Osmoregulation Performance and Kidney Transplant Outcome

Abstract

Visual Abstract

Background 

Kidney transplant recipients have an impaired ability to dilute urine but seldom develop baseline hyponatremia before ESRD. Although hyponatremia is a risk factor for adverse events in CKD and in kidney transplant recipients, it remains unclear whether subtler alterations in osmoregulation performance are associated with outcome.

Methods 

We studied a single-center prospective cohort of 1258 kidney transplant recipients who underwent a water-loading test 3 months after transplant to determine osmoregulation performance. Measured GFR (mGFR) was performed at the same visit. A group of 164 healthy candidates for kidney donation served as controls. We further evaluated the association of osmoregulation performance with transplantation outcomes and subsequent kidney function.

Results 

Unlike controls, most kidney transplant recipients failed to maintain plasma sodium during water loading (plasma sodium slope of −0.6±0.4 mmol/L per hour in transplant recipients versus −0.12±0.3 mmol/L per hour in controls; P<0.001). Steeper plasma sodium reduction during the test independently associated with the composite outcome of all-cause mortality and allograft loss (hazard ratio [HR], 1.73 per 1 mmol/L per hour decrease in plasma sodium; 95% confidence interval [95% CI], 1.23 to 2.45; P=0.002) and allograft loss alone (HR, 2.04 per 1 mmol/L per hour decrease in plasma sodium; 95% CI, 1.19 to 3.51; P=0.01). The association remained significant in a prespecified sensitivity analysis excluding patients with hyperglycemia. In addition, a steeper plasma sodium slope 3 months after transplantation independently correlated with lower mGFR at 12 months (β=1.93; 95% CI, 0.46 to 3.41; P=0.01).

Conclusions 

Reduced osmoregulation performance occurs frequently in kidney transplant recipients and is an independent predictor of renal outcome.

Kidney transplantation is the preferred treatment for patients with ESRD, because it provides higher survival rates and a better quality of life compared with dialysis.¹ Despite the progress in understanding the multiple processes affecting the allograft kidney, renal function decline and allograft loss remain significant concerns.² The identification of unrecognized targetable processes promoting allograft damage or extrarenal complications remains an important challenge. Dysfunction of the osmoregulatory machinery may represent such risk factor.

The brain-kidney axis plays a central role in the maintenance of constant cell volume through the tight control of extracellular fluid tonicity, which is mainly determined by plasma sodium concentration (P_Na).³ Variations of extracellular tonicity are sensed by a specialized hypothalamic region, which regulates water and salt intake through thirst and salt appetite, as well as renal water excretion through the modulation of vasopressin secretion.³ An increase in plasma tonicity activates hypothalamic osmoreceptors present in the organum vasculosum of the lamina terminalis and the subfornical organ, to promote thirst and vasopressin release.⁴ Through its binding to the V2 receptor (V2R), vasopressin stimulates tubular water reabsorption by increasing the expression of the water channel aquaporin 2 (AQP2) and its insertion in the apical membrane of the principal cells of the collecting duct.⁵ In addition, V2R stimulation increases sodium and urea transport into the medulla, thereby increasing medullary osmolality which is the driving force for water reabsorption in the collecting duct.^6,7 Conversely, in response to water loading, vasopressin secretion is abolished and water diuresis rapidly occurs. The molecular identity of central osmoreceptors remains debated. Both the stretch responsive transient receptor potential cation channel subfamily V member 1 (TRPV1)⁸ and the sodium voltage-gated channel NaX^4,9 have been proposed to mediate osmosensation. However, the invalidation of the genes encoding these proteins in rodents minimally affects osmoregulatory capacity.^10–12 In addition to central osmoreceptors, multiple neurosensory cues that anticipate osmotic changes are integrated by the central nervous system to appropriately promote or impede vasopressin secretion before the occurrence of P_Na modification.¹³

Compromise of osmoregulation has important consequences for health. Hyponatremia due to excessive vasopressin signaling is associated with pejorative outcomes in various pathologic states such as congestive heart failure or cirrhosis.¹⁴ In addition, fluctuation in P_Na may have deleterious consequences for the brain¹⁵ and failure to maintain P_Na is associated with cognitive decline in the elderly.¹⁶ Epidemiologic studies have further linked increased vasopressin level with GFR decline. Indeed, high levels of vasopressin or plasma copeptin, a surrogate marker for vasopressin secretion, are associated with accelerated renal function decline in distinct populations.^17–21 In rats submitted to subtotal nephrectomy or streptozocin-induced diabetic nephropathy, infusion of a V2R agonist increases proteinuria, plasma creatinine, and renal damage, whereas V2R inhibition significantly reduces albuminuria.^22–24 These data suggest a causal role for V2R signaling in CKD progression that echoes the well established deleterious effect of V2R stimulation in autosomal dominant polycystic kidney disease.²⁵

CKD affects the osmoregulatory function of the kidney. First, reduced GFR is associated with reduced urine concentration even at high-vasopressin plasma levels. This defect is the consequence of functional and structural alterations of the counter-current mechanisms which normally allow the maintenance of a corticomedullary osmotic gradient as well as of a reduction of V2R expression in the kidney.^26,27 Conversely, CKD also leads to a reduced ability to produce diluted urine in response to water loading.²⁸ Although the precise mechanisms remain unclear, this phenomenon appears to be linked to excessive vasopressin signaling. Indeed, in patients with CKD, impaired urine dilution in response to water loading correlates with higher level of vasopressin either at baseline or during the test, suggesting the inability to adequately suppress plasma vasopressin.²⁸ Experiments in vasopressin-deficient rats after subtotal nephrectomy indicate that, in the absence of vasopressin, GFR reduction only marginally affects urine dilution capability.²⁴ Similarly, patients with autosomal dominant polycystic kidney disease, with either reduced or preserved GFR, achieve a similar urine osmolality in response to V2R blockage.²⁹ Despite the reduced capacity to dilute urine and the high-vasopressin levels, patients with CKD seldom develop overt hyponatremia before ESRD.³⁰

Although occasional case reports have demonstrated that altered osmoregulation can have deleterious consequences in kidney transplant recipients (KTRs),^31–34 the literature regarding hyponatremia and osmoregulation in this population is remarkably scarce.³⁵ Similar to the general CKD population, hyponatremia associates with adverse outcome in KTRs, but remains a rare condition.³⁶ Yet a small, observational study suggested that a reduced ability to excrete water is not infrequent in KTRs.³⁷ However, the precise prevalence of osmoregulation defects and their effect on renal outcome have not been studied in this population.

To address these questions, we took advantage of the moderate and prolonged water-loading protocol used to obtain hourly urine collection during measured GFR (mGFR) measurement. We hypothesized that defective osmoregulation would result in a defect and/or a delay in urine dilution leading to a reduction in P_Na. We compared the modification of P_Na, urine flow rate, and urine osmolarity (U_osmo) in a large prospective cohort of KTRs and in a control group of healthy candidates to kidney donation that underwent GFR measurement 3 months after transplantation or before kidney donation. We determined the factors associated with impaired osmoregulation capacity in KTRs and studied whether altered osmoregulation capacity is associated with allograft loss, mortality, and subsequent mGFR.

Methods

Study Population

All patients were transplanted and followed in the department of renal transplantation at Necker Hospital, Paris, France. According to our standard follow-up protocol, all patients with a functioning allograft underwent GFR measurement 3 months after transplantation. All but two patients tested from October of 2007 to June of 2016 were included in the study. The two excluded patients were missing ≥1 P_Na value. A control group of healthy adults consisted of candidates for kidney donation tested during the same period as the KTRs. Candidates for kidney donation with mGFR<60 ml/min per 1.73 m² were excluded.

Data Collection

Clinical data were obtained from a prospective database, Données Informatiques Validées en Transplantation (DIVAT clinical prospective cohort, official website: www.divat.fr; registration number: 1016618). Data were entered at specific points for each patient (at day 0, 3 months, and 1 year post-transplantation); follow-up was updated annually thereafter. Research assistants under the supervision of transplant clinicians prospectively entered all data, and the database underwent an annual external audit for data accuracy. For each patient, follow-up data after 1 year were entered on the basis of a complete review of the medical charts and the collection of clinical events.

The collected data are computerized in real time and include information concerning the donor (age, sex, deceased or living donor, cold ischemia time), the recipient (age, sex, primary cause of CKD), the transplantation procedure, the immunosuppression regimen, the occurrence of delayed graft function (defined as the need for hemodialysis during the first week after transplantation), and the occurrence of graft loss as defined by retransplantation or a return to long-term dialysis or death. The clinician in charge of the patients reported death or allograft loss to the research assistant implementing the DIVAT database. A history of acute rejection was defined as the occurrence of an acute rejection episode requiring treatment before GFR measurement. In addition, we prospectively collected height, body weight, BP, and standard biochemical parameters at the time of GFR measurement.

Water-Loading Test

The water-loading test was performed in the day hospital of our physiology department by a team of four nurses specialized in renal physiology testing. A 24-hour urine collection was obtained the day before the test. Patients were deprived of water overnight (from midnight to 8 am). Upon arrival, blood and urine samples were collected and patients were asked to drink 6 ml/kg of water over 30 minutes, followed by 150 ml of water every hour. Hourly blood and urine samples were collected for 5 hours.

Biochemistry

Total calcium, phosphate, urea, standard plasma, and urine ion concentration as well as urinary proteinuria and creatinine were measured by standard biochemical methods. Ionized calcium was measured on an ABL800 analyzer (Radiometer). GFR was measured by iohexol clearance or inulin clearance when allergy to iohexol was suspected on the basis of a patient’s history, as previously described.³⁸ C-terminal FGF23 plasma concentration was measured by ELISA (Immutopics, Inc., San Clemente, CA). Serum parathyroid hormone concentration was measured with an immunochemiluminescent assay performed on the Elecsys analyzer (Roche Diagnostics). Calcitriol and 25-hydroxy vitamin D (25-OH-vitD) concentrations were measured with radioimmunoassay (DiaSorin, Stillwater, MN and IDS Ltd, Boldon, UK). The assays were carried out in a single laboratory throughout the study. Two laboratories of Necker hospital were involved: the biochemistry department (standard biochemistry) and the physiology department (GFR measurement and hormone testing).

P_Na Slopes, Osmolarity, and Renal Free Water Clearance Calculation

P_Na slopes were calculated using the linear regression of the six successive P_Na values measured during the water-loading test and time. Plasma osmolarity (P_osmo), U_osmo, and free water clearance were calculated from P_Na, plasma urea (P_urea), plasma glucose (P_glucose), urine urea (U_urea), urine sodium (U_Na), and urine potassium (U_K) concentrations (mmol/l) using the following formulas:

Free water clearance at time point T=(1−[2×U_osmoT/(P_osmoT-1+P_osmoT)])×urine flow rate_T.³⁹

Statistical Analyses

Data were analyzed using R version 3.5.0 and JMP 9.0.0 (SAS institute). Data are expressed as mean±SD for normally distributed continuous variables and as median (25th–75th percentiles) for continuous variables with a skewed distribution. Categoric variables are expressed as number (percentage).

We analyzed the association of osmoregulation and transplantation outcome. We measured the magnitude of P_Na decrease during water loading as a means to assess the severity of osmoregulation defects. We used the slope of the linear regression of P_Na and time during mGFR measurement (P_Na slope) as a continuous variable reflecting the magnitude of the alteration of osmoregulation kinetics. We used P_Na slope because it takes into account all P_Na measurements performed during the test. We correlated the association of P_Na slope tertiles with death or the composite outcome of death or allograft loss using cumulative incidence functions and log-rank test. To analyze the association of P_Na slope tertile and allograft loss, we took into account death as a competing risk using the Fine and Gray inverse probability of censoring weighting technique.⁴⁰ We used univariate and multivariable Cox proportional hazards or Fine and Grays models to determine relevant parameters for death and the composite outcome of allograft loss and death or allograft loss taking into account death as a competitive risk, respectively.⁴⁰ We tested 14 variables in addition to P_Na slope: recipient age, sex, body mass index, smoking and diabetic status, systolic BP, mGFR measurement, and C-terminal FGF23 level (log-transformed) 3 months after transplantation, as well as donor age and type (cadaveric or living) and HLA graft mismatch (<3 or ≥3), history of acute rejection, pre-emptive transplantation, and delayed graft function. The proportionality assumption was checked for each parameter using the Schoenfeld residuals test. Variables that did not satisfy the assumption were used as stratifying variables in the models with transformation if needed. Association between covariates was systematically tested and significant interactions were included in the final models. Fewer events limited the number of covariates that could be included in the separate competing risk analyses of death-censored allograft loss. Therefore, we only included variables that were associated with the outcome in univariate analysis and/or clinically pertinent. In the sensitivity analysis, in order to address whether the association between P_Na slope and transplantation outcome is independent of the presence of hyperglycemia, we repeated the main analyses after excluding patients with either missing blood glucose measurement (n=51) or a blood glucose >11 mmol/l (n=329) at any of the time points used for P_Na slope determination. All clinical and biologic parameters included in the multivariable models were measured/assessed at the inclusion of the patients in the study (i.e., at GFR measurement 3 months after KT).

Because of variations in our policy regarding urinary protein excretion measurement during the study period, urinary protein-to-creatinine ratios were only available on the day of GFR measurement in 929 patients. We therefore performed additional multivariable analyses in this population to assess whether the association between P_Na slope and specific outcomes was dependent on urinary protein excretion rate.

In a complementary analysis, we studied the association of P_Na slope at 3 months with the mGFR at 12 months after transplantation (12-month mGFR). We used univariate and multivariable linear regression models to determine significant associations. In a univariate analysis, we tested the association of 12-month mGFR with 14 variables in addition to P_Na slope: donor age and status (cadaveric or living); HLA graft mismatch (<3 or ≥3); pre-emptive transplantation; delayed graft function; history of acute rejection; and recipient age, sex, body mass index, smoking and diabetes status, systolic BP, mGFR, and 25-OH-vitD level.

Results

KTRs Have Altered Osmoregulation in Response to Water Loading

From October of 2007 to June of 2016, 1258 KTRs and 164 healthy candidates to kidney donation meeting the inclusion criteria underwent mGFR measurement associated with water loading. The characteristics of these patients are given in Table 1 and Supplemental Table 1. We first compared the global variations of mean P_Na, urine flow, and urine osmolality during the exploration in the two populations (Figure 1). Fasting morning mean P_Na concentration was slightly yet significantly lower in KTRs (139±3 mmol/l versus 140±2 mmol/l; P<0.001). Controls and KTRs showed marked differences in their response to water loading. In controls, after an initial decrease, mean P_Na remained stable (139±2 mmol/l). In KTRs, mean P_Na decreased throughout the test to reach 136±3 mmol/l. At baseline, hyponatremia (P_Na<135 mmol/l) was not observed in controls but was observed in 75 KTRs (6%). At the end of the test, P_Na<135 mmol/l was observed in 365 KTRs (29%) but only one of the controls (Supplemental Figure 1). Mean urine osmolality decreased and mean urine flow rate increased in both groups. However, these changes were delayed and less pronounced in KTRs compared with controls (Figure 1). In patients with diabetes mellitus, high blood glucose moves water out of cells and thus decreases P_Na.⁴¹ To exclude this phenomenon, we restricted our analysis to the 878 patients with P_glucose constantly <11 mmol/l during the exploration and observed essentially the same results (Supplemental Figure 2, Supplemental Table 2). Taken together, these results demonstrate that, compared with healthy adult subjects, KTRs have reduced osmoregulation capacity with a decreased ability to excrete free water, lowering P_Na.

Table 1. - KTR population characteristics according to P_Na slope tertiles

Characteristic	All Patients (n=1258)	Tertile 1 (n=419)	Tertile 2 (n=419)	Tertile 3 (n=420)	P Value a
PNa slope, mmol/l per hour	−0.62 (−0.91–0.33)	−0.22 (−0.33–0.06)	−0.61 (−0.71–0.54)	−1.04 (−1.26–0.91)
Age at transplantation, yr	51±14	45±14	50±14	55±14	<0.001
Male sex	749 (60%)	254 (61%)	263 (63%)	232 (55%)	0.07
Initial nephropathy					0.85
Cystic kidney disease	214 (17%)	70 (17%)	70 (17%)	74 (18%)
GN	175 (14%)	58 (14%)	63 (15%)	54 (13%)
FSGS	85 (7%)	27 (6%)	33 (8%)	25 (6%)
Diabetic nephropathy	109 (9%)	29 (7%)	31 (7%)	49 (12%)
Vascular nephropathy	71 (6%)	19 (5%)	30 (7%)	22 (5%)
CAKUT	52 (4%)	18 (4%)	19 (5%)	15 (4%)
Uropathy	45 (4%)	17 (4%)	14 (3%)	14 (3%)
Systemic lupus	30 (2%)	12 (3%)	8 (2%)	10 (2%)
Toxic nephropathy	22 (2%)	5 (1%)	6 (1%)	11 (3%)
Genetic glomerular diseases	25 (2%)	10 (2%)	8 (2%)	7 (2%)
Hemolytic and uremic syndrome	21 (2%)	9 (2%)	5 (1%)	7 (2%)
Myeloma	3 (0%)	1 (0%)	1 (0%)	1 (0%)
Others	117 (9%)	43 (10%)	38 (9%)	36 (9%)
Unknown	289 (23%)	101 (24%)	93 (22%)	95 (23%)
Pre-emptive transplantation	184 (15%)	75 (18%)	54 (13%)	55 (13%)	0.07
First transplantation	1219 (97%)	407 (97°%)	404 (4%)	408 (97%)	0.79
Active smoking	146 (88%)	50 (12%)	52 (12%)	44 (10%)	0.66
Diabetes mellitus	395 (31%)	104 (25%)	126 (30%)	165 (39%)	<0.001
Living donor	302 (24%)	138 (33%)	98 (24%)	66 (16%)	<0.001
Donor age, yr	54±16	47±16	54±17	58±16	<0.001
Cold ischemia duration, min	1007 (480–1403)	900 (150–1245)	1031 (529–1500)	1068 (761–1502)	<0.001
Delayed graft function	273 (22%)	70 (17%)	95 (23%)	108 (26%)	0.006
HLA mismatch A, B, and DR ≥3	821 (67%)	254 (63%)	283 (69%)	284 (69%)	0.08
History of treated acute rejection	168 (13%)	52 (12%)	57 (14%)	59 (14%)	0.77
Body mass index, kg/m2	24±4	24±4	24±5	25±5	0.17
Systolic BP, mm Hg	134±17	132±16	133±17	136±17	0.006
Diastolic BP, mm Hg	76±10	78±10	77±10	75±11	<0.001
mGFR, ml/min per 1.73 m2	55±16	61±16	54±16	50±16	<0.001
Hemoglobin, g/dl	11.9±1.6	12.1±1.5	11.9±1.6	11.7±1.6	0.001
Ionized calcium, mmol/l	1.25±0.09	1.26±0.11	1.26±0.09	1.24±0.1	0.05
Total calcium, mmol/l	2.39±0.17	2.41±0.15	2.40±0.16	2.38±0.19	0.05
Serum phosphate, mmol/l	0.90±0.25	0.89±0.24	0.91±0.24	0.91±0.27	0.45
25-OH-vitD, µg/l	21 (14–28)	21 (13–28)	21 (15–28)	22 (14–29)	0.91
Calcitriol, µg/l	48 (33–69)	48 (33–73)	50 (34–69)	47 (32–64)	0.22
Parathyroid hormone, ng/l	103 (67–162)	94 (67–151)	107 (65–158)	105 (71–184)	0.06
FGF23, RU/ml	100 (67–146)	90 (63–125)	100 (64–146)	113 (77–181)	0.001
Immunosuppressive regimen					0.02
Tacrolimus	17 (83%)	352 (87%)	319 (81%)	311 (80%)
mTOR inhibitor	9 (1%)	2 (0%)	1 (0%)	6 (2%)
Cyclosporine	182 (15%)	50 (12%)	69 (17%)	16 (16%)
Belatacept	17 (1%)	2 (0%)	6 (2%)	9 (2%)

Categoric variables are described as numbers (%) and continuous variables as mean±SD or median (interquartile range), as appropriate. CAKUT, congenital anomalies of the kidney and urinary tract; 25-OH-vitD, 25-hydroxy vitamin D.

^aP value represents tests of significance from t test, Mann–Whitney test, or chi-squared test, as appropriate.

Figure 1.: Kidney transplant recipients have altered osmoregulatory response to water loading: variation of mean P_Na, urine flow rate, urine osmolality, and renal free water clearance during water loading in KTRs (red) and healthy candidates to kidney donation (controls in blue). Bars indicate 95% confidence intervals. t test: *P<0.05, **P<0.01, ****P<0.001.

We next analyzed the osmoregulation of individual patients. Inspection of individual P_Na trajectories during the test revealed important interindividual heterogeneity in KTRs. Indeed, as exemplified in Figure 2A, some patients maintained a constant P_Na concentration, whereas others experienced a linear decrease of P_Na indicating impaired osmoregulation. Of note, a subset of patients with baseline hyponatremia maintained a constant P_Na during the test, suggesting a reset osmostat (Figure 2A). We then focused on the slope of the linear regression of P_Na and time (P_Na slope) to quantitatively assess osmoregulation in individual patients (Figure 2A). The density plots of P_Na slope in controls and KTRs are shown in Figure 2B. Whereas control patients had a narrow distribution of P_Na slope with a mean around 0 mmol/l per hour (−0.12±0.3 mmol/l per hour), KTRs showed a wider distribution centered on a mean P_Na slope of −0.6±0.4 mmol/l per hour. Indeed, half of KTRs but only 7% of controls had a P_Na slope below −0.6 mmol/l per hour. As expected, in KTRs, P_Na slope correlated significantly with minimal urine osmolality (r=−0.32; P<0.001; Figure 2C) and urine flow rate (r=0.41; P<0.001; Figure 2D), indicating that a steep reduction in P_Na during water loading was associated with a reduced ability to decrease urine osmolality and to increase urine flow (i.e., with a reduced ability of the kidney to excrete free water). The 73 patients with baseline P_Na<135 mmol/l (6%) did not display significantly steeper P_Na slopes than patients with baseline P_Na>135 mmol/l (−0.59±0.54 mmol/l per hour versus −0.64±0.43 mmol/l per hour; P=0.44). In contrast, the 365 patients that reached a P_Na<135 mmol/l at the end of the test showed a steeper P_Na slope than the rest of the cohort (−0.89±0.43 versus −0.53±0.39; P<0.001). Together, these results show that an important fraction of KTRs present altered osmoregulatory response to water loading. In these patients, the impaired ability of the renal allograft to rapidly excrete water leads to a progressive reduction in P_Na. Our results further indicate that the magnitude of this defect can be quantitatively assessed using P_Na slope.

Figure 2.: P_Na slope is an indicator of osmoregulation performance. (A) Examples of P_Na variations in four patients with normal or low initial P_Na and normal (gray lines) or altered osmoregulation kinetics (green lines). The solid lines represent observed P_Na variations. The dashed lines represent the linear regression of P_Na concentration and time. The slopes of the regression lines are indicated on the right. (B) Distribution of P_Na slopes in KTRs and control patients. (C and D) Linear regression of P_Na slopes and the minimal urine osmolality observed during the test (C) or the mean urine flow rate observed throughout the test (D).

Risk Factors for Altered Osmoregulation in KTRs

We then investigated the determinants of altered osmoregulation in KTRs. The characteristics of the KTR population according to tertiles of P_Na slope are presented in Table 1. Compared with the higher tertiles (tertiles 1 and 2), patients with the steepest P_Na slope (tertile 3) were older and had lower diastolic and higher systolic BP, lower mGFR, lower hemoglobin, lower total and ionized blood calcium, and higher FGF23 concentrations. They were more likely to have diabetes mellitus and a prior history of delayed graft function. Furthermore, the following characteristics of donors were associated with the lowest P_Na slope tertile: older age, deceased donor, and increased cold ischemia duration. Multivariable linear regression revealed that lower mGFR, older donor age, diabetes mellitus, and, to a lesser extent, older recipient age were independent predictors of a steep P_Na slope (Table 2).

Table 2. - Factors associated with P_Na slope

Characteristic	Standardized β (95% CI)	P Value
Donor type (cadaveric)	−0.06 (−0.12 to 0)	0.06
Donor age, yr	−0.12 (−0.19 to −0.05)	0.002
Recipient age, yr	−0.09 (−0.17 to −0.01)	0.02
Delayed graft function (no)	0.04 (−0.02 to 0.1)	0.21
mGFR, ml/min per 1.73 m2	0.12 (0.05 to 0.19)	0.001
Diabetes mellitus (yes)	−0.08 (−0.14 to −0.02)	<0.01
Systolic BP, mm Hg	−0.02 (−0.09 to 0.05)	0.55
Diastolic BP, mm Hg	0.05 (−0.02 to 0.11)	0.14
Hemoglobin, g/dl	0.05 (−0.01 to 0.11)	0.11
logFGF23 C-terminal, RU/ml	−0.05 (−0.11 to 0.01)	0.10
Immunosuppressive regimen (belatacept)	0.06 (−0.1 to 0.22)	0.45
Immunosuppressive regimen (cyclosporine)	0.08 (−0.08 to 0.24)	0.35
Immunosuppressive regimen (mTOR inhibitors)	−0.14 (−0.33 to 0.05)	0.16

Multivariable linear regression using P_Na slope as the dependent variable. 95% CI, 95% confidence interval.

Association of Baseline Osmoregulation with Transplantation Outcome

The median follow-up time was 49 months (25th–75th percentiles: 22–80 months). At the end of the study, 29 patients were lost to follow-up, 83 patients had lost their allograft, 120 patients had died with a functioning allograft, and 11 had died after losing their allograft. Cumulative incidence functions for mortality, graft loss, and their composite according to tertiles of P_Na slope are shown in Figure 3. In univariate analysis, steeper P_Na slope tertiles were associated with a higher incidence of death, allograft loss, and the composite outcome. We further used univariate and multivariable Cox modeling to assess the hazard ratio of adverse outcome associated with P_Na slope (Table 3). After adjustment for multiple potential confounding variables, P_Na slope remained significantly associated with the composite outcome and with allograft loss alone but not with death. P_Na slope remained significantly associated with the composite outcome and allograft loss in additional Cox models taking into account the urinary protein-to-creatinine ratio that was available for 929 patients (Supplemental Table 3). In a sensitivity analysis, we focused on the population that did not show hyperglycemia during the exploration (i.e., a blood glucose superior to 11 mmol/l at any of the time points). In this subpopulation, P_Na slope remained significantly associated with allograft loss and the composite outcome (Supplemental Table 3, Table 3).

Figure 3.: Steeper P_Na slopes associate with greater incidence of mortality and allograft loss in KTRs. (A–C) Cumulative incidence plots of P_Na slope tertiles and the composite outcome of all-cause mortality and allograft loss (A), all-cause mortality (B), and allograft loss (C).

Table 3. - Cox uni- and multivariable modeling of the association of continuous P_Na slope values with transplantation outcome

Outcome and studied population	Per 1 Unit Decrease of PNa Slope	P Value
Composite outcome of all-cause mortality and allograft loss
All patients
Crude hazard ratio	2.54 (1.85 to 3.50)	<0.001
Adjusted hazard ratio a	1.73 (1.23 to 2.45)	0.002
Restricted to patients without uncontrolled diabetes
Crude hazard ratio	3.32 (2.18 to 5.06)	<0.001
Adjusted hazard ratio b	2.47 (1.51 to 4.04)	<0.001
Outcome of all-cause mortality
All patients
Crude hazard ratio	2.45 (1.65 to 3.64)	<0.001
Adjusted hazard ratio a	1.52 (0.99 to 2.34)	0.06
Restricted to patients without uncontrolled diabetes
Crude hazard ratio	3.30 (1.93 to 5.64)	<0.001
Adjusted hazard ratio b	2.20 (1.13 to 4.30)	0.02
Outcome of allograft loss with mortality as a competing risk
All patients
Crude hazard ratio	2.60 (1.58 to 4.26)	<0.001
Adjusted hazard ratio c	2.04 (1.19 to 3.51)	0.01
Restricted to patients without uncontrolled diabetes
Crude hazard ratio	3.12 (1.69 to 5.74)	<0.001
Adjusted hazard ratio d	2.73 (1.39 to 5.34)	0.003

Data are shown as hazard ratio (95% confidence interval).

^aAdjusted for: patient age, diabetes, smoking status, donor status and age, pre-emptive transplantation, delayed graft function, mGFR, body mass index, systolic BP, and logFGF23.

^bAdjusted for: patient age, smoking status, donor status and age, pre-emptive transplantation, delayed graft function, mGFR, body mass index, systolic BP, logFGF23, and HLA mismatch.

^cAdjusted for: patient age, smoking status, donor status and age, pre-emptive transplantation, mGFR, body mass index, systolic BP, and logFGF23.

^dAdjusted for: patient age, smoking status, donor status and age, mGFR, logFGF23, and HLA mismatch.

Association of P_Na Slope with Subsequent mGFR

Of 1258 patients that underwent mGFR measurement and P_Na slope 3 months after transplantation, 943 (75%) had a second measurement of mGFR 9 months later (i.e., 12-month mGFR). Table 4 presents the association of baseline P_Na slope with 12-month mGFR using univariate and multivariable analyses. A steep P_Na slope 3 months after transplantation associated with a lower 12-month mGFR independently of baseline mGFR and other potential confounding variables (Supplemental Tables 4 and 5, Table 4).

Table 4. - Association of baseline (3 mo) P_Na slope with 12-mo mGFR

Studied population	β Coefficient	P Value
All patients
Crude	9.22 (6.96 to 11.48)	<0.001
Adjusted a	1.93 (0.46 to 3.41)	0.01
Patients with available urinary protein-to-creatinine ratio
Crude	9.54 (6.96 to 12.12)	<0.001
Adjusted b	1.88 (0.18 to 3.59)	0.03

Univariate and multivariable linear regression analysis using 12-mo mGFR as the dependent variable. Data are shown as β coefficient (95% confidence interval).

^aAdjusted for recipient age, sex, body mass index, diabetes, active smoking, systolic BP, 3 mo-mGFR, 25-OH-vitamin D level, donor age, donor status, history of acute rejection, HLA mismatch, delayed graft function, and pre-emptive transplantation.

^bAdjusted for recipient age, sex, body mass index, diabetes, active smoking, systolic BP, 3 mo-mGFR, 25-OH-vitamin D level, donor age, diabetic status, history of acute rejection, HLA mismatch, delayed graft function, pre-emptive transplantation, and urinary protein-to-creatinine ratio.

Discussion

In this retrospective analysis of a large prospective cohort of KTRs, reduced osmoregulation during a standardized water-loading test 3 months after transplantation was independently associated with both allograft loss and reduced allograft function 12 months after transplantation.

We used a standardized water-loading test, which was performed during a routine mGFR measurement 3 months after kidney transplantation to assess osmoregulation in KTRs. This water-loading protocol differs from the usual test used to assess urine dilution capacity.²⁸ Although the latter involves a single administration of a large amount of water followed by hourly urine collection to assess water excretion rate, our protocol consisted of the repeated administration of moderate water load over 5 hours. Notably, whereas a single water load does not result in a sustained reduction of P_Na over time, we observed that the repeated administration of water induced a progressive reduction in P_Na in a substantial fraction of KTRs, unveiling a defect in osmoregulation.³ Using the slope of the linear regression of P_Na over time to quantitatively assess osmoregulation capability in a large population of KTRs and controls, we observed that, contrary to healthy adults, an important fraction of KTRs failed to maintain P_Na during water loading. Failure to maintain P_Na correlated with a reduced and delayed ability to excrete diluted urine, indicating altered renal osmoregulatory response. Furthermore, a steep P_Na slope was associated with the development of overt hyponatremia in 29% of the KTRs. Our results further sustain that the assessment of P_Na slope is a reliable indicator of osmoregulation performance in KTRs. Of note, P_Na slope is not affected by urine collection errors, including urine loss or incomplete bladder voiding, which represents a major pitfall for the proper estimation of more complex parameters used to assess renal response to water loading, such as free water clearance.

One important methodologic aspect of our study is that we focus on P_Na and not on plasma osmolality to assess osmoregulation. Although the precise nature of the parameter(s) targeted by osmoregulation remains debated, the effects of different solutes on thirst and vasopressin secretion have long been studied.^42–45 The results of these seminal experiments indicate that, in most conditions, P_Na is more relevant for osmoregulation than osmolality. Indeed, in KTRs as in most patients, the main determinants of plasma osmolality are P_Na, urea, and glucose. Although modification of P_Na triggers osmoregulatory response resulting in homeostatic adjustment of thirst and vasopressin secretion, this is not the case for P_urea or glucose.⁴³ Contrary to sodium, urea almost freely permeates the plasma membrane⁴⁶; therefore, urea does not induce sustained water movement across the cell membrane and is less efficient than sodium in inducing vasopressin secretion.^43,45 Contrary to urea, high blood glucose does induce water movement across the cell membrane, at least in cells with low glucose premeability.³⁸ However, it is well established that high blood glucose per se does not trigger vasopressin secretion or thirst in healthy subjects or patients with diabetes.^42–44 Therefore, P_Na maintenance more accurately reflects osmoregulation performance than osmolality. For instance, large fluctuations in blood glucose levels are not rare in KTRs with poorly controlled diabetes. In these cases, the corresponding modifications of osmolality reflect altered glycoregulation and not osmoregulation.

Having established that a large fraction of KTRs have an impaired osmoregulatory response to water loading, we investigated the determinants of this defect and found that reduced mGFR, increased donor age, and diabetes were the main independent factors associated with steeper P_Na slopes. In patients without transplants, these three conditions have been associated with increased vasopressin plasma levels and reduced ability to excrete water.^18,28,47 The mechanisms responsible for increased vasopressin plasma level in these conditions remain unclear. In aging rats, increased vasopressin release is in part due to an age-related increase in IL-6 production by microglial cells, resulting in higher sensitivity of vasopressin-producing cells to P_Na.⁴⁸ Whether similar mechanisms exist in KTRs, or more generally in CKD, has not been assessed. Furthermore, the effect of renal disease on the recently identified neural circuitries allowing the anticipation of P_Na decrease to regulate vasopressin secretion remain unknown.¹³

A major finding of this study is the link between altered osmoregulation and renal outcome in KTRs. Indeed, we observed that altered osmoregulation performance 3 months after transplantation independently associated not only with allograft loss, but also with a reduced mGFR at 12 months. To date, the concept that osmoregulatory mechanisms could affect allograft fate was only supported by reports showing an association between high copeptin (as a surrogate measure of increased vasopressin secretion) or reduced P_Na level and GFR decline⁴⁹ or adverse outcomes,³⁶ respectively. However, these studies did not establish a direct link between osmoregulation and KTR outcome. Although copeptin is secreted in equimolar amounts to vasopressin, this molecule does not represent a good surrogate for vasopressin level in KTRs. Indeed, with decreased GFR, copeptin levels increase to a higher extent than vasopressin levels, although the mechanism responsible for this difference remains unclear.^50,51 Therefore, a high copeptin level in KTRs may not reflect a proportional increase in the vasopressin level and the significance of increased copeptin in this population remains to be fully understood. Conversely, hyponatremia may be due either to a defect of osmoregulation or to a reset of the physiologic set point for P_Na (reset osmostat). Our results do not indicate that baseline hyponatremia is a reliable marker for altered osmoregulation performance. First, fasting hyponatremia is a rather rare condition in KTRs (6% in our cohort) compared with the high prevalence of osmoregulation defects revealed by this study. Second, we observed that baseline hyponatremia was not associated with a steeper P_Na decrease, suggesting that the ability of hyponatremic KTRs to cope with water loading is essentially similar to normonatremic KTRs. In contrast to the measurement of baseline copeptin and P_Na, our study, which assessed P_Na kinetics in response to water loading, directly links decreased osmoregulation capability with adverse renal outcome and GFR decline in KTRs. Because this study is observational in nature, we can only speculate on the putative mechanisms underlying this association. Considering the established role of V2R signaling in renal disease progression, it can be hypothesized that altered osmoregulation kinetics in KTRs reflect persistent vasopressin signaling that leads, in turn, to kidney damage. However, in addition to vasopressin signaling, several hormonal and kidney-specific aspects, such as decreased fluid delivery to the collecting duct or decreased sodium reabsorption in the distal tubule, may reduce the kidneys ability to excrete water.^52–54 Thus, independently of vasopressin, a reduced ability of the kidney to promptly excrete water could reflect specific tubular injury that affects urine dilution capacity and predates kidney function decline. Because we did not measure baseline vasopressin level or its decrease during water loading, our study does not provide insight regarding the mechanisms underlying defective osmoregulation in KTRs, and additional investigations are required to understand the molecular bases underlying this phenomenon.

An important question raised by our data is the effect of osmoregulation defects on the daily variation of P_Na and its subsequent effect on patient health. Indeed, KTRs are advocated to maintain high water intake to prevent dehydration, especially during the early post-transplantation period. Our results suggest that such a policy will produce substantial daily fluctuation in P_Na, at least in a fraction of KTRs. We did not find any consistent association between defective osmoregulation and death. However, the consequences of important daily P_Na fluctuations on subtler outcome such as cognitive decline need to be assessed.

Our study has potential limitations. First, as mentioned above, we did not measure plasma vasopressin during water loading. In addition, the monocentric design of our study, which includes only KTRs, may hinder the extrapolation of the results to the general CKD population. In addition, because of the specificity of our institution, which combines pediatric and adult nephrology departments, our KTR population includes an unusually high proportion of genetic kidney diseases and a low proportion of patients with diabetes. However, the simplicity of P_Na slope measurement should allow replication studies in centers using mGFR to monitor KTRs or patients with CKD. Finally, as stated above, the observational nature of this study does not allow for inferring causality, only association.

Our study has, however, significant strengths. This is the first study to assess osmoregulation performance in a large, prospective cohort of KTRs, with an important follow-up allowing the use of multivariable analysis to identify independent predictors of transplantation outcome.^55,56 We further used multiple adjustment variables in our analysis, including mGFR and FGF23, two important factors affecting the results of multivariable analysis aiming at the identification of factors affecting transplantation outcome.^55,56 In addition, we used a functional dynamic test assessing the ability of the brain-kidney axis to maintain constant plasma tonicity rather than a single measurement of baseline parameters. Finally, the inclusion of multiple analytic strategies and complementary outcome measurement regarding allograft functions (allograft loss and 12-month GFR) gives consistency to our results.

In conclusion, our study identified defective osmoregulation dynamic as a novel prognostic factor influencing kidney transplantation outcome. Whether the molecular events underlying defective osmoregulation are directly involved in the pathophysiologic processes leading to allograft dysfunction remains to be established.

Disclosures

Dr. Brazier reports personal fees from Amgen, outside of the submitted work. Prof. Legendre reports other from Novartis, other from AStellas, and other from CSL Behring, outside of the submitted work. Dr. Bienaimé reports receiving honoraria from Otsuka for two lectures on the pathophysiology of autosomal dominant polycystic kidney disease in 2017. All of the remaining authors have nothing to disclose.