Factors Associated with Metabolic Acidosis in Patients Receiving Parenteral Nutrition

NEPHROLOGY 2007; 12, 3–7 doi:10.1111/j.1440-1797.2006.00748.x

Original Article

Factors associated with metabolic acidosis in patients receiving parenteral nutrition

I-CHIEH TSAI,1 JENQ-WEN HUANG,2 TZONG-SHINN CHU,2 KWAN-DUN WU2 and TUN-JUN TSAI2

1Department of Internal Medicine, Taipei Hospital, Department of Health, Hsin-Chuang City and 2National Taiwan University Hospital, Taipei, Taiwan

SUMMARY: Aims: Metabolic acidosis is a common problem after infusion with chloride-based parenteral nutrition. However, it is unknown whether the occurrence of metabolic acidosis is related to this regimen of therapy or to patient- specific risk factors. Methods: Patients receiving parenteral nutrition from July to December 2003 at this hospital were included for a retrospective study. Patients were excluded who had illnesses that were potentially related to acid-base disorders. The remaining patients were divided on the basis of parental nutrition they had received: a chloride- base regimen group, and an acetate-based therapy group. Biochemical character and blood gas data were analysed. Continuous variables were analysed by t-test. Categorical variables were assessed by chi-squared test. Independent determinants for bicarbonate decline were analysed using forward stepwise multiple linear regression analysis. Results: There were 29 patients (17 women, 12 men) who received chloride-based regimen and 26 patients (16 women, 10 men) took acetate-based therapy. The acetate group had significantly higher baseline serum creatinine and blood urea nitrogen than chloride group. The blood pH, CO2, bicarbonate and base excess were significantly lower after receiving chloride-based therapy; while these changes were not observed in acetate- based therapy group. However, the serum creatinine and blood urea nitrogen levels were not statistically different. With multiple-stepwise linear regression, parenteral nutrition formula and estimated creatinine clearance are independent predictors of bicarbonate decline. Conclusion: The acetate-base regimen can decrease the occurrence of metabolic acidosis after parenteral nutrition. In addition, the risk of acidosis is higher in patients with impaired renal function.

KEY WORDS: acetate, chloride, creatinine, metabolic acidosis, parenteral nutrition.

Metabolic acidosis can occur in patients receiving additional chloride ions for pH adjustment had also been parenteral nutrition via both underlying disease and in the discussed in rabbits4 and preterm neonates.5 composition of the nutritional solutions.1 Acidosis caused We observed an increase in the incidence of metabolic by parenteral nutrition is related to (i) interruption of meta- acidosis after introducing a new chloride-based parenteral bolic pathway of carbohydrate and fat; (ii) metabolism of nutrition formula (Aminomix 1) to this hospital. Sodium cationic amino acids2 and sulphur-containing amino acids;3 bicarbonate and even more haemodialysis were required to (iii) the titratable acidity of solutions, although this factor is treat the acidosis. However, these therapeutic effects were not favoured by most authors;2 (iv) deficiency of thiamine or transitory and metabolic acidosis quickly returned. As we biotin; and (v) injection of D-fructose. Besides, the changed the formula to an acetate-based regimen (Paren- aid), the serum bicarbonate level elevated. The most different composition of these two formulas is the concentration Correspondence: Associate Professor Tzong-Shinn Chu, Depart- ratio of acetate and chloride. (Table 1) The purpose of the ment of Internal Medicine, National Taiwan University Hospital. No. present study was to identify whether metabolic acidosis was 7, Chung-Shan South Road, Taipei 100, Taiwan. Email: significantly related to the regimen of therapy. In addition, [email protected] we analysed the roles of other clinical and biochemical Accepted for publication 9 October 2006. variables in order to determine if certain risk factors exist in © 2007 The Authors the occurrence of metabolic acidosis during administration Journal compilation © 2007 Asian Pacific Society of Nephrology of parenteral nutrition. 4 I-C Tsai et al.

Table 1 Composition of parenteral nutrition solutions Paren-aid No. 2 Paren-aid No. 3 Aminomix 1 Electrolyte (mEq/L) Na 30 30.4 50 K303030 Ca – 20 5 Mg 8 8 5 Cl 30.4 33 100.11 Phosphate 12.5 – 15 Acetate 17.5 30 – Zn – – 0.08 Glucose (g/L) 250 250 200 Amino acid (g/L) – – 50 kCal/L 1000 1000 1025 mOsm/L 1518 1550 1769 pH 4.5–6 4.5–6 4.8–5.2

METHODS other patients with high anion gap metabolic acidosis or mixed high anion gap metabolic acidosis/metabolic alkalosis We performed a retrospective analysis on the records of patients receiv- were also excluded. In addition, six patients with hypercap- ing parenteral nutrition in our hospital for more than 10 days during nia respiratory acidosis (CO2 > 44 mmHg) were excluded. the 6-month period, July 2003 to December 2003. Patients who had The remaining 55 patients were divided into two groups. one of the following conditions after parenteral nutrition, which could Group I included 29 patients who had received chloride- result in an acid-base disorder were excluded: (i) severe diarrhea; based (Aminomix 1) parenteral nutrition and group II (ii) pure high anion gap metabolic acidosis; (iii) mixed high anion gap metabolic acidosis and metabolic alkalosis; and (iv) included 26 patients who had received acetate-based (Paren-aid) regimen. As shown in Table 1, the acetate- PaCO2 > 44 mmHg. based solution did not contain amino acids. For this reason, patients in this group received additional amino acid solu- Definitions tions, and the compositions of them were displayed in Table 2. Ten patients took amino 12X, 13 patients took The anion gap was calculated by using the equation: [Na+] - – – aminoleban, and three patients received moriamin-SN ([Cl ] + [HCO3 ]). This value was adjusted downward as reduction in solutions. the anion gap of 2.5 meq/L for every 1 g/dL declined from 4 g/dL in the serum albumin concentration. A mixed metabolic disorder was – Clinical characteristics at the beginning of therapy detected by use of the following ratio: Danion gap/Dplasma HCO3 6 concentration. Analyses of the demographic data comparing the two groups After excluding patients with (i) severe diarrhea; (ii) pure high of patients found no statistical differences in age, sex distri- anion gap metabolic acidosis; (iii) mixed high anion gap metabolic bution, or estimated creatinine clearance between the acidosis and metabolic alkalosis; and (iv) PaCO2 > 44 mmHg; we – 7 groups. However, statistical analyses found that group II defined metabolic acidosis as HCO3 smaller than 22 meg/L. Estimated creatinine clearance was calculated via Cockcroft- patients had a statistically significant higher baseline serum

Gault equations as follows: ClCr(ml/min) = [(140 - age) ¥ BW(kg)]/ creatinine, blood urea nitrogen level than group I (Table 3). [72 ¥ serum Cr (mg/dL)] ¥ 0.85 (for women)8 Biochemical status after receiving parenteral nutrition

Statistical analysis Patients with chloride-based formula had signiﬁcantly lower

pH, CO2, bicarbonate, and base excess than the acetate- The statistical program SPSS was used to analyse sample data. Descrip- based formula group after receiving parenteral nutrition. 1 tive analyses are reported as means SD unless otherwise specified. Chloride and sodium levels were higher in the chloride- The t-test was used to analyse continuous variables, while categorical based formula group. However, the serum creatinine and variables were assessed by the chi-squared test. A P-value of less than 0.05 was considered statistically significant. Linear regression analysis blood urea nitrogen levels were not different statistically was used to find predictive variables for the occurrence of fall in bicar- between these two groups of patients (Table 3). The inci- bonate level after parenteral nutrition. dence of metabolic acidosis was significant higher in chloride-based group (Table 4).

RESULTS In patients with renal function impairment

There were 68 patients initially recruited into this study. We classiﬁed patients according to pretreatment estimated Four patients were excluded for severe diarrhea, and three CCr into two groups; estimated CCr larger than 30 mL/min

Table 2 Composition of amino acid solutions Components Amino 12X Aminoleban Moriamin-SN Aminomix 1 Amino acids (g/L) L-Isoleucine 5.97 9 5.6 2.5 L-Leucine 11.38 11 12.5 3.7 L-Lysine HCl 9.80 7.6 4.125 L-Lysine Acetate 12.4 L-Methionine 4.33 1 3.5 2.15 L-Phenylalanine 9.74 1 9.35 2.55 L-Tryptophan 1.87 0.7 1.3 1 L-Threonine 5.04 4.5 6.5 2.2 L-Valine 6.90 8.4 4.5 3.1 L-Cysteine 0.23 0.3 1 L-Tyrosine 0.57 0.35 L-Arginine HCL 14.88 7.3 7.9 6 L-Histidine HCl 7.06 3.2 6 1.5 L-Alanine 8.21 7.5 6.2 7.5 L-Aspartic Acid 2.02 3.8 L-Glutamic Acid 1.02 6.5 L-Proline 10.63 8 3.3 7.5 L-Serine 4.67 5 2.2 Glycine 15.68 7 Xylitol 50.00 L-Malic acid 3.065 Electrolyte (mEq/L) Na 150 14 <550 Cl 150 94 0 100.11 Acetate 0 0 60 0 pH 5.0–7.0 5.5–6.5 5.5–6.5 4.8–5.2

Table 3 Laboratory data before and after parenteral nutrition Group I (Aminomix 1) Group II (Paren-aid) Before Infusion After Infusion Before Infusion After Infusion Mean 1 SD Mean 1 SD Mean 1 SD Mean 1 SD Age 67.45 1 13.89 64.46 1 14.22 Sex (M : F) 17:12 16:10 Estimated CCr (ml/min) 41.06 1 23.66 – 35.89 1 25.10 – pH 7.44 1 0.07††† 7.36 1 0.08*** 7.45 1 0.05 7.43 1 0.05 CO2 (mmHg) 32.50 1 4.36††† 25.56 1 6.86*** 31.37 1 4.83† 35.05 1 5.52 O2 (mmHg) 150.33 1 72.78 128.12 1 41.76 129.21 1 63.83 118.11 1 48.28 – HCO3 (meq/L) 21.92 1 3.52††† 15.03 1 5.89*** 21.58 1 3.98 23.35 1 4.55 BE (meq/L) -1.08 1 4.09††† -8.57 1 6.58*** -1.29 1 4.21 -0.10 1 4.51 Cr (mg/dL) 1.59 1 0.93* 1.84 1 1.17 2.38 1 1.72 2.12 1 1.49 BUN (mg/dL) 31.85 1 20.38**††† 55.66 1 34.13 50.85 1 29.14 53.35 1 33.81 Na (mmol/L) 137.94 1 7.18 139.14 1 6.98* 142.30 1 9.54†† 135.34 1 4.47 K (mmol/L) 3.59 1 0.86†† 4.31 1 0.68 3.90 1 0.50 4.13 1 0.59 Cl (mmol/L) 104.77 1 5.65††† 112.65 1 8.90*** 109.74 1 10.72†† 103.74 1 5.88 AG (meq/L) 11.47 1 4.86 12.38 1 3.98** 11.03 1 5.12† 7.99 1 4.18 Hb (g/dL) 10.96 1 2.05**††† 10.14 1 1.35 9.69 1 1.23 10.11 1 1.26 Albumin (g/dL) 2.67 1 0.67* 2.72 1 0.52* 3.05 1 0.52 3.02 1 0.47

Inter-group Comparison: *P < 0.05; **<0.01; ***<0.001. Intra-group Comparison: †P < 0.05; ††<0.01; †††<0.001.

Table 4 The incidence of metabolic acidosis in patients with renal function impairment Estimated CCr Acidosis in Group I Acidosis in Group II P-value <30 mL/min 100% (12/12) 30.77% (4/13) P < 0.001 >30 mL/min 58.82% (10/17) 23.08% (3/13) P = 0.05 Total 75.86% (22/29) 26.92% (7/26) P < 0.001

Table 5 Predictors analysis by linear regression been discussed many times. Early parenteral nutrional solutions used protein hydrolysates, but did not achieve B 1 SE P-value adequate nitrogen retention.9 However, metabolic acidosis Formula -9.70 1 1.36 <0.001 was not a problem until the use of amino acid solutions. Estimated CCr - 0.87 1 0.03 <0.01 Parenteral nutrition preparations have been adjusted to a lower pH value to prevent Maillard’s reaction induced by R2 = 0.537. D-glucose and amino acids in the fluid.4 During the manu- facturing process, some parenteral nutrition solutions have a higher pH value after heat sterilization. Therefore, acidifi- and smaller than 30 mL/min respectively. The incidence of cation should be added to the solution to maintain the metabolic acidosis was significant higher in chloride-based quality of the formula. Hydrochloric acid (non-metabolized) regimen when estimated CCr < 30 mL/min but borderline and acetic acid (metabolized) are used to adjust the pH of significance with estimated CCr > 30 mL/min (Table 4). commercial solutions. The other role of these anions is to balance cationic amino acids or other ions. Plasma chloride and bicarbonate levels have an inverse relationship due to Intra-group comparisons reciprocal renal tubular reabsorption.10 Sodium and its Group I attendant anions are reabsorbed in the proximal tubule. It is possibly that tubular reabsorption is relatively equal for the anions presenting in the lumen, that is, whether chloride or Blood pH, CO2, bicarbonate and base excess were significantly lower while blood urea nitrogen, chloride and sodium bicarbonate. Thus, a diminished absorption of bicarbonate were significantly higher after receiving chloride-based might be compensated by an increased absorption of chlo- 11 parenteral nutrition. No statistical difference was noted ride as a partner to sodium. Therefore, as the plasma chlo- between serum creatinine levels before and after treatment ride level increases, the plasma bicarbonate concentration (Table 3). would be expected to decrease leading to hyperchloremic metabolic acidosis. In contrast, acetate is a metabolized anion. Acetate con- Group II sumes hydrogen ions upon complete oxidation.1 Alterna- tively, infusion of salts of metabolized carboxylic acids will The blood pH and bicarbonate levels between before and slowly increase bicarbonate ion concentration. Thus, the after receiving acetate-based formula were not statistically hydrogen ions from metabolism of cationic amino acids can different. The blood CO2 was significantly higher while be neutralized by acetate. In our group II patients using chloride and sodium were significantly lower after receiving acetate-based formula, there were three amino acid solu- acetate-based parenteral nutrition (Table 3). tions. As shown in Table 2, all of them and aminomix 1 have cationic amino acids (lysine, histidine and arginine). The difference among these amino acid solutions and ami- Predictors analysis nomix 1 are acetate and chloride ratio and the component of amino acids. As previously mentioned, metabolism of For prediction of fall in bicarbonate level, we examined the cationic amino acids and sulphur-containing amino acids following variables: parenteral nutrition formula, age, sex, (i.e. methionine, cysteine and cystine) generate an acid estimated creatinine clearance, serum creatinine, blood urea load, while metabolism of anionic amino acids (i.e. nitrogen, albumin and haemoglobin level prior to parenteral glutamate and aspartate) consumes hydrogen ions. The net nutrition therapy. Using multiple-stepwise linear regression, acid equilibrium attributed to amino acids in these two we determined that parenteral nutrition formula and esti- groups of patients was calculated. Although the net acid mated creatinine clearance were independent predictors of equilibrium was higher in group I, there was no statistical bicarbonate decline (Table 5). difference between the groups. Regardless of the amino acid solution received, there was no significant bicarbonate DISCUSSION decline after parenteral nutrition in group II patients. It is the acetate-based components in an electrolyte solution As the appearance of parenteral nutrition as a practical that prevents acidosis after parenteral nutrition therapy in reality in 1968, the occurrence of metabolic acidosis had patients with renal function impairment.

As the kidney is the major organ to excrete acid load, the REFERENCES ability to excrete hydrogen ions begins to fall when the GFR is less than 40–50 mL/min.6 Both decreased titratable 1. Kushner RF. Total parenteral nutrition associated metabolic aci- acidity and reduced bicarbonate reabsorption may contrib- dosis. J. Parenter. Enteral Nutr. 1986; 10: 306–10. 2. Heird WC, Dell RB, Driscoll JM, Grebin B. Metabolic acidosis ute to the decline in net acid excretion. In group I patients, resulting from intravenous alimentation mixtures containing syn- the serum creatinine level was not different statistically thetic amino acids. N. Engl. J. Med. 1972; 287: 943–8. before and after chloride-based parenteral nutrition, 3. England BK, Mitch WE. Acid-base, fluid, and electrolyte aspects of however, the bicarbonate decline was statistically signifi- parenteral nutrition. In: Kokko JP, Tanen RL (eds). Fluids and cant. As described above, there are acid loads resulting from Electrolytes. Philadelphia, PA: WB Saunders, 1996; 787–801. parenteral nutrition. Patients with renal insufficiency 4. Sugiura SI, Inagaki K, Noda Y, Nagai T, Nabeshima T. Acid load cannot compensate for the increased hydrogen load by during total parenteral nutrition: Comparison of hydrochloric acid increasing renal acid excretion. According to our analysis, and acetic acid on plasma acid-base balance. Nutrition 2000; 16: estimated creatinine clearance rather than serum creatinine 260–63. level may predict the occurrence of metabolic acidosis. This 5. Peters O, Ryan S, Matthew L, Cheng K, Lunn J. Randomised controlled trial of acetate in preterm neonates receiving parenteral result reminds us again that serum creatinine level may not nutrition. Arch. Dis. Child. Fetal Neonatal Ed. 1997; 12: F12–F15. reflect actual renal function in malnourished patients. 6. Rose BD, Post TW. Metabolic acidosis. In: Rose BD, Post TW The incidence of metabolic acidosis is significant higher (eds). Clinical Physiology of Acid-Base and Electrolyte Disorders. Sin- in patients receiving chloride-based parenteral nutrition. gapore: McGraw-Hill, 2001; 578–646. The risk is especially elevated in patients with impaired 7. Rose BD, Post TW. Regulation of acid-base balance. In: Rose BD, renal function. Estimated creatinine clearance could predict Post TW (eds). Clinical Physiology of Acid-Base and Electrolyte Dis- the occurrence of metabolic acidosis. The role of insufficient orders. Singapore: McGraw-Hill, 2001; 327. acetate content seemed to be more important than chloride 8. Cockcroft DW, Gault MH. Prediction of creatinine clearance load in the occurrence of metabolic acidosis. In conclusion, from serum creatinine. Nephron 1976; 16: 31–41. renal function should be evaluated before parenteral nutri- 9. Patel D, Anderson GH, Jeejeebhoy KN. Amino acid adequacy of parenteral casein hydrolysate and oral cottage cheese in patients tion. Once renal function impairment is noted, chloride- with gastrointestinal disease as measured by nitrogen balance and based parenteral nutrition should be avoided. Regular blood aminogram. Gastroenterology 1973; 65: 427–37. monitoring of blood gas levels is mandatory for patients 10. Caldwell MD, O’neill JA, Meng HC et al. Evaluation of a new receiving parenteral nutrition. amino acid source for use in parenteral nutrition. Ann. Surg. 1977; 185: 153–61. ACKNOWLEDGEMENTS 11. Sartorius OW, Roemmelt JC, Pitts RF et al. The nature of the renal compensations in ammonium chloride acidosis. J. Clin. This research was supported by grants from the Ta-Tung Invest 1949; 28: 423–39. Kidney Foundation and Mrs Hsiu-Chin Lee Kideny Research Foundation, Taipei, Taiwan.