Ecacy of tolvaptan on nephrotic patients with diuretic-resistant edema: a pilot study

Background Tolvaptan (TLV), a vasopressin type 2 receptor antagonist, is an effective drug for heart failure without worsening renal function. However, the effect of TLV on nephrotic syndrome remains unknown. This study aims to assess the association between response to TLV and clinical parameters including pathological evaluation in nephrotic patients with diuretics-resistance. Methods For prospectively enrolled patients, TLV was added for 14 days after renal biopsy performed. We evaluated the effect of TLV on urine output (UO) and body weight (BW) and the correlation of UO change with BW, urine and blood measurements and pathological evaluation. Pathological evaluation included glomerular sclerosis, tubulointerstitial injury, interstitial brosis and the degree of aquaporin 2 (AQP2) positivity in collecting ducts.


Background
Tolvaptan (TLV) is an antagonist of vasopressin type 2 receptor and promotes water diuresis. The rst study reports that TLV is effective for the syndrome of inappropriate secretion of antidiuretic hormone (SIADH) to correct hyponatremia 1 . And now TLV plays an important role in treating patients with acute and chronic heart failure refractory to other diuretics 2 . Responders to TLV experienced not only a better clinical course but better prognosis in both short-and long-term observation 3 . Moreover, TLV has no adverse effect on neurohormonal activation or renal activation while other diuretics can exaggerate neurohormonal imbalance and cause vasoconstriction with the reduction of renal perfusion. The previous clinical trial showed TLV helped weight loss and better relief of dyspnea without worsening renal function in heart failure patients 4 .
Nephrotic syndrome is also a state of uid overload with decreased urine volume and often accompanied with renal dysfunction. Although the pathophysiology of sodium and water retention in nephrotic syndrome has not been precisely clari ed, activation of the renin-angiotensin aldosterone system and sympathetic nervous system is referred as a part of causes. Under lled circulation to kidneys activates these neurohormonal systems and may worsen renal function 5 . Vasopressin type 2 receptor antagonists, which are not expected to cause neurohormonal imbalance, will be an ideal agent to treat hypervolemia in nephrotic syndrome without worsening renal function 6 .
Although some case reports showed TLV was effective in patients with nephrotic syndrome 7,8 , it has not been fully elucidated which nephrotic patients are responsive to TLV. Also, nephrotic patients sometimes have renal impairment 9,10 , which is one of the major risk factors for resistance to TLV 11 . Thus, it is required to evaluate whether nephrotic syndrome is responsive to TLV and whether the response to TLV can be predicted by urine, blood or pathological test.
This study aims to evaluate 1) responsiveness for TLV treatment and 2) associating factors including serum and urinary biochemical parameters, urinary aquaporin (AQP) 2, serum TLV concentration, renal pathological ndings, and renal AQP2 expression for TLV effectiveness, in patients with diureticsresistant nephrotic syndrome.

Patient enrollment
The inclusion criteria were (a) nephrotic syndrome and (b) uid overload resistant to diuretics. Nephrotic syndrome was de ned as > 3.5 g/day proteinuria and hypoalbuminemia (< 3.0 g/dL). Diuretic resistance was de ned that volume overload remained even when furosemide dose was 40 mg or more. The exclusion criteria were patients with allergy to vaptans, anuria (< 200 mL/day), di culty to free water intake, hypernatremia (> 147 mEq/L), or pregnancy.

Study Protocol
This was a prospective exploratory study conducted from January through December in 2014. The study design is shown as Supplemental Figure S1. Evaluation of the degree of edema, chest X-ray, echocardiogram, and renal biopsy was performed at the screening phase before TLV administration. Then TLV 7.5 mg per day was added on enrolled patients for 14 days. During the trial, other diuretics than TLV were continued with unchanged doses. If urine output did not increase or symptoms were not improved for two days after 7.5 mg of TLV administration, the dose of TLV could be increased to 15 mg as needed. During TLV administration, water intake was not restricted to prevent volume depletion.
Body weight, daily urine output, blood pressure and pulse rate were checked every day. Serum and urine osmolality, serum sodium and creatinine were measured before TLV administration, and at 4hr, 8hr, day2, 3, 5, 8, and 15 after TLV medication.
Patients were followed up until day 15 if TLV was effective and safe. Follow up was nished at the time when symptoms were not improved or exaggerated and continuing trial was considered as a risk during the trial.
This study protocol was approved by Tokushukai Group Institutional Review Board (TGE00343-024) and adhered to the Declaration of Helsinki and to CONSORT guidelines. Informed consents were obtained before the study registration by written form. This study was also registered in UMIN-CTR (UMIN000011763).
Urine AQP2 before TLV administration was determined using sandwich enzyme-linked immunosorbent assay (human AQP2 ELISA kit, LSI Medience) as described previously 12,13 . Serum TLV concentration was evaluated at 4 hour after TLV administration on day 1, and determined using a validated highperformance liquid chromatography-tandem mass spectrometry method at Toray Research Center, Inc.
Details have been described previously 14 . Plasma ADH before TLV administration was determined using commercially available enzyme-linked immunosorbent assay (ELISA) kits

Pathological evaluation
To evaluate the collagen deposition, 2 µm sections of para n-embedded tissue were subjected to Masson Trichrome staining (MTS) by routine procedures. Stained sections were examined by Olympus BX50 epi uorescence microscope equipped with a digital camera DP73 (Tokyo, Japan). To estimate brotic area, computer-aided morphometric analysis on MTS sections was performed as described previously 15 . Brie y, a grid containing 117 (13 × 9) sampling points was superimposed on images of a cortical high-power eld (x400). The number of grid points overlying the MTS-positive area was counted and expressed as a percentage of all sampling points. For each kidney specimen, 10 randomly selected non-overlapping elds were analyzed.
Tubulointerstitial injury in the cortex was analyzed histomorphometrically by counting the number of tubules that demonstrated vacuolar degeneration, chromatin condensation of tubular nuclei, tubular dilatation and thickened and/or wrinkled tubular basement membranes divided by the number of total tubules per eld (× 200) in 10 randomly selected cortical elds per cross Sect. 17 .
To compare the histological features in participants, three patients with minor glomerular abnormality and normal renal function were evaluated as controls (N = 3). The pathological evaluation was performed by well-trained nephrologists.
Immunohistochemical staining and evaluation of AQP2 immunostaining Two micrometer of formalin-xed specimen were used for AQP2 immunostaining. After depara nization, endogenous peroxidase activity was blocked by 0.3% H 2 O 2 in methanol for 20 minutes at room temperature. Then sections were incubated with Rabbit anti-human AQP2 (C-17) antibody (1:100 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) for primary antibody for an hour at room temperature. Bound antibodies were detected with peroxidase-conjugated goat anti-rabbit antibody (Histo ne Simple Stain MAX PO; Nichirei Biosciences, Tokyo, Japan) using diaminobenzidine-tetrahydrochloride (DAB) (Simple Stain DAB solution; Nichirei Biosciences, Tokyo, Japan) as the substrate for 30 minutes at room temperature. The sections were counterstained with hematoxylin. The normal part of the whole kidney with clear cell carcinoma was used for normal control.
We de ned the proportion of AQP2 staining as the proportion of the number of stained cells divided by the number of all cells in a sectional collecting duct. The example in Supplemental Figure S2 shows ve AQP2-positive cells (red arrows) of twelve cells so that the proportion of AQP2 staining is 0.42. AQP2 positivity was evaluated in all collecting ducts in each kidney specimen in each patient.

Statistical analysis
Data were presented as median (interquartile). Wilcoxon ranked test was used for rejecting the null hypothesis that the change of UO and BW was not signi cant. Wilcoxon ranked test was used when comparing between time-serial data on osmolality and electrolytes in urine and serum because all data were not normally distributed. The existence of correlation between two continuous variables was determined using Spearman correlation. We did statistical analyses by JMP Pro 11 for Windows (SAS Institute Japan, Tokyo). A signi cant difference was de ned as p-value < 0.05.

Results
Baseline characteristics Ten patients with nephrotic syndrome were enrolled and Table 1 describes their baseline characteristics.
Two patients already received 140 and 160 mg furosemide and others 40 mg furosemide before TLV. Diabetic nephropathy was the majority in the causes of nephrotic syndrome and 9 patients had renal impairment. No one had very low cardiac output (EF < 30%). The median of urine osmolality was lower than the predictive value for responsiveness to TLV in the previous report 18 . Five patients nished follow-up before day 15 because of no improvement of symptoms (N = 2) or discharge in alive (N = 3). The remaining ve patients were followed up until day 15.
The pathological evaluation was exhibited in Supplemental Figure S3. When compared with patients with minor glomerular abnormality and normal renal function, nephrotic patients had more glomerular sclerosis and tubulointerstitial injury, while interstitial brosis was not signi cantly different between 2 groups. Osmolality and electrolytes in urine and serum after TLV administration The time-dependent course of osmolality and electrolytes in spot urine, and serum are depicted in Supplemental Figure S4. Spot urine osmolality and electrolytes signi cantly decreased four and eight hours after TLV administration and returned to baseline at day 2 or after. On the other hand, except the increase of serum osmolality at day 2 and serum chloride at day 2-3, serum osmolality and electrolytes were unchanged during TLV administration. All patients did not suffer from any electrolyte disorders including hypernatremia (> 145 mEq/L) and did not have worsening renal function with repeatitive TLV administration.
Effect and safety of TLV in patients with nephrotic syndrome Urine output at day 1 exhibited a signi cant increase from baseline and urine output after day 1 was not signi cantly different from baseline. Body weight was signi cantly reduced at the last follow-up after TLV administration (Fig. 1).
When the correlation with urine output and weight change at last follow-up was evaluated, the change of urine output at day 1 was signi cantly correlated with weight change at last follow-up (R 2 = 0.52, p = 0.02).
No association with urine output increase by TLV and various measurements in urine and blood.
Although some measurements in urine and blood changed after TLV administration (Supplemental Figure S4), the change of UO at day 1 was not correlated with osmolality and electrolytes in urine and serum before TLV administration (Table 2). Also, the change of urine output on day 1 did not correlate with urine AQP2 before administration. As same as urine AQP2, the concentration of TLV and plasma ADH were not related to the change of urine output. (Table 3) Table 2: The correlation of urine output change at day 1 with urine and serological measurements.
None of them were associated with urine output.  Pathological evaluation revealed the degree of AQP2 staining was related to urine output Then urine output change was evaluated from the pathological point. The proportion of AQP2 staining was strongly correlated with urine output change at day 1 while the degree of brosis, glomerular sclerosis and tubulointerstitial injury were not signi cantly correlated with urine output change at day 1 ( Table 3).

Discussion
Nephrotic syndrome is common kidney disease. Some nephrotic patients have refractory edema resistant to near-maximum doses of diuretics 19,20 . Also, volume correction and the use of diuretics may deteriorate renal function 21,22 . Optimal treatment for nephrotic edema without worsening renal function is an important issue. An antagonist to V2 receptor, TLV, blocks water reabsorption via aquaporin 2 in cortical collecting ducts without activating the renin-angiotensin-aldosterone axis and the sympathetic nervous system 6 . TLV is expected as a novel drug for nephrotic edema. However, to date, it remains to be clari ed who is responsive to TLV though some researchers reported the effect of TLV on nephrotic edema. Then our study challenged to elucidate the association of responsiveness to TLV with pathological evaluation and serological or urinary markers. And the novel ndings are as below. First, urine output change at day 1 was correlated with the degree of weight loss at the nal. Second, the proportion of AQP2 staining was correlated with the increase of urine output by TLV while urine and blood measurements including uAQP2 and TLV were not correlated. And nally, TLV could be used safely for some nephrotic patients without adverse effects in the short term.
Previous reports showed that responders to TLV in heart failure patients had a signi cant increase in urine volume soon after administration and improved clinical course with signi cant weight loss 23 . As similar to the previous studies, our study demonstrated that weight loss was also signi cantly related to the increase of urine volume at day 1 in nephrotic syndrome. This indicated that urine volume soon after TLV administration was enough to evaluate whether it was responsive to TLV and the clinical course was improved in nephrotic syndrome patients.
A transient increase in urine output on day 1 was not re ected in a concomitant weight loss on day 2. This undesirable result is attributed to arbitrary water intake. In general, uid restriction is needed for volume control irrespective of the use of diuretics. However, at the time when this study was conducted, Pharmaceuticals and Medical Devices Agency in Japan did not recommend uid restriction to avoid volume depletion. Moreover, we could not predict the urine output increase so that we could not manage volume control appropriately. But the result in this study revealed the effect of TLV and we would manage the proper uid restriction.
Urine AQP2, urine osmolality change and serum TLV concentration can be predictors for TLV e cacy in heart failure 18,24,25 . However, these measurements did not correlate with urine output change and weight loss in our study despite detected urine AQP2, the reduction of urine osmolality and elevated TLV concentration. Moreover, serum ADH was not correlated with the effect of TLV. There were no precise explanations, but some reasons could be speculated. First, evidence was not enough about the association with urine AQP2 and pathological AQP2 staining in nephrotic syndrome though AQP2 expression in kidney tissues and net urine AQP2 were increased in a previous report 26 . Urine output in nephrotic patients sometimes decreases and urine AQP2 concentration needs to be adjusted when it is interpreted. However, AQP2 is located and excreted in principal cells of collecting ducts. Urine creatinine or other ltration makers were not enough as calibrations because they do not re ect the dynamics of collecting ducts precisely. Thus, urine AQP2 concentration did not correlate with AQP2 staining and urine output change in nephrotic patients. Second, the majority in our study had renal impairment, unlike previous reports. Patients with decreased eGFR disable to dilute or concentrate urine despite elevated vasopressin concentration 27 , which indicates collecting ducts in the diseased kidney are reluctant to respond to vasopressin and V2 receptor antagonists. Serum ADH is slightly higher in CKD patients, which indicates that the response to ADH would be sluggish. Although the mechanism of urine dilution and concentration is not clearly understood, it is considered that urine osmolality, serum ADH or TLV concentration were not enough to predict the aquauresis by TLV.
Although the nding in this study indicates that it requires a renal biopsy to predict urine output change by TLV in nephrotic patients, renal biopsy is impractical only for the purpose. Biopsy has several risks such as ank pain, infection and bleeding. A few days are necessary for the result of the biopsy. Thus, the nding in this study may makes little clinical sense and a novel, easier manner will be expected for predicting the responsiveness. However, although it may not give a suggestion to our clinical practice immediately, we believe that this nding in this study is academically important. Our ndings are not totally consistent with previous studies because we found that the effect of tolvaptan retained for patients with kidney dysfunction, but was correlated with AQP2 in biopsied kidney, not urine and blood parameters like the previous studies. Thus, we consider this nding should be described in this article to further understand the physiology of diuresis under kidney impairment.
The limitation of this study is a small exploratory study in a single center for the short term. The population in this study was limited to nephrotic patients with diabetic nephropathy and CKD stage 3-4. The pathophysiology of diabetic nephropathy and nephrotic patients may be different as indicated in the previous report 28 . It is postulated that nephrotic syndrome results from permeability factors affecting podocyte function such as cardiotrophin-like cytokine 1, which is not regarded as a cause of diabetic nephrophaty 29,30 . Therefore, this study cannot tell that tolvaptan is effective for idiopathic nephrotic syndrome and further studied are required. Second, the number of enrolled patients could not be reached to the target size (N = 20) during the trial. More detail evaluation for the larger population and other types of nephrotic syndrome is needed. The long term effect of TLV should be evaluated in further study. Third, this study was an exploratory study, not a randomized trial with controls of conventional therapy. Further randomized comparative trials are needed to validate the e cacy of tolvaptan. Fourth, low dose TLV in this study may be a potential limitation. According to the report on the pharmacokinetics of TLV, even higher dose TLV have less effect on urine excretion in advanced kidney disease 31 . Therefore, higher dose might have increased urine output in some patients without any improvement of symptom in this study.

Conclusion
Our exploratory prospective study provided that TLV increased urine output in nephrotic patients and the increase of urine output by TLV was only correlated with AQP2 positivity in kidney tissue. Urinary and serological measurements were not correlated with the effect of TLV, and no clinical parameters except kidney biopsy specimen may predict the responsiveness for TLV in patients with nephrotic syndrome.

Declarations
Ethics approval and consent to participate: This study protocol was approved by Tokushukai Group Institutional Review Board (TGE00343-024) and adhered to the Declaration of Helsinki. Informed consents were obtained before the study registration by written form. Author's contribution RM and TO conceived and designed the study, analyzed the data, and drafted the manuscript. YM, KI, KM, and MO conducted the data collection. SH, TI, KK and SK provided additional guidance for the analysis. All authors revised the manuscript critically for important intellectual content and gave nal approval of the manuscript.