An embryo-fetal survival and development study was conducted to augment the toxicity database for alpha-glycosyl isoquercitrin (AGIQ), a generally recognized as safe (GRAS) additive and flavor in food and beverages. In Phase I, 24 naturally mated New Zealand white (NZW) female rabbits per group were administered AGIQ by oral gavage at 0, 250, 500, or 1000 mg/kg/day once daily during gestation days 6–28, followed by necropsy. There was no evidence of maternal or fetal toxicity except for equivocal findings of unilateral absent kidney and ureter in one and two unrelated fetuses at 500 and 1000 mg/kg/day, respectively. To more thoroughly assess fetal kidney/ureter development, in Phase II groups of time mated NZW rabbits were administered AGIQ at 0, 500, or 1000 mg/kg/day, under the same conditions as Phase I. No occurrences of absent kidney/ureter were noted in the AGIQ-treated Phase II dams or fetuses; although, one control fetus had unilateral missing kidney/ureter. Given the lack of reproducibility following treatment with AGIQ in Phase II using 48 animals per group, the missing kidney/ureter observations in Phase I were considered unrelated to treatment. Since oral gavage administration of AGIQ to pregnant female NZW rabbits at dose levels of 250, 500, or 1000 mg/kg/day was well-tolerated with no adverse treatment-related effects on the maternal animal, pregnancy, or the developing conceptus, the no-observed-adverse-effect-level (NOAEL) for maternal toxicity and embryo-fetal survival, growth, and development was 1000 mg/kg/day.
Introduction
Alpha-glycosyl isoquercitrin (AGIQ) (Figure 1) is an enzymatically modified form of the natural flavonol isoquercitrin (quercetin-3-O-ß D-glucoside), derived from rutin and used in Japan as an additive or flavor ingredient in various beverages and foods. Although quercetin and its glycosides have demonstrated anti-inflammatory, pain-reducing, and cardioprotective properties1–6 and therefore have been promoted as antioxidant dietary supplements to consumers, their poor miscibility in water and limited absorption have hindered their broad application to the food and beverage industry.7 However, the enzymatic modification resulting in AGIQ has been shown to enhance the solubility and bioavailability of isoquercitrin.8,9
Figure 1. Chemical formula of alpha-glycosyl isoquercitrin.
Previous toxicity assessments have shown that AGIQ is safe, non-carcinogenic, and non-genotoxic,5,7,10–13 but some of the earlier studies used incompletely characterized AGIQ and/or did not adhere to current Good Laboratory Practice (GLP).6,10,12,13 Due to the current effort to expand the use of AGIQ in the consumer food and beverage industry, recent GLP-compliant studies of high-purity AGIQ (including comprehensive genotoxic assessment,7 10-day and 4-week studies in preweaning Göttingen minipigs,14 and a 90-day study in rats)5 were conducted to augment the older toxicity database. These studies have generally confirmed the safety profile of AGIQ.
The current study using highly purified AGIQ and New Zealand White (NZW) rabbits was conducted to assess the potential effects of the compound on embryo-fetal development, growth, and survival in a non-rodent species for registration purposes. The objective of the initial dose-response phase of the study (Phase I) was to assess the potential of AGIQ to induce prenatal developmental toxicity after maternal exposure via oral gavage during the critical period of organogenesis and fetal development. A subsequent investigational phase (Phase II) was conducted using 48 animals in each group to verify equivocal Phase I findings of unilateral absent fetal kidney/ureter in one and two (unrelated) fetuses at the mid and high dose levels, respectively. This study provides a base of non-rodent developmental toxicity data obtained in accordance with current developmental toxicity testing guidelines (Phase I)15–17 and GLP (both phases)18–21 for human risk assessment of orally administered AGIQ.
Materials and methods
Test article
AGIQ (>97% pure, 0.13% quercetin, lot no. 170727) was supplied as a yellow to yellow-orange powder by San-Ei Gen F.F.I., Inc., Osaka, Japan.
Dosing formulations were prepared by dissolving AGIQ in purified water at the target dose concentrations of 25, 50, and 100 mg/mL. Representative samples of each formulation concentration prepared for administration during each phase were analyzed on two occasions for achieved concentration of AGIQ. A previous GLP-compliant validation study (Envigo Study No. NL85VW, 21 June 2018; data not presented) found that solutions of AGIQ in purified water at concentrations ranging from 1 to 200 mg/mL were homogeneous and stable for 1 day when stored at ambient temperature (15–25°C) and 15 days when stored refrigerated (2–8°C). Based on these results, the formulations in this study were prepared daily, stored at ambient temperature until use, and administered within 4 h of preparation.
Animal husbandry and mating
Female NZW rabbits (obtained from Envigo RMS UK) were used for this research. For Phase I, 96 sexually mature, virgin females uniquely identified by an ear tag were supplied (88 assigned to study and 8 serving as potential replacements). For Phase II, 144 time-mated females (all assigned to study) were supplied in four deliveries on GD 1 after mating at the supplier. Phase II animals were uniquely identified by a microchip inserted shortly after arrival. Each cage label was color-coded according to group and was numbered uniquely with cage number, study number, and the identity of the occupant.
The rabbits were allowed a 19-day (Phase I) or 5-day (Phase II) period of acclimation to the facility conditions prior to allocation to the study (mating in Phase I [GD 0] or day of arrival [GD 1] in Phase II). Female rabbits were cohabited with stock NZW bucks of established fertility at the performing laboratory (Phase I) or the supplier (Phase II). Males and females were not closely related. Phase I females were assigned to group and cage position in the sequence of observed natural mating (females mating on the same day were evenly distributed among the groups) and Phase II females were randomly assigned to group and cage position, evenly distributed among the groups. After mating, each female was injected intravenously with 25 i.u. luteinizing hormone. The day of observed mating was designated GD 0. Allocation was controlled to prevent any stock male from providing more than one mated female in each treated group and to prevent more than one sibling female in each group, where possible.
The animals were 19–23 (Phase I) or 16–20 (Phase II) weeks of age at the start of the study (GD 0/1). The basal diet, Teklad 2930 Diet, was restricted to 150 g/animal/day during acclimation up to 1 week prior to the onset of mating (Phase I) and 200 g/animal/day thereafter (Phase I and II). In addition to the basal diet, a small supplement of autoclaved hay was given on a daily basis to promote gastric motility and a small amount of chopped fresh vegetables were given twice weekly. Consumption of hay and vegetables were monitored qualitatively but not quantitatively. The animals were allowed unrestricted access to potable drinking water from the public supply, supplied via polycarbonate water bottles equipped with sipper tubes and supplementary water bowls in each cage, throughout the study. Water bottles/bowls were changed at appropriate intervals. All animals were housed individually (except while paired for mating (Phase I) in suspended cages fitted with perforated floor panels and plastic resting platforms. Undertrays lined with absorbent paper were changed at least three times per week. All animals were maintained on a 14-hour daily photoperiod (10 hours dark) at an environmental temperature of 15–21°C and relative humidity of 45–70%. Environmental enrichment for each animal consisted of an Aspen chew block (soft white untreated wood block), a stainless-steel key ring attached to the cage, and nesting paper placed into each cage starting at post-mating GD 20 to allow expression of nesting behavior.
Phase I and II were conducted in an AAALAC accredited facility (Covance Laboratories Limited, Eye, UK (Study JJ43CT)) in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012.22 The number of animals used was the minimum that was consistent with scientific integrity and regulatory acceptability, consideration having been given to the welfare of individual animals in terms of the number and extent of procedures to be carried out on each animal. The NZW rabbit was chosen as the test model based on the availability of Historical Control Data (HCD) in the performing laboratory and the acceptability of this species and strain to regulatory agencies.
Experimental design
This study was conducted in two phases (Table 1): (I) a regulatory guideline-compliant embryo-fetal survival and development study and (II) a follow-up confirmatory investigation of equivocal fetal morphological findings noted in Phase I.
In a preliminary dose range-finding study conducted in the same laboratory, administration of AGIQ to 6 females per group was well-tolerated at 250, 500, and 1000 mg/kg/day, but was not tolerated at 2000 mg/kg/day and this group of animals was terminated early on GD 21/22 due to marked body weight loss and persistently low food consumption in some females. The urine from all treated animals was colored yellow and stained the cage tray paper and animal fur yellow/orange/ brown and the amniotic fluid/sacs in 1/6 animals at 250 mg/kg/day and 1/6 animals at 500 mg/kg/day, but none at 1000 mg/kg/day, were colored yellow. The strong coloration was attributed to the color of AGIQ (yellow to yellow-orange powder). At 1000 mg/kg/day, overall body weight gain was low and food intake was marginally low. Reproductive parameters did not appear to be affected by treatment with AGIQ.
Based on these preliminary results, female rabbits in Phases I (22/group) and II (48/group) were administered AGIQ in purified water once daily during GD 6–28 (inclusive) at approximately the same time each day via oral gavage at a dose volume of 10 mL/kg (calculated from the most recently recorded scheduled body weight) using a suitably graduated cylinder and rubber catheter inserted via the mouth. Phase I dose levels were 0 (purified water vehicle), 250, 500, and 1000 mg/kg/day and Phase II dose levels were 0, 500, and 1000 mg/kg/day. Surviving females in each phase were euthanized on GD 29 for maternal and fetal examinations.
Maternal evaluation (Phases I and II)
Clinical observations were conducted daily during the acclimation period and at least twice daily during the study for evidence of ill-health or reaction to treatment. Detailed observations for signs associated with dosing were conducted daily during the treatment period prior to dosing, 1–2 hours after completion of dosing, and as late as possible in the working day. A detailed physical examination of each animal was also conducted on GD 0 (Phase I only), 1 (Phase II only), 6, 12, 18, 23, and 29 to monitor general health. Maternal body weights were collected during acclimation and on the day of mating (GD 0; Phase I), on arrival (GD 1; Phase II), and on GD 3 and 6–29 (daily). Food consumption was measured daily during the study. All surviving adult females were euthanized on GD 29 via an intravenous injection of sodium pentobarbitone and all viable fetuses were euthanized via a subcutaneous injection of sodium pentobarbitone. One control and two dosed rabbits were euthanized for welfare reasons and all were pregnant.
Detailed necropsies were conducted for Phase I females, including full macroscopic examination of tissues (including the kidney and ureters) and visual examination of all external features and orifices. Any abnormality in the appearance or size of any organ and tissue (external and cut surface) was recorded and the required tissue samples were preserved in appropriate fixative.
Postmortem examinations for Phase II females were limited to confirmation of pregnancy status and examination for the presence/absence of kidney/ureters and any abnormalities in these organs.
Reproductive assessment (Phases I and II)
For Phase I, the uterus of each dam was excised and gravid uterine weights (including cervix and ovaries) were obtained and recorded. For each ovary/uterine horn, corpora lutea were counted and the number and location of viable and nonviable fetuses, early and late resorptions, and total number of implantation sites were recorded. For apparently nonpregnant animals and apparently empty uterine horns, the absence or number of uterine implantation sites was confirmed by visual examination.
For Phase II females that survived to term, all live fetuses were examined only for the presence/absence of kidney/ureters and any abnormalities in these organs. Fetuses with macroscopic findings or abnormalities in these organs were fixed in neutral-buffered formalin and retained. Representative photographs were taken of at least one male and one female fetus from each group to document the renal/ureter relationship and any fetus(es) with renal/ureter abnormalities.
Fetal macropathology (Phase I)
All viable fetuses and placentae were dissected from the uterus, individually weighed, identified within the litter using a coding system based on their position in the uterus, examined externally with abnormalities recorded, sampled as appropriate, and retained in 10% neutral-buffered formalin fixative. All fetuses were subjected to a gross internal examination of the viscera of the neck, thorax, and abdominal cavities. The sex of each fetus was also recorded. Approximately one-half of the eviscerated fetuses were decapitated and the heads were initially stored in Bouin’s fluid. The remaining eviscerated fetuses and torsos were fixed in Industrial Methylated Spirit. Fetal heads fixed in Bouin’s fluid were subjected to freehand serial sections, which were examined for soft tissue abnormalities. The fetuses and torsos fixed in Industrial Methylated Spirit were processed and stained with Alizarin Red S, then assessed for skeletal development and abnormalities.
Findings from external, visceral, and skeletal examination of fetuses were classified, according to severity and incidence, as either major abnormalities, minor abnormalities, or variants. Major abnormalities are normally rare, definitely detrimental to normal subsequent development, and possibly lethal (e.g., partially open eyelids or absent kidney/ureter). Minor abnormalities are minor differences from normal that are detected relatively frequently, are considered to have little detrimental effect, and may be a transient stage in development (e.g., bipartite centrum or dilated ureter). Variants are alternative structures or stages of development occurring regularly in the control population (e.g., number of ribs and thoracolumbar vertebrae or incomplete ossification of fifth and sixth sternebrae).
Statistical analysis
All statistical analyses were conducted for minimum significance levels of 5% and 1%, comparing each AGIQ-treated group to the appropriate control group by phase and sex. Maternal and fetal developmental toxicity endpoints were analyzed using the maternal animal or the litter as the experimental unit, as appropriate. For Phase I reproductive assessment (i.e., corpora lutea, implantations, resorptions, litter size, live young, and pre- and post-implantation losses) and fetal, litter, and placental weight data, group mean values and standard deviations were calculated using individual litter mean values, as appropriate. Standard deviations were not calculated for derived data, such as levels of pre- and post-implantation loss, or for the incidence of resorbing fetuses where the distribution of these findings commonly does not conform to the normal statistical distribution.
Continuous data variables (mean maternal body weights, body weight changes, and food consumption), gravid uterine weight and adjusted body weight, corpora lutea, implantations, litter size, live young, and placental, litter, and fetal body weights were subjected to a parametric analysis if Bartlett’s test for variance homogeneity23 was not significant at the 1% level. For pretreatment data, analysis of variance was used to test for any group differences. Where this analysis was significant (p < 0.05), intergroup comparisons were made using t-tests, with the error mean square from the one-way analysis of variance. For all other comparisons the F1 approximate test was applied. This test was designed to detect significant departure from monotonicity of means when the main test for the comparison of the means is a parametric monotonic trend test, such as Williams’ test.24,25 The test statistic compared the mean square, NMS, for the deviations of the observed means from the maximum likelihood means, calculated under a constraint of monotonicity with the usual error mean square, EMS. The null hypothesis was that the true means were monotonically ordered. The test statistic was F1 = NMS/EMS, which can be compared with standard tables of the F-distribution with 1 and error degrees of freedom. If the F1 approximate test for monotonicity of dose-response was not significant at the 1% level, Williams’ test for a monotonic trend was applied. If the F1 approximate test was significant, suggesting that the dose-response was not monotone, Dunnett’s test26,27 was performed instead.
For the data variables described above, a non-parametric analysis was performed if Bartlett’s test was still significant at the 1% level following both logarithmic and square-root transformations. For pretreatment data, the Kruskal–Wallis test28,29 was used to test for any group differences. Where this analysis was significant (p < 0.05), intergroup comparisons using Wilcoxon rank sum tests30 were made. For all other comparisons, the H1 approximate test, the non-parametric equivalent of the F1 test described above, was applied. This test was designed to be used when the main test for comparison of the means is a non-parametric monotonic trend test, such as Shirley’s test (Shirley 1977).31 The test statistic compared the non-monotonicity sums of squares, NRSS, for the deviations of the observed mean ranks from the maximum likelihood mean ranks with the non-parametric equivalent of the error sums of squares, ERSS = N(N + 1)/12. The test statistic was H1 = NRSS/ERSS, which can be compared to standard tables of the χ2-distribution with 1 degree of freedom. If the H1 approximate test for monotonicity of dose-response was not significant at the 1% level, Shirley’s test for a monotonic trend was applied. If the H1 approximate test was significant, suggesting that the dose-response was not monotone, Steel’s test32 was performed instead.
For litter size, litter survival indices, and gravid uterine, placental, litter, and fetal body weight data (Phase I only), if 75% of the data (across all groups) were the same value, for example c, Fisher’s exact tests33 were performed. Treatment groups were compared using pairwise comparisons of each dose group against the control both for 1) values <c versus values ≥c and for 2) values ≤c versus values >c, as applicable.
Pre- and post-implantation loss and fetal sex ratio (Phase I only) were analyzed by a generalized mixed linear model with binomial errors, a logit link function, and litter as a random effect (Lipsitz 1991).34 Each treated group was compared to the control group using a Wald chi-square test. For pre-implantati