toxicity [1]. In contrast, preclinical studies usually employ doses based on convention within a research group, on published studies that may or may not have reported optimisation experiments or on quick optimisation steps in which the methods are varied [2,3]. In addition, while in clinical studies further dose escalation is allowed by extensive supportive care measures such as intravenous hydration, anti-emetics, antihistamines and corticosteroids; this is usually lacking in animal studies [2]. Together, this may result in both the use of subtherapeutic dosages and excess use of animals.
There is a clear dose-response relationship between chemotherapy and tumour regression in preclinical studies [4] and in the clinical setting [5]. Furthermore, there are many secondary antitumour effects that depend on chemotherapy dose, for example immune stimulatory potential of dendritic cells [6]; production of IL-17 by peripheral blood and splenic CD4 + T cells [7]; antiangiogenic effects [8,9] or depletion of regulatory T cells [10][11][12][13]. Therefore, the dose used in preclinical studies may significantly affect translation into clinical trials [14]. Human MTDs are often well predicted by animal studies. A meta-analysis of the preclinical and subsequent clinical development phases of 25 cancer drugs showed that rodent toxicology generally provided a safe and reliable way of assessing starting dosages in humans and adequately predicted potential side effects [15]. It is reasonable then that preclinical studies should make use of MTD regimes. However, to our knowledge, there has not been a systematic study done to determine the MTD for chemotherapeutics from each of the canonical classes. As part of the clinical development pathway LD50 values (median lethal dose) have often been determined, giving some indication of where the MTD will be. However, many of these studies were done decades ago in mouse strains that are often no longer used in cancer research, while the tolerability to chemotherapy varies considerably between mouse strains [16][17][18]. This compromises the extrapolation of those MTDs to currently standard mouse strains. We therefore aimed to create a murine cancer chemotherapy MTD resource in BALB/c and C57BL/6 mice, which would both reduce individual dose optimising investigations and allow standardized dosing strategies in preclinical cancer research. We chose weight loss and clinical score (Additional file 1: Table S2) as endpoints, because in previous murine studies weight loss was by far the most common dose-limiting toxicity (81%), followed by clinical signs, such as neurotoxicity or diarrhoea [15]. This allowed us a straightforward way of assessing toxicity that we think is universally relevant. We also tested whether the single-dose MTD could be readily extrapolated to repeated cycles and whether there were differences in MTDs between mouse strains. Lastly, because the corticosteroid dexamethasone and the 5-hydroxytryptamine 3 (5-HT 3 ) receptor antagonist ondansetron are commonly administered in conjunction with chemotherapy to reduce nausea and anorexia in patients, we determined the effect of these drugs on the MTD in mice.

Mice
Female BALB/c and C57Bl/6 J mice were obtained from the Animal Resources Centre (Murdoch, Western Australia) and housed under specific pathogen free (SPF) conditions (M-block Animal Facility and Harry Perkins Bioresources Facility, Queen Elizabeth II Medical Centre, The University of Western Australia). Mice were between 8 and 10 weeks of age for these studies. All experiments were conducted according to the University of Western Australia Animal Ethics Committee approvals (Protocols RA/3/100/1139, RA/3/ 100/1217) and the Harry Perkins Institute for Medical Research Animal Ethics Committee (AEC029-2015) and the code of conduct of the National Health and Medical Research Council of Australia. Mice were fed rat and mouse cubes (Speciality Feeds, Perth, WA Australia) and housed on aspenchipsAB3 bedding (Datesand, Manchester UK). The animal facility temperature was kept between 21°C and 22°C.

Chemotherapy dosing
The following chemotherapies were used in these studies: 5-fluorouracil (5-FU), bleomycin, cisplatin, cyclophosphamide (CY), docetaxel, doxorubicin, etoposide phosphate, gemcitabine, irinotecan, vinorelbine and were obtained from the pharmacy department at Sir Charles Gardiner Hospital, Perth, Australia. Further details are available in Additional file 1: Table S1. All mice were dosed intraperitoneally (i.p) using a 29G insulin syringe. Chemotherapy was prepared and diluted under sterile conditions in either phosphate buffered saline (PBS) or 0.9% sodium chloride as per manufacturer's instructions. Where possible, chemotherapy was made to a dilution whereby a 20 g mouse would receive a 100 μl i.p injection.

Determination of MTD
To determine the MTD, we used two endpoints: weight loss and clinical score. Clinical signs [15,19] were scored by observing activity, appearance and body condition with a maximum of 2 points going to each (0, normal; 1 slight deviation from normal; 2, moderate deviation from normal, Additional file 1: Table S2). The starting doses were based on literature review, taking a dose that was reportedly safe to administer in one or more publications. Doses were escalated incrementally in steps of not more than 50% of the original dose until any mice met the primary endpoint of either >15% weight loss or reached a clinical score > 2. When either of these endpoints was met, dose escalation was ceased and the prior dose was set as the MTD. Euthanasia criteria included weight loss ≥20% or clinical score ≥ 3, and in such cases the previous identified safe dose was set as the MTD. Mice were monitored daily until both weight and clinical condition returned to baseline.

Repeated cycles of chemotherapy
To determine the effect of repeated chemotherapy dosing, BALB/c mice were dosed i.p. with either 4 mg/ kg or 6 mg/kg cisplatin or 10 mg/kg vinorelbine. Once mice had recovered to 100% of their starting weight or a clinical score of 0, a second MTD was given. Dosing was repeated for a total of 3 cycles of dosing.
Effect of supportive care on chemotherapy-induced weight-loss BALB/c mice were dosed i.p. with cisplatin at MTD in combination with dexamethasone (DBL dexamethasone sodium phosphate, Hospira, Mulgrave VIC Australia), ondansetron (Ondansetron-Claris, Claris, Burwood NSW Australia) or both. Dexamethasone and ondansetron were diluted with sterile water for injection and dosed on the day of chemotherapy administration and for 3 days post chemotherapy. Dose titrations were conducted for dexamethasone while ondansetron was dosed at the commonly reported dose of 1 mg/kg in mice [20].

Statistical analysis
To determine a difference of 15% loss from starting weight, with a standard deviation of 5%, assuming an α of 0.05 and P 0.80, the sample size was calculated to be three mice per group using a paired t-test analysis.
Statistical significance of the weight loss nadir in mice treated with chemotherapy with or without supportive care was determined using a student's t-test.

Determination of MTD of selected chemotherapeutics
We first determined the MTD of a range of chemotherapeutics from each class in BALB/c mice. We found a clear correlation between dose and weight loss and/or clinical score for all chemotherapeutics (Fig. 1). The established MTDs for a single dose were: 5-FU 125 mg/ kg, bleomycin 30 mg/kg, cisplatin 6 mg/kg, cyclophosphamide 300 mg/kg, docetaxel 130 mg/kg, doxorubicin 7.5 mg/kg, etoposide 75 mg/kg, gemcitabine 700 mg/kg, irinotecan 240 mg/kg and vinorelbine 10 mg/kg. These dosages along with those commonly reported in the literature can be found in Table 1. Weight loss profiles differed between chemotherapeutics. The majority of drugs caused acute weight loss that returned to baseline within 10 days, with the exception of doxorubicin at 10 mg/kg, which led to an extended period (46 days) of weight loss. While this was not more than our endpoint of 15%, the fact that it did not return to baseline led us to set the previous dose of 7.5 mg/kg as the MTD. While weight loss was the primary dose-limiting toxicity for most chemotherapeutics, 5-FU, etoposide, gemcitabine and irinotecan endpoints were based upon clinical score. With gemcitabine, for example, mice became lethargic within minutes of administration. Although the clinical score improved after several hours, at single doses above 700 mg/kg, it remained at or above 3 for an extended period and so this dose was determined as the MTD.

Reproducibility of MTD between mouse strains
To assess the reproducibility of the MTD across mouse strains, we repeated dosing of two chemotherapeutics from distinct classes (vinca alkaloids and platinum-based compounds) in the C57BL/6 J strain (Fig. 2). We found similar weight loss in C57BL/6 J mice for vinorelbine at the BALB/c MTD of 10 mg/kg (Fig. 2a). Cisplatin at the BALB/c MTD of 6 mg/kg showed slightly less severe weight loss in the C57BL/6 J strain (Fig. 2b). However, when cisplatin dose was increased to 8 mg/kg (Fig. 2c), one of the three mice exceeded the 20% weight loss cutoff. The MTD for both tested chemotherapeutics was therefore similar for both strains.

Repeated cycles of MTD chemotherapy
Since patients receive multiple courses of chemotherapy in the clinic and because toxicity to chemotherapeutics can be cumulative, we determined the effect of repeated dosing of chemotherapy at MTD. We gave 3 cycles at the single-dose MTD of 6 mg/kg for cisplatin and 10 mg/kg for vinorelbine. Vinorelbine was well tolerated at 10 mg/kg without any cumulative weight loss or deterioration in clinical score after repeated cycles (Fig. 3a). With repeat cisplatin administration at the single dose MTD, weight loss exceeded 20% in the second dosing cycle, (Fig. 3b). A lower dose of 4 mg/kg cisplatin was well tolerated and 3 cycles could be given without additional weight loss (Fig. 3c).

Effect of supportive care on chemotherapy-induced weight loss
In the clinical setting extensive supportive care measures are used so that higher doses can be tolerated. We therefore investigated the effect of supportive care with dexamethasone and ondansetron on both weight loss and clinical score of mice (Fig. 4). Ondansetron is a serotonin 5-HT 3 receptor antagonist that is used as an antiemetic to prevent chemotherapy-induced nausea and vomiting. Ondansetron was not effective in reducing weight loss induced by cisplatin (Fig. 4a). Dexamethasone, a corticosteroid also used in patients as an antiemetic, was dose titrated and showed that higher dosages led to greater weight loss than cisplatin alone, however, doses below 1 mg/kg were effective at countering chemo induced weight loss with 0.2 mg/kg the most effective dose (Fig. 4b). In patients, ondansetron is often combined with dexamethasone to reduce nausea and vomiting, even in situations where ondansetron alone is not effective [21]. We tested the combination in mice and found that the benefit of dexamethasone was lost when ondansetron was added into the treatment regimen (Fig. 4c).

Discussion
Chemotherapy administration to patients is governed by strict guidelines relating to dosage and scheduling as determined in dose-optimising phase I studies. Yet these same standards are often not applied to in vivo preclinical studies [2,3]. When we searched the literature for chemotherapy dosages in order to inform related studies, we found that unlike the clinical situation, dosages varied widely between studies. For example, doxorubicin is reportedly used at dosages varying between 2 and 3 mg/kg [22,23] and 10-12 mg/kg [24,25]. In addition, in some studies, doxorubicin is given intratumourally [26][27][28], which may provide interesting mechanistic information, but does compromise translatability. A further discrepancy is with the definition of low and high dose chemotherapy. Low-dose cyclophosphamide has often been studied in regards to its capacity to deplete regulatory T cell (Treg) [29]. However, the concentrations used are quite varied, with doses ranging from 30 to 200 mg/kg, even though it has been shown that specific depletion of Tregs occurs at 20 mg/kg but not at 200 mg/kg [30]. Thus this unclear definition of 'low-dose' could lead to misinterpretations of preclinical findings, and thus hamper translation of these studies into the clinic.
This study was undertaken to provide some guidance on the MTD of chemotherapy in mice for future studies. We chose to use practical measurements that can be applied easily to any research setting, and which have been validated in previous studies [15]. By following this strategy, we found the MTD of some drugs to be quite different from commonly used dosages in the literature. The largest differences were seen with docetaxel and gemcitabine and to a lesser extent with 5-FU and irinotecan. Docetaxel had an MTD of 130 mg/kg, far higher than the commonly used dose of 16 to 33 mg/kg found in the literature [31][32][33]. The MTD of gemcitabine was 700 mg/kg, five-fold higher than those used by others (and our own group previously [34]) with the most common dose being 120 mg/kg [35][36][37]. It should also be noted however that many similar MTDs were found between this study and others. Cisplatin is commonly dosed at 5-6 mg/kg [34,38,39], which is concordant with the MTD of 6 mg/kg found in this study. Similarly, vinorelbine is often administered at 10 mg/kg [40][41][42], the same dose reported here.
We anticipate that other research groups may want to use the MTDs described here as a starting point for their own studies, potentially reducing the number of animals needed to optimize protocols. However, there are some limitations to our studies. Firstly, although we found that the MTD for both cisplatin and vinorelbine were similar for C57BL/6 J and BALB/c mice, the BALB/ c mice did show slightly more weight loss for the tested chemotherapeutics. This suggests that care should be taken when transposing dosages between strains, including immunodeficient strains. Secondly, we determined the MTD when given as a single dose. Although we found that multiple cycles of vinorelbine at single-dose MTD was very well tolerated, this was not the case for cisplatin. A dose reduction was needed to maintain the weights of the animals when giving multiple dosages. This suggests that some further optimization steps will be needed when using the dosages as described, depending on the required scheduling regimen and the mouse  strain used and possibly the emetogenicity of the chemotherapeutic [43]. Furthermore, small differences between research centres may affect either the response to or toxicity from chemotherapy. These include temperature [44], time of dosing [45], sex [46], microbiome [47], and age [48] which should all be considered before beginning experimental studies.
Of interest, some of the MTDs that we determined are very close to or sometimes even over the reported LD50 [49]. Explanations for this could be related to the abovementioned factors, and particularly with the mouse strain used for determining the LD50 [49]. For example, the LD50 for irinotecan was originally determined in ICR mice, whereas we used BALB/c mice, which also in the primary paper reportedly could tolerate higher dosages of irinotecan, similar to C57BL/6 mice [16]. Similarly, we found that cisplatin could be safely dosed at 6 mg/kg, and that 1/3 mice lost more than 20% weight after 8 mg/kg, while the reported LD50 of cisplatin is 6.6 mg/kg [49]. However, this LD50 is based on studies in DBA mice [50]; in other strains, dosages as high as 18 mg/kg have been reported [51]. These data underscore the importance of considering mouse background when interpreting preclinical chemotherapy dosages.
The final aim of this study was to investigate the effect of two common supportive care agents, ondansetron and dexamethasone. Both drugs are used in cancer patients to reduce chemotherapy and radiotherapyinduced nausea and vomiting, with dexamethasone also used to maintain weight in some circumstances [52,53]. The anti-emetic effect of dexamethasone, a synthetic glucocorticoid, is not well understood, although several mechanisms have been put forward, such as antiinflammatory effects, normalisation of the hypothalamicpituitary-adrenal axis, and effects on serotonin [54]. Indeed, we found that also in mice dexamethasone showed a dose-dependent effect on chemotherapy-induced weight loss, with the optimum dosage at 0.2 mg/kg, which is within the dose range that is used in patients in this context [55]. Ondansetron is a serotonin 5-HT 3 receptor antagonist used for the treatment of chemotherapyinduced nausea [56]. Chemotherapeutics induce the release of serotonin in the small intestine, which binds 5-HT 3 receptors and induces emesis. Ondansetron outcompetes serotonin, preventing receptor binding and therefore acting as an effective anti-emetic [57]. It is used to prevent nausea and vomiting after chemotherapy or radiotherapy in humans, but also in company animals such as dogs and cats [58]. Two 5- Fig. 2 Maximum tolerated doses in C57BL/6 J mice. Individual body weights as a percentage of original for C57BL/6 J mice (n = 3 per group) treated with (a) vinorelbine 10 mg/kg, (b) cisplatin 6 mg/kg or cisplatin 8 mg/kg (c) Fig. 3 Effect of repeated cycles of chemotherapy given at MTD. Individual body weights as a percentage of starting weight for BALB/c mice (n = 3 per group) treated with repeated cycles of (a) vinorelbine 10 mg/kg, (b) cisplatin 6 mg/kg and (c) cisplatin 4 mg/kg. Doses were repeated when mice recovered to 100% of their starting weight. Arrows indicate dosing of respective chemotherapeutics. For vinorelbine this was day 0, 7, 14; for cisplatin 6 mg/kg this was day 0, 9; for cisplatin 4 mg/kg this was day 0, 8,16 HT 3 subunits have been identified in mice, namely A and B subunits while in humans there are five, A-E [59]. The primary binding of ondansetron to 5-HT 3 is via the subunit A receptor [60] and so this provided some rational for investigation in our supportive care studies. However, our results show that ondansetron alone is ineffective in reducing cisplatin-induced weight loss or improving clinical condition in mice. This is perhaps not completely unexpected, as ondansetron primarily regulates the vomit reflex along with nausea [55]. Since rodents are not able to vomit, it might be expected that only an improved appetite, not decreased vomiting would result in reduced weight loss. However, surprisingly, we found that ondansetron abolished the beneficial effect of dexamethasone on preventing chemotherapy-induced weight loss in mice. This is striking since in the clinical setting it is common practice to combine ondansetron with dexamethasone as this combination is better at emetic control than ondansetron alone [21]. Our data suggest that this combination should not be used in mice for this indication.

Conclusion
Together, these data constitute a resource for other researchers investigating cytotoxic chemotherapy in mice, using the identified MTDs as a starting point for their studies.

Additional file
Additional file 1: Additional information regarding the clinical severity score used to determine wellbeing of the mice and a comprehensive list of the chemotherapeutic drugs used in this study. (PDF 82 kb)

Availability of data and materials
The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.
Authors' contributions WJA, WJL and RAL and developed the experimental strategy and designed the experiments. WJA and DEH conducted the experiments. WJA, WJL and RAL analysed and interpreted the data and performed statistical analysis. DEH, BWR and AN contributed to discussions, Fig. 4 Effect of supportive treatment on cisplatin-induced weight loss. Individual maximum body weight loss as compared to starting weight for mice treated with cisplatin at MTD with or without ondansetron (a). Maximum weight loss shown for cisplatin 6 mg/kg plus dexamethasone at varied dosages (b) and cisplatin 6 mg/kg plus ondansetron 1 mg/kg and dexamethasone 0.2 mg/kg (c). Depicted are mean weight loss with SEM, n = 3 for all groups. **p < 0.01 interpretation of the data and revision of the manuscript. WJA, WJL and