Long-term safety of dietary salt: A 5-year ProspEctive rAndomized bliNded and controlled stUdy in healThy aged cats (PEANUT study)

2.1 Study population and husbandry

Twenty-six aged, neutered, domestic shorthair research colony cats were screened for eligibility (Figure 1). At baseline, cats had to be healthy based on standard veterinary evaluations, kidney ultrasonography, standard echocardiography, and conventional cardiac Doppler examination and had to be temperamentally cooperative.

Cats were housed in their randomized groups in an indoor research facility (Unité de Nutrition et d'Endocrinologie, Oniris, F-44307, Nantes Cedex, France) with a 12-h light/dark cycle, and controlled temperature (18°C-21°C) and ventilation (250 m³/h, 12 h/d). From screening onward, cats were fed a maintenance dry expanded diet with a sodium content of 2.3 g/Mcal (Veterinary Cat Neutered, Young Male, Royal Canin SAS, Aimargues, France; Table 1), which was the same diet they had received for most of their life after neutering. Cats had free access to water.

2.2 Study design

Our study was a prospective, randomized, blinded, and controlled trial of a high-salt diet for up to 60 months. The protocol was reviewed and approved by the animal care and use committee of “Pays-de-la-Loire” (file CEEA.2012.152) and by Royal Canin's ethics committee. The study complied with conditions approved by the French Ministry of Agriculture and the guidelines of the Guide for Care and Use of Laboratory Animals.²²

Eligible cats were stratified into 3 subsets based on cardiac 2-dimensional (2D) color tissue Doppler imaging (TDI) at baseline (normal, subnormal, and abnormal).²⁰ Subnormality indicated the presence of regional postsystolic contraction waves for the left ventricular (LV) free wall (LVFW), the interventricular septum (IVS), or both without any other alteration. Abnormality indicated mild to moderate regional diastolic alterations characterized by an early to late diastolic velocity ratio (E : A ratio) < 1.²³ In each subset, cats were ranked and paired based on glomerular filtration rate (GFR). Cats in each pair were randomized to the control or high-salt diet by coin flip. Blinding to diet allocation is detailed in the Data S1.

The high-salt (test) diet contained 3.26 ± 0.30 g/Mcal metabolizable energy (ME; mean, SD) of sodium (1.3% as fed) and 5.71 ± 0.28 g/Mcal ME of chloride (2.27% as fed; Table 1). The control diet had a sodium and chloride content of 1.02 ± 0.16 g/Mcal ME (0.35%) and 2.26 ± 0.33 g/Mcal ME (0.70%), respectively (Table 1). Cats had individual access via programmed collars to 70 g/day of their allocated diet, equivalent to the energy requirements of a 4.8 kg cat in good body condition (100 × bw^0.67).

In the 2-year study reported earlier, we included 3-month and 6-month follow-ups; in the 5-year study, we elected to report only the yearly follow-ups.^{20, 21} Procedures and analyses detailed below were conducted at baseline and months 12, 24, 36, 48, and 60, except for body weight measured weekly and food consumption measured daily.

Cats that completed the study remained on their randomized diet and received standard veterinary care.

2.3 Body condition score

Monthly body condition scoring was planned with a 5-point scale: 1 = emaciated or very thin, 3 = ideal, and 5 = obese or grossly obese.²⁴ This measurement was mistakenly neglected for the first 6 months.

2.4 Hematology, blood biochemistry, and hormone assay

Blood samples were taken by jugular venipuncture and cephalic vein microsampling.²⁵ Plasma biochemistry was performed using dry-slide technology (Vitros 250 Chemistry System, Ortho-Clinical Diagnostics, Raritan, New Jersey). The measurements included albumin, bilirubin, glucose, cholesterol, triglycerides, sodium, potassium, chloride, total carbon dioxide, calcium, phosphate, total protein, urea and creatinine concentrations, and alanine aminotransferase and alkaline phosphatase activities.^26-28 Serum total thyroxine concentration was measured using a validated chemiluminescent immunoassay (Chemiluminescent Immulite 2000, DPC, Los Angeles, California).^26-28

2.5 Urinalyses

Urine samples were collected by cystocentesis. Spot urinalysis included urine specific gravity (USG) measured by medical refractometer (Atago, Fisher Bioblock, Illkirch, France) and urine protein/creatinine ratio (UPC).^{5, 29}

2.6 Glomerular filtration rate

Glomerular filtration rate was determined from the plasma clearance of a nominal dose of 20 mg/kg exogenous creatinine.²¹ Creatinine concentrations before and 5 minutes, 30 minutes, and 1, 2, 3, 5, and 8 hours after an IV bolus of creatinine (anhydrous creatinine, Sigma Chemical Co., St. Louis, Missouri) were analyzed using noncompartmental methods (WinNonlin version 5.2 Pharsight, Mountain View, California). Glomerular filtration rate was calculated as the creatinine dose divided by the area under the creatinine concentration versus time curve; unless otherwise stated, GFR was weight adjusted.

2.7 Blood pressure

Systolic arterial blood pressure (SABP) was measured indirectly in awake cats by the same trained operators using a Doppler system (811-BL Parks Medical Electronics Inc, ALOHA, Oregon).^{30, 31} Systemic arterial hypertension was defined as an average SABP (from 5 readings) >160 mmHg.^{21, 31}

2.8 Renal ultrasonography

A single experienced veterinary echographer (IT) blinded to the diet assignment measured the ultrasonographic aspect, length, width, and height of both kidneys. Resistive indices of renal and interlobar arteries were measured from a longitudinal section of each kidney by Doppler examination using an ultrasonographic unit with a microconvex 8 MHz probe (Biosound MyLab30 Universal Medical Systems Inc, Bedford Hills, New York).²¹ The resistive index was calculated as (peak systolic velocity − end-diastolic velocity) divided by peak systolic velocity.

2.9 Standard echocardiography and conventional Doppler examination

Standard transthoracic echocardiography and conventional Doppler examinations were performed by an experienced board-certified cardiologist (VC) blinded to diet assignment, in awake standing cats, with continuous ECG monitoring, using an ultrasonographic unit (Vivid 7 Dimension, General Electric Medical System, Waukesha, Wisconsin) equipped with 2 phased-array transducers (4-8 MHz and 4.5-11.5 MHz). All procedures and methodologies have been validated earlier.^{20, 30, 32, 33} End-diastolic and end-systolic LV internal diameters, LVFW, and IVS thicknesses were measured from the right parasternal short-axis view with 2D-guided M-mode echocardiography.^{30, 34} Two-dimensional measurements included aortic (Ao) and left atrial (LA) diameters at end-diastole and the end-diastolic subaortic IVS thickness from the right parasternal 5-chamber view at the level of the attachment of the left chordae tendineae to the mitral valve leaflets.^{23, 30} Pulsed wave Doppler mode variables were peak systolic aortic flow velocity and peak early and late diastolic mitral flow velocities (mitral E and A waves, respectively) from the left apical 5- and 4-chamber views, respectively, and isovolumic relaxation time (IVRT), defined as the time interval between the end of aortic flow velocity and the onset of transmitral flow, taken from the left apical 5-chamber view. Left atrial enlargement was defined as end-diastolic LA:Ao >1.2²⁵ and LV enlargement as internal LV diameter > the 95% prediction interval assessed according to body weight from a large population of healthy cats, with LV shortening fraction considered normal if within the corresponding 95% prediction interval.³⁵ The hypertrophic myocardial phenotype was defined as at least 1 2D or M-mode end-diastolic LV myocardial wall thickness ≥6 mm.³⁶

2.10 Two-dimensional color tissue Doppler imaging

Examinations by 2D color tissue doppler imaging (TDI) were undertaken by the same operator blinded to the diet assignment using the same ultrasonographic unit as for standard echocardiography in awake standing cats with continuous ECG monitoring. Digital images were analyzed with specialist software (Echopac Dimension, General Electric Medical System, Waukesha, Wisconsin). Full details are published elsewhere.^{20, 23} In brief, peak myocardial velocities resulting from radial LVFW motion were measured in systole and in early and late diastole (S, E and A waves, respectively) in the subendocardial and subepicardial LVFW segments from the right parasternal ventricular short-axis view.²⁰ Peak systolic and early and late diastolic longitudinal velocities were measured with the standard left apical 4-chamber view in 3 myocardial segments (2 from the LVFW at the base and the apex and 1 from the IVS at the base).²⁰ The TDI diastolic E:A ratio was calculated for each of these 5 myocardial segments. Radial systolic myocardial velocity gradient (MVG) was the difference between subendocardial and subepicardial systolic velocities, and longitudinal systolic MVG was the difference between basal and apical systolic LVFW velocities. Mean heart rate was calculated from ECG monitoring during each radial and longitudinal TDI examination.

2.11 Necropsy examinations

Necropsy examinations, including histopathology, were conducted (JA, LD) on cats that died naturally or were euthanized during the study because of any severe disease resulting in unacceptably poor quality of life.

2.12 Retrospective case review of renal function

After the study end, an expert veterinary nephrologist (JE) blinded to the diet assignment and necropsy findings evaluated serial results of plasma creatinine and urea concentrations, spot USG, and body weight to identify cats with evidence of deteriorating renal function.

2.13 Statistical analysis

Two groups of 10 cats powered the study to detect a 20% change in GFR and blood pressure. There was 85% power to detect effect sizes of 0.4 ± 0.3 mL/min/kg for GFR and 20 ± 15 mmHg for SABP.

Statistical analyses were performed with R Core Team 2022. Descriptive statistics were expressed as median (interquartile range [IQR]) unless otherwise stated. Glomerular filtration rate, blood creatinine concentrations, and SABP, echocardiographic, Doppler, and TDI variables were analyzed for changes over time using linear mixed models (LMMs). Log or rank transformation was used as appropriate to meet statistical assumptions of LMMs (normally distributed residuals and homoscedasticity). Diet, time, and their interaction were fixed effects, and animals were a random effect; time was treated as a continuous variable. The LMMs for body weight (mean per cat for each year) had an additional fixed effect of sex. Effect sizes for GFR, blood creatinine concentrations, SABP, and imaging variables were calculated by pairwise comparisons at each time point with Cohen's d effect size for nontransformed or log-transformed variables, and r Rosenthal's effect size for rank-transformed variables. P ≤ .05 was considered significant.

Kaplan-Meier analysis was performed on events of study discontinuation, such as meeting exclusion criteria, death, or euthanasia for a severe condition. Study survival (ie, ongoing continuation in the study) was compared between diet groups using a log-rank test. The disease frequencies were compared between groups using the Chi-squared test.

3.1 Study population

Twenty healthy neutered cats were enrolled, and 5 male and 5 female cats were randomized to each diet (Figure 1). Groups were similar at baseline for all study variables (Tables 2, 3, and S1-S4). Age and body weight were (median ± [IQR]) 11.0 years (8.0-11.6) and 5.06 kg (4.32-5.64), respectively, for the control group and 11.6 years (11.5-11.6) and 4.51 kg (3.90-4.98), respectively, for the test group. Baseline 2D color TDI was normal for 8 cats, subnormal for 6 cats, and abnormal for 6 cats.^{20, 23} Baseline GFR was 1.9 mL/min/kg (1.8-2.1) in the control group and 1.9 mL/min/kg (1.7-2.1) in the test group.

3.2 Study survival and discontinuations

Figure 1 summarizes study discontinuations. No significant difference was found between diet groups in the probability of remaining in the study (Figure 2, P = .18). Time in the study was 38.7 months (28.6-48.2) for the control group and 51.4 months (45.7-59.0) for the test group. All other statistical analyses excluded data at month 60 because only 4 cats remained in the study.

Necropsy examinations were performed on 6 of 7 cats euthanized during the study because of severe disease, 3 cats that died naturally, 4 cats that completed the 60-month study and were later euthanized, and 1 of 6 cats that were excluded during the study (Figure 1 and Table S5). No cats were euthanized for, or died from, cardiovascular disease or CKD during the study. Two cats were removed from the study because of a cardiovascular or renal disorder: a control cat (case 8; 15.1 years old) excluded after 51.0 months because of persistent systemic arterial hypertension and 1 cat in the test group (case 12; 15.9 years old) that was excluded after 50.5 months because of azotemia (plasma creatinine concentration >180 μmol/L [>2 mg/dL]). The case review found deteriorating renal function indicative of early CKD in 4 nonazotemic cats (2 in each diet group [cases 7, 17, 19, and 25]), which was supported by necropsy findings of CKD. Of these 4 cats, 2 completed the 60-month study (1 control cat euthanized 1.9 months after study end and 1 cat in the test group euthanized 3.6 months after study end, both for CKD), 1 control cat died from lymphoma, and 1 cat in the test group was excluded for megacolon and euthanized 4.4 months later. One cat in the test group (case 13) had no biochemical evidence of renal deterioration during the study but was euthanized 3.5 years after study end with CKD complicated by pneumonia.

3.3 Hematology, plasma biochemistry, and serum thyroxine

Hematologic variables were mainly within the reference intervals (Table S1). Platelet counts often were slightly lower than reference levels with no apparent pattern, except for 1 cat excluded after month 24 with lymphoma (case 4) and another considered to have early CKD (case 19).

Biochemistry variables also were mainly within reference intervals (Table S2). Median sodium concentrations were slightly below the reference interval at months 24 and 36 in the control group and at months 0, 24, 48, and 60 in the test group.

Serum total thyroxine concentrations remained within the laboratory reference interval except for 1 cat in the test group that was diagnosed with hyperthyroidism (serum thyroxine, 7.7 μg/dL) and left the study after month 36.

3.4 Indicators of renal function

Descriptive statistics, rates of change, and effect sizes for renal function variables are presented in Tables 2 and S6. Glomerular filtration rates tended to decrease over time but no significant difference was found in rate of decrease between groups (P = .51) or in effect sizes except at month 12. In all cats, GFR was ≥1 mL/min/kg except 1 cat in the test group (case 25) that had GFR 0.9 mL/min/kg at month 60 and was diagnosed with CKD after the study end. This cat was diagnosed with early CKD in the case review (Table S5). The single cat that developed azotemia (case 12) and the 4 nonazotemic cats identified in the case review with early signs of CKD (cases 7, 17, 19, and 25) had final GFRs between 0.9 and 1.2 mL/kg/min. Of the 15 other cats, 14 had final GFR ≥1.5 mL/min/kg and 1 had GFR of 1.3 mL/min/kg at month 60 (case 13). No significant difference was found between the diet groups in the frequency of CKD (P = .99).

Plasma creatinine concentration decreased over time in the test group, but there was no significant difference compared to the control (P = .58; Table 2) or significant effect size at any time point (Table S6). The case review attributed trends in plasma creatinine and urea concentrations in individual cats to hyperthyroidism, CKD, or early CKD, except for 1 control cat (case 22) with unexplained decreasing USG (2 values <1.030).

Urine specific gravity values decreasing to <1.030 were apparent in cats diagnosed with CKD or considered to have early CKD in the case review (Table S5). Median UPC was slightly below or above 0.2 in both diet groups at all time points.

3.5 Blood pressure

No significant change in SABP over time was found in or between groups (P = .59; Table 2) or in effect sizes at any time point (Table S6). Median SABP for each group was <160 mmHg at all time points (Table 2). The single control cat (case 8) with SABP >160 mmHg (190 mmHg at month 48) was discontinued because of overt systemic hypertension with no biochemical evidence of renal dysfunction (no necropsy examination). In the test group, 3 cats had at least 1 SABP measurement >160 mmHg. One of these cats had SABP of 191 mmHg at month 48 and was diagnosed with hyperthyroidism with no evidence of coexisting renal disease (case 9, no necropsy examination). The second cat (case 13) had SABP of 162 mmHg at month 36, but no evidence of renal disease was found during the study, although GFR decreased at month 60 to 1.3 mL/min/kg (Table S5). When this cat was euthanized because of CKD and aspiration pneumonia 3.5 years after study completion (aged 15.5 years), necropsy examination identified cardiovascular abnormalities consistent with hypertension secondary to CKD. The third cat had SABP of 162 mmHg at months 48 and 60 and case review showed evidence of early CKD, verified by necropsy examination when the cat was euthanized because of CKD 3.6 months after study end. No significant difference was found between the diet groups in frequency of hypertension (P = .99).

3.6 Cardiac and renal imaging

All 2D and M-mode echocardiographic variables (n = 9) remained within reference intervals throughout the study for all cats.^{23, 35} The end-diastolic LVFW thickness tended to decrease over time, but the rate was not significant in or between groups (P = .43; Table 3), and effect sizes were not significant except at month 36 (Table S6). The end-diastolic IVS thickness decreased significantly over time in both groups, but no significant difference was found between them in rate of change (P = .30) or significant effect size at any time point.

Similarly, rates of change of all other 2D and M-mode variables were not significantly different between the 2 groups, except for end-systolic LVFW thickness and end-systole IVS thickness (Table 3). No systolic anterior mitral valve motion, leading to LV outflow tract obstruction, was detected in any cat using both 2D and M-modes. No arrhythmias were detected by ECG.

Peak systolic aortic flow velocity and IVRT remained within reference intervals throughout the study for all cats, with no significant differences between groups in rates of change (Table 3).^{20, 37} The mitral E : A ratio decreased significantly over time in both groups with median values <1 at month 48, but the rate of change was not significantly different between them (P = .56), and effect sizes were nonsignificant (Table S6).

Several TDI E : A ratios decreased significantly (Table S3). In the test group, these were the E : A ratio for the radial LVFW motion measured in the subendocardial and subepicardial segments and the E : A ratio for the longitudinal IVS motion at the base. In the control group, significant decreases were observed in the E : A ratio for the LFVW longitudinal motion at the base and the S wave at the IVS base. The only rates of change that were significantly different between groups were the E : A ratio for the radial LVFW motion measured in the subepicardial segment (P = .01;effect sizes were not significant [Table S6]) and the longitudinal IVS S wave (P = .01;significant effect sizes at months 36 and 48). Descriptive statistics and rates of change are presented for renal resistive indices in Table S4. Rates of change were not significantly different between diet groups. Some individual results were above the reference interval (>72)³⁸ at various time points in 5 control cats (including 1 cat at 2 time points) and 5 test cats (3 cats at 2 time points).

3.7 Food consumption, body weight, and body condition score

Food consumption decreased significantly over time in both groups, but the rate of change was not significantly different between them (−3.6 g/year [95% CI: −6.0 to −1.2, P = .001] for the control group and −1.2 g/year [−3.6 to −0.36, P = .05] for the test group; P = .12; Table 4). Body weight decreased significantly in both diet groups, but the rate of change was not significantly different between them (−93 g/year [−152 to −34, P = .003] vs −78 g/year [−124 to −32, P = .01]; P = .68; Table 4). Median body condition score (5-point scale) (BCS) was 3/5 at months 12, 24, 36, and 48 in both diet groups except for month 48 when the control group had a median BCS of 2/5 (Table 4).