Elevated Serum Chloride Level Contributes to the Poor Prognosis of IgA Nephropathy

Finding reliable prognostic factors is crucial for IgA nephropathy (IgAN). Here, we determined the relationship between prognosis of IgAN with serum chloride. Primary IgAN diagnosed by renal biopsy from January 1, 2015 to April 1, 2019 were recruited. Patients were divided into lower group and higher group based on the best cut-off value of survival receiver operating characteristics (ROC) curves. The baseline clinicopathological characteristics were retrospectively compared. Cox proportional hazard models were used to demonstrate the prognostic value of serum chloride in IgAN. Prognosis prediction model was built by multivariate Cox regression. Results Compared to higher group , age (cid:0) 24-hour urinary protein and serum creatinine(Cr) in the lower group were signicantly lower, hemoglobin(Hb) (cid:0) albumin were signicantly higher(all P<0.05),the degree of endothelial cell proliferation (E) and renal tubule atrophy or renal interstitial brosis (T) were signicantly lighter (all P<0.05). Univariate and multivariate Cox analysis revealed that serum chloride ≥ 105.4 mmol/l was an independent risk factor for the prognosis of IgAN(P<0.05). Serum chloride, Cr, T, hypertension, and Hb were screened out as features in predictive prognosis model. The c-index of the model was 0.85 (cid:0) 0.82 and 0.77 for 1 (cid:0) 2 and 3 years respectively, and brier scores were 0.06 (cid:0) 0.09 and 0.16 respectively. independent prognosis of IgAN. A predictive prognosis model including serum chloride, Cr, T, hypertension and Hb exhibited a relatively good prediction that serum chloride disorder will affect the clinical and pathological indicators and prognosis of IgAN. The main objective of this study is to determine the relationship between serum chloride and disease severity as well as prognosis in patients with IgAN.


Introduction
Primary IgA nephropathy (IgAN) is the most common type of idiopathic glomerulonephritis (GN) throughout the world and the main cause of ESRD in patients with primary glomerular disease [1]. To date, predictors of prognosis in IgAN based on clinical ,pathological, genetic, and noninvasive biological markers have been gradually discovered [2]. However, these parameters either not sensitive enough, or traumatic and expensive. Thus, we are committed to identifying a more convenient and accurate predictor of prognosis in IgAN.
A variety of electrolyte and acid-base alterations predictably occur with progressive loss of kidney function [3][4][5][6]. Most of the disorders are intricately linked to morbidity and mortality of kidney disease patients [7], especially potassium imbalance, metabolic acidosis and derangement of bone mineral metabolism. Chloride ion is the highest anion in the extracellular uid, which plays a crucial role to regulate amount of functions in human body [8] including the maintenance of osmotic pressure, acid-base balance, muscular activity, and the movement of water between uid compartments. More and more studies have begun to recognize the importance of chloride, especially with the excavation and exploration of acid-base [9,10] and chloride ion channels [11][12][13]. Little of studies have assessed the relationship between serum chloride and the renal prognosis of IgAN, but we can get some clues from the relationship between serum chloride and kidney injury. Based on previous studies, serum chloride disorder may adversely affect renal function.
Compared with hypochloremia, it is currently believed that hyperchloremia is more closely related to the severity of kidney injury [14,15]. High chloride can cause thromboxane release [16] and enhance the response of renal vasoconstrictors, such as angiotensin [17]. In addition, high chloride can induce a glomerular feedback mechanism in dense plaques, causing contraction of afferent arterioles, contraction of the glomerular mesentery and reductions in glomerular ltration rate [18]. In addition, high chloride often accompany with metabolic acidosis [9,10]. Perchloric acidosis increases the production of endothelin-1 and aldosterone, leading to tubulointerstitial in ammation and injury, thereby accelerating the progression of CKD [7] . . Although rare, there are reports revealing that low chloride is also associated with kidney damage [15]. Therefore, we hypothesize that serum chloride disorder will affect the clinical and pathological indicators and prognosis of IgAN. The main objective of this study is to determine the relationship between serum chloride and disease severity as well as prognosis in patients with IgAN.

Study Population
Patients were identi ed from a retrospective, unit-institutional database. Patients with a biopsy-based diagnosis of primary IgAN between January 1, 2015 and April 1, 2019 by the First A liated Hospital of Zhengzhou University were enrolled in the study. The inclusion criteria included the following: (1) primary IgAN was diagnosed by renal biopsy, (2) follow-up time was greater than 6 months, (3) no glucocorticoid immunosuppressant was used before renal biopsy, (4) an initial estimated glomerular ltration rate (eGFR) ≥ 15 mL/min/1.73 m2 at the time of renal biopsy. The following exclusion criteria were employed: (1) patients with secondary IgA nephropathy, including chronic hepatitis B, Henoch-Schonlein purpura, ankylosing spondylitis, systemic lupus erythematosus, and rheumatoid arthritis; (2) the number of glomeruli in renal biopsy specimens was less than 10; (3) clinicopathological data were incomplete; (4) eGFR<15 ml·min-1 (1.73 m2)-1; (5) patients with acute kidney failure at the time of renal biopsy.
Clinical, biochemical, and histopathological data collection The baseline clinicopathological data were recorded at the time of the renal biopsy. Hypertension was de ned as systolic BP(SBP) ≥140mmHg, diastolic BP(DBP)≥90mmHg, or use of antihypertensive drugs. Oxford classi cation of each patients was scored by two renal pathologists blinded to clinical data [19] . Immuno uorescence results were scored as follows: 0 represents -/+, 1 for +, 2 for ++, 3 for +++, and 4 for ++++. The degree of vascular injury (A) was assessed as follows: 0 indicates no obvious abnormality, 1 represents simple vascular wall thickening, and 2 represents vascular wall thickening and other lesions, such as brinoid necrosis and vitreous degeneration. Renal tubular necrosis was scored as follows: 0 indicates no necrosis, and 1 indicates necrosis. The treatment plan was based on KDIGO guidelines [20] .
In order to analyze prognosis, double serum creatinine Cr levCl from baseline data, ESRD, or greater than 30% eGFR decline were used as the de nition of the composite end point [21] . ESRD was defined as eGFR≤15 ml/min per 1.73 m2 or need for RRT (including hemodialysis, peritoneal dialysis, or renal transplantation) for the purpose of this study.

Statistical methods
SurvivalROC package in R version 3.6.3 was used to calculate the best cut-off point of serum chloride. Normally distributed quantitative variables were expressed as means±SDs, and based on homogeneity of variance, the t-test or corrected t-test were used for comparisons between groups. Non-normally distributed features were expressed as medians and interquartile ranges, and the Wilcoxon rank sum test was used for comparisons between groups. Quantitative data were expressed by frequency and percentage, and the chi-square test was used for the comparison between groups. Unadjusted and the multivariable-adjusted Cox proportional hazards models were used to analyze the relationship between serum chloride and the prognosis of IgAN. Age, sex, hypertension, 24-H UP, Cr, Oxford classi cation scores were adjusted in multivariable-adjusted Cox proportional hazards models. To further expand the applicability of the result, we built a predictive prognosis model. Clinicopathological variables associated with survival were assessed a priori based on univariate Cox regression, clinical importance, and predictors identi ed in previously published articles [22,23]. And forward-backward selection with the Akaike information criterion (AIC) was used to identify variables for the models. Therefore, serum chloride, hypertension, Cr, hemoglobin (Hb), and renal tubule atrophy or renal interstitial brosis (T) were included in the model. The internal stability of the model was tested using the bootstrap approach. The model performance was evaluated based on the predictive accuracy for individual outcomes (discriminating ability) and the accuracy of point estimates of the survival function(calibration). SPSS 21 software (SPSS Inc., Chicago, IL, USA) and R software (version 3.6.3, http://www.Rproject.org) were used for statistical analysis. Values of P less than 0.05 were considered statistically signi cant.

Demographic and Clinicopathological Characteristics
From January 1 2015 to April 1, 2019, 394 patients with primary IgAN were recorded. The clinical characteristics at baseline were summarized in Table 1. In total, 212 (54%) males and 182 (46%) females were included in the study, and the mean age at biopsy was 35±12years. 154 (39%) patients had hypertension. The mean serum chloride was 104.10±3.47 mmol/l. 267 (68%) patients accepted steroid and immunosuppressant therapy. Patients were followed for 14 (9-24 months) months, and 46 patients reached the composite endpoint.
Grouping based on blood chloride levels According to survivalROC, The areas under the receiver operating characteristics curve(AUCs) were 0.63, 0.70, and 0.61 for 1, 2 and 3 years respectively, and the best cut-off points were 105.50 for 1 year, 105.40mmol/l for 2 and 3 years(shown in Fig. 1). To summarize the results, the patients were divided into two groups according to the cut-off value 105.40 mmol/l because no signi cant difference in the subsequent statistical analysis results was noted regardless of whether 105.40 mmol/l or 105.50 mmol/l serum chloride was use as the cut-off point. Lower group had 247 patients and higher group had 159 patients.

Relationships between the serum chloride and demographic, clinical, and histopathological data
The clinical characteristics of the 394 patients in the lower group and the higher group are shown in Table 1 In addition, we found that blood urea nitrogen, complement 3, sodium, calcium, magnesium, and immunoglobulin G also exhibited signi cant differences between the two groups. No signi cant difference in other parameters were noted between the two groups.

Predictive prognosis model based on clinical and histopathological parameters and internal validation
We incorporated serum chloride <105.40mmol/l or ≥105.40mmol/l , hypertension (yes or no), Cr, uric acid(UA), 24-H UP, serum phosphorus, Hb, T (T0, T1 or T3), mesangial hypercellularity [M(M0 or M1)], segmental glomerulosclerosis/adhesion [S(S0 or S1)], and the degree of vascular injury [A(A0, A1 or A2)] as prognostic features. All these parameters were reduced to the most useful 5 potential predictors (serum chloride group, Cr, T, hypertension and Hb) for survival using forward-backward selection with the AIC. A nomogram based on the prognosis model was constructed to estimate 1-3 years renal survival (shown in Fig. 2). Then, bootstrap validation with 200 resampling was employed for internal validation. The discriminative ability of the nal model was assessed using C statistics. The c-index of this model was 0.82 (95% CI 0.68-0.94) for 1 year; 0.85 (95% CI 0.76-0.95) for the 2 years and 0.77 (95% CI 0.52-0.99) for the 3 years. Calibration was evaluated using calibration plots and brier scores. The Brier score of this model was 0.06 (95% CI 0.04-0.09) for the rst year, 0.09 (95% CI 0.06-0.14) for the second year and 0.16(0.06, 0.30) for the third year (shown in Fig. 3).

Discussion
Our study demonstrated that serum chloride levels at the time of renal biopsy is a sensitive and convenient index to predict and identify adverse renal outcomes.
Early identi cation or prediction of poor prognosis in patients with IgAN is often very di cult but critical. In recent years, an increasing number of IgAN prognosis indicators have been explored [24]. Among them, hypertension, massive proteinuria, renal impairment, albumin, and severe histological ndings have been accepted widely [25]. However, these indicators also have some disadvantages, such as limited sensitivity or analysis methods are traumatic and expensive. Currently, biomarkers and genetic indictors are being explored gradually [26][27][28]. However, they are relatively expensive, which may impose an economic burden on some patients. Therefore, we are committed to identifying a convenient, cheap, and highly sensitive and speci c indicator.
MARY A demonstrated that hypertonic NaCl in dogs lead to a transient renal vasodilation that was probably related to plasma hypertonicity followed by sustained renal vasoconstriction and reduced eGFR. Thromboxane played a very important role in this progression [16]. TANAKE demonstrated that in stroke-prone spontaneously hypertensive rats (SHRSP), Cllikely ampli ed microangiopathy by exacerbating hypertension and potentially increasing plasma renin activity (PRA), which potentially on account of constricting the renal afferent arteriole [17]. In addition, chloride can induce a glomerular feedback mechanism in dense plaques that involves contraction of afferent arterioles, contraction of the glomerular mesentery and reductions in the glomerular ltration rate [29]. In addition, a high level of chloride often accompanies with metabolic acidosis. Metabolic acidosis caused by the loss of carbonate in the gastrointestinal tract and kidneys often leads to hyperchloremia [8]. Evidence on the relationship between metabolic acidosis and IgAN prognosis remains limited. However, perchloric metabolic acidosis is associated with accelerated CKD progression and elevates all-cause mortality, which could provide some clues [30, 31].
We used survivalROC to divide patients into two groups. Although a range of statistical methods are available for cut-off point selection[32, 33], but disease status was considered a xed characteristic of the study subject in classic analysis [34] . There were researches revealed that time-dependent ROC could assist in developing eligibility criteria for clinical trials [21,34]. In this study, we considered survival time, so chose survivalROC to group.
Comparing the baseline demographic, clinical and histopathological data and treatment modalities in the two groups, we found the clinical and pathological indicators of IgAN were more serious when the Cllevel ≥105.4 mmol/l,.
Subsequently, we performed a prognostic analysis. According to univariate and multivariate COX regression, the serum chloride ≥105.4 mmol/l was an independent risk factor for IgA nephropathy. This result also suggested that 105.4 mmol/l serum chloride level was a very important threshold for IgAN.
Clinically, due to differences in testing techniques, the normal high limit of serum chloride is within the range of 106-107 mmol/l. In this study, considering the prognosis of patients with IgAN, the normal high limit of serum chloride level was further accurate to 105.4 mmol/l. This nding suggested that clinicians should pay close attention to the change of serum chloride level when treating IgAN, and it is ideal to ensure the level is below 105.4 mmol/l.
In addition, we also established a clinical predictive prognosis model including Cr, serum chloride <105.4mmol/l or ≥105.4mmol/l , hypertension, T, Hb. Our predictive prognosis model was demonstrated a good discriminative ability according to c-index. However, calibration plots did not demonstrate very good agreement.
We thought more deeply about the above results. Firstly, in clinical work, 0.9% NaCl solution is widely used, while which contains more than physiological amounts of chloride ions [35]. This can cause hyperchloremia and metabolic acidosis [36,37]. Based on the consideration of the in uence of serum chloride level on IgAN, can we replace the NaCl solution with other solutions or restrict the use of NaCl solution? There is no research in this area yet. But there are some researches that give us some inspiration [18,38]. Malley [38]used a double-blind method to compare the results of renal transplantation patients using lactated Ringer's solution and 0.9% normal saline. The results showed that the incidence of hyperchloremia and metabolic acidosis was higher in the 0.9% saline group. And Yunos [18]proved that restricting the use of chorine-rich uids in tertiary ICU could decreased the incidence of acute kidney injury and renal replacement therapy requirement. Therefore, we guess that the same result will be produced in IgAN. But it still depends on further con rmation. Secondly, 99.1% of chloride ions are reabsorbed in the kidney [29]. Is it possible to improve the prognosis of IgAN patients with high blood chloride levels by reducing the kidney's reabsorption of chloride ions? This sparked our interest to nd pathways to inhibit the reabsorption of chloride ions. There are many mechanisms related to the reabsorption of chloride ions in the kidney [8,39].Among them, the most important process occurs in the latter half of the proximal tubule, where chloride ions are transported into the cell through the Cl --HCO3exchanger at the top membrane of the epithelial cell. The chloride ions that enter the cell are transported to the intercellular uid through the K + -Clsymporter of the basilar membrane, and then absorbed into the blood. In addition, chloride ions can also be reabsorbed through paracellular pathway [39] . Therefore, we speculate whether it is possible to inhibit the reabsorption of chloride ions by inhibiting the transmembrane transport of chloride ions in the proximal renal tubules or the paracellular pathway, thereby reducing the blood chloride level of IgAN, and improving the prognosis of IgAN patients. However, this speculation has yet to be further veri ed.
In addition, many hazards are associated with reduced blood chloride levels [8]. We modeled chloride as a continuous variable using restricted cubic splines before this research to obtain a complete picture of the relationship between blood chloride and renal outcome of IgAN(shown in Fig. 4).
Interestingly, a nonlinear association was observed for IgAN progression. This curve has an asymmetrical U-shape, and patients with serum chloride>105 mmol/L exhibited distinct associations with poor renal outcome(P<0.001). Moreover, although IgAN patients with serum chloride <101 mmol/L tended to have less favorable renal survival, the relationship was not statistically signi cant P=0.70 (Table 3). Thus, our study did not consider the impact of reduced blood chloride levels on the prognosis of IgAN.
Our study has certain limitations. First, this was a retrospective single-center study, and our sample size was relatively small. In addition, the follow-up time was relatively short. Although we used rigorous statistical methods to correct confounding factors, some bias may affect the robustness of our results. In addition, it is worth noting that serum chloride levels may be affected by excessive use of saline solution during treatment. Unfortunately, our study did not delve into the causes of elevated blood chloride levels, so we could not determine whether elevated blood chloride levels were caused by the disease itself or by rehydration during treatment. In addition, our clinical predictive model does not include external validation to demonstrate its universality to a large population and other races. Although our study proves that low blood chloride levels are not a risk factor for IgA nephropathy, in general, low blood chloride levels are also harmful.