Glycemic Control and Burnt-Out Diabetes in ESRD
Csaba P. Kovesdy,*† Jong C. Park,‡ and Kamyar Kalantar-Zadeh‡§
*Division of Nephrology, Salem Veterans Affairs Medical Center, Salem, Virginia, †Division of Nephrology,
University of Virginia, Charlottesville, Virginia, ‡Division of Nephrology and Hypertension, Harold Simmons
Center for Chronic Disease Research and Epidemiology, Los Angeles Biomedical Research Institute at
Harbor-UCLA Medical Center, Torrance, California, and §David Geffen School of Medicine at UCLA, Los
Angeles, California
ABSTRACT
Treatment of early diabetes mellitus, the most common cause
of chronic kidney disease (CKD), may prevent or slow the
progression of diabetic nephropathy and lower mortality and
the incidence of cardiovascular disease in the general diabetic
population and in patients with early stages of CKD. It is
unclear whether glycemic control in patients with advanced
CKD, including those with end-stage renal disease (ESRD)
who undergo maintenance dialysis treatment is beneficial.
Aside from the uncertain benefits of treatment in ESRD, hypo-
glycemic interventions in this population are also complicated
by the complex changes in glucose homeostasis related to
decreased kidney function and to dialytic therapies, occasion-
ally leading to spontaneous resolution of hyperglycemia and
normalization of hemoglobin A1c levels, a condition which
might be termed ‘‘burnt-out diabetes.’’ Further difficulties in
ESRD are posed by the complicated pharmacokinetics of
antidiabetic medications and the serious flaws in our available
diagnostic tools used for monitoring long-term glycemic
control.We review the physiology and pathophysiology of glu-
cose homeostasis in advanced CKD and ESRD, the available
antidiabetic medications and their specifics related to kidney
function, and the diagnostic tools used to monitor the severity
of hyperglycemia and the therapeutic effects of available treat-
ments, along with their deficiencies in ESRD. We also review
the concept of burnt-out diabetes and summarize the findings
of studies that examined outcomes related to glycemic control
in diabetic ESRD patients, and emphasize areas in need of
further research.
There are currently�400,000 Americans with chronic
kidney disease (CKD) stage 5, a.k.a., end-stage renal
disease (ESRD), who are dependent on maintenance
hemodialysis (MHD) (�92%) or continuous peritoneal
dialysis (CPD) (�8%) for their survival (1). According
to US Renal Data System (USRDS) estimates, the
number of ESRD patients will surpass one-half million
during the 2010s (2). These individuals experience a low
quality of life, high hospitalization rates and a 20–25%
annual mortality rate, in spite of recent advances in
dialysis treatment and technique (1). Approximately
two-thirds of all dialysis patients die within 5 years of
initiation of dialysis treatment, a 5-year survival worse
than that expected in themajority of patients with cancer
(3). Almost half of all CPD andMHD patients presum-
ably die of CV diseases (1).
Diabetes mellitus (DM) is the leading cause of CKD
including ESRD in the US and most industrialized
nations (1,4). The proportion of incident diabetic
dialysis patients in the US has increased from<30% in
late 1980s to 50% or higher in 2006 (1). Most studies
indicate higher comorbidity and poorer outcomes in
diabetic dialysis patients as compared with nondiabetic
subjects (5–7), but a large observational study of Medi-
care beneficiaries showed that while diabetic patients
without CKD have significantly worse survival than
their nondiabetic counterparts, among those with CKD
the mortality is almost equally high in both diabetic and
nondiabetic patients (8).
While strict glucose control in patients with normal
kidney function and early stages of DM can result in
significant clinical benefits including improved cardio-
vascular mortality (9,10) and better renal outcomes
(11–16), similar interventions in patients with more
advanced DM have produced disappointing results,
including in some cases worse outcomes in patients
who achieved strict glucose control (17–19). As a result
a more nuanced view of glycemic control may be
emerging, with individual patient characteristics deter-
mining what therapeutic targets and treatment regi-
mens to be pursued.
The issue of long-term outcomes as a function of
diabetes control is especially complicated in patients
with ESRD in whom there are complex changes affect-
ing glucose homeostasis, independent of the usual
diabetic pathophysiology. It is currently unclear whether
the altered glucose homeostasis in ESRD combined with
medical management of DM has any significant bearing
Address correspondence to: Csaba P. Kovesdy, MD, Salem
VAMC (111D), 1970 Roanoke Blvd., Salem, VA 24153,
Tel.: +1-540-982-2463, Fax: +1-540-224-1963, or e-mail:
csaba.kovesdy@va.gov.
Seminars in Dialysis—Vol 23, No 2 (March–April) 2010
pp. 148–156
DOI: 10.1111/j.1525-139X.2010.00701.x
ª 2010 Wiley Periodicals, Inc.
IMPROVING OUTCOMES FOR DIABETIC PATIENTS ON DIALYSIS
148
on clinical outcomes of diabetic dialysis patients (20).
Indeed, some diabetic ESRDpatients experience sponta-
neous improvement in their glycemic control that can
occasionally return them to normoglycemia. The occur-
rence of such ‘‘burnt-out’’ diabetes in ESRD (21) may
necessitate the dissociation of the diagnosis of DM
(which includes individuals with blood sugar levels rang-
ing from normal to elevated) from that of hyperglyce-
mia, which may occur in some diabetic patients, but
could also be a de novo ESRD-related phenomenon
(22). We review the unique features of DM in ESRD,
examine available data on the importance or lack
thereof of glycemic control in achieving better survival
in this patient population, and examine potential causes
and consequences of burnt-out diabetes. We also
briefly review ESRD-specific aspects of antidiabetic
pharmaceuticals.
Glycemic Control in ESRD: The Concept of
Burnt-Out Diabetes
While diabetes is one of the most important causes of
CKD, the gradual decline in kidney function in itself
causes significant changes that alter glucose homeostasis
in patients with kidney disease (23). Abnormal glycemic
control has been described as a complication of
advanced CKD (24) independent of underlying diabetic
status, as even nondiabetic patients with CKD can dis-
play mild fasting hyperglycemia and abnormal glucose
tolerance (22). Interestingly enough although, in some
patients with establishedDMadecline in insulin require-
ments and even spontaneous hypoglycemia can also
occur with advancing stages of CKD (22,25). A recent
2-year cohort study of 23,618 diabetic hemodialysis
patients showed that 33% of them had A1c levels <6%
(26). Although in this cohort higher A1c values were
incrementally associated with increased death risk after
controlling for demographics and other confounders,
low A1c, especially if below 5%, was also associated
with poor survival. This nationally representative study
shows that approximately one-third of all prevalent dia-
betic hemodialysis patients in the United States have a
normal to low A1c level despite the fact that they
sustained ESRD as a result of diabetic nephropathy
(26).
The reason for these alterations in glucose homeosta-
sis are multifactorial and involve various mechanisms
related to both decreased kidney function and dialytic
therapies (Table 1) (23). Once GFR declines below
15–20 ml ⁄min, the renal clearance of insulin decreases
significantly (22). Concomitantly, hepatic clearance of
insulin also declines, likely because of the inhibitory
effect of unidentified uremic toxin(s), as the initiation of
dialysis seems tomitigate this effect (22). Counterbalanc-
ing diminished insulin clearance is lower insulin produc-
tion and increased insulin resistance. The exact reason(s)
for the diminished insulin secretion is unclear; it may be
because of hyperparathyroidism and activated vitamin
D deficiency, as insulin secretion appears to improve
after the treatment of hyperparathyroidism and after
administration of activated vitamin D (27–31).
Although activated vitamin D can lower parathyroid
hormone (PTH) levels, its effect on insulin secretion is
believed to be independent of the PTH lowering effect
(31), and may be one of the explanations for the
improved outcomes associated with these medications
(32,33).
The cause of increased insulin resistance in ESRD is
also not fully clarified. It may be related to another
unspecified uremic toxin possibly acting on the muscle
tissue (34–36), as insulin sensitivity is improved with
dialysis (37–41). The consequences of insulin resistance
and deficiency in ESRD are complex and reach
beyond glucose homeostasis, as they are associated
with muscle protein breakdown through the ubiqu-
itin–proteasome pathway (42,43) via suppression of
phosphatidylinositol-3 kinase (44,45), thus contribut-
ing to a state of protein-energy wasting which may
also play a major role in the higher mortality seen in
this population. Decreased kidney function is also
associated with deficient renal gluconeogenesis (46),
which along with malnutrition, deficient catecholamine
release and impaired renal insulin degradation and
clearance can contribute to a lower threshold for
clinical hypoglycemia (25).
Aside from the salutary effects of dialytic therapies
on insulin production and sensitivity (37–40), these
treatments can complicate the management of diabe-
tes by the glucose load provided by both hemo- and
peritoneal dialysis. The latter especially can result in
significantly higher glucose loads if higher dialysate
concentrations are required to achieve ultrafiltration
goals. The impact of the often significant glucose load
delivered by peritoneal dialysis [which can be as much
as 10–30% of total energy intake (47)] is complex. In
spite of the added nutritional value of glucose the
total nutrient intake of these patients is often below
ideal (48), possibly because of a loss of appetite
related to continuous glucose absorption (49–51) and
the mechanical effects of large filling volumes (in the
case of PD) (52). Glucose absorbed during dialysis
could at the same time lower patients’ energy
TABLE 1. Possible causes of the ‘‘burnt-out diabetes’’ in
maintenance dialysis patients (adapted from ref. 21)
1. Decreased renal clearance of insulin
2. Decreased hepatic clearance of insulin
3. Impaired renal insulin degradation
4. Increased insulin half-lifea
5. Decline in renal gluconeogenesis
6. Deficient catecholamine release
7. Other impacts of uremia on glucose homeostasis
8. Diminished food intake because of anorexia,
diabetic gastroparesis, etc.
9. Protein-energy wasting (malnutrition–inflammation complex)
10. Loss of body weight and fat mass
11. Comorbid conditions
12. Hypoglycemia during hemodialysis treatments
13. Effects of peritoneal dialysis on glucose metabolism
14. Prescribed medications
15. Imposed dietary restrictions
16. Apparently low A1c because of confounding by
uremia or anemia
aBecause of conditions other than those under 1 through 3.
GLYCEMIC CONTROL IN ESRD 149
expenditure (53), limit amino acid losses and stimu-
late insulin secretion (54).
The practical significance of the listed changes in
glucose homeostasis in ESRD remains unclear. While
changes inducing worsening hyperglycemia would
result in heightened efforts to better control the blood
sugar of such patients, it is less clear what approach
has to be taken in the considerable number of dialysis
patients whose A1c decreases to levels approaching
normalcy (26). It is not entirely clear what distinguishes
such patients with normoglycemic or ‘‘burnt-out’’ dia-
betes from their counterparts who continue to display
hyperglycemia, and studying them may offer further
insights into the natural course of diabetes not only in
dialysis patients but also in other populations such as
those suffering from chronic illnesses (55) or in the
elderly, who may experience an altered course of this
disease. While in today’s benchmark-driven environ-
ment a lower A1c may be perceived as advantageous,
we believe that patients with burnt-out diabetes may
on the contrary be at higher risk for poor outcomes
including mortality. The advantages of a normal blood
sugar level likely take a very long time to become mani-
fest (56,57), and on the short run these patients may in
fact be more prone to develop clinically relevant
hypoglycemic episodes. Observational studies in dialy-
sis patients indeed suggest that patients with the lowest
A1c levels suffer significantly higher mortality rates
(26), thus emphasizing the need to relate to burnt-out
diabetes as a complication of a disease (ESRD) rather
than as a benefit of it.
Diagnostic Tests of Glycemic Control in ESRD
There have been ongoing questions about what the
most reliable marker of long-term glycemic control is in
these patients. HbA1c has long been the assay of choice
for determining long-term glycemic control among dia-
betic dialysis patients.
Glycated hemoglobin refers to a series of minor
hemoglobin components formed by the adduction of
various carbohydrate molecules to hemoglobin. The
minor fractions include HbA1a, HbA1b, and HbA1c
(58). HbA1c, the largest fraction, is the most consistent
index of the prevailing ambient concentration of circu-
lating glucose (59). Glycation rate is determined by
temperature, pH, hemoglobin concentration, glucose
concentration, and length of exposure to glucose.
There are various glycated hemoglobin assays avail-
able with normal ranges that vary between laborato-
ries so that test results may not be directly
comparable.
HbA1c measurements can be confounded in the ure-
mic milieu, by formation of carbamylated hemoglobin
(60) or by metabolic acidosis which increase the rate
of HbAlc formation (61), but the practical significance
of these have been questioned (62,63). Other factors that
could affect HbAlc levels in patients on dialysis include a
shortened erythrocyte life span, blood transfusion, and
accelerated erythropoesis because of routine use of
erythropoietin. However, the life span of erythrocyte is
close to normal in well dialyzed patients treated with
ertythropoesis stimulating agents and routine blood
transfusions are rarely needed nowadays. HbA1c was
found to underestimate glucose measurements in
diabetic patients on hemodialysis because of anemia and
use of erythropoietin compared with glycated albumin
(64–67); there have been recent suggestions to use a
correction factor based on the degree of anemia and the
erythropoietin dosage (65). Aside from all the above
limitations, HbAlc is still considered to be a reasonable
index of glycemic control even in patients with ESRD
(68–71).
The uses of other marker of long-term glycemic con-
trol that are devoid of HbAlc’s drawbacks in dialysis
patients have been proposed. Fructosamine is formed
by the nonenzymatic reaction between fructose and
ammonia or an amine and can form when the carbonyl
group of glucose reacts with an amino group of a pro-
tein. Therefore, fructosamine is a general measure of
glycated serum proteins, 90% of which is represented
by glycated albumin (72). Fructosamine level correlates
well with measurements of mean blood glucose and
HbAlc (73); it has therefore been suggested as an alter-
native method for monitoring glycemic control in dia-
betes patients. Fructosamine level reflects integrated
glycemia during a period of 2–3 weeks compared with
2–3 months for HbAlc (74). Thus, fructosamine may
be a better choice in situations when HbA1c cannot be
reliably measured. As with HbAlc, fructosamine level
may be less reliable in renal failure (69,75), and may
not be superior to HbA1c in diabetic patients with
ESRD (69,71). High urate levels often seen in patients
with CKD may interfere with the fructosamine assay
and levels may be falsely low with decreased protein
levels such as nephrotic syndrome or liver disease, and
also affected by increased protein turnover related to
dialysis (20). Use of this marker is further hampered by
its lack of availability in routine practice and the lack
of established reference levels.
Glycated albumin and the percentage of glycated
albumin have also received much attention lately as
a potentially more accurate measure of glycemic con-
trol in diabetic hemodialysis patients. Glycated albu-
min was a more accurate measure of glycemic
control in two recent studies (64,66). In both of these
studies, HbAlc underestimated blood glucose levels in
diabetic hemodialysis patients mainly because of
erythropoietin use and anemia. In addition to being
a significant marker of hyperglycemia in diabetic
patients on hemodialysis, higher glycated albumin is
also associated with diabetic complications such as
increased arterial stiffness (76), peripheral vascular
calcification (77), and increased cardiovascular mor-
bidity and shortened survival in diabetic patients with
ESRD (78,79). It is unclear, however, what therapeu-
tic target level of glycated albumin has to be used
for glycemic control, and in what stages of CKD its
application has to replace or supplement HbAlc.
Glycated albumin has limitations in diabetic patients
on peritoneal dialysis because of increased albumin
turnover, and in nonoliguric patients with significant
proteinuria.
150 Kovesdy et al.
Antidiabetic Drugs in ESRD
The complex pathophysiology of glucose and insulin
homeostasis and the imperfect diagnostic and monitor-
ing tools available in patient with ESRD make therapy
aimed at glycemic control in these patients challenging.
Therapeutic interventions are made even more compli-
cated by concomitant comorbidities such as sepsis, mal-
nutrition, liver disease, and congestive heart failure that
could exacerbate hypoglycemia (25), and also by the
pharmacokinetic alterations caused by the decrease (or
lack) of kidney function. The medications used to treat
DMcan be affected in CKDandESRD (Table 2).
Insulin Therapy
Insulin requirements usually diminish as kidney
function declines, but no uniform recipe can be
provided for dose adjustment, which has to be individ-
ualized. Patients receiving PD are in a unique situation
as they can receive insulin by injection into their dialy-
sate. One advantage of this approach is that intraperi-
toneal delivery of insulin provides a more reliable and
more physiologic delivery of insulin when compared
with the subcutaneous route. Direct insulin delivery to
the liver stimulates endogenous insulin secretion and
inhibits hepatic gluconeogenesis and ketogenesis when
compared with the subcutaneous route (80), and has
been reported to result in better insulin sensitivity (38).
Peritoneal insulin delivery can result in fewer hyper-
and hypoglycemic episodes and results in smaller diur-
nal variation in blood glucose levels (81,82). Potential
disadvantages of intraperitoneal insulin delivery
include the possibility of an increased incidence of
peritonitis (83), lowered plasma HDL levels (84), and
rare reports of subcapsular hepatic steatonecrosis (85)
and malignant omentum syndrome (86). Controlled
clinical trials would be needed to fully describe the net
impact of intraperitoneal insulin administration in PD.
Oral (and Other) Antidiabetics
The number of noninsulin type antidiabetic medica-
tions (which are no longer exclusively oral, Table 2)
has proliferated in the past decade. These medications
depend on renal excretion to various extents; hence
they may or may not require adjustments with advanc-
ing stages of CKD. Biguanides (metformin) represent
a special category in that they have to be withheld in
patients with an even modest decline of kidney func-
tion (the exact degree of which remains ill-defined),
because of the rare but potentially fatal lactic acidosis
caused by the accumulation of this drug (87). The
other hypoglycemic agents may or may not require
dose adjustments, as shown in Table 2. Unfortunately,
there is no uniformity in how kidney function is quan-
tified