Animals
The current investigation involved 24 female Beagle dogs, which were studied from birth to 24 months of age. We expected to have groups that present a significant difference of FM% between two grades of BCS. Assuming that the difference of FM% between two grades of BCS is approximately 5% [19], the sample size was established considering 3 pairwise comparisons (corresponding to BCS 5, 6 and 7) of FM% means with a first type error of 0.5 and a power of 0.8. The maximum sample size per group obtained was 7.47, therefore 8 dogs were allocated to each of the three groups [20]. Furthermore, the sample size of 24 dogs was suitable for both the capacity of Oniris’ facilities to guarantee animal welfare and to ensure that the sampling workload could be conducted in reliable conditions by one person in order to avoid manipulation bias.
They were the offspring of 10 litters (10 mothers and 7 fathers), were housed by litter in the same breeding centre (Isoquimen SL., Barcelona, Spain), weaned at 10 weeks, and neutered at the age of 8 months.
All dogs received an annual veterinary check-up and were vaccinated against canine distemper, canine adenovirus type 2, canine parainfluenza virus, canine parvovirus, rabies, and received worming treatments, at 2.5, 4, 12, and 22 months of age. A registered veterinarian was available to carry out additional veterinary treatments if required, and had the authority to withdraw dogs from the study if any adverse events occurred. Two dogs (one at 12 and one at 14 months of age) reached a BCS of 8/9 prior to the end of the study. Then, they were rationed in order to maintain a stable body weight until study completion.
Housing & diet
Throughout the study the dogs lived in the same environment and were fed in the same manner. All diets were supplied by Royal Canin (Royal Canin SAS, Aimargues, France). Prior to weaning (at 10 weeks of age), puppies were housed by litter with their mother in the breeding centre of Isoquimen SL, and had free access to mother’s milk and dry diet. This dry diet, Medium Starter (protein = 30%DM, fat = 22%DM, 4010 kcal/kg or 16.8 MJ/kg) was available ad libitum for both mothers and puppies, making an accurate assessment of the puppies’ nutrition source (dry diet vs maternal milk) impossible.
After weaning, dogs were relocated to Oniris (Nantes, France) and were housed in pairs. Each pair was housed in an outdoor enclosure of 4 m2 that included a sheltered place to sleep. Each dog was fed individually ad libitum for 3.5 h per day whilst the partner was temporarily removed from the enclosure. From weaning to 10.5 months of age, the dogs were given a dry diet formulated to meet growth requirements, Pediatric Junior Dog (protein =29%DM, fat = 20%DM, 3900 kcal/kg or 16.3 MJ/kg). From 10.5 months of age, in order to avoid too much excess weight gain after spaying, the dogs were fed with a moderate calorie dry maintenance diet (Neutered Adult; protein = 28%DM, fat = 11%DM, 3260 kcal/kg or 13.6 MJ/kg).
Individual food intake (g/day) was recorded daily (except on weekends) from weaning to 24 months of age, on the same calibrated electronic weigh scale (Ohaus Europe, Greifensee, Switzerland; accurate to within 0.2 g). The energy intake was corrected for metabolic body weight (EI; kcal/BW0.75) or for fat-free mass (EIFFM; kcal/FFM) which was calculated as follows:
$$ {EI}_{FFM}\ \left[{kcal.FFM}^{- 1}\right] = \frac{food\ intake\ \left[{kg.d}^{-\mathrm{1}}\right]\times energy\ content\left[{kcal.kg}^{-\mathrm{1}}\right]}{FFM\left[ kg\right]} $$
(1)
Dogs were walked on a leash for at least 15 min twice a week and had access to 1 h/day of free time in a closed garden of 400 m2 enriched with agility equipment.
Dogs had free access to water throughout the study.
Biometric assessment
Early-life data were provided by the breeding centre, including age and BW of parents at mating, parity, weight gain during gestation, litter size and BW of each puppy at birth.
Prior to weaning, puppies were weighed every 2 weeks. Post-weaning, BW was recorded weekly on the same electronic weigh scale (Mettler-Toledo SAS, Viroflay, France; accurate to within 50 g). The withers height was measured at 24 months of age. The morphometric estimation of body fat described by Burkholder and Toll [21] has been validated only on adult dogs, so a the pelvic circumference (PC) and patella-to-calcaneus (PCL) was measured every 2 months from 7 months of age, when dogs had a morphology closer to adult one. The body condition score (BCS) was evaluated monthly from 7 months of age by the same investigator, using a 9-point scale (1 for emaciated, 9 for morbidly obese) as recommended by the WSAVA [22].
The body composition was determined by isotopic dilution (deuterium oxide) at 6, 9, 12, 15 and 24 months of age. Food was withheld for 20 h before and water from 1 h before to 3 h after a subcutaneous tracer injection (physiological saline 2H2O solution (99.9% 2H/H; Euriso-top, Saint-Aubin, France), 0.5 g/kg), to achieve body water equilibration. Venous blood samples were collected in sterile ethylenediaminetetraacetic acid (EDTA) tubes before and 3 h after injection of the isotope. Total body water was determined in two steps. Firstly, the deuterium enrichment of plasma water was determined by Fourier-transform infrared on a Vector 33-type spectroscope (Brücker SA, Wissembourg, France) as previously described [23]. The deuterium enrichment (2H/H) was used to calculate the dilution space of the isotope, which indicates the total body water content after correction for proton exchanges with non-aqueous molecules [24]. Finally, the fat-free mass (FFM) in dogs was calculated with a canine specific hydration rate [25]:
$$ F F M\ \left[ kg\right] = \frac{Total\ body\ water\ \left[ kg\right]}{0.744} $$
(2)
The proportion of fat mass (FM%) was calculated as the difference between BW and FFM, divided by BW.
Given that the ideal BW in Beagles should be composed of approximately 80% FFM and 20% FM [21], the ideal weight would be FFM × 1.25. The percentage of excess weight according to estimated ideal weight at 24 months of age was calculated as follows:
$$ Excess\ weight\ \left[\%\right]=\frac{BW\ \left[ kg\right]-\left( FFM\ \left[ kg\right] \times 1.25\right)}{FFM\ \left[ kg\right]\times 1.25} \times 100 $$
(3)
Blood sampling
Blood samples were taken from dogs after 20 h of food deprivation, every 2 months from 3 to 17 months of age in order to measure plasma levels of glucose, appetite-related hormones (insulin, ghrelin, leptin and insulin-like growth factor 1 [IGF-1]), and stress markers (cortisol and prolactin).
Additional blood samples were taken every 2 months until 9 months of age, and subsequently every 3 months until 15 months, in order to measure levels of markers of inflammation (C-reactive protein [CRP], adiponectin, interleukin [IL-] 6, IL-8, IL-10 and tumour necrosis factor alpha [TNFα]).
At 7 and 13 months of age, in order to follow the post-prandial plasma kinetics of glucose, insulin, ghrelin and peptide YY3–36 (PYY), blood was collected immediately before a meal, and then 15, 30, 60, 90, 120 and 150 min after the meal. To avoid the influence of meal size and eating duration [26], dogs were given 10 min of access to a meal of their standard diet providing 130 kcal/kg metabolic BW (BW0.75) or 544 kJ/BW0.75 according to the recommendations for kennel dogs [27].
In all cases, blood was collected in heparin-coated sterile vacutainers. Plasma was separated by centrifugation at 5000 g for 10 min, then aliquoted and stored at –20 °C in sealed vials until analyses were completed.
Assays
Glucose was assayed immediately after collection by AlphaTRACK 2, a validated portable canine blood glucose meter (Abbott Animal Health, Abbott Park, IL, USA) using capillary or heparin-venous blood [28].
Plasma insulin and insulin-like growth factor 1 (IGF1) were assayed, as previously used in dogs by immunoradiometric assay (IRMA) [29] and radioimmuno assay (RIA) [30] using human kits (Insulin IRMA KIT, Beckman Coulter, Nyon, Swiss; IGF-1 RIA-CT, Mediagnost, Reutlingen, Germany). Active ghrelin concentration were assayed by enzyme-linked immunosorbent assay (ELISA), using a human kit (Human Acylated Ghrelin Express ELISA, BioVendor, Brno, Czech Republic, validated in dogs [31]). The total PYY and leptin concentrations were assayed by a human PYY ELISA kit and a canine leptin ELISA kit, respectively (Millipore, St. Charles, MO, USA). Cortisol concentration was assayed by a cortisol human RIA kit (Demeditec, Kiel, Germany) which was internally validated (coefficients of variation on 3 levels: A: 60 nmol; B: 200 nmol and C: 550 nmol; inter-assay A: 5%, B: 8%, C: 5%; intra-assay A: 12%, B: 8%, C: 14%). Prolactin concentrations were assayed by a canine prolactin ELISA kit, (Demeditec, Kiel, Germany).
Plasma CRP concentrations were measured using a specific solid phase sandwich immunoassay (Canine C-reactive Protein Assay, Tridelta Development Limited, County Kildare, Ireland). Adiponectin levels were determined using a high-sensitive human adiponectin ELISA kit (Human Adiponectin ELISA High sensitivity, BioVendor, Brno, Czech Republic; validated in dogs [32]). Plasma concentrations of canine IL-6, IL-8 and IL-10 and TNFα were assayed by specific ELISA kits (Quantikine ELISA Canine IL-6, IL-8, IL-10, TNFα, R&D Systems Inc., Minneapolis, MN, USA).
Plasma insulin to glucose concentrations ratio (I:G) was calculated in the unfed and postprandial state as follows:
$$ I: G=\frac{insulin\left[\mu U. m{L}^{-1}\right]}{fastingglucose\left[ mg. d{L}^{-1}\right]} $$
(4)
Energy expenditure assessment
Energy expenditure was determined by indirect calorimetry at 4, 7, 10 and 16 months of age, as validated in dogs by Pouteau et al. [33], with the following minor modifications. Food was withheld for 20 h, after which dogs were placed in a metabolic chamber (60 × 66.5 × 65 cm) for 4 h. The chamber was connected to a breath gas-exchange monitor (Quark RMR, Cosmed, Rome, Italy), which was calibrated at the start and then hourly, using a standard gas mixture. The system was an open-circuit ventilated by atmospheric air, pumped through the metabolism chamber at a flow rate of approximately 8 L/min adjusted for each dog at each age. The rate of flow of CO2 production and O2 consumption was measured every 5 s. The energy expenditure (kcal/d) was calculated using the abbreviated Weir formula [34]:
$$ Energy\ expenditure\ \left[ kcal.{d}^{-1}\right]=\left(1.11\kern0.75em \times \kern0.5em r C{O}_2\left[ L.{d}^{-1}\right]+3.94\kern0.75em \times \kern0.5em r{O}_2\left[ L.{d}^{-1}\right]\right) $$
(5)
After an approximately 40-min equilibration period, the energy expenditure was averaged on rolling 20-min periods. The resting energy expenditure (REE) was assumed to correspond to the lowest rolling mean of the energy expenditure during the 4 h of measurement, when the dog was calm but not asleep.
The REE was corrected (i) for metabolic BW (BW0.75) at 4 and 7 months and expressed as REE (kcal/BW0.75) and (ii) for FFM at 7, 10 and 16 months and expressed as REEFFM (kcal/FFM), both determinations being performed within a window of 30 days. The activity level was not measured.
Data analysis
All statistical analyses were performed using R software (R Foundation for Statistical Computing, Vienna, Austria) [35]. Graphs were prepared using GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA).
In order to distinguish dogs in three distinct groups, a principal component analysis (PCA) was performed on quantitative and active variables: FM%, FFM and PC, on dogs aged of 24 months. This was followed by a hierarchical clustering classification (HCPC) which was realized using Ward algorithm with euclidean distance on the two first principal components. The three identified groups were named IW, OW1 and OW2. The BCS is a qualitative variable, so it was included as supplementary qualitative variable in the PCA.
In order to assess the impact of parental and gestational factors on FM%, FFM and PC, and parental characteristics were included as supplementary variables in the PCA.
In order to determine the age at which groups became significantly different, additional PCAs were conducted on FM%, FFM and PC at the ages of 6, 9, 12 and 15 months. Confidence ellipses (95% confidence level) were constructed for the three groups in each PCA.
Mean and standard deviation (SD) were computed for each variable at 24 months of age in the three groups. The independence between the groups and the parents was assessed by Fisher’s exact test.
In order to characterize the growth throughout the study period, individual growth curves, defined as BW over time were plotted. Growth curves usually show a sigmoid profile and are fitted by a Gompertz function [36]:
$$ B{W}_t= B{W}_{max}\times {e}^{-\alpha \times {e}^{- kt}} $$
(6)
where:
t is age in weeks,
BWt is body weight at time t in kg,
BWmax is the maximum body weight, also named mature body weight,
α is an expression of the ratio between mature and birth body weight (BWb; \( \alpha = ln\left(\frac{B{ W}_{max}}{B{ W}_b}\right) \))
k is the maturation rate, which corresponds to the velocity to reach the adult body weight and e is Euler number.
The growth period could be divided into two periods by the point of inflection: the first period corresponding to an increasing growth rate and the second period to a decreasing growth rate. The maximum of weight velocity, calculated as \( \frac{B{ W}_{max}\times k}{e} \) occurs at the point of inflection (PI) of coordinates \( \left({t}_{PI}=\frac{ \ln \left(\alpha \right)}{k};\kern0.5em B{W}_{PI}=\frac{B{W}_{max}}{e}\right) \).
In order to assess the growth parameters (BWmax, α and k) of each group, the growth data collected from 0 to 18 months of age were fitted by the Gompertz function using nonlinear mixed effect model tools (saemix package in R software) with the dog as random effect.
To compare the evolution of BW of the groups during specific life-stage periods, we split the study duration into five periods. Linear mixed effects models were performed on age, BW, and groups over the following periods: before weaning (0 to 2.5 months of age), pseudo-linear growth (2.5 to 6 months of age), before spaying (6 to 8 months of age) and after spaying (8 to 10.5 months of age). Given that the normal weight in 2-week old neonatal puppies is twice that at birth [37], GR2W, % was calculated. We also calculated the relative weight gain after sterilisation (%) to assess the susceptibility to gain weight after the surgery. Post-prandial kinetic data (0–150 min) was transformed into area under the curve (AUC) values based on the difference from baseline. In order to interpret energy intake and expenditure, resting energy balance was calculated by subtracting REEFFM from EIFFM.
To take into account the repeated measurements for individual dogs in time, linear mixed effects models (nlme package in R software) were performed with dogs as random term.
These models enabled us to assess
-
i)
the interactions between groups and age on BW, FFM, FM%, PC, energy intake and expenditure or on hormonal concentrations,
-
ii)
the correlation between parental factors and biometric data at 24 months of age,
-
iii)
the correlation between the growth rate in the first 2 weeks of life (GR2W) and either groups or FM% at 24 months of age.
Independency and normal distribution for residuals and random effects were checked by graphical tools as described in the theory of mixed models effects [38].
Logistic regression analysis was performed on the three groups by pairs (IW–OW1; OW1–OW2, IW–OW2) to attempt to discriminate the groups by
-
i)
parental and gestational variables,
-
ii)
GR2W,
-
iii)
BCS, FM% or FFM prior to 2 years of age.
For each identified discriminant factor, a discriminant value was deduced from logistic regression results and was used as a cut-off to differentiate the groups. The goodness of fit was explored through an analysis of deviance table.
In each model, multiple comparisons were taken into account and adjusted p-values were calculated by single-step method, proposed in “multcomp” package of R software. All p-values were compared to α = 0.05 to establish significant differences.