Acid–base disturbances in nephrotic syndrome: analysis using the CO2/HCO3 method (traditional Boston model) and the physicochemical method (Stewart model)

Background The Stewart model for analyzing acid–base disturbances emphasizes serum albumin levels, which are ignored in the traditional Boston model. We compared data derived using the Stewart model to those using the Boston model in patients with nephrotic syndrome. Methods Twenty-nine patients with nephrotic syndrome and six patients without urinary protein or acid–base disturbances provided blood and urine samples for analysis that included routine biochemical and arterial blood gas tests, plasma renin activity, and aldosterone. The total concentration of non-volatile weak acids (ATOT), apparent strong ion difference (SIDa), effective strong ion difference (SIDe), and strong ion gap (SIG) were calculated according to the formulas of Agrafiotis in the Stewart model. Results According to the Boston model, 25 of 29 patients (90%) had alkalemia. Eighteen patients had respiratory alkalosis, 11 had metabolic alkalosis, and 4 had both conditions. Only three patients had hyperreninemic hyperaldosteronism. The Stewart model demonstrated respiratory alkalosis based on decreased PaCO2, metabolic alkalosis based on decreased ATOT, and metabolic acidosis based on decreased SIDa. We could diagnose metabolic alkalosis or acidosis with a normal anion gap after comparing delta ATOT [(14.09 − measured ATOT) or (11.77 − 2.64 × Alb (g/dL))] and delta SIDa [(42.7 − measured SIDa) or (42.7 − (Na + K − Cl)]). We could also identify metabolic acidosis with an increased anion gap using SIG > 7.0 (SIG = 0.9463 × corrected anion gap—8.1956). Conclusions Patients with nephrotic syndrome had primary respiratory alkalosis, decreased ATOT due to hypoalbuminemia (power to metabolic alkalosis), and decreased levels of SIDa (power to metabolic acidosis). We could detect metabolic acidosis with an increased anion gap by calculating SIG. The Stewart model in combination with the Boston model facilitates the analysis of complex acid–base disturbances in nephrotic syndrome.


Introduction
Several methods for analyzing acid-base disturbances have been proposed. One method is the traditional Boston model based on the CO 2 /HCO 3 method [1,2]. Another is the Copenhagen-Danish model based on base excess (BE) [3,4]. A third method is the Stewart model based on the physicochemical method [5][6][7]. The Boston model makes it easy for beginners to understand and analyze acid-base disturbances. However, without the use of compensatory formulas, this model is limited in the setting of complex conditions such as respiratory abnormalities and two or more concurrent metabolic abnormalities. The BE model is useful only in metabolic acidosis. In contrast, the Stewart model is theoretically superior to the other models mentioned above in complex situations because it is based on physicochemical data. However, it is very difficult for physicians to understand the underlying Abstract Background The Stewart model for analyzing acid-base disturbances emphasizes serum albumin levels, which are ignored in the traditional Boston model. We compared data derived using the Stewart model to those using the Boston model in patients with nephrotic syndrome. Methods Twenty-nine patients with nephrotic syndrome and six patients without urinary protein or acid-base disturbances provided blood and urine samples for analysis that included routine biochemical and arterial blood gas tests, plasma renin activity, and aldosterone. The total concentration of non-volatile weak acids (A TOT ), apparent strong ion difference (SIDa), effective strong ion difference (SIDe), and strong ion gap (SIG) were calculated according to the formulas of Agrafiotis in the Stewart model. Results According to the Boston model, 25 of 29 patients (90%) had alkalemia. Eighteen patients had respiratory alkalosis, 11 had metabolic alkalosis, and 4 had both conditions. Only three patients had hyperreninemic hyperaldosteronism. The Stewart model demonstrated respiratory alkalosis based on decreased PaCO 2 , metabolic alkalosis based on decreased A TOT , and metabolic acidosis based on decreased SIDa. We could diagnose metabolic alkalosis or acidosis with a normal anion gap after comparing delta A TOT [(14.09 − measured A TOT ) or (11.77 − 2.64 × Alb (g/dL))] and delta SIDa [(42.7 − measured SIDa) or concepts and make calculations using the relevant formulas in clinical practice. The Stewart model emphasizes the importance of albumin (Alb), which carries a negative charge in serum and influences acid-base balance; Alb is not taken into account by the traditional Boston model. There have been no reports about acid-base disturbances in nephrotic syndrome. In this study, we compared data derived using the Stewart model to data from the traditional Boston model in the context of acid-base disturbances in patients with nephrotic syndrome.

Materials and methods
This study was approved by the Ethics Committee of Aichi Medical University .

Patient characteristics (Table 1)
We enrolled 29 patients with nephrotic syndrome and 6 patients without urinary protein or acid-base disturbances as a control group. All patients were Japanese. They were admitted to Aichi Medical University Hospital between August 2014 and August 2015. The diagnosis of nephrotic with endocapillary proliferative glomerulonephritis (endo-PGN), and 2 with crescentic glomerulonephritis (crescentic GN). The remaining 13 patients did not undergo kidney biopsy due to comorbidities or refusal to provide consent. Based on clinical and laboratory findings, these patients were suspected of having the following diagnoses: MCNS (n = 2), MN (n = 3; 1 with primary MN and 2 with MN secondary to malignancy and bucillamine), lupus nephritis (n = 1), essential thrombocytosis (n = 1), diabetes mellitus (n = 2), systemic amyloidosis (amyloid light chain type) (n = 1), anti-myeloperoxidase (MPO) antineutrophil cytoplasmic antibody (ANCA)-associated nephritis (n = 1), and no primary disease (n = 1). The median levels of urinary protein, serum total protein (TP), and serum Alb were 7.0 g/day (IQR, 4.6-10.7), 4.  (Table 1).

CO 2 /HCO 3 method (traditional Boston model)
Initially, we defined acidemia or alkalemia based on a pH of 7.40. Metabolic or respiratory disturbances, and acidosis or alkalosis, were determined using the CO 2 /HCO 3 method. In respiratory alkalosis, we diagnosed acute or chronic respiratory alkalosis by comparing actual HCO 3 to estimated HCO 3 using Kellum's compensatory formula [10]. Thus, estimated HCO 3 = 0.2 × (actual PaCO 2 ) + 17 in acute respiratory alkalosis and estimated HCO 3 = 0.5 × (actual PaCO 2 ) + 5 in chronic respiratory alkalosis. When there was a discrepancy between actual and estimated HCO 3 , we determined if additional metabolic acidosis or alkalosis was present.

Relationships between albumin and A TOT and between cAG and SIG
We created scatter plots with albumin or cAG on the x-axis and A TOT or SIG on the y-axis. We also derived correlation coefficients using Excel (Microsoft Corporation, Redmond, WA).

Relationship between eGFR and SIG or (SIG − lactate)
We created scatter plots with eGFR on the x-axis, and SIG or (SIG − lactate) on the y-axis. (SIG − lactate) represents unknown non-volatile acids other than lactate. We also derived correlation coefficients using Excel (Microsoft Corporation).

Relationship between IgG and A TOT , SIDa, SIDe, and SIG
We created scatter plots with IgG on the x-axis and A TOT , SIDa, SIDe, or SIG on the y-axis. We also derived correlation coefficients using Excel (Microsoft Corporation).

Traditional model (Boston model)
The median pH was slightly alkalotic at 7.43 (IQR, 7.42-7.45) (Tables 2, 3); 25 of 29 patients had a pH of more than 7.40, while 4 had a pH of less than 7.40. The median PaCO 2 was 34.9 Torr (30.8-40.0). Twenty-two patients had PaCO 2 levels of less than 40 Torr, while 7 had levels between 40 and 44.7 Torr. The median A-aDO 2 was 33.9 Torr [28. 4-53.6]. Given that the normal limit of A-aDO 2 is less than 20 Torr, 26 of 29 patients (90%) in the high age group had impaired blood gas diffusion. The median PaO 2 was 84.2 Torr [70. 3-99.8]. These results suggest that hyperventilation occurs in most nephrotic patients. Since we routinely exclude pulmonary embolism and infection using laboratory data and chest computed tomography, hypoalbuminemia due to nephrotic syndrome seems to induce pulmonary interstitial edema, which in turn increases A-aDO 2 . Hyperventilation then occurs in order to maintain PaO 2 levels. Primary respiratory alkalosis is essential in patients with nephrotic syndrome. The median HCO 3 concentration was 23.7 mEq/L [21.1-26.0]. In cases of primary respiratory alkalosis, HCO 3 concentrations should be decreased as a result of compensatory mechanisms. Actual HCO 3 concentrations that are higher than expected suggest the coexistence of metabolic alkalosis. In nephrotic syndrome, the analysis of acid-base disturbances using the Boston method is limited because it requires the use of Kellum's compensatory formula for respiratory alkalosis [10] or the concept of cAG.

Serum AG (Tables 2, 3)
The median AG was 8.6 mEq/L (IQR, 7.0-10.8), which was significantly lower than the normal range of 12 ± 2 mEq/L. The median cAG, which adjusts for serum albumin using the formula cAG = AG + 2.5 × (4.4 − measured albumin (g/dL)), was 15.  (Table 3). (Table 4) PRA values were within the normal range (0.3-2.9 ng/ mL/h) in 19 patients, below normal in 5 patients, and above normal in 5 patients. PAC was within the normal range (30-160 pg/mL) in 15 patients. Hypoaldosteronism was observed in 11 patients, and 3 patients had hyperreninemic hyperaldosteronism. These results demonstrated that hypoaldosteronism and normoaldosteronism are the predominant conditions in nephrotic syndrome.
Low PaCO 2 levels, such as those observed here, contribute to the development of alkalosis (Fig. 1).

A TOT (mEq/L)
A TOT represents the total concentration of non-volatile weak acids. It depends on albumin and phosphate levels.
The median A TOT was 7.13 (mEq/L) (IQR, 6.00-7.83), because hypoalbuminemia secondary to nephrotic syndrome directly interferes with A TOT values. The median delta A TOT , calculated by [14.09 (mean value of control group) − A TOT ], was 6.96 (6.26-8.09). This lower value of A TOT contributed to metabolic alkalosis (Fig. 1).  (Fig. 1). In individual patients, delta A TOT (alkalosis) greater than delta SIDa (acidosis) results in metabolic alkalosis. Inversely, when delta SIDa is greater than delta A TOT , the patient has metabolic acidosis with a normal anion gap.

SIDe (mEq/L)
The median SIDe was 29.9 mEq/L (IQR, 27.1-32.9), which was lower than the mean value in the control group (38.5 mEq/L).

SIG (mEq/L)
SIG reflects the accumulation of various non-volatile acids, including lactate or uremic substances. Increased AG suggests metabolic acidosis with an increased anion gap. The median SIG was 6.6 mEq/L (IQR, 4.4-8.2), which was higher than 3.9 mEq/L (3.4-5.1) in the control group. The mean ± SD in the control group was 4.2 ± 1.4 mEq/L (normal range: 1.4-7.0).

Lactate (mmol/L = mEq/L) and SIG (mEq/L)
The median lactate was 1.3 mmol/L (IQR, 1.0-1.9), higher than 0.9 mmol/L (0.8-1.0) in the control group. The mean was 0.9 mmol/L (normal range: 0.8-1.0). Lactate levels greater than 1.0 mmol/L were observed in 20 of 29 patients (69%), and 12 out of 29 patients had metabolic acidosis with an increased anion gap. (SIG − lactate) represents the accumulation of non-volatile acids except for lactate. Seven of 12 patients had mainly lactate accumulation. Three had accumulation of lactate and non-volatile acids such as uremic toxins. Two of 12 patients had accumulation of nonvolatile acids except for lactate.

Relationship between serum albumin and A TOT
There was a strong significant relationship between serum albumin and A TOT : A TOT = 2.6425 × Alb + 2.3323 (R 2 = 0.91851) (Fig. 2a). We can easily calculate and estimate A TOT using serum albumin alone. When the serum albumin level is 4.4 g/dL, A TOT will be 14.0, a normal value (14.09). If the albumin level is 1.5 g/dL, A TOT will be 6.30. In addition, delta A TOT , which represents the power to metabolic alkalosis, is calculated using the formula 11.77 − 2.64 × Alb (g/dL) (Fig. 1).

Relationship between IgG and A TOT , SIDa, SIDe, and SIG
Levels of IgG, a positively charged protein, might influence acid-base balance. The relationship between IgG

Discussion
This study revealed 4 specific findings about acid-base disturbances in nephrotic syndrome. First, primary respiratory alkalosis occurs due to respiratory dysfunction, with elevated A-aDO2 caused by pulmonary interstitial edema due to hypoalbuminemia. Second, the comparison between delta A TOT and delta SID helps to determine whether metabolic alkalosis or metabolic acidosis with a normal anion gap is present. Third, we can estimate A TOT using serum albumin, SIDa as Na + K − Cl, and SIG using cAG. Fourth, 90% of patients had hyporeninemic or normoreninemic hypoaldosteronism, not hyperreninemic hyperaldosteronism. The above findings suggest that hyperchloremic metabolic acidosis in nephrotic syndrome is due to hyporeninemic hypoaldosteronism, with a shift of K into the intracellular space due to alkalemia.
The Boston model is easy for beginners to understand and use to analyze acid-base disturbances. However, without the use of compensatory formulas described by Kellum [10], this model is limited in the setting of complex conditions such as respiratory abnormalities and 2 or more concurrent metabolic abnormalities. It is very hard not only for beginners, but also for nephrologists to understand and calculate the effects of compensatory mechanisms. On the  , d). Six out of 9 patients with eGFR less than 29 mL/min had metabolic acidosis with an increased anion gap. Five out of 6 patients had accumulation of lactate and unknown non-volatile acids (i.e., uremic toxins). One patient had increased levels of lactate alone. On the other hand, increased lactate levels were observed in patients with eGFR greater than 30 mL/min, who have metabolic acidosis and an increased anion gap other hand, the Stewart model is theoretically superior to the Boston model in complex situations because it is based on physicochemical data. Changes after treatment such as supplementation of solution or albumin can be evaluated and be estimated using the Stewart model. The present study revealed that the formula from Agrafiotis et al. [12] is the most reliable; several manuscripts including those by Nguyen et al. [8], Restegar [9], and Kishen et al. [11], contained incorrect formulas. A lack of consensus on formulas for the Stewart model limits its use. Masevicius and Dubin [14] claimed that the Stewart approach is cumbersome, requires more determinations and calculations, and is more time-consuming and expensive. They only need to continue with the proper use of the Boston model. However, this study suggested that a combination of the Boston and Stewart models would be important to consider in complicated cases such as mixed metabolic disorders and hypoalbuminemia.
In the Steward model, the first step in the approach to analyzing acid-base disturbances in nephrotic syndrome is calculating delta A TOT and delta SIDa. The second step is comparing delta A TOT and delta SIDa values. When delta A TOT (14.1 − measured A TOT ) is greater than delta SIDa (49.2 − measured SIDa), we can predict the presence of metabolic alkalosis. Inversely, when delta A TOT is smaller than delta SIDa, metabolic acidosis with a normal anion gap is likely to be present. Thus, we can easily distinguish between metabolic alkalosis and metabolic acidosis using delta A TOT and delta SIDa values. This study also demonstrated a strong relationship between A TOT and serum albumin and between SIG and cAG. We can estimate A TOT and delta A TOT using the following formulas: A TOT = 2.6425 × Alb + 2.3323 (R 2 = 0.91851) and delta A TOT = 11.77 − 2.64 × Alb (g/dL) (power to alkalosis). Since SIDa is calculated as (Na + K − Cl), the formula for delta SIDa is (49.2 − measured SIDa), which indicates the power to acidosis. In addition, we can identify another metabolic acidosis with an increased anion gap using the formula SIG = 0.9463 × cAG − 8.1956 (R 2 = 0.91057). (more than 7.0). The reliability of the estimated formula is 89.7% (26/29) in metabolic alkalosis or acidosis with a normal anion gap and 86.2% (25/29) in metabolic acidosis with an increased anion gap (data not shown). Our formulas will help [who?] understand mixed acid-base disturbances in nephrotic syndrome and renal failure.
The present study revealed that patients with eGFR less than 29 mL/min/1.72 m 2 mainly have accumulation of lactate and uremic toxins. On the other hand, patients with eGFR greater than 30 mL/min/1.72 m 2 have metabolic acidosis with an increased anion gap as a result of increased lactate levels alone. Systemic edema due to nephrotic syndrome from MCNS or FSGS may influence the production of lactate.
Regarding the mechanism of edema in nephrotic syndrome, the underfill and overfill hypotheses have been proposed [15]. According to the underfill hypothesis, the renin-angiotensin-aldosterone system is activated by decreased intravascular blood volume, which in turn induces sodium retention and causes edema. However, the present study demonstrated that only 10% of patients have hyperreninemic hyperaldosteronism, while 90% had hypoaldosteronism or normoaldosteronism. These data support the overfill hypothesis [16,17]. Fluid management should involve administration of albumin and loop diuretics. Anti-aldosterone drugs such as spironolactone and eplerenone should be avoided. Recently, acetazolamide and hydrochlorothiazide followed by furosemide have been reported to be more effective for the treatment of refractory nephrotic edema [18]. However, the influence of these treatments on acid-base disturbances in nephrotic syndrome should be further studied because acetazolamide induces metabolic acidosis by increasing urinary excretion of Na + and HCO 3 − .

Conclusion
We showed that patients with nephrotic syndrome have primary respiratory alkalosis, decreased A TOT due to hypoalbuminemia (power to metabolic alkalosis), decreased SIDa (power to metabolic acidosis), and increased SIG, suggesting that they accumulate non-volatile acids such as lactate, uremic toxins, or other acids. Using both the Stewart and Boston methods facilitates the analysis of complex acid-base disturbances in primary and secondary nephrotic syndrome.

Compliance with ethical standards
Conflict of interest The authors have declared that no conflict of interest exists.
Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee at which the studies were conducted (Aichi Medical University 14-164) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent Informed consent was obtained from all individual participants included in the study.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.