Beyond bicarbonate: complete acid base assessment in patients receiving intermittent hemodialysis

Transcription

1 Nephrol Dial Transplant (2017) 32: doi: /ndt/gfw022 Advance Access publication 21 March 2016 Beyond bicarbonate: complete acid base assessment in patients receiving intermittent hemodialysis Marco Marano 1, Stefano Marano 2 and F. John Gennari 3 1 Hemodialysis Unit, Maria Rosaria Clinic, Pompeii, Naples, Italy, 2 Department of Information and Electrical Engineering and Applied Mathematics, University of Salerno, Fisciano, Salerno, Italy and 3 University of Vermont, UVM Medical Center, 1 South Prospect St., Burlington, VT, USA Correspondence and offprint requests to: F. John Gennari; fgennari@uvm.edu ABSTRACT Background: Acid base assessments in hemodialysis patients have been limited almost entirely to measurements of total CO 2 concentration, and assumptions have been made about the presence of acid base disorders. To gain a fuller understanding of the acid base status of stable hemodialysis patients, we analyzed measurements of pco 2, ph and HCO 3 obtained in a cohort of chronic stable hemodialysis patients over a 5-year period. Methods: We reviewed acid base measurements taken predialysis from fistula blood in 53 outpatients receiving hemodialysis thrice weekly between 2008 and In these patients, ph and pco 2 were measured using an onsite blood gas analyzer, and HCO 3 was computed. Relevant clinical and laboratory data were obtained from medical records. Factors affecting serum HCO 3 were identified. Simple and mixed acid base disorders were diagnosed using accepted rules. Results: Serum HCO 3 was affected by age, normalized protein catabolic rate, interdialytic weight gain and length of interval between treatments. As expected, metabolic acidosis was the most common acid base disorder, but respiratory acid base disturbances, as simple or complex disorders, were found in 41% of the measurements. Respiratory alkalosis was seen more frequently than respiratory acidosis, but the latter disorder was more commonly associated with serious comorbidities. Conclusions: Respiratory acid base disorders are an important component of the acid base abnormalities seen in hemodialysis patients and are not identified by measuring total CO 2 concentration; hence, complete acid base measurements are needed to determine the components of hemodialysis patients acid base status that are contributing to mortality risk. Keywords: acid-base disorders, chronic hemodialysis, chronic renal failure, hemodialysis, blood gas analysis INTRODUCTION Over the last decade, concerns have been raised about the acid base status of patients receiving intermittent hemodialysis [1 6]. Despite these concerns, most of the published information about acid base equilibrium has been limited to measurements of serum total CO 2. These measurements have provided no insight into systemic ph nor into the ventilatory response to the prevailing systemic ph. In the absence of unstable blood pressure or tissue perfusion, blood ph is directly and predictably related to systemic and cell ph, and the latter is critical for enzyme function and cellular metabolism. In addition, recent studies have provided evidence that blood ph may be key for assessing mortality risk in hemodialysis patients [3]. Thus, to evaluate acid base status more completely in patients receiving hemodialysis, we have reviewed the measurements of ph, pco 2 and HCO 3 made over the course of 5 years in patients treated at a single dialysis unit. This cohort all received the same dialysis bath prescription, enabling a clear assessment of the patient s acid base response to exposure to the same dialysis bicarbonate and acetate concentration. The results indicate the presence of a variety of unrecognized respiratory and mixed acid base disorders, which may contribute to mortality risk in these patients. MATERIALS AND METHODS Acid base measurements were reviewed in 60 stable patients who had all been receiving hemodialysis treatments three times weekly for at least 3 months in a single dialysis unit (Maria Rosaria Clinic, Pompeii, NA, Italy). All patients were dialyzed during this period of time using a bath (Diasol, Baxter) containing HCO 3, 35; acetate,3;na þ,138;cl,109;ca þþ,1.5; Mg þþ, 0.5; glucose, 5.55 and K þ, 2 (or occasionally 3) mmol/l after final dilution. To minimize bias, we excluded from further VC The Author Published by Oxford University Press Downloaded from onhttps://academic.oup.com/ndt/article-abstract/32/3/528/ behalf of ERA-EDTA. All rights reserved. 528

2 analysis one patient for whom only a single measurement was available and six patients who had more than twice the average measurements of the remaining population. The final group included 53 patients and 362 samples. The number of samples per patient ranged from 2 to 14 (average 7). All samples were collected predialysis over a 5-year period ( ) and were obtained both before the long interval (3 days) and short interval between treatments (2 days). If oxygen saturation was <97%, a pulse oximeter was placed to ensure that the blood was equivalent to arterial blood. Blood ph and pco 2 were measured immediately using a blood gas analyzer in the dialysis unit (Roche OMNI S4, Roche Diagnostics, Mannheim, Germany) and HCO 3 was calculated from these measurements using the Henderson equation. System calibrations and quality control measurements were automatically performed by the analyzer. Patient diseases and lab data are those reported in the medical records. All the studied patients were followed until December 2014 (2 years after completion of the measurements) and allcause mortality was recorded. Six patients received kidney transplants during this interval, and none of these patients died. A portion of the data was published in an earlier report for use in developing a new formula to evaluate the ventilatory response to metabolic acidosis in hemodialysis patients [7]. Statistics Our database contains measurements from individual blood samples and aggregated data for each patient. For the latter analysis, all measurements collected at different times from the same patient were averaged to yield a single data point for that patient. In both cases, average values and standard deviations of the pertinent population were computed. The effects of different factors on predialysis serum HCO 3 were analyzed by dividing the population into two equal size classes, larger or smaller than Table 1. Patient characteristics Age (years) a Gender Male 33 Female 20 Time on dialysis (months) a Weight (kg) a Interval weight gain (kg) a % Body weight a Primary cause of ESKD Diabetes mellitus 10 (18.9%) Hypertension 13 (24.5%) Glomerulonephritis 6 (11.3%) Urologic disorder 5 (9.4%) Cystic kidney disease 2 (3.8%) Multiple myeloma 2 (3.8%) Vasculitis 2 (3.8%) Amyloidosis 1 (1.9%) Unknown 12 (22.6%) Comorbidities Coronary artery disease 22 (41.5%) Peripheral vascular disease 17 (32.1%) Diabetes mellitus 14 (26.4%) Systolic heart failure (EF <50%) 15 (28.3%) COPD b 17 (32.1%) Hepatic cirrhosis 2 (3.8%) a Mean 6 SD. b Chronic obstructive pulmonary disease. the median, and observing the average value in each class. Linear regression analysis was carried out to assess the relationship between continuous variables of interest and predialysis serum HCO 3. Statistical significance was assessed by the P-value from a two-tailed unequal-variance two-sample t-test and by v 2 test and Pearson s correlation. RESULTS The patients in the analyzed group comprised 33 men and 20 women (Table 1). All were Caucasian. The mean age was 64.3 years. The mean body weight was 67.9 kg, with an average weight gain between treatments of 2.1 kg (3.1% of body weight). The average time on dialysis was 64 months. In 43% of the patients, the cause of end-stage kidney disease (ESKD) was either diabetes or hypertension. Cardiovascular disease was a comorbid feature in the majority of patients. Table 2 shows the mean acid base values for all 362 measurements. Mean predialysis ph was 7.37, pco 2 was 37.1 mmhg and calculated HCO 3 was 21.0 mmol/l. The mean total CO 2 was 22.5 mmol/l and the mean base excess (BE) was 3.9 mmol/l (calculated in 261 samples). Figure 1 shows the distribution of values for ph, pco 2 and HCO 3 ; 10.5% of samples showed serum HCO 3 values >24 mmol/l, and 21.3% of samples were <19 mmol/l. As shown in Tables 3 and 4, predialysis serum HCO 3 was influenced by a variety of factors. Table 3 shows linear regression analysis relating predialysis serum HCO 3 with age and normalized protein catabolic rate (npcr) for the aggregated data. Age was directly related to the serum HCO 3 and npcr (assessed within 30 days of the acid base measurement) was inversely correlated with this value. With regard to the influence of interdialytic weight gain on predialysis serum HCO 3,no effect was noted during the short interval between treatments, but an inverse relationship was discernable after the long interval (Table 3). Table 4 shows the effects of age, npcr and the interval between treatments on predialysis serum HCO 3.For the first two variables, the aggregated data have been divided into two groups at the median value. As can be seen, predialysis serum HCO 3 was 0.9 mmol/l higher in measurements obtained in those older than the median age compared with those who were younger, and predialysis serum HCO 3 was 1.4 mmol/l lower in those patients with npcr >1.04 g/kg body wt/ day. When all the measurements were considered, predialysis Table 2. Average acid base values for all 362 measurements a,b ph pco 2 (mmhg) HCO 3 (mmol/l) Total CO 2 (mmol/l) Base excess (mmol/l) c po 2 (mmhg) a One hundred and thirty-four obtained after long interval between treatments, 228 after short interval. b Mean 6 SD. c Calculated in 261 samples. Acidbaseassessmentinhemodialysispatients 529

3 FIGURE 1: Histograms of the distribution of predialysis ph, pco 2 and HCO 3. Table 3. Regression analysis for continuous variables affecting predialysis serum HCO 3 Variables Slope Intercept R 2 Pearsoncoefficient P- value Age a npcr Weight gain b npcr, normalized protein catabolic rate. a Obtained in aggregated data (by patient), N ¼ 53 for age, N ¼ 48 for npcr. b During long interval between treatments, N ¼ 134. Table 4. Factors influencing pre-dialysis serum HCO 3 NumberMean 6 SD (mmol/l) D (mmol/l)p-value Age 69.5 years a Age >69.5 years < npcr a,b npcr > Short interval (48 h) Long interval (72 h) a Aggregated data by patient. b Normalized protein catabolic rate (mg/kg body weight per day). serum HCO 3 was 0.9 mmol/l lower after the long interval between treatments than after the short interval (P ¼ ). To evaluate the acid base observations in these hemodialysis patients further, we have arbitrarily compared them with published normal values (61 SD) for humans with functioning kidneys for the purpose of classification (Table 5). The normal values used are ph , pco mmhg and HCO mmol/l 8]. Our classification yielded eight groups with regard to acid base status, based on the criteria shown in Table 5. The table shows the number of measurements and patients in each of the categories. Of the 362 measurements, all three acid base values were within the normal range in only 17 instances. When the measurements were averaged and aggregated for each patient, only 4 of the 53 patients were classified as normal. Not surprisingly, the largest abnormal group was classified as having metabolic acidosis, comprising 140 measurements and 22 patients. If the patients classified as having metabolic acidosis combined with respiratory abnormalities are added to this group, the number increases to 38 of the 53 patients (Table 5). Only nine of the measurements showed values consistent with metabolic alkalosis, and no patients averaged out to have this disorder. Respiratory acidosis (defined as a pco 2 >42 mmhg) was present in 47 measurements in five patients, with three of them classified as also having metabolic acidosis (see Table 5). Respiratory alkalosis was present in 18 patients, 72% of whom also had metabolic acidosis. In total, respiratory disorders were found in 41.4% of the measurements and in 43.4% of the averaged data per patient. In patients classified as having either respiratory acidosis or alkalosis in one or more of their measurements, we looked at comorbidities to determine whether the abnormal pulmonary response was due to an underlying illness. By v 2 analysis, measurements that showed respiratory acidosis were significantly more likely to be associated with systolic heart failure and chronic obstructive pulmonary disease (COPD) than were measurements that were classified as metabolic acidosis (P < for both) or respiratory alkalosis (P ¼ and P ¼ 0.001, respectively). The same trend was observed in the aggregated patient data, but due to the limited number of patients in these classes, none of the differences reached statistical significance. Although the numbers are too small to analyze statistically, it is noteworthy to look at mortality trends in the patients with various acid base disorders. During the 2-year follow-up period, 4 of the 5 patients with respiratory acidosis died, whereas only 8 of the 18 patients with respiratory alkalosis and 9 of the 22 patients with metabolic acidosis died. From a clinical perspective, a more useful way to view these data is to look at serum HCO 3 as a continuous function and to evaluate for each measurement of HCO 3 whether the secondary pco 2 response was appropriate. Because virtually all HCO 3 values were 24 mmol/l, we used the slope of the empiric rule of thumb for metabolic acidosis to evaluate the secondary ventilatory response: 1.2 ¼ DpCO 2 /DHCO 3 [9]. This slope is indicated by the solid line in Figure 2. Thisruleof thumb is associated with the lowest error in the prediction of ventilatory response to metabolic acidosis in hemodialysis patients [7]. Linear regression of our data, in fact, shows a slope of 1.22, almost identical to the rule of thumb value. Using generally accepted guidelines, we have used the generous range of 65 mmhg from the expected pco 2 for any given value of HCO 3 to identify abnormal values. Using this rule, we identified 48 values indicating abnormal CO 2 retention (respiratory acidosis) 530 M. Marano et al.

4 Table 5. Acid base disturbances based on normal values for individuals with functioning kidneys a Class Criteria Acid base status Number of samples Number of patients 1 ph Normal 17 (4.7%) 4 (7.5%) HCO mmol/l pco mmhg 2 ph<7.38 Metabolic acidosis 140 (38.7%) 22 (41.5%) HCO 3 <22 mmol/l pco 2 42 mmhg 3 ph>7.42 Metabolic alkalosis 9 (2.5%) 0 HCO 3 >26 mmol/l pco 2 38 mmhg 4 ph<7.38 Respiratory acidosis 27 (7.5%) 2 (3.8%) HCO 3 22 mmol/l pco 2 >42 mmhg 5 ph>7.42 Respiratory alkalosis 37 (10.2%) 5 (9.4%) HCO 3 26 mmol/l pco 2 <38 mmhg 6 ph<7.38 Metabolic and respiratory acidosis 20 (5.5%) 3 (5.7%) HCO 3 <22 mmol/l pco 2 >42 mmhg 7 ph Metabolic acidosis and respiratory alkalosis 64 (17.7%) 13 (24.5%) HCO 3 <22 mmol/l pco 2 <38 mmhg 8 ph>7.42 Metabolic and respiratory alkalosis 2 (0.5%) 0 HCO 3 >26 mmol/l pco 2 <38 mmhg 9 Not meeting above criteria b 46 (12.7%) 4 (7.5%) a Normal ranges: ph, ; pco 2, mmhg; HCO 3, (see Marano et al. [7]). b Most just missing normal, with only one acid base value abnormal. Table 6. Subset of measurements with predialysis blood ph 7.40 (N ¼ 100 in 12 patients) Class a Samples (%) Patients (%) 1-Normal 5 (5.0%) 2 (16.7%) 3-Metabolic alkalosis 9 (9.0%) 0 5-Respiratory alkalosis 37 (37.0%) 5 (41.7%) 7-Respiratory alkalosis and metabolic acidosis 24 (24.0%) 4 (33.3%) 8-Metabolic and respiratory alkalosis 2 (2.0%) 0 9-Not meeting criteria 23 (23.0%) 1 (8.3%) a ph, ; pco2 (mmhg) ; HCO 3 (mmol/l), In this subset, there were no instances of metabolic or respiratory acidosis. FIGURE 2: Plot of pco 2 versus HCO 3. Individual measurements are depicted with different symbols according to the classes shown in Table 5. The solid line represents the slope of the rule of thumb, 1.2 ¼ DpCO 2 /DHCO 3, and the two dashed lines represent 65 mmhg. with associated metabolic acidosis. Figure 3 depicts the relationship between serum HCO 3 and ph in all the measurements, highlighting the wide range of serum HCO 3 values in patients with ph levels DISCUSSION and 23 values indicating abnormal hyperventilation (respiratory alkalosis). Based on the recent publication of data showing an increased mortality risk in patients with predialysis blood ph values 7.40 [3], we looked at the subset of measurements and patients who had predialysis ph values in this range. As shown in Table 6, this group comprised 100 measurements in 12 patients. For this group, 9 of the 12 had respiratory alkalosis, either as an isolated primary disorder or as a mixed disorder Our results indicate that hemodialysis patients can have an array of acid base abnormalities and resultant variations in ph for any given predialysis serum HCO 3. Respiratory acid base abnormalities were present in 41% of measurements and in 23 patients in the aggregated data. Our data illustrate the presence of a more complex acid base picture in hemodialysis patients than reflected by measurements of serum total CO 2 alone. Yamamoto et al. [3] have also shown a wide array of ph values for any given predialysis serum HCO 3. They stressed that Acidbaseassessmentinhemodialysispatients 531

5 FIGURE 3: Plot of ph versus HCO 3. Individual measurements are depicted with different symbols according to the classes shown in Table 5. from a physiological perspective, ph is most likely to be the important determinant of mortality risk. They did not fully characterize the acid base disorders in their population, so one cannot determine the role of respiratory acid base abnormalities in the variations in ph they observed, but they were likely present. In our population, variations in ph are clearly due to wide variations in ventilation, including primary respiratory acid base disorders, as well as variations in the expected ventilatory response to changes in serum HCO 3. Respiratory acidosis and alkalosis are likely to be associated with an increase in mortality risk, and these disorders cannot be diagnosed from isolated measurements of serum total CO 2 in patients with ESKD since no secondary changes in serum HCO 3 occur with these disorders in the absence of kidney function [10]. As shown in Figure 2, the presence of significant respiratory acidosis was found to occur across a wide range of serum HCO 3 values, but was particularly evident in those with serum HCO 3 values that were mmol/l. The patients with these measurements had a significantly higher incidence of COPD and congestive heart failure, and these disorders were likely responsible for the respiratory acidosis. The presence of cardiac and pulmonary disease with significant CO 2 retention identifies a group with higher mortality risk despite the absence of severe metabolic acidosis. In our population, only one of five patients with respiratory acidosis survived in the 2 years of follow-up after the measurements were obtained, but further studies are needed to fully characterize this risk. Respiratory alkalosis, in contrast, occurred in patients with normal and low serum HCO 3 values (Figure 2), and these patients did not show a significant increase in cardiac or pulmonary disorders compared with patients without respiratory alkalosis, although the majority had significant comorbidities. The cause of the hyperventilation in these patients could simply be due to anxiety at the time of sampling, but this disorder could also reflect more serious disorders such as underlying cerebral, cardiac or respiratory diseases. If one examines the subset of our measurements in which predialysis blood ph was 7.40, a level associated with increased mortality risk [3], more than half of the measurements and three-quarters of the patients had respiratory alkalosis (Table 6). Figure 3 shows a plot of serum HCO 3 versus ph and highlights the range of serum HCO 3 values that were associated with high predialysis ph values. Most of these measurements reflect the presence of respiratory alkalosis. The high prevalence of respiratory disorders in our measurements raises a concern about making adjustments in bath HCO 3 based solely on the predialysis serum total CO 2.Ascan be appreciated in Figure 3, approximately one-fourth of the measurements in patients with predialysis serum HCO 3 <20 meq/l have ph values >7.40, due to a complicating respiratory alkalosis. Increasing bath HCO 3 in these patients could worsen their alkalemia and increase the risk of cardiac arrhythmias. On the other end of the spectrum, almost half of the measurements in patients with predialysis serum HCO 3 >25 meq/l have ph values <7.40, due to a complicating respiratory acidosis. Reducing bath HCO 3 in these patients could worsen their acidemia, exacerbating muscle catabolism and bone disease. Based on our findings and these considerations, we recommend that measurements of ph and pco 2,aswellasserumHCO 3, be undertaken prior to initiating any change in bath HCO 3 in hemodialysis patients. It is not surprising that the majority of our measurements fall into the category of metabolic acidosis compared with the lower limit for normal serum HCO 3 in individuals with normal kidney function. Bath HCO 3 is the primary determinant of predialysis blood HCO 3, and this parameter has been set empirically in most treatment centers to achieve a predialysis serum total CO 2 in the mmol/l range in the majority of patients. It is important to emphasize that this value represents a nadir from the high level immediately postdialysis. This point is reflected by our measurements: serum HCO 3 was almost 1 mmol/l lower after the long interval between treatments due to continued acid production (Table 4). Variations in the nadir value also are caused by variations in the rate of endogenous acid production in the interval between treatments [1, 5]. Consistent with prior studies, we found an inverse correlation between npcr and predialysis serum HCO 3 in our patients (Table 3). A reduction in the nadir value in patients gaining more weight between treatments was also found in our patients, consistent with prior studies [11]. Greater than 80% of the measurements and 90% of the patients in our population had predialysis HCO 3 values between 18 and 24 mmol/l. These values correspond to a range of serum total CO 2 values from 19 to 25 mmol/l, consistent with results in prior large cohort studies [4, 5]. There were 38 measurements with serum HCO 3 <18 mmol/l, but when the data were aggregated, only three patients had an average value this low. There were also 38 measurements with serum HCO 3 >24 mmol/l. However, in the aggregated data, only five patients fell into this category. The number of patients with predialysis total CO 2 values >25 mmol/l has increased over the last 20 years, even in dialysis units in which bath prescription has not changed, and these patients appear to have a higher mortality risk [2, 5]. This increase may be due to decreased dietary intake in the increasingly elderly population in our dialysis units and correspondingly lower rates of endogenous acid 532 M. Marano et al.

6 production. In our patients, in fact, a clear effect of age on predialysis serum HCO 3 was evident. Measurements in patients above the median age of 69.5 years showed a predialysis serum HCO mmol/l higher than in those below the median age (Table 4). Our study is limited by the fact that it is a retrospective review of our measurements. Also, our study is limited by the patient population, which is from a single dialysis unit and all the patients were Caucasian. Nonetheless, the results illuminate the complexity of the acid base response to dialysis and should stimulate further studies to clarify the spectrum of acid base abnormalities that can occur. In summary, our measurements indicate a wide array of acid base abnormalities in patients receiving intermittent hemodialysis treatments. Unrecognized respiratory acid base abnormalities, and particularly respiratory alkalosis, appear to account for a significant fraction of the abnormalities seen. Further studies with complete acid base measurements in the dialysis population need to be undertaken to define more fully the components of the patient s acid base status that are contributing to mortality risk. CONFLICT OF INTEREST STATEMENT None declared. REFERENCES 1. Gennari FJ. Very low and high predialysis serum bicarbonate levels are risk factors for mortality: what are the appropriate interventions? Semin Dial 2010; 23: Lisawat P, Gennari FJ. Approach to the hemodialysis patient with an abnormal serum bicarbonate concentration. Am J Kidney Dis 2014; 64: Yamamoto T, Shoji S, Yamakawa T et al. Predialysis and postdialysis ph and bicarbonate and risk of all-cause and cardiovascular mortality in longterm hemodialysis patients. Am J Kidney Dis 2015; 66: Bommer J, Locatelli F, Satayathum S et al. Association of predialysis serum bicarbonate levels with risk of mortality and hospitalization in the Dialysis Outcomes and Practice Patterns Study (DOPPS). Am J Kidney Dis 2004; 44: Wu DY, Shinaberger CS, Regidor DL et al. Association between serum bicarbonate and death in hemodialysis patients: is it better to be acidotic or alkalotic? Clin J Am Soc Nephrol 2006; 1: Tentori F, Karaboyas A, Robinson BM et al. Association of dialysate bicarbonate concentration with mortality in the dialysis outcomes and practice patterns study (DOPPS). Am J Kidney Dis 2013; 62: Marano M, D Amato A, Marano S. A very simple formula to compute pco 2 in hemodialysis patients. Int Urol Nephrol 2015; 47: Berend K, de Vries AP, Gans RO. Physiological approach to assessment of acid-base disturbances. NEnglJMed2014; 371: Bushinsky DA, Coe FL, Katzenberg C et al. Arterial PCO 2 in chronic metabolic acidosis. Kidney Int 1982; 22: Gennari FJ. Acid-base disorders in dialysis patients. In: FJ Gennari, HJ Adrogue, JH Galla, NE Madias (eds). Acid-Base Disorders and Their Treatment. Boca Raton, FL, USA: Taylor & Francis, 2005: Fabris A, LaGreca G, Chiaramonte S et al. The importance of ultrafiltration on acid-base status in a dialysis population. ASAIO Trans 1988; 34: Received: ; Editorial decision: Acidbaseassessmentinhemodialysispatients 533