Lower bicarbonate level is associated with CKD progression and all-cause mortality: a propensity score matching analysis.

BACKGROUND: Although metabolic acidosis is known as a potential complication of chronic kidney disease (CKD), there is limited information concerning the association between metabolic acidosis and clinical outcomes.
METHODS: Five hundred fifty-two patients referred to renal division of Iwata City Hospital from 2015 to 2017 were included as a retrospective CKD cohort, and finally 178 patients with CKD stage III or IV and 20 to 80 years of age were analyzed. We examined the association between serum bicarbonate (HCO3-) levels and clinical outcomes using Kaplan-Meier methods after the matching of baseline characteristics by propensity scores.
RESULTS: Of 178 patients with CKD, patients with lower HCO3- levels (N = 94), as compared with patients with higher HCO3- levels (N = 84), were more likely to be male (P < 0.05), had more severe CKD stages (P < 0.05), more frequent use of renin-angiotensin system inhibitor (P < 0.05) or uric acid lowering agent (P < 0.001), heavier body weight (P < 0.001) and lower estimated glomerular filtration rate (P < 0.05). In Kaplan-Meier analysis after propensity score matching, the incidence of composite outcome as the doubling of serum creatinine level from baseline, end-stage kidney disease requiring the initiation of dialysis, or death from any causes was significantly fewer in the higher HCO3- group than the lower HCO3- group (N = 57 each group, P = 0.016).
CONCLUSIONS: Lower HCO3- level is significantly associated with the doubling of serum creatinine level, end-stage kidney disease or all-cause mortality in patients with CKD. TRIAL REGISTRATION: This study was registered with the Clinical Trial Registry of the University Hospital Medical Information Network ( http://www.umin.ac.jp/ , study number: UMIN000044861 ).

Background

Metabolic acidosis is one of the first recognized complications in patients with chronic kidney disease (CKD) [1-3]. In patients with CKD, there is a direct correlation between the decline of estimated glomerular filtration rate (eGFR) and the reduction in serum bicarbonate (HCO3−) levels, which is thought to be caused by the inability of the kidney to synthesize ammonia, regenerate HCO3−, and excrete hydrogen ions (H+) [4].

Observational studies have shown an independent association between serum HCO3− levels and adverse kidney outcomes [5-7], and small interventional trials of alkali supplementation or dietary modification have reported potential benefits of raising serum HCO3− levels in patients with CKD [8-10]. On the other hand, there is limited information concerning the significant association between metabolic acidosis and clinical outcomes.

In this study, we examined the effects of metabolic acidosis on kidney disease progression and mortality in patients with CKD using the method of propensity score matching analyses.

Materials and methods

Study design

We conducted a retrospective observational cohort study. We included all patients referred to Iwata City Hospital (Shizuoka, Japan) from January 1, 2015 to December 31, 2017. Eligible patients were adults (more than 20 years of age), although patients over 80 years of age were excluded due to their life expectancy.

This study was approved by the institutional ethics committee (approval number: 2021–020) and conducted in accordance with the Declaration of Helsinki. Because of its retrospective nature, informed consent was waived. This study was registered with the Clinical Trial Registry of the University Hospital Medical Information Network (http://www.umin.ac.jp/, study number: UMIN000044861).

Data collection and parameters analyzed

Clinical information including comorbidities and laboratory findings was collected by the access to patient’s records. Complete blood count was measured using XN-3000™ hematology analyzer (Sysmex Corporation, Hyogo, Japan) and serum electrolytes, albumin, low-density lipoprotein (LDL) cholesterol, creatinine, and C-reactive protein (CRP) levels were measured using Ci16200 auto-analyzer (Canon Medical Systems, Tochigi, Japan). Serum HCO3− levels were measured using ABL90 FLEX blood gas analyzer (Radiometer medical ApS, Denmark). Normal range of serum HCO3− levels were generally from 22 to 29 mEq/L [11, 12].

The demographic characteristics included age, sex, chronic kidney disease (CKD) stage, original kidney disease, the history of cardiovascular disease (CVD), medication, systolic blood pressure and body weight. Blood parameters included hemoglobin, albumin, LDL cholesterol, CRP, estimated glomerular filtration (eGFR), albumin-corrected calcium (Ca)-phosphorus (Pi) product and serum HCO3− level. Urine parameters included spot urine proteinuria score by dipstick (coded as four grades of 0 to 4 according to 0, 1+, 2+, 3+, and 4+ and as 0.5 if ±).

Outcome

The all patients were followed until January 1, 2020 or the incidence of the composite outcome, whichever occurred first. The outcome, assessed in a time-to-event analysis, was a composite of the doubling of serum Cr level from baseline, end-stage kidney disease requiring the initiation of dialysis, or death from any causes.

Statistical analyses

Data were expressed as the mean ± standard deviation (SD) for continuous variables with normal distributions or the median and interquartile range (25th to 75th percentiles) for data with skewed distributions. The threshold for statistical significance was set at P < 0.05. Comparisons between two groups were performed using the Mann-Whitney U-test for continuous variables or the chi-squared test for categorical variables.

To adjust the differences in the baseline characteristics and perform the following analyses, propensity score matching was employed [13, 14]. To calculate the propensity score, the target threshold of serum HCO3− level was set at 25.3 mEq/L based on its median value. The probability to over this threshold, as the propensity score, was determined by a multivariate logistic regression using the aforementioned 16 baseline characteristics including age, sex, CKD stage, original kidney disease, the history of CVD, medication, systolic blood pressure, body weight, hemoglobin, albumin, LDL cholesterol, CRP, eGFR, albumin-corrected Ca-Pi product and urinary protein score. A nearest-neighbor matching algorithm was used to match patients within a caliper width of 0.03 of the logit of the propensity scores. Outcome curves were constructed with Kaplan-Meier method and compared with the log-rank test. For Kaplan-Meier method, we analyzed the composite outcome as the doubling of serum Cr level from baseline, end-stage kidney disease requiring the initiation of dialysis, or death from any causes by the time to the first event. All statistical analyses were performed using IBM SPSS statistical software, version 26.0 (IBM SPSS, Tokyo, Japan).

Results

Patients

During 3 years from January 1, 2015 to December 31, 2017, 552 patients were referred to renal division of Iwata City Hospital for the evaluation of proteinuria, hematuria or renal insufficiency. Among 552 patients, 247 patients were CKD stage III or IV and 20 to 80 years of age, although 69 patients were not measured serum HCO3− levels. Finally, 178 patients were included in the following analyses (Fig. 1).

Fig. 1

Study flow chart. Abbreviation: CKD, chronic kidney disease

Patient’s characteristics

In 178 patients analyzed, the mean age was 65 ± 10 years old and 104 patients (58.4%) were male. The original diseases were as follows: overt diabetic nephropathy in 18 patients (10.1%), chronic glomerulonephritis in 55 patients (30.9%), nephrosclerosis in 81 patients (45.5%), and polycystic kidney disease in 3 patients (1.7%). The mean eGFR was 41 ± 13 mL/min/1.73 m2 and the median value of serum HCO3− levels was 25.3 mEq/L (Table 1).

Table 1

Patient characteristics

All variables are expressed as the mean ± SD or the median and interquartile range (25th to 75th percentiles)

Abbreviations: ADPKD autosomal dominant polycystic kidney disease; Ca calcium; CKD chronic kidney disease; CRP C-reactive protein; CVD cardiovascular disease; eGFR estimated glomerular filtration rate; HCO bicarbonate; LDL low-density lipoprotein; pi inorganic phosphorus; RAS renin-angiotensin system; SD standard deviation

Next, we divided those 178 patients into lower and higher HCO3− groups according to the median value of serum HCO3− levels (N = 94 and 84, respectively, Table 2). Patients with lower HCO3− levels, as compared with patients with higher HCO3− levels, were more likely to be male (P < 0.05), had more severe CKD stages (P < 0.05), more frequent use of renin-angiotensin system (RAS) inhibitor (P < 0.05) or uric acid lowering agent (P < 0.001), heavier body weight (P < 0.001) and lower eGFR (P < 0.05).

Table 2

Patient characteristics of lower (serum HCO3− levels ≤25.3 mEq/L) and higher HCO3− groups (25.3 < mEq/L) before and after propensity score matching

All variables are expressed as the mean ± SD or the median and interquartile range (25th to 75th percentiles)

Comparisons between lower and higher HCO3− groups were performed using the Mann-Whitney U test or chi-square test

Abbreviations: ADPKD autosomal dominant polycystic kidney disease; Ca calcium; CKD chronic kidney disease; CRP C-reactive protein; CVD cardiovascular disease; eGFR estimated glomerular filtration rate; HCO bicarbonate; LDL low-density lipoprotein; pi inorganic phosphorus; RAS renin-angiotensin system; SD standard deviation

Adjustment of patient’s characteristics with propensity score matching

Next, to compare the clinical outcomes between lower and higher HCO3− groups under the similar background, we employed the method of propensity score matching. After the propensity score matching, the differences of patient’s characteristics at the baseline between lower and higher HCO3− groups were adjusted (N = 57 each group, Table 2).

The all patients were followed until January 1, 2020 and were evaluated the incidence of clinical outcomes. The mean follow-up period was 36 ± 12 months. In 114 patients after propensity score matching (N = 57 each group), the incidence of composite outcomes as the doubling of serum Cr level from baseline, end-stage kidney disease requiring the initiation of dialysis, or death from any causes was significantly less in the higher HCO3− group than the lower HCO3− group, occurring in 2 patients (3.5%) and 10 patients (17.5%), respectively (P = 0.016, Fig. 2).

Fig. 2

Kaplan-Meier curves of lower (serum HCO3− levels ≤25.3 mEq/L) and higher HCO3− groups (25.3 < mEq/L) after propensity score matching

Discussion

In the present study, we showed that lower HCO3− level was significantly associated with the incidence of composite outcomes as the doubling of serum creatinine level, end-stage kidney disease requiring the initiation of dialysis and all-cause mortality in patients with CKD, after the matching for baseline characteristics and laboratory data using propensity scores.

Experimental studies in animals have showed that metabolic acidosis contributes to progression of CKD [15, 16], and observational studies in humans have found the association between metabolic acidosis and worsening kidney function [5-7]. A small number of interventional trials have also showed that sodium bicarbonate (NaHCO3) supplementation can slow the rate of kidney disease progression [9, 17].

Several mechanisms how metabolic acidosis accelerates the progression of CKD are proposed. Nath et al. [18] have reported that metabolic acidosis stimulates ammonia production that increases acid excretion, but leads to ammonia-induced complement activation and deposition of C3 and C5b-9 that can cause tubulo-interstitial injury. Wesson et al. [19, 20] have showed that metabolic acidosis in CKD stimulates the production of intra-kidney paracrine hormones including angiotensin II, aldosterone, and endothelin that mediate the immediate benefit of increased acid excretion, but their chronic upregulation promotes inflammation and fibrosis. It was also proposed that the new renal HCO3− synthesis in response to metabolic acidosis leads to calcium precipitation in the kidney [21].

One of key findings in this study is that serum HCO3− levels in some patients of lower HCO3− group were within the normal range (22–25.3 mEq/L). At present, the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines recommends to start the administration of alkali when serum HCO3− levels are < 22 mEq/L, to maintain the value within the normal range [11]. On the other hand, previous study showed that H+ retention of variable severity has already occurred even if serum HCO3− levels are within the normal range in patients with CKD, which is referred as subclinical metabolic acidosis [22-24]. Epidemiologic studies have showed that CKD patients with low-normal HCO3− levels have higher risk of eGFR decline and mortality than those with high-normal HCO3− levels [6, 25]. A randomized study in CKD patients with normal HCO3− levels has also showed that sodium bicarbonate (NaHCO3) treatment preserved their kidney function after 10 years compared with sodium chloride (NaCl) administration or usual care [26]. Therefore, our result suggests the requirement of earlier intervention before overt metabolic acidosis develops, although both the precise prevalence of subclinical metabolic acidosis and optimal range for the correction of serum HCO3− levels remain to be clarified. Further studies are needed to clarify these issues.

Another key finding in this study is that all-cause mortality was included in the composite outcomes we set. In accordance with our result, epidemiological studies have showed that higher HCO3− levels within the normal range are associated with better survival [6, 25]. On the other hand, there is few data of the effect of alkali on the mortality in patients with CKD [27]. Because it was reported that metabolic acidosis can cause insulin resistance, skeletal muscle breakdown or cardiovascular disease as well as the progression of kidney diseases [7, 28, 29], our result that patients with lower HCO3− levels had poorer prognosis might be reasonable. Further interventional studies are needed to identify the relationships between metabolic acidosis and mortality in patients with CKD.

Our study has several limitations. First, due to the retrospective study design, a longitudinal causal relationship cannot be established. However, to minimize this limitation, we employed the method of propensity score matching analyses in this study. Second, because of the relatively small number of patients in our cohort, the generalizability of our conclusions remains unclear.

Conclusion

In conclusion, we showed here that CKD patients with lower HCO3− levels, as compared with those with higher HCO3− levels, were more likely to progress their kidney disease and have poorer prognosis. Therefore, metabolic acidosis may be a modifiable risk factor for the progression of CKD.