Metabolic Acidosis in Advanced Renal Failure:

Differences Between Diabetic and Nondiabetic Patients

Francisco Caravaca, MD, Manuel Arrobas, MD, Jose L. Pizarro, MD, and Juan F. Espárrago, MD

● Metabolic acidosis is almost invariably a consequence of advanced renal failure, although its severity can vary
widely. To evaluate the determinants of the severity of metabolic acidosis, with special interest in determining if
there is any difference in the prevalence and severity of metabolic acidosis between patients with and without
diabetes, 113 predialysis patients with renal failure were studied. Criteria for inclusion onto the study were:
creatinine clearance (Ccr)/1.73 m2 less than 30 mL/min, no alkali therapy within the previous 30 days, and the
absence of respiratory diseases. Forty-eight patients had diabetes (33 patients with diabetic nephropathy). The
following data were analyzed: demographics; cause of renal failure; hematocrit; serum urea, creatinine, uric acid,
albumin, glucose, hemoglobin A1c, bicarbonate, sodium, potassium, chloride, calcium, phosphorus, and alkaline
phosphatase levels; anion gap; urinary protein excretion; Ccr/1.73 m2; half of the sum of creatinine and urea
clearances (Ccr-Cu); protein-equivalent nitrogen appearance (PNA); and whether the patients received diuretics (75
patients), angiotensin-converting enzyme inhibitors (54 patients), and/or calcium channel blockers (55 patients).
After the exclusion of eight patients because of hypochloremia (three patients with and five patients without
diabetes), mean serum bicarbonate levels were significantly greater in patients with diabetes than in the rest of the
patients (20.7 ⴞ 2.3 v 18.2 ⴞ 2.3 mmol/L; P ⴝ 0.0001). The mean anion gap (mmol/L) was also significantly less in
patients with than without diabetes (19.70 ⴞ 3.65 v 22.35 ⴞ 3.64; P ⴝ 0.003). Eleven of 105 patients had serum
bicarbonate levels of 23 mmol/L or greater (9 patients with and 2 patients without diabetes). Pure elevated anion gap
followed by mixed (high anion gap and hyperchloremia) were the most common types of metabolic acidosis
observed in both groups. There were no differences in PNA, diuretic treatment, or vomiting history between patients
with and without diabetes. By multiple logistic regression analysis, the best determinants for a serum bicarbonate
level greater than 19 mmol/L were: the diagnosis of diabetic nephropathy (odds ratio, 0.107; P ⴝ 0.0002), Ccr-Cu
(odds ratio, 0.824; P ⴝ 0.014), and age (odds ratio, 0.966; P ⴝ 0.046). In conclusion, patients with diabetes with
advanced renal failure showed a less severe metabolic acidosis, which cannot be explained by gastrointestinal
hydrogen ion losses, drugs, or reduced protein catabolic rate. Patients with diabetes may have a more efficient
extrarenal generation of bicarbonate than end-stage renal failure patients without diabetes.
娀 1999 by the National Kidney Foundation, Inc.
INDEX WORDS: Chronic renal failure; diabetes; metabolic acidosis.

M ETABOLIC ACIDOSIS is a common con-

sequence of chronic renal insufficiency.1-4
The failure to maintain net acid excretion leads
to the exhaustion of extracellular buffers. Re-
duced serum bicarbonate levels, along with an
elevated anion gap with or without hyperchlore-
mia, are the characteristic biochemical features
of uremic chronic metabolic acidosis, which is
more frequently observed when the glomerular
filtration rate decreases to less than 25 to 30
mL/min.
The severity of metabolic acidosis can vary
widely among uremic patients with a similar
degree of renal dysfunction.1-4 Kidney diseases
From the Servicio de Nefrologı́a, Hospital Regional Uni-
versitario Infanta Cristina, Badajoz, Spain.
Received July 7, 1998; accepted in revised form October
30, 1998.
Address reprint requests to Francisco Caravaca, MD,
Servicio de Nefrologı́a, Hospital Infanta Cristina, 06080
Badajoz, Spain. E-mail: fcaravacam@senefro.org
娀 1999 by the National Kidney Foundation, Inc.
0272-6386/99/3305-0011$3.00/0
associated with more severe tubulointerstitial
damage can be accompanied by more severe
acidosis in the early stages of renal failure.5,6
However, when renal function decreases to less
than a critical point, the inability of the kidney to
excrete acid is the main mechanism of metabolic
acidosis.1-4 When this stage is reached, such
factors as a reduction in dietary acid load, hydro-
gen ion losses through the gastrointestinal tract
with superimposed metabolic alkalosis, exog-
enous administration of alkali, or the contribu-
tion of extrarenal nonbicarbonate buffers7,8 may
potentially modulate the severity of metabolic
acidosis.
It has previously been observed that a small
proportion of patients with end-stage renal fail-
ure can have normal serum bicarbonate levels
despite no overt therapeutic intervention.8 Most
of these patients have diabetes mellitus. How-
ever, the mechanisms by which these patients are
able to maintain normal bicarbonate concentra-
tions remain unclear.
This study was designed to evaluate whether

892 American Journal of Kidney Diseases, Vol 33, No 5 (May), 1999: pp 892-898
METABOLIC ACIDOSIS IN ADVANCED RENAL FAILURE
there is a difference in the prevalence and sever-
ity of metabolic acidosis between patients with
and without diabetes with advanced renal failure.
PATIENTS AND METHODS
The study group consisted of 113 patients (mean age ⫾
standard deviation [SD], 63 ⫾ 14 years; 60 women) succes-
sively recruited from those referred to our service because of
advanced renal failure from October 1996 to February 1998.
To be included on the study, patients were required to have a
creatinine clearance corrected for a body surface area of 1.73
m2 (Ccr/1.73 m2) of less than 30 mL/min, no intake of alkali
(bicarbonate, acetate, citrate, or carbonate) within the previ-
ous 30 days, and the absence of pulmonary diseases. All
were in stable condition with no overt clinical events at the
time of the study that could explain an acute imbalance of
the acid-base status, such as heart failure, hemorrhage, or
sepsis.
The causes of renal failure were primary glomerulonephri-
tis (25 patients), diabetic nephropathy (33 patients), tubuloin-
terstitial disease (13 patients), renal vascular disease (16
patients), adult polycystic kidney disease (7 patients), un-
known origin (13 patients), or others (6 patients). Apart from
the 33 patients with diabetic nephropathy (25 with type I
diabetes mellitus), another 15 patients had diabetes mellitus
(13 with type II diabetes mellitus). The causes of renal
insufficiency in patients with diabetes without diabetic ne-
phropathy were primary glomerulonephritis (3 patients),
ischemic nephropathy (8 patients), interstitial nephritis (2
patients), and adult polycystic kidney disease (2 patients).
Renal biopsy was not routinely performed in most patients
diagnosed with diabetic nephropathy, and the clinical crite-
ria for this diagnosis were slowly progressive renal failure
with severe proteinuria in patients with a history of type I
diabetes for more than 15 years duration or for more than 10
years in patients with type II diabetes associated with hyper-
tension, morphologically normal kidneys, and other compli-
cations of diabetes (mainly retinopathy or neuropathy).
Primary glomerulonephritis was diagnosed by kidney bi-
opsy in all patients. Interstitial nephritis was also diagnosed
by kidney biopsy in two patients with diabetes. Ischemic
nephropathy as the cause of renal failure in patients with
diabetes was diagnosed when severe renal vascular occlu-
sion with unilateral or bilateral ischemic atrophy was associ-
ated with mild to moderate proteinuria without other overt
diabetic complications.
Seventy-five patients received diuretics (furosemide), 54
patients received angiotensin-converting enzyme (ACE) in-
hibitors, and 55 patients received calcium channel blockers.
No patients were receiving steroidal or nonsteroidal anti-
inflammatory drugs.
In the morning, after fasting for at least 8 hours, the
following biochemical parameters were determined in blood
samples: urea, creatinine, sodium, potassium, chloride, cal-
cium, phosphorus, alkaline phosphatase, magnesium, albu-
min, uric acid, and glucose levels (Hitachi Autoanalyzer,
Boehringer, Germany); hematocrit; hemoglobin; and venous
bicarbonate levels (IL1306 gas analyzer; Intrumentation
Laboratory, Milano, Italy). Heparinized venous blood samples
for bicarbonate determinations were collected under anaero-
893
bic conditions; pH and PCO2 were analyzed in the blood
sample within 15 minutes after collection. The concentration
of bicarbonate in plasma was calculated from the pH and
PCO2 using a pK (constant dissociation) of 6.1 and a solubil-
ity factor of 0.0301. Hemoglobin A1c levels were determined
in all patients with diabetes. In a 24-hour urine sample, urea,
creatinine, and protein excretion were also determined. Ccr
and the half of the sum of creatinine and urea clearances
(Ccr-Cu) corrected for a body surface area of 1.73 m2 were
determined as a measure of glomerular filtration rate. Pro-
tein equivalent to nitrogen appearance (PNA), an indirect
method for assessing protein intake, was determined by the
24-hour urinary urea nitrogen excretion, following the com-
bined formulas of Cottini et al9 and Maroni et al,10 as
described by Bergström et al.11 Anion gap was calculated by
the conventional formula: [(Na ⫹ K) – (Cl ⫹ HCO3)], and
the result was corrected downward for plasma albumin
concentrations (2.5-mEq/L reduction in anion gap for each
1-g/dL reduction in plasma albumin level).1
Statistical Analysis
The results are expressed as mean ⫾ SD. The differences
between the means of continuous variables were analyzed
with the unpaired t-test (two-tailed) when two subgroups
were compared and with the one-way analysis of variance
(ANOVA) when the data from more than two subgroups
were compared. Newman-Keuls multiple range test was
used with the ANOVA to ascertain the significance of the
difference between the mean of two groups. The differences
between the proportions of discrete variables were analyzed
with the chi-squared test.
Multiple logistic regression analysis (forward condi-
tional) was used to determine the variables that fitted the best
equation for the presence of a serum bicarbonate level of 19
mmol/L or greater (0) or less than 19 mmol/L (1; dependent
variable). The independent variables included in the model
were: age; sex; hematocrit; Ccr/1.73 m2; Ccr-Cu; serum
calcium, phosphorus, and albumin levels; PNA; diagnosis of
diabetes mellitus (1) versus no diabetes (0), diagnosis of
type I diabetes mellitus versus type II diabetes or no diabe-
tes; diagnosis of diabetic nephropathy versus other, glomer-
ulopathy versus other, vascular disease versus other, intersti-
tial disease versus other; ACE inhibitor, diuretic, or calcium
channel blocker therapy; and a history of vomiting or diar-
rhea within the previous 2 weeks.
P less than 0.05 was considered statistically significant.
RESULTS
The biochemical data for the studied group are
listed in Table 1. Mean serum bicarbonate levels
were significantly greater in patients with diabe-
tes than in the rest of the patients (21.09 ⫾ 2.61 v
18.58 ⫾ 2.88 mmol/L; P ⬍ 0.0001) and the
mean anion gap (mmol/L) was significantly less
in patients with diabetes than in the rest (19.69 ⫾
3.56 v 22.31 ⫾ 3.56; P ⫽ 0.002). However, eight
patients had hypochloremia (five patients with-
out and three patients with diabetes). Seven of the
894 CARAVACA ET AL

Table 1. Biochemical Characteristics
for the 113 Patients Included on the Study
Serum urea (mg/dL)
Serum creatinine (mg/dL)
Uric acid (mg/dL)
Creatinine clearance (mL/min/1.73 m2) 13.00 ⫾ 4.44
Ccr-Cu (mL/min/1.73 m2)
Hematocrit (%)
Chloride (mmol/L)
Bicarbonate (mmol/L)
Anion gap (mmol/L)
Calcium (mg/dL)
Phosphorus (mg/dL)
Magnesium (mg/dL)
Alkaline phosphatase (IU/mL)
Albumin (g/dL)
eight patients were receiving diuretics and had a
recent vomiting history, suggesting a superim-
posed metabolic alkalosis. For this reason, these
eight patients were excluded from the study.
The difference between mean serum bicarbon-
ate levels in the remaining 45 patients with
diabetes and 60 patients without diabetes was
also statistically significant (20.79 ⫾ 2.38 v
18.20 ⫾ 2.61 mmol/L; P ⬍ 0.0001), as was the
mean anion gap (19.70 ⫾ 3.65 v 22.35 ⫾ 3.64;
P ⫽ 0.003).
The type of metabolic acidosis was classified
according to bicarbonate, chloride, and anion
gap serum values. Table 2 lists the number and
percentage of patients with each type of meta-
bolic acidosis and the differences between pa-
tients with diabetes (type I and type II) and
without diabetes. Pure elevated anion gap fol-
lowed by mixed (elevated anion gap and hyper-
chloremia) were the most common types of meta-
187 ⫾ 54
bolic acidosis observed in both groups. Nine of
5.75 ⫾ 1.81 the 11 patients with normal serum bicarbonate
7.24 ⫾ 2.01 levels had diabetes (seven patients with type I
diabetes). Conversely, patients without diabetes
9.48 ⫾ 3.12
had a greater prevalence of mixed metabolic
30.2 ⫾ 5.6
106 ⫾ 4 acidosis than patients with diabetes.
19.6 ⫾ 3.0 Patients were classified into three subgroups
21.2 ⫾ 2.7 according to whether they had no diabetes, diabe-
9.75 ⫾ 0.73 tes with another nephropathy, or diabetic ne-
5.38 ⫾ 1.29
phropathy. Figure 1 shows the box plot of mean
2.14 ⫾ 0.43
233 ⫾ 112 serum bicarbonate levels in each subgroup and
3.57 ⫾ 0.40 the statistical differences among them. Patients
diagnosed with diabetic nephropathy had greater
serum bicarbonate levels, and a greater propor-
tion of patients had normal serum bicarbonate
levels (26% of patients with diabetic nephropa-
thy v 7% of patients with diabetes with another
nephropathy v 3% of patients without diabetes).
Both diabetic subgroups had significantly greater
serum bicarbonate levels than the subgroup with-
out diabetes. Mean serum bicarbonate levels,
however, showed no statistically significant dif-
ference between both subgroups with diabetes.
Patients with diabetic nephropathy had a signifi-
cantly lower mean plasma albumin concentra-
tion than patients with diabetes with another
nephropathy or patients without diabetes (3.32 ⫾
0.41 v 3.64 ⫾ 0.35 v 3.68 ⫾ 0.35 g/dL); how-
ever, there were no differences in mean serum

Table 2. Type of Metabolic Acidosis According

to Whether Patients Had Diabetes

Type of Metabolic Acidosis

No Pure Pure
Abnor- Anion Hyper-
malities Gap chloremic Mixed

Nondiabetic 2 (3%) 30 (48%) 3 (5%) 25 (42%)

Diabetic
Total 9 (20%) 22 (49%) 5 (11%) 9 (20%)
Type I 7 10 3 5
Type II 2 12 2 4 Fig 1. Serum bicarbonate levels in patients without
diabetes, patients with diabetes with another nephropa-
Total 11 (10%) 52 (50%) 8 (8%) 34 (32%)
thy, or patients with diabetic nephropathy. One-way
ANOVA among the three subgroups: F, 15.84; P F
NOTE. Patients with hypochloremia are excluded. Cutoff 0.0001. *P F 0.0001, diabetic nephropathy v nondia-
values considered for the definition of metabolic acidosis betic. **P F 0.001, diabetic with another nephropathy v
types: bicarbonate ⬍ or ⱖ23 mmol/L; chloride ⱖ or ⬍107 nondiabetic. There is no statistically significant differ-
mmol/L; anion gap ⱖ or ⬍18 mmol/L. ence between both diabetic subgroups.
METABOLIC ACIDOSIS IN ADVANCED RENAL FAILURE
potassium concentrations (5.37 ⫾ 0.56 v 5.15 ⫾
0.68 v 5.25 ⫾ 0.58 mmol/L), alkaline phospha-
tase levels (211 ⫾ 85 v 231 ⫾ 125 v 243 ⫾ 120
mIU/mL), or PNA (0.94 ⫾ 0.33 v 0.87 ⫾ 0.17 v
0.83 ⫾ 0.24 g/kg/24 h) among the three groups.
Twenty-eight of the 31 patients with diabetic
nephropathy and 7 of the 14 patients with diabe-
tes with another nephropathy were on insulin
therapy. There were no differences in fasting
serum glucose (134 ⫾ 53 v 139 ⫾ 29 mg/dL) or
hemoglobin A1c (6.8% ⫾ 1.4% v 6.5% ⫾ 0.9%)
levels between patients with diabetes with dia-
betic nephropathy or with another nephropathy.
Twenty-five patients had been diagnosed with
type I diabetes and 20 patients with type II
diabetes. Patients with type I diabetes were youn-
ger than patients with type II diabetes (58 ⫾ 15 v
69 ⫾ 9 years; P ⫽ 0.009), had a slightly better
mean renal function (Ccr-Cu, 11.1 ⫾ 3.4 v 9.1 ⫾
2.4 mL/min; P ⫽ 0.03), greater mean serum
bicarbonate level (21.46 ⫾ 2.00 v 19.95 ⫾ 2.60
mmol/L; P ⫽ 0.03), and lower mean anion gap
(18.62 ⫾ 2.67 v 21.04 ⫾ 4.28; P ⫽ 0.02). There
were no significant differences in plasma albu-
min, potassium, or alkaline phosphatase levels;
PNA; or whether patients received diuretics,
ACE inhibitors, or calcium channel blockers.
Searching for confounding variables that could
better explain the difference observed in serum
bicarbonate levels between patients with and
without diabetes, no significant differences in
age (63 ⫾ 14 v 63 ⫾ 15 years), hematocrit
(30.1% ⫾ 6% v 29.9% ⫾ 5%), diuretic treatment
(27 of 45 v 41 of 60 patients), PNA (0.92 ⫾ 0.29
v 0.83 ⫾ 0.24 g/kg/d), or any recent history of
vomiting (2 of 45 v 1 of 60 patients) or diarrhea
(1 of 45 v 2 of 65 patients) were observed.
However, patients with diabetes were more fre-
quently treated with ACE inhibitors (30 of 45 v
21 of 60 patients; P ⫽ 0.0013), and they had a
slightly better mean renal function (Ccr/1.73 m2,
13.96 ⫾ 4.54 v 12.34 ⫾ 4.45 mL/min; P ⫽
0.070; Ccr-Cu, 10.21 ⫾ 3.18 v 8.97 ⫾ 3.12
mL/min; P ⫽ 0.049).
Patients with diabetes receiving ACE inhibi-
tors (30 of 45 patients) had mean serum bicarbon-
ate levels similar to those of patients with diabe-
tes not receiving ACE inhibitors (20.7 ⫾ 2.2 v
20.8 ⫾ 2.6 mmol/L). Those patients with diabe-
tes who received ACE inhibitors had a not statis-
tically significant better mean renal function (Ccr-
895
Cu, 10.7 ⫾ 3.3 v 9.1 ⫾ 2.6 mL/min; P ⫽ 0.09)
and a greater mean serum potassium concentra-
tion (5.41 ⫾ 0.57 v 5.10 ⫾ 0.61 mmol/L; P ⫽
0.10).
As expected, both Ccr and Ccr-Cu correlated
directly with serum bicarbonate levels (r ⫽ 0.30;
P ⫽ 0.0018). Nevertheless, at any level of glo-
merular filtration rate, patients with diabetes had
greater serum bicarbonate levels than the rest of
the patients (Fig 2).
By multiple logistic regression analysis, the
variables that entered in the best equation for the
presence of a serum bicarbonate level of 19
mmol/L or greater (0) or less than 19 mmol/L (1)
were: Ccr-Cu, age, and diagnosis of diabetic
nephropathy (Table 3).
DISCUSSION
The results of the present study can be summa-
rized as: (1) metabolic acidosis was a very com-
mon feature in these patients with advanced
renal failure, with a severity related to the sever-
ity of renal insufficiency; (2) patients with diabe-
tes, and mainly those diagnosed with diabetic
nephropathy or type I diabetes, had a lower
prevalence or a less severe degree of metabolic
acidosis than the rest of the patients; and (3) this
Fig 2. Correlation between the half of the sum of
creatinine and urea clearances with serum bicarbon-
ate levels. Regression equation for the whole group
(line not shown): y ⴝ 16.7 ⴙ 0.27 ⴛ (r ⴝ 0.30; P ⴝ
0.0015). Regression equation for the diabetic sub-
group (●; grey line): y ⴝ 19.1 ⴙ 0.16 ⴛ (r ⴝ 0.22; P ⴝ
0.14). Regression equation for the nondiabetic sub-
group (䊐, solid line): y ⴝ 16.2 ⴙ 0.22 ⴛ (r ⴝ 0.27; P ⴝ
0.036).
896
Table 3. Multivariate Logistic Regression Analysis
for the Presence of a Serum Bicarbonate Level
Less Than 19 mmol/L
Log Odds 95% CI
Variable Likelihood Ratio Odds Ratio P common in patients with diabetes than in those
Ccr-Cu ⫺59.23 0.824 0.962-0.705 0.010
Age ⫺58.20 0.966 0.999–0.934 0.034
Diabetic
nephropathy ⫺65.00 0.107 0.347–0.033 0.0001
Abbreviation: CI, confidence interval.
finding cannot be explained by the use of drugs,
reduced protein catabolic rate, or overt gastroin-
testinal hydrogen ion losses.
These results are consistent with those ob-
served by Wallia et al,12 who showed that among
those few patients who did not have reduced
serum bicarbonate levels despite end-stage renal
failure, diabetic nephropathy was the most com-
mon cause of renal failure.
A reduction in protein catabolic rate or gastro-
intestinal hydrogen ion losses could not explain
the greater bicarbonate levels factored for renal
function in patients with diabetes compared with
the rest of the uremic patients. Therefore, the
major question that arises from this observation
is: how are patients with diabetes able to gener-
ate more bicarbonate or neutralize more acid
than the rest of the patients with end-stage renal
failure with a similar degree of renal dysfunction?
When glomerular filtration rate decreases to
less than a critical point, total ammonium excre-
tion diminishes despite the adaptive increase in
ammonia production per nephron.1-4,13 This pro-
gressive loss of ammoniagenic capacity leads to
the failure of net acid excretion and a positive
balance of hydrogen ions, which deplete extracel-
lular fluid bicarbonate. However, a typical char-
acteristic of uremic metabolic acidosis is that
bicarbonate concentrations are maintained within
relatively stable ranges (approximately 12 to 18
mmol/L) despite progressive accumulation of
acid.1-4 This finding suggests that extrarenal non-
bicarbonate buffers must exist to neutralize en-
dogenous and exogenous acids escaping urinary
excretion. Bone and skeletal muscle, two of the
main targets of the untoward effects of chronic
metabolic acidosis, act as extrarenal buffers.7,8
Nevertheless, it seems unlikely that these alterna-
tive buffers are able to maintain normal or near-
normal serum bicarbonate concentrations.14
CARAVACA ET AL
Patients with diabetes are more prone to de-
velop ketosis than other uremic patients. In addi-
tion, hyporeninemic hypoaldosteronism with its
associated type IV renal tubular acidosis is more
without diabetes.15 Thus, a more severe degree of
acidosis would be expected in patients with dia-
betes. Notwithstanding, the results of the present
study contradict this assumption.
Based on the observation that obese subjects
with fasting ketoacidosis excrete less ketoacid
anions and decrease net acid production when
challenged by an inorganic acid load, Hood et
al16 hypothesized that systemic acid-base status
may exert feedback control over hydrogen ion
production by inhibiting ketoacid anion produc-
tion. This hypothesis was based on two assump-
tions: that urinary ketoacid excretion reflected its
endogenous production, and on the conventional
definition of net acid balance (the sum of urinary
NH4⫹ and the titratable acidity minus HCO3–
loss). However, ketoacid anions are potential
bicarbonates, on account of their conversion to
bicarbonate when oxidized.17 Using a definition
of net acid excretion in which the component of
bicarbonate loss was expanded to include poten-
tial bicarbonate (ketoacid anions) in the urine,
Kamel et al18 showed that the rate of net acid
excretion was greater in fasting ketoacidotic sub-
jects who ingested an inorganic acid load be-
cause of a much lower rate of excretion of
ketoacid anions. As renal fractional reabsorption
of ketoacid anions increased and their serum
levels were unchanged, they concluded that the
conversion of ketoacid anions to bicarbonate
may be one of the main mechanisms maintaining
acid-base balance in this condition.
Theoretically, advanced renal dysfunction may
further favor ketoacid anion retention. This poten-
tial source of extrarenal bicarbonate may be a
plausible explanation for the less severe degree
of metabolic acidosis observed in patients with
diabetes.
Other organic anions processed by the gut
could potentially result in the addition of bicar-
bonate to body fluids. These organic anions (ace-
tic, propionic, and butyric), generated from bac-
terial anaerobic metabolism of neutral foodstuffs
remaining in the intestinal lumen, are almost
completely reabsorbed from the intestinal lu-
men.19 Although patients with diabetes fre-
METABOLIC ACIDOSIS IN ADVANCED RENAL FAILURE
quently have intestinal problems caused by de-
creased intestinal motility with secondary
intraluminal bacterial overgrowth (diabetic enter-
opathy), it is not known whether these intestinal
abnormalities could increase the fermentative
processes that eventually may lead to a larger
generation of organic anions.
Mixed metabolic acidosis was less often ob-
served in patients with diabetes than in the rest of
the patients. Patients with diabetes included on
the study did not have hypochloremia or extrare-
nal chloride losses. In keeping with the hypoth-
esis previously suggested, generation of bicarbon-
ate from ketoacid anion oxidation would force
the urinary excretion of chloride with ammo-
nium,19 therefore helping to maintain chloride
levels within the normal range.
The development of secondary hyperparathy-
roidism has been shown to be less severe in
patients with diabetes than in the rest of the
uremic patients.20,21 As previously mentioned,
sustained metabolic acidosis demands the use of
extrarenal nonbicarbonate buffers, with bone as
one of the major sources. Metabolic acidosis
may aggravate secondary hyperparathyroidism
by several mechanisms.7,22,23 Thus, it may be
possible that a less severe degree of metabolic
acidosis during the course of renal insufficiency
may contribute to less severe forms of secondary
hyperparathyroidism in aging patients and those
with diabetes. This possibility should be taken
into consideration in future studies of the patho-
genesis of the increasingly diagnosed adynamic
form of bone disease in the end-stage renal
failure population.
In conclusion, patients with diabetes with ad-
vanced renal failure show a less severe degree of
metabolic acidosis that cannot be explained by
gastrointestinal hydrogen ion losses, use of drugs,
or reduced protein catabolic rate. This finding
suggests that an efficient extrarenal generation of
buffers exists in patients with diabetes. Further
studies of ketoacid anions as a potential source of
extrarenal bicarbonate generation in patients with
diabetes with advanced renal failure should be
undertaken.
REFERENCES
1. Rose BD: Clinical Physiology of Acid-Base and Elec-
trolyte Disorders (ed 4). New York, NY, McGraw-Hill,
1994, pp 562-565
897
2. Warnock DG: Uremic acidosis. Kidney Int 34:278-
287, 1988
3. Widmer B, Gerhardt RE, Harrington JT, Cohen JJ:
Serum electrolyte and acid-base composition. The influence
of graded degrees of chronic renal failure. Arch Intern Med
139:1099-1102, 1979
4. Hakim RM, Lazarus JM: Biochemical parameters in
chronic renal failure. Am J Kidney Dis 11:238-247, 1988
5. Sabatini S, Kurtzman NA: Enzyme activity in obstruc-
tive uropathy: Basis for salt wastage and the acidification
defect. Kidney Int 37:79-84, 1990
6. Chan JC: Acid-base disorders and the kidney. Adv
Pediatr 30:401-471, 1983
7. Bushinsky DA: The contribution of acidosis to renal
osteodystrophy. Kidney Int 47:1816-1832, 1995
8. Del Canale S, Fiaccadori E, Coffrini E, Vitali P, Ronda
N, Antonucci C, Arduini U, Guariglia A: Uremic acidosis
and intracellular buffering. Scand J Urol Nephrol 20:301-
306, 1986
9. Cottini EP, Gallina DL, Dominguez JM: Urea excre-
tion in adult humans with varying degrees of kidney malfunc-
tion fed milk, eggs, or an amino acid mixture: Assessment of
nitrogen balance. J Nutr 103:11-19, 1973
10. Maroni BJ, Steinman TI, Mitch WE: A method for
estimating nitrogen intake of patients with chronic renal
failure. Kidney Int 27:58-65, 1985
11. Bergström J, Fürst P, Alvestrand A, Lindholm B:
Protein and energy intake, nitrogen balance and nitrogen
losses in patients treated with ambulatory peritoneal dialy-
sis. Kidney Int 44:1048-1057, 1993
12. Wallia R, Greenberg A, Piraino B, Mitro R, Puschett
JB: Serum electrolyte patterns in end-stage renal disease.
Am J Kidney Dis 8:98-104, 1986
13. MacClean AJ, Hayslett JP: Adaptive change in ammo-
nia excretion in renal insufficiency. Kidney Int 17:595-606,
1980
14. Oh MS: Irrelevance of bone buffering to acid-base
homeostasis in chronic metabolic acidosis. Nephron 59:7-
10, 1991
15. Grande Villoria J, Macias Nunez JF, Miralles JM, De
Castro del Pozo S, Tabernero Romo JM: Hyporeninemic
hypoaldosteronism in diabetic patients with chronic renal
failure. Am J Nephrol 8:127-137, 1988
16. Hood VL, Danforth E Jr, Horton ES, Tannen RL:
Impact of hydrogen ion on fasting ketogenesis: Feedback
regulation of acid production. Am J Physiol 242:F238-F245,
1982
17. Brown JC, Packer RK, Knepper MA: Role of organic
anions in renal response to dietary acid and base loads. Am J
Physiol 257: F170-F176, 1989
18. Kamel KS, Ethier JH, Stinebaugh BJ, Schloeder FX,
Halperin ML: Removal of an inorganic acid load in subjects
with ketoacidosis of chronic fasting. Kidney Int 38:507-511,
1990
19. Cohen RM, Feldman GM, Fernandez PC: The bal-
ance of acid, base, and charge in health and disease. Kidney
Int 52:287-293, 1997
20. Pei Y, Hercz G, Greenwood C, Segre GV, Manuel A,
Saiphoo C, Fenton S, Sherrad DJ: Renal osteodystrophy in
diabetic patients. Kidney Int 44:159-164, 1993
21. Sugimoto T, Ritter C, Morrisey J, Hayes C, Slatopol-
898
sky EA: Effects of high concentration of glucose on PTH
secretion in parathyroid cells. Kidney Int 37:1522-1527,
1990
22. Coen G, Manni M, Addari O, Ballanti P, Pasquali M,
Chicca S, Mazzaferro S, Napoletano I, Sardella D, Bonucci
E: Metabolic acidosis and osteodystrophic bone disease in
CARAVACA ET AL
predialysis chronic renal failure: Effect of calcitriol treat-
ment. Miner Electrolyte Metab 21:375-382, 1995
23. Kleeman CR: The role of chronic anion gap and/or
nonanion gap acidosis in the osteodystrophy of chronic renal
failure in the predialysis era: A minority report. Miner
Electrolyte Metab 20:81-96, 1994