• Open Access

Effect of Dietary Carbohydrate, Fat, and Protein on Postprandial Glycemia and Energy Intake in Cats

Authors


Corresponding author: Prof. J.S. Rand, School of Veterinary Science, The University of Queensland, Brisbane, Qld 4072; Australia; e-mail: j.rand@uq.edu.au.

Abstract

Background

Reducing carbohydrate intake is recommended in diabetic cats and might also be useful in some healthy cats to decrease diabetes risk.

Objective

To compare postprandial glucose and insulin concentrations and energy intakes between cats fed diets high in protein, fat, or carbohydrate.

Animals

Twenty-four lean cats with normal glucose tolerance.

Methods

In a prospective randomized study, each of 3 matched groups (n = 8) received a different test diet for 5 weeks. Diets were high in either protein (46% of metabolizable energy [ME]), fat (47% ME), or carbohydrate (47% ME). Glucose and insulin were measured during glucose tolerance, ad libitum, and meal-feeding tests.

Results

During ad libitum feeding, cats fed the high-carbohydrate diet consumed 25% and 18% more carbohydrate than cats fed diets high in fat and protein, respectively, and energy intake was highest when the high-fat and high-protein diets were fed. Regardless of the feeding pattern, cats fed the high-carbohydrate diet had 10–31% higher peak and mean glucose compared with both other diets; peak glucose in some cats reached 10.4 mmol/L (188 mg/dL) in cats fed 47% ME carbohydrate and 9.0 mmol/L (162 mg/dL) in cats fed 23% ME.

Conclusions and Clinical Importance

High-carbohydrate diets increase postprandial glycemia in healthy cats compared with diets high in fat or protein, although energy intake is lower. Avoidance of high- and moderate-carbohydrate diets can be advantageous in cats at risk of diabetes. Maintenance energy requirements should be fed to prevent weight gain when switching to lower carbohydrate diets.

Abbreviation
ME

metabolizable energy

The diet and lifestyle of domestic cats has changed over the last 20–30 years, and they are increasingly kept indoors and are sedentary.[1] Cats are now typically fed energy-dense food that is moderate to high in fat and carbohydrate content, and commonly fed in excess of their daily energy requirements.[2] In some populations, two of every 3 cats are overweight or obese.[3] The overweight condition in cats is associated with increased risk for the development of diabetes mellitus, among other diseases.[4] As in humans, lifestyle and dietary factors are involved in the development of diabetes.[3, 5] Type 2 diabetes in humans and diabetes in cats are very similar diseases clinicopathologically. In humans, over 50% of patients with type 2 diabetes are undiagnosed, and prediabetes, defined as impaired glucose tolerance or impaired fasting glucose, is 2–4 times more common than diabetes.[5] It is unknown if these statistics are true for cats. The reported prevalence of diabetes in cats has increased,[6] and it is likely there are many cats with undiagnosed prediabetes and diabetes.1,2

There is no consensus in veterinary medicine on the best diet for the prevention of diabetes in cats, with both moderate-carbohydrate, low-fat, high-protein, high-fiber diets recommended to prevent or manage obesity, and low-carbohydrate, moderate-fat, high-protein diets recommended to reduce postprandial hyperglycaemia.[7, 8] Minimizing the increase in glucose concentration after a meal and the subsequent demand on beta-cells to secrete insulin is a primary goal for the management of prediabetic and diabetic human patients.[9] In humans, it is more important (but also more difficult) to normalize postprandial hyperglycemia, as compared to fasting glucose concentrations.[9] Chronic hyperglycemia increases the demand on the beta cells to secrete insulin, and chronic hyperinsulinemia is associated with eventual beta-cell failure and type 2 diabetes.[10] The International Diabetes Federation defines postprandial hyperglycemia in humans as a plasma glucose concentration of greater than 7.7 mmol/L (140 mg/dL),[11] and glucose toxicity can cause impaired beta-cell function at glucose concentrations that are only 1 mmol/L (18 mg/dL) higher than normal.[12-14] Given that the upper cutpoint for normal fasting blood glucose is similar in humans and cats,[5, 15-17] and that cats have a very prolonged postprandial period (≥12 hours),[17] postprandial blood glucose concentrations >7.7 mmol/L (140 mg/dL) might also be detrimental in cats.[18, 19] Diets with moderate to high levels of carbohydrate (6.7–14.5 g carbohydrate/100 kcal; 23–50% metabolizable energy [ME]) result in peak and mean 24-hour postprandial glucose concentrations greater than 8 mmol/L (144 mg/dL) in some cats.[16-18] Of concern was that a diet with 14.5 g/100 kcal (51% ME) resulted in peak glucose concentrations as high as 10.8 mmol/L (194 mg/dL) in lean cats, and 13.4 mmol/L (241 mg/dL) after moderate weight gain (mean body condition score 6.3 out of a 9 point scale), which is considered in the diabetic range for cats.[20] Overweight and obese cats have increased mean concentrations and duration of the postprandial glucose and insulin response.[16, 18] Therefore, reduced carbohydrate diets might be useful in cats that are at risk of diabetes, especially obese, physically inactive cats, European-origin Burmese cats, diabetic cats in remission, and cats with intrinsically low insulin sensitivity. Low-carbohydrate, moderate-fat, high-protein diets decrease postprandial blood glucose concentrations, insulin requirements, or both compared with moderate-carbohydrate, low-fat diets.[7, 21] However, there is still controversy regarding carbohydrate content in feline diets, and in most reported studies comparing diets with differing carbohydrate content, other nutritional differences between the diets confounded comparisons.

The aim of this study was to compare postprandial glucose and insulin concentrations and energy intakes between healthy cats fed diets high in protein, fat, or carbohydrate. Importantly, the knowledge gained from this study informs the debate about the effect of dietary carbohydrate content on postprandial glucose and insulin concentrations, and energy intake. The results are relevant to cats at risk of diabetes, especially those with insulin resistance, impaired fasting glucose, and/or impaired glucose tolerance.

Materials and Methods

Overview

A randomized controlled trial was performed. Each cat was allocated to one of 3 groups, matched for sex, body weight, and mean glucose and insulin concentrations at times 0, 60, and 120 minutes measured during a simplified intravenous glucose tolerance test,[22] performed when cats were consuming the standard kennel diet, before feeding the “washout” diet (a commercially available premium feline maintenance diet).3 The washout diet was fed to all 24 cats for 5 weeks, and baseline testing was performed in the 5th week (Fig 1). Each group was then randomly allocated by random draw to one of 3 test diets high in one of protein, fat or carbohydrate, and fed the diet for 5 weeks. Each cat (and group) received only one of the test diets. Composition of the washout and test diets is shown in Table 1. Metabolic tests were performed in the 5th week of consuming the washout diet (1 week before feeding the test diets), and in the 5th week of feeding the test diets. During each of the 2 metabolic test weeks, plasma glucose and insulin concentrations were measured during a glucose tolerance test, ad libitum feeding test and meal-feeding test (once daily feeding). Energy intake was measured during ad libitum feeding tests.

Table 1. Nutrient composition and ingredient list of the standard maintenance (“washout”) diet and the 3 test diets fed to 3 groups of 8 clinically healthy cats, to evaluate the effect of diets high in protein, fat, or carbohydrate on glucose and insulin concentrations, and energy intakes
VariablesDietary Nutrient Composition
 Caloric Basis (g/100 kcal)
WashoutaHigh ProteinHigh FatHigh CarbohydrateWashoutaHigh ProteinHigh FatHigh Carbohydrate
  1. DM, dry matter; ME, metabolizable energy.

  2. a

    Iams Adult Cat Chicken Dry Food.

  3. b

    Metabolizable energy calculated using the modified Atwater factors, NRC, 1985.

  4. c

    Metabolizable energy calculated using the equation proposed by the NRC, 2006.

  5. d

    Ingredients used for diet formulation remained the same between all test diets with the quantity of macronutrients changing to allow variation in the distribution of calories to facilitate diets that were high in either protein, fat, or carbohydrate.

Key values (as-fed basis %)b
Moisture10.06.56.67.4    
Protein34.046.731.326.57.212.17.17.0
Fat21.010.823.311.24.52.85.32.9
Carbohydrate25.527.532.349.15.47.17.312.9
Crude fiber2.51.31.31.40.50.30.30.4
Ash7.07.25.24.51.51.91.21.2
Key values (DM %)
Protein35.649.933.628.6    
Fat23.311.624.912.1    
Carbohydrate30.029.434.652.9    
Approximate energy density (ME)
Total (kcal/kg); NRC 1985b4,4753,5164,2053,599    
Total (kcal/kg); NRC 2006c4,6983,8514,3973,809    
Protein (%)c29.047.026.026.0    
Fat (%)c46.026.047.027.0    
Carbohydrate (%)c25.027.027.048.0    
Main ingredientsdChicken by-product meal
Corn grits
Corn meal
Ground grain sorghum
Dried beet pulp (sugar removed)
Fish meal
Natural chicken flavor
Dried egg product
Chicken fat (preserved with mixed tocopherols)
Figure 1.

Timetable for the study.

Animals and Dietary Treatments

Twenty-four neutered adult cats (12 male, 12 female) were used in the trial. All cats were healthy based on physical examination and routine hematological, serum biochemical, and urine analyses. Mean body weight was 4.9 kg (range 3.6–6.0 kg). All cats had an ideal body condition score of 3 as assessed on a scale of 1 (underweight) to 5 (obese).[23] Based on visual assessment, all cats were estimated to be between 2 and 6 years old, although accurate ages were unknown. The Animal Experimentation Ethics Committee of the University of Queensland, Australia, approved the protocol for this study. After the trial, cats were re-homed.

Each test diet provided approximately 50% of energy from the test macronutrient, and 25% of energy from each of the other 2 test macronutrients (Table 1). All diets were dry extruded formulations. Both the washout diet and test diets were fed for 5 weeks. Sources of macronutrients and other ingredients were the same for the washout and test diets. Protein source was predominantly chicken, and carbohydrate source was corn and sorghum (Table 1).

Cats were fed the washout and test diets once daily to maintain body weight within 10% of their pretrial weight; mean amounts fed to maintain body weight in individual cats ranged from 76 to 93 kcal/(kg body weight)0.67 (45–55 kcal/kg body weight) over the duration of the study. To maintain body weight at the pretrial weight, food bowls were weighed daily before each feeding to calculate the amount of food consumed in 24 hours, and cats were weighed weekly, and amount fed adjusted weekly. Cats were allowed free access to water at all times.

Metabolic Testing

Metabolic tests were performed in the week before feeding the test diets (while consuming the washout diet), and again in the 5th week of feeding the test diets. Metabolic testing was performed as follows: day 1, general anesthesia and jugular catheter placement followed by a meal and removal of food bowls at least 12 hours before the intravenous glucose tolerance test; day 2, intravenous glucose tolerance test followed by ad libitum feeding; day 3 ad libitum feeding; day 4, ad libitum feeding test for 12 hours followed by removal of food; day 5, meal fed and uneaten food removed 30 minutes later; day 6, 24-hour meal-feeding test (Fig 1).

A jugular catheter4 was placed into 1 jugular vein under general anesthesia using a modified Seldinger technique 24 hours before the glucose tolerance test in each of the 2 test weeks. During catheter placement cats were anesthetised with propofol,5 given as an initial bolus dose of 6–7 mg/kg followed by additional doses of 5–10 mg as required. The catheter line and port were held in place via a neck bandage, which was checked daily. Catheters were flushed twice daily with heparinized saline (20 IU of heparin/mL in 0.9% saline solution) to maintain patency until their removal at the end of the meal-feeding test.

Intravenous Glucose Tolerance Test

Blood samples (4 mL) were collected before (−30 and 0 minutes) and at 2, 5, 10, 15, 30, 45, 60, 90, 120, and 180 minutes after 1.0 g/kg body weight glucose6 administration.[24]

Ad Libitum Feeding Test

The ad libitum feeding test was used to mimic the feeding pattern observed in cats when food is freely available. For 36 hours before the test, cats were allowed unrestricted access to food in excess of what could be consumed, to facilitate an ad libitum eating pattern typical of domestic cats characterized by multiple (10–20) small meals daily.[25] At the start of the test, fresh food was provided in excess of what could be consumed within 12 hours, and blood samples (4 mL) were collected before (−30 and 0 minutes) and at 1, 2, 3, 4, 6, 8, 10, and 12 hours after initial placement of food.

Meal-Feeding Test

The meal-feeding test was used to mimic the feeding pattern commonly observed in overweight cats fed limited energy to lose weight, and in some ideal weight cats fed at maintenance energy requirements to prevent weight gain. Typically in these cats the majority of the food is eaten shortly after feeding. This feeding pattern also occurred naturally in our cats during the study; cats were fed once daily at maintenance energy requirements to prevent weight gain. Before the meal-feeding test, food was withheld for 23.5 hours. During the meal-feeding test, cats ate 90–100% of the 12-hour ad libitum intake as a single meal in 0.5 hour, after being fed at time 0. This was facilitated by feeding 85 kcal/kg body weight0.67 (50 kcal/kg body weight) 24 hours before the test and withdrawing uneaten food 30 minutes after feeding. Blood samples (4 mL), were collected before (−30 and −5 minutes) and at 1, 2, 3, 4, 6, 8, 10, 12, 15, 18, and 24 hours after feeding.

Sample Handling and Analysis

Blood samples from each of the tests were handled similarly. Samples were placed into sterile EDTA vacuettes containing the proteinase inhibitor, Trasylol,7 added to the vacuettes at 0.05 mL per mL of whole blood. After collection, samples were kept on ice for 10–15 minutes until centrifugation at 1,500 × g for 8 minutes. After separation, plasma samples were split and stored in 500 μL vials at −70°C until assayed for glucose and insulin concentration.

Glucose was measured in plasma using an automated glucose analyzer.8 Insulin was measured using a commercially available kit,9 validated for the detection of feline insulin.

Calculations and Statistical Analysis

For the ad libitum feeding test, 1 cat (cat 17) had an implausibly high insulin concentration recorded at 4 hours (300.6 μU/mL). As all other values for this cat were within the expected range, this value was disregarded. Glucose disappearance coefficients (Kglucose) and glucose half-lives (T1/2) were calculated for each cat with data from the intravenous glucose tolerance test using linear regression of the semi-logarithmic plot of glucose concentration versus time between 15 and 90 minutes after glucose administration.[26] Mean 24-hour concentrations for glucose and insulin (or mean 12-hour concentrations for the ad libitum feeding test) were calculated for each cat as areas under the curve (from 0 to 24 hours calculated using the trapezoidal method,[27] and expressed as (mg/dL)–hours and (μU/mL)–hours, respectively), divided by 24 or 12 as appropriate. For each test, “baseline” glucose concentrations were calculated for each cat as the mean of the −30 and −5 minute values.

Each cat was considered to have exceeded its baseline glucose concentration if blood glucose concentration at one or more timepoints was greater than the sum of the cat's individual baseline concentration and the 90% range of differences. Times to first exceed baseline and to return to baseline were estimated by linear interpolation. Cats that did not exceed baseline concentration were excluded from analyses of times to return to baseline. The 90% range of differences was calculated using previously reported methodology based on the variance of the 2 baseline samples within cat[28] but with the variance calculated as the residual mean square from analysis of variance after fitting cat. Concentrations of insulin at −30 and −5 minutes were log transformed before estimation of the 90% range of differences.

Energy intakes in the ad libitum feeding tests were calculated for each cat as the amount of food eaten from initial placement of food until 12 hours later, multiplied by the energy density of the diet on an “as fed” basis.[29] Energy density on a dry matter basis was calculated using the modified Atwater factors as proposed by the National Research Council (NRC) and using the formula proposed by the NRC in 2006.[29, 30]

Baseline, peak, nadir (minimum values after feeding) and mean 24-hour concentrations or, for the ad libitum feeding test, mean 12-hour concentrations, glucose disappearance coefficients (Kglucose) and glucose half-lives (T1/2) and energy intakes were calculated for each cat and compared between diets using linear regression with the cat as the unit of analysis. Insulin concentrations were log-transformed before analyses. Distributions of beta-coefficients were estimated using the nonparametric bootstrap, with 1,000 replications. Times to peak and times to return to baseline were compared between diets using Kaplan–Meier analyses with log-rank tests with the cat as the unit of analysis. For cats that had not returned to baseline by the final timepoint at which blood was collected, times to return to baseline were right-censored at that time. Overall P values and P values for pair-wise comparisons between the 3 diets were calculated. P values were not adjusted for multiple pair-wise comparisons as all comparisons were of a priori interest.

Means for each of glucose and log-transformed insulin during the ad libitum and meal-feeding tests were compared between diets and timepoints using generalized estimating equations, with cat fitted as the grouping (or panel) variable, and with diet, timepoint, and the interaction between these fitted as fixed effects. The cat-timepoint was the unit of analyses where each measurement timepoint for each cat constituted one cat-timepoint. Normal error distributions, identity link functions, exchangeable correlation structures, and the Huber/White/sandwich (“robust”) estimator of variance were used. P values for interactions between diet and timepoint were all <.001, providing substantial evidence that effects of diet (ie, differences between concentrations) varied between timepoints, and that concentrations over time differed by diet. Accordingly, the interaction terms were retained in the models and pair-wise comparisons then performed: (1) between diets at each timepoint, and (2) between each timepoint and baseline within each diet. P values were adjusted for multiple comparisons using the Benjamini-Hochberg step-up False Discovery Rate method in WinPepi (version 11.15).[31] All other statistical analyses were performed using the statistical software package Stata (version 11.1).10 Wald P values were used for all linear regression and generalized estimating equation analyses. Significance was determined as P values ≤.05.

Results

Effects of Diet on Energy Intake during ad libitum Feeding

After consumption of the “washout” diet for 5 weeks, mean (±SD) energy intakes over 12 hours in the ad libitum feeding test before cats were fed the 3 test diets were 108 (±35), 96 (±38), and 134 (±41) kcal/kg body weight0.67 for cats subsequently fed the high-protein, fat, and carbohydrate diets, respectively.

Energy intake (mean ± SD kcal/kg body weight0.67) varied significantly between diets (overall P = .026) during the ad libitum feeding test with higher intake for the high-protein diet (13 ± 39) compared to the high-carbohydrate diet (91 ± 30; pair-wise P = .012), and a similar trend for higher intake with the high-fat diet (121 ± 36; pair-wise P = .059). There was no significant difference in energy intake between the high-protein and high-fat diets (pair-wise P = .531; difference between means (high-fat diet minus high-protein diet): −117; 95% CI: −4,726 to 2,413), although the high-fat diet had the highest and the high-protein diet the lowest energy density (Table 1). Results were similar after adjusting (separately) for washout diet energy intake, for body weight at that time, and for body weight after the test diets had been fed for 4 weeks (results not shown). Mean intakes of energy supplied by carbohydrate (mean ± SD kcal/kg body weight0.67) were 36 ± 11 (high protein), 33 ± 10 (high fat), and 44 ± 14 (high-carbohydrate diets) kcal/kg body weight0.67. Although these did not differ significantly between diets (overall P = .174; pair-wise P values >.06), numerically, cats fed the high-carbohydrate diet consumed 25% and 18% more carbohydrate than the cats fed the high-fat and high-protein diets, respectively. When calculated on a kcal/kg basis it was 27% and 19% higher.

Effects of Diet on Postprandial Glucose Variables after 5 Weeks of Feeding

Glucose variables in the ad libitum and meal-feeding tests were not statistically different between groups after consumption of the washout diet for 5 weeks, before the consumption of the test diets (results not shown). After consumption of the test diets for 5 weeks, baseline glucose concentrations were higher for the high-carbohydrate diet compared to both the high-protein (< .001) and high-fat (= .022) diets in the ad libitum feeding test; however, cats were not fasted and were eating ad libitum for 36 hours before the test (Table 2). In contrast, baseline glucose concentrations were similar for each diet for the meal-feeding test where food was withheld for 23.5 hours before the sample collection (overall = .392).

Table 2. Arithmetic mean ± SEM (SD; range) for plasma glucose variables and glucose to insulin ratios during an intravenous glucose tolerance test, a 12-hour ad libitum feeding test and a 24-hour meal feeding test in 3 groups of cats (n = 8 in each group) after being fed a diet high in protein, fat, or carbohydrate for 4 weeks, and median (range) of times to peak and to return to baseline. Means with different superscripts differ significantly at the .05 level
VariableHigh ProteinHigh FatHigh CarbohydrateOverall P Value
Intravenous glucose tolerance test
Baseline glucose concentration (mg/dL)97 ± 4 (12; 4–119)98 ± 5 (14; 86–127)105 ± 7 (19; 83–138).636
Peak glucose concentration (mg/dL)872 ± 52 (148; 712–1,119)896 ± 16 (44; 832–958)885 ± 19 (54; 820–976).835
Nadir glucose concentration (mg/dL)78 ± 2 (5; 71–85)94 ± 18 (51; 2–189)97 ± 11 (30; 73–169).155
Mean glucose concentration (mg/dL)276 ± 19 (53; 206–340)304 ± 17 (49; 238–397)292 ± 12 (34; 253–343).534
Kglucose (%/min)1.3 ± 0.2 (0.5; 0.6–2.0)1.4 ± 0.1 (0.3; 1.0–1.8)1.3 ± 0.1 (0.2; 1.0–1.5).713
Glucose T1/2 (min)62 ± 9 (27; 35–119)53 ± 4 (11: 39–70)55 ± 3 (7; 46–67).618
Ad libitum feeding test
Baseline glucose concentration (mg/dL)92 ± 2a (4; 85–97)96 ± 3a (8; 87–110)107 ± 4b (11; 97–132)<.001
Baseline glucose concentration (mmol/L)5.1 ± 0.1a (0.2; 4.7–5.6)5.3 ± 0.2a (0.4; 4.8–6.1)5.9 ± 0.2a(0.6; 5.4–7.3) 
Peak glucose concentration (mg/dL)99 ± 3a (9; 91–114)104 ± 3a (10; 90–117)128 ± 9b (26; 110–191).012
Peak glucose concentration (mmol/L)5.5 ± 0.2a (0.5; 5.1–6.3)5.8 ± 0.2a (0.6; 5.0–6.5)7.1 ± 0.5b (1.4; 6.1–10.6) 
Nadir glucose concentration (mg/dL)83 ± 2a (7; 74–95)90 ± 2b (5; 82–97)102 ± 6c (13; 91–132).003
Nadir glucose concentration (mmol/L)4.6 ± 0.1a (0.4; 4.1–5.3)5.0 ± 0.1b (0.3; 4.6–5.4)5.7 ± 0.3c (0.7; 5.1–7.3) 
Mean glucose concentration (mg/dL)91 ± 2a (7; 85–101)97 ± 3a (7; 87–110)115 ± 8b (23; 100–170).008
Mean glucose concentration (mmol/L)5.1 ± 0.1a (0.4; 4.7–5.6)5.4 ± 0.2a (0.4; 4.8–6.1)6.4 ± 0.4b(1.3; 5.6–9.4) 
Glucose to insulin ratio (mg/dL)/(μU/mL)4 ± 0.3 (1; 2–6)4 ± 0.2 (0.7; 3–5)5 ± 0.5 (2; 3–8).264
Meal feeding test
Baseline glucose concentration (mg/dL)82 ± 2 (5; 74–91)82 ± 2 (5; 73–89)90 ± 7 (15; 74–120).392
Baseline glucose concentration (mmol/L)4.6 ± 0.1 (0.3; 4.1–5.1)4.6 ± 0.1 (0.3; 4.1–4.9)5.0 ± 0.4 (0.8; 4.1–6.7) 
Peak glucose concentration (mg/dL)115 ± 8a (21; 93–163)122 ± 5a (14; 96–141)151 ± 7b (19; 128–188).001
Peak glucose concentration (mmol/L)6.4 ± 0.4a (1.2; 5.2–9.0)6.8 ± 0.3a (0.8; 5.3–7.8)8.4 ± 0.4b (1.1; 7.1–10.4) 
Time to peak glucose concentration (h)6 (3–24)6 (6–12)6 (3–15).952
Mean glucose concentration (mg/dL)95 ± 3a (10; 84–111)98 ± 1a (4; 89–101)119 ± 6b (16; 104–155)<.001
Mean glucose concentration (mmol/L)5.3 ± 0.2a (0.6; 4.7–6.2)5.4 ± 0.1a (0.2; 4.9–5.6)6.6 ± 0.3b (0.9; 5.8–8.6) 
Glucose to insulin ratio (mg/dL)/(μU/mL)7 ± 0.4 (1; 6–10)6 ± 0.4 (1; 5–8)8 ± 1 (3; 5–12).147
Time for glucose to return to baseline (h)

23 (14 to >24)

3 of 8 cats had not returned to baseline by 24 hours

20 (12 to >24)

1 of 8 cats had not returned to baseline by 24 hours

24 (11 to >24)

4 of 8 cats had not returned to baseline by 24 hours

.376

The lowest glucose concentrations during the ad libitum feeding test were higher for the high-carbohydrate diet compared to both the high-protein (= .002) and high-fat (= .043) diets. Lowest glucose concentrations were higher for the high-fat diet compared to the high-protein diet (= .024) in the ad libitum feeding test.

For both ad libitum and meal-feeding tests, consumption of the high-carbohydrate diet resulted in 10–31% higher mean glucose concentrations and peak glucose concentrations than diets high in either protein or fat (Figs 2, 3; Table 2). In general, cats fed the high-carbohydrate diet had the highest peak and mean glucose concentrations, and those fed the high-protein diet had the lowest in both feeding tests (P for overall effect of diet <.001 to .012); pair-wise P (high carbohydrate compared to high protein <.001 to .004) (Tables 2, 3). The greatest difference in glucose concentrations between cats fed the different diets was for peak glucose concentration in cats meal-fed the high-carbohydrate diet (8 mmol/L; 151 mg/dL); which was 31% higher than for the high-protein diet (6 mmol/L; 115 mg/dL) (pair-wise P < .001). Similarly, the greatest difference in mean glucose concentration between cats fed the different diets was for cats meal-fed the high-carbohydrate diet (7 mmol/L; 119 mg/dL), which was 25% higher than for the high-protein diet (5 mmol/L; 95 mg/dL) (P = .001). In summary, although cats fed the high-protein and high-fat diets ingested approximately 43% and 31% more energy, respectively, than those fed the high-carbohydrate diet, their baseline, highest, lowest, and mean glucose concentrations in the ad libitum feeding test were significantly lower than for cats fed the high-carbohydrate diet.

Table 3. Means ± SD,a differences between means/ratios of means and associated confidence intervals and P values for mean and peak glucose and insulin concentrations, and glucose to insulin ratio
DietDifference/Ratiob(95% CI)P Value
High ProteinHigh FatHigh Carbohydrate
  1. a

    Arithmetic means are reported for glucose variables and glucose to insulin ratio; geometric means are reported for insulin variables because these required log-transformation before analyses and the resulting effect estimates (Euler's number raised to the power of the regression coefficients) can be interpreted as ratios of geometric means.

  2. b

    Difference between arithmetic means for glucose variables and glucose to insulin ratio; ratio of geometric means for insulin variables.

Ad libitum-feeding test
Peak glucose (mg/dL) (mmol/L)
99 ± 9 (5.5 ± 0.5)104 ± 10 (5.8 ± 0.6) 5.13(−4.03–14.29).272
99 ± 9 128 ± 26 (7.1 ± 1.4)28.72(9.37–48.06).004
 104 ± 10128 ± 2623.59(3.67–43.50).020
Mean glucose (mg/dL) (mmol/L)
91 ± 7 (5.1 ± 0.4)97 ± 7 (5.4 ± 0.4) 5.65(−1.25–12.54).108
91 ± 7 115 ± 23 (6.4 ± 1.3)23.92(7.55–40.29).004
 97 ± 7115 ± 2318.27(1.46–35.09).033
Peak insulin (µU/mL)
26.1 ± 5.229.4 ± 6.2 1.13(0.93–1.36).228
26.1 ± 5.2 30.9 ± 9.61.18(0.93–1.51).170
 29.4 ± 6.230.9 ± 9.61.05(0.82–1.35).684
Mean insulin (µU/mL)
21.4 ± 14.425.0 ± 17.0 1.17(0.96–1.43).125
21.4 ± 14.4 25.0 ± 12.61.17(0.88–1.56).272
 25.0 ± 17.025.0 ± 12.61.00(0.77–1.31).991
Glucose to insulin ratio (mg/dL/µU/mL)
4.333.91 −0.42(−1.15–0.31).258
4.33 4.700.37(−0.80–1.54).537
 3.914.700.79(−0.31–1.89).160
Meal feeding test
Peak glucose (mg/dL) (mmol/L)
115 ± 21 (6.4 ± 1.2)122 ± 14 (6.8 ± 0.8) 6.47(−9.98–22.91).441
115 ± 21 151 ± 19 (8.4 ± 1.1)35.74(17.07–54.40)<.001
 122 ± 14151 ± 1929.27(13.46–45.08)<.001
Mean glucose (mg/dL) (mmol/L)
95 ± 10 (5.3 ± 0.6)98 ± 4 (5.4 ± 0.2) 3.13(−3.84–10.09).379
95 ± 10 119 ± 16 (6.6 ± 0.9)23.94(11.37–36.51)<.001
 98 ± 4119 ± 1620.82(9.39–32.24)<.001
Peak insulin (µU/mL)
22.0 ± 5.021.9 ± 4.7 1.00(0.83–1.20).962
22.0 ± 5.0 24.3 ± 9.81.10(0.84–1.44).475
 21.9 ± 4.724.3 ± 9.81.11(0.85–1.44).449
Mean insulin (µU/mL)
13.9 ± 2.316.0 ± 2.3 1.15(0.99–1.33).073
13.9 ± 2.3 15.6 ± 4.71.12(0.89–1.41).340
 16.0 ± 2.315.6 ± 4.70.97(0.77–1.23).829
Glucose to insulin ratio (mg/dL/µU/mL)
6.876.20 −0.67(−1.70–0.36).204
6.87 7.971.10(−0.93–3.13).288
 6.207.971.77(−0.22–3.76).082
Figure 2.

Plasma glucose concentrations (mean ± SEM) during a 12-hour ad libitum feeding test in 3 groups of cats (n = 8 in each group), after being fed a diet high in protein, fat, or carbohydrate for 4 weeks. Cats were fed ad libitum for the 36 hours preceding and during the test. Within a timepoint, means with different letters (a or b) differed at the .05 level. NB: Values for variables derived from graphs may appear different from those in tables. Variables in tables are calculated from individual cat data whereas figures show mean data for each timepoint.

Figure 3.

Plasma glucose concentrations (mean ± SEM) during a 24-hour meal-feeding test in 3 groups of cats (n = 8 in each group) after being fed a diet high in protein, fat, or carbohydrate for 4 weeks. Food was withheld for 23.5 hours before testing. Within a timepoint, means with different letters (a or b) differed at the .05 level.

Cats meal-fed the high-fat diet had significantly lower glucose peak and mean concentrations than those eating the high-carbohydrate diet (P < .001), but peak and mean concentrations were not significantly different between cats eating the high protein and high-fat diets (Figs 2, 3; Tables 2, 3). In cats fed the high-fat diet, the greatest difference in peak glucose concentration was in the meal-feeding test (mean of peak values 7 mmol/L; 122 mg/dL), which was 20% lower than for the high-carbohydrate diet (8 mmol/L; 151 mg/dL) (P < .001, Table 2).

A number of cats in all 3 test groups did not return to baseline glucose concentrations after 24 hours in the meal-feeding test (3/8 cats fed the high-protein diet, 1/8 cats fed the high-fat diet, and 4/8 cats fed the high-carbohydrate diet). The median times to peak glucose concentration were 6 hours in the meal-fed cats for all diets (P = .952), and median times to return to baseline were 23, 20, and 24 hours (P = .376) for the high protein, high fat, and high-carbohydrate diets, respectively (Table 2).

After consumption of the test diets for 5 weeks, there was no significant difference for any variable between diets during the intravenous glucose tolerance test (Table 2).

Effects of Diet on Postprandial Insulin Variables after 5 Weeks of Feeding

Insulin variables during the ad libitum and meal-feeding tests were similar between groups after consumption of the “washout” diet for 5 weeks, before consumption of the test diets (results not shown).

After consumption of the test diets for 5 weeks, there was some evidence that baseline insulin concentrations were higher for the high-fat diet compared to the high-protein diet (P for overall effect of diet = .056; pair-wise P [high fat compared to high protein] = .022) in the ad libitum feeding test; cats were fed ad libitum for 36 hours before sampling (Table 4).

Table 4. Arithmetic mean ± SEM (SD; range) for plasma insulin variables during an intravenous glucose tolerance test, a 12-hour ad libitum feeding test and a 24-hour meal-feeding test in 3 groups of cats (n = 8 in each group) after being fed a diet high in protein, fat, or carbohydrate for 4 weeks, and median (range) of times to peak and to return to baseline. Means with different superscripts differ significantly at the .05 level
VariableHigh ProteinHigh FatHigh CarbohydrateOverall P Value
Intravenous glucose tolerance test
Baseline insulin concentration (µU/mL)10.3 ± 0.9 (2.7; 7.1–13.6)10.6 ± 0.5 (1.4; 8.4–12.1)11.3 ± 1.1 (3.0; 8.3–16.0).795
Peak insulin concentration (µU/mL)28.9 ± 2.0 (5.7; 20.5–36.4)37.6 ± 7.7 (22.2; 23.6–91.1)36.8 ± 4.1 (11.6; 18.8–52.4).259
Nadir insulin concentration (µU/mL)7.9 ± 1.4 (4.0; 1.6–15.6)7.2 ± 0.9 (2.6; 3.3–12.6)7.7 ± 0.8 (2.2; 3.4–11.0).893
Mean insulin concentration (µU/mL)19.8 ± 1.4 (3.9; 14.0–24.9)22.9 ± 2.2 (6.3; 15.6–37.1)23.5 ± 2.5 (7; 11.4–30.4).387
Ad libitum feeding test
Baseline insulin concentration (µU/mL)20.2 ± 1.6a (4.5; 14.1–26.6)25.5 ± 1.7b (4.8; 18.3–33.2)27.0 ± 3.8a,b (10.9; 14.2–46.1).056
Peak insulin concentration (µU/mL)26.5 ± 1.8 (5.2; 20.5–35.9)30.0 ± 2.2 (6.2; 22.0–38.2)32.1 ± 3.3 (9.6; 17.2–48.8).294
Nadir insulin concentration (µU/mL)17.4 ± 1.4 (3.9; 12.2–21.9)20.8 ± 1.9 (5.5; 11.7–28.0)20.8 ± 3.4 (9.6; 9.4–42.0).424
Mean insulin concentration (µU/mL)21.9 ± 1.8 (5.0; 15.5–53.8)25.4 ± 1.6 (4.7; 18.8–31.7)26.5 ± 3.5 (9.8; 14.0–46.2).275
Meal feeding test
Baseline insulin concentration (µU/mL)11.0 ± 1.1 (3.2; 7.8–16.6)10.6 ± 1.0 (2.9; 6.1–14.2)9.2 ± 1.2 (3.3; 6.2–14.8).397
Peak insulin concentration (µU/mL)22.5 ± 1.6 (4.6; 14.4–27.2)22.2 ± 1.3 (3.6; 17.0–26.9)25.5 ± 2.9 (8.1; 12.6–36.4).734
Time to peak insulin concentration (h)5 (1–8)6 (2–18)6 (4–12).161
Mean insulin concentration (µU/mL)14.1 ± 0.8 (2.3; 9.6–16.8)16.1 ± 0.8 (2.3; 12.5–19.5)16.2 ± 1.7 (4.7; 8.8–22.5).186
Time for insulin to return to baseline (h)

13 (3 to >24)

1 of 7 cats had not returned to baseline by 24 hours

17 (3 to >24)

2 of 8 cats had not returned to baseline by 24 hours

21 (9 to >24)

2 of 8 cats had not returned to baseline by 24 hours

.436

The greatest numerical difference in baseline insulin concentration was for cats fed ad libitium the high-carbohydrate diet; the arithmetic mean was 27.0 μU/mL, which was 34% higher than for the high-protein diet (20.2 μU/mL; P for pair-wise comparison after log transformation of data = .117) (Table 4).

For both ad libitum and meal-feeding tests, there was a consistent trend for peak insulin concentration to be highest following consumption of the high-carbohydrate diet but these differences were not statistically significant. The high-fat diet produced insulin concentrations that were more similar to the high-carbohydrate diet than high-protein diet (Table 4, Fig 4, 5).

Figure 4.

Plasma insulin concentrations (mean ± SEM) during a 12-hour ad libitum feeding test in 3 groups of cats (n = 8 in each group) after being fed a diet high in protein, fat, or carbohydrate for 4 weeks. Cats were fed ad libitum for the 36 hours preceding and during the test. Within each timepoint, no means differed at the .05 level.

Figure 5.

Plasma insulin concentrations (mean ± SEM) during a 24-hour meal-feeding test in 3 groups of cats (n = 8 in each group) after being fed a diet high in protein, fat, or carbohydrate for 4 weeks. Food was withheld for 23.5 hours before testing. Within each timepoint, no means differed at the .05 level.

The greatest numerical difference in mean insulin concentration for cats fed ad libitum was for the high-carbohydrate diet (arithmetic mean = 26.5 μU/mL; Table 4), which was 21% higher than for the high-protein diet (21.9 μU/mL; P for pair-wise comparison after log transformation of data = .272) (Table 4).

Median times to peak insulin concentration were 5, 6, and 6 hours (P = .161) in the meal-fed cats for the high-protein, high-fat, and high-carbohydrate diets respectively. The highest insulin concentrations were 26.5, 30.0, and 32.1 μU/mL in the cats fed ad libitum, and 22.5, 22.2, and 25.5 μU/mL in meal-fed cats (overall P values .294 and .734) for the high-protein, high-fat, and high-carbohydrate diets, respectively (Table 4). Time to return to baseline insulin concentration in the meal-feeding test was 13, 17, and 21 hours (P = .436) for the high-protein, high-fat, and high-carbohydrate diets, respectively (Table 4).

Effects of Diet on Postprandial Glucose to Insulin Ratios after 5 Weeks of Feeding

Glucose to insulin ratios in the postprandial period were highest for the high-carbohydrate diet and lowest for the high-fat diet, with the pair-wise difference between these diets showing a trend toward significance for the meal-feeding test (P = .082; Table 3).

Discussion

Our study showed that in healthy lean cats, consumption of a diet with approximately 50% of energy from carbohydrate (12.9 g/100 kcal) resulted in higher peak and mean blood glucose concentrations for 4–18 hours after eating, compared with consumption of a high protein or a fat diet containing approximately 25% of energy from carbohydrate (4.5–5.4 g/100 kcal). Cats consuming the high-protein diet consumed more energy than cats consuming the high-carbohydrate diet. However, the overall effect was that cats fed the high-carbohydrate diet consumed 25% more carbohydrate than cats fed the high-fat diet, and had 19–25% higher mean and 23% higher peak glucose concentrations. Similarly, cats fed the high-carbohydrate diet consumed an 18% higher carbohydrate load than cats fed the high-protein diet, and had 25–30% higher mean and 30% higher peak glucose concentrations. This finding has important implications for cats predisposed to diabetes, for example obese and European-origin Burmese cats, and cats with insulin resistance. It is also likely important for cats already exhibiting some degree of beta cell failure resulting in impaired glucose tolerance—some obese cats and the majority of diabetic cats in remission.[18, 32] Decreasing the glucose load from a meal has been shown to be important for achieving remission in diabetic cats.[21] Although our study compared a high-carbohydrate diet with a moderate-carbohydrate diet, further decreasing carbohydrate content to ≤12% ME has been shown to increase remission rates in diabetic cats. In fact, the highest remission rates (>80%) have only been reported using diets with approximately ≤6% of energy from carbohydrate.[33, 34] The results of our study highlight the likely mechanism for these observations—decreasing dietary carbohydrate load reduces the postprandial blood glucose elevation, which in cats is prolonged.[17, 18] In our study, blood glucose remained above baseline for 11 to >24 hours in cats fed the high-carbohydrate diet. For all 3 diets tested, some cats had peak blood glucose values more than 1 mmo/L (18 mg/dL) above the upper reference range for cats (<6–6.5 mmol/L; 108–117 mg/dL).[35] Of concern was in some cats fed the high-carbohydrate diet, peak glucose concentration (10.4 mmol/L; 150 mg/dL) was in the range reported for diabetic cats.[20] Even in some cats consuming the high-protein diet, peak glucose concentration was as high as 9.0 mmol/L (162 mg/dL). This finding might explain why, in diabetic cats, a diet with 3.5 g/100 kcal ME (12% ME) from carbohydrate resulted in higher remission rates than one with 7.6 g/100 kcal ME (24% ME).[21] Although the diets in the latter study differed in ingredient sources, fiber content, and micronutrients, which could have accounted for some of the differences observed, the carbohydrate content differed markedly between them, and 2 mechanisms likely explain the findings. Firstly, the higher postprandial glucose concentrations from the higher carbohydrate load make it less likely that a cat with very marginal beta cell function can secrete sufficient insulin to maintain euglycemia. In addition, the ongoing effect of glucose toxicity suppresses insulin secretion and prevents recovery of sufficient beta cell function to maintain euglycemia. The magnitude of the increase in postprandial glucose concentrations would likely be even greater had overweight and obese cats been studied.[16, 18]

The increase in blood glucose concentrations after a carbohydrate load is accentuated in cats compared to dogs and humans.[5, 19, 36] The relative carbohydrate intolerance of cats might be because of a number of unique features associated with glucose metabolism in this species, and could also contribute to their extended postprandial period of 8–15 hours compared to 2–3 hours for humans[5] and 3–6 hours for dogs.[37] The gluconeogenic pathway is almost always permanently “switched on.”[38] Cats have markedly reduced or absent glucokinase concentrations,[39] and rely on low-capacity hexokinase to clear a glucose load.[40] Further, delayed gastric emptying,11 reduced small intestinal disaccharidase activity,[41] and reduced and delayed insulin secretion compared with dogs[36] are some of the factors that result in cats having reduced capacity to decrease glucose concentrations after the ingestion of a high-carbohydrate load.[19, 36]

The lower blood glucose concentrations in our cats fed the lower carbohydrate diets is consistent with previous findings,[16, 40] however, others have not reported differences, largely because of the methodology used,[42-45] or the difference in carbohydrate content was not as great.[46] It was not surprising that we found no differences in glucose tolerance test results between the diets, because the glucose tolerance test is performed after a 24-hour fast, and the glucose load is the same for all cats. For there to be differences, it would imply that there was a physiological adaption to a given dietary carbohydrate load that was not modifiable in the short term for a different carbohydrate load. Therefore, the glucose tolerance test is not a sensitive test for determining the daily impact of diets on glucose and insulin concentrations; its use is to assess glucose tolerance status. Similarly, time 0 glucose concentrations were not different between diets for the meal-feeding test because cats had been fasted for 23.5 hours. In contrast, for the ad libitum feeding test, cats had access to food continuously before time 0, and hence their time 0 glucose concentrations reflected the effect of diet on postprandial glucose concentrations, which were significantly different between diets.

There were also 2 trends worth noting: firstly, postprandial insulin concentrations were consistently higher after the high-carbohydrate diet was consumed compared to the high-protein diet, but were similar to the high-fat diet. The small group size (n = 8), higher CV for the insulin assay compared to glucose assay, and the variability in the actual amount of food consumed by each cat likely contributed to the variability in insulin results and lack of statistical significance. Secondly, glucose to insulin ratio was highest for the high-carbohydrate diet and lowest for the high-fat diet. The finding of a trend to the highest glucose to insulin ratio for the high-carbohydrate diet might reflect the higher glucose concentrations, secondary to the high carbohydrate load coupled with reduced clearance of blood glucose in cats.[19, 36] An additional contributing factor was that insulin concentrations in cats fed the high-fat diet were similarly high to those fed the high-carbohydrate diet. This finding might be related to stimulation of incretin secretion by free fatty acids in the high-fat diet, which increases insulin secretion.[47, 48]

In our study, dietary protein did not appear to be a substantial driver of postprandial insulin concentrations in cats compared to dietary carbohydrate. This is despite amino acids being secretagogues for insulin in dogs, cats and humans.[49, 50] However, this finding may have been because of the postprandial amino acid concentrations being insufficient to fully stimulate insulin secretion.[51] Therefore, to reduce the postprandial increase in glucose and the subsequent demand for postprandial insulin secretion, restricting carbohydrate content appears most effective in cats. Our results are consistent with results from studies in humans and dogs, which indicate that carbohydrates are the principal nutrients involved in determining the magnitude of the postprandial changes in plasma glucose and insulin.[52, 53] Based on the collective effects in our study of diet on peak and mean glucose and insulin concentrations, substitution of protein for carbohydrate would be expected to lead to lower postprandial glucose and insulin concentrations compared to substitution with fat.

Obesity is a recognized risk factor for development of diabetes,[4, 54] and in our study, cats fed the high-protein and high-fat diets consumed more energy than the high-carbohydrate diet. Cats consuming the high-protein diet consumed 9% more energy than cats eating the high-fat diet, but the difference did not reach statistical significance. Although previous studies have reported high-fat diets fed ad libitum are associated with obesity in cats,[44, 54, 55] the energy consumption of the high-protein diet was less expected. However, in a recent study, when cats were fed diets of the same energy density and similar fat content, those eating a low-carbohydrate, moderate-fat, high-protein diet gained more weight than those on a high carbohydrate, moderate-fat, low-protein diet, so it was unknown which macronutrient had the greatest effect on weight gain.[16] Weight gain was reported to be caused by the higher energy efficiency of the low-carbohydrate, moderate-fat, high-protein diet.[16] The finding in our study that cats consuming the high-carbohydrate diet had the lowest energy intake was expected because cats are reported to limit their total energy intake when consuming a high-carbohydrate diet.[40] Cats have a ceiling for carbohydrate intake, which limits ingestion when consuming high carbohydrate foods and constrains them to deficits in protein and fat intake, relative to their targets for these macronutrients.[40] The authors suggested this “carbohydrate ceiling” can reflect the cats' adaption to a carnivorous diet. In the same study, cats overshot protein intake relative to their protein target, presumably to gain energy (a limiting resource).[40] The similar energy intakes observed in our study with the high protein and high-fat diets, and lower intake with the high-carbohydrate diet, might have simply reflected that cats fed the high-protein and high-fat diets were eating to achieve a similar perceived required energy intake, and cats fed the high-carbohydrate diet had lower energy intake because of the need to limit carbohydrate intake to a ceiling, which was reported to be 75 kcal/cat/day (approximately 30% of cats' maintenance energy requirements) for cats fed a high-carbohydrate diet for a week. Our cats consumed on average 129 kcal/cat/day (approximately 53% of the study cats' maintenance energy requirements) of carbohydrate, after being fed the high-carbohydrate diet for 5 weeks, and we have previously reported that cats fed a high-carbohydrate diet for longer periods have even higher levels of carbohydrate intake. For example, cats fed for 4 weeks at maintenance energy requirements and 8 weeks ad libitum, consumed 150 kcal/cat/day of carbohydrate (approximately 88% of the study cats' maintenance energy requirements), double the carbohydrate ceiling reported in cats fed a high-carbohydrate diet for 1 week. This suggests that over time, the carbohydrate ceiling is higher if cats are more adapted to the diet.

Our finding of the highest energy consumption with the high-protein diet and lowest with the high-carbohydrate diet might explain why there is no clear evidence from epidemiological studies of an effect of dietary composition on the predisposition to diabetes. Higher consumption of the high-protein and high-fat diets leading long term to obesity might counter the beneficial effect of lower glucose concentrations. Indeed, findings from a recent study in cats suggested that feeding a diet with 50% of energy from carbohydrate produced similar adverse effects on mean postprandial blood glucose and insulin concentrations as short-term moderate weight gain, although other dietary differences might have confounded results.[16] One epidemiological study has reported a dietary effect on risk of diabetes. In a study among feline patients in the United Kingdom, consumption of a mix of wet and dry foods was associated with a lower risk for diabetes, compared to only dry diets (typically high carbohydrate, moderate fat, and protein), or only wet diets (typically lower in carbohydrate and higher in fat and protein).[56] In that study, cats fed wet diets were 3 times more likely to develop diabetes than cats fed mixed diets; cats fed dry diets had 2 times the risk, suggesting the adverse effect of obesity was greater than the adverse effect of increased postprandial glycemia on risk of diabetes. Although wet foods have on average lower carbohydrate content than dry foods, the sauces may contain simple sugars, which would be expected to exacerbate postprandial glycemia. Well-designed studies are urgently required to better understand dietary risk factors for feline diabetes. In our study, cats were fed ad libitum for only 48 hours, and intake measured in the last 12 hours. The effects of fat, protein and carbohydrate on energy intake and risk of diabetes need to be evaluated over a longer time to determine if this finding is true over longer periods.

An important factor that differentiates our study from others is that all diets used contained the same ingredients and did not differ in carbohydrate source because this may cause some variation in blood glucose concentration.[18] The carbohydrate source used in our study (corn/sorghum) produces a lower glycemic response in cats than rice, which is commonly used as a starch source in feline diets.[18] Fiber sources were also the same between diets (mostly soluble fiber from beet pulp). Fiber content was similar between all test diets (0.3–0.5 g/100 kcal), and comparable in amount to a low-fiber, low-carbohydrate diet (0.1 g/100 kcal) associated with increased remission rates in diabetic cats, when compared to a high fiber, moderate-carbohydrate diet (3.1 g/100 kcal).[21] There are almost no data in the literature on the effect of fiber on postprandial glucose concentrations in cats. One study reported that a high-fiber diet (3.9 g/100 kcal) promoted better glycemic control in diabetic cats than a diet lower in fiber (0.5 g/100 kcal).[57] However, results were confounded by differences in carbohydrate content between diets, as the low-fiber diet had higher carbohydrate content (38% DM) than the high-fiber diet (27% DM). In our study, the carbohydrate load may not be responsible for the entirety of the differences seen between the diets used in our study, and other unidentified effects of protein and fat may have influenced the results, including incretin effects and effects of fat and carbohydrate on insulin sensitivity. However, it is not possible to test just 1 effect in isolation from the others, because changing 1 macronutrient inevitably changes others—lowering carbohydrate content must result in increases in either fat, protein, or both. It is clear that further studies are needed in this area of feline nutrition to gain better understanding of the effects of dietary macronutrients on energy intake and postprandial glucose concentrations, and the subsequent risk for obesity and diabetes.

In conclusion, our study shows that a diet with approximately 50% of energy (12.8 g/100 kcal) from carbohydrate results in significantly higher postprandial blood glucose concentrations for a median of 24 hours after eating compared to diets with approximately 25% of energy from carbohydrate (7.1–7.3 g/100 kcal) and 50% from either fat or protein. However, even diets with 25% of energy from carbohydrate result in peak blood glucose concentrations as high as 9 mmol/L (162 mg/dL) after eating in lean cats with normal glucose tolerance. Given the frequency of impaired glucose tolerance in obese cats, diets with carbohydrate content <25% ME are recommended to minimize postprandial glucose and insulin response, and to avoid potential glucose toxic damage to beta cells. However, the high-protein diet was associated with the highest energy intake, and so cats at risk of diabetes should be fed at maintenance energy requirements to avoid obesity. Further research is required to determine the optimum level of dietary carbohydrate, fat, and protein for cats.

Acknowledgments

The authors thank the Iams Company, Procter & Gamble, Lewisburg, OH 45338, USA for contributing funding towards the study and Lyn Knott for help with sample analysis, Delisa Appleton and Linda Fleeman for technical assistance and advice and Caitlin McGuckin for assistance with writing and editing.

Conflict of Interest Declaration: Authors disclose no conflict of interest.

Footnotes

  1. 1

    Reeve-Johnson MK, Rand JS, Anderson S, et al. Determination of reference values for casual blood glucose concentration in clinically healthy, aged cats measured with a portable glucose meter from an ear or paw sample. J Vet Intern Med 2012;26:755 (abstract)

  2. 2

    Gottlieb S, Rand JS, Marshall RD. Diabetic cats in remission have mildly impaired glucose tolerance. J Vet Intern Med 2011;25:682–683 (abstract)

  3. 3

    Iams Adult Cat Chicken Dry Food; Iams Company, Lewisburg, OH

  4. 4

    18 gage × 8 cm polyurethane jugular catheter; Cook Veterinary Products, Bloomington, IN

  5. 5

    Diprivan 10 mg/mL; AstraZeneca S.p.A., Caponago, Italy

  6. 6

    Glucose Pfizer, West Ryde, NSW, Australia

  7. 7

    Aprotinin, Trasylol, Kallikrein Inactivator, 10,000 U/mL; Bayer, Pymble, NSW, Australia

  8. 8

    Olympus 400 Biochemistry Analyzer; Beckman Coulter Australia Pty Ltd, Lane Cove, NSW, Australia

  9. 9

    Phadeseph Insulin RIA; Pharmacia and Upjohn Diagnostics AB, Uppsala, Sweden

  10. 10

    StataCorp, College Station, TX

  11. 11

    Coradini M, Rand JS, Morton JM, Filippich LJ. Delayed gastric emptying may contribute to prolonged postprandial hyperglycemia in meal-fed cats. J Vet Intern Med 2006;20:726–727 (abstract)

Ancillary