Professional Documents
Culture Documents
the peripheral blood. Because this clinical association has ferritin, total iron-binding capacity, hemoglobin elec-
not been reported before, the mechanism involved in this trophoresis, hemoglobin A2 quantification, haptoglobin,
phenomenon is unknown. In light of these observations, G6PD, folate, vitamin B12, thyroid function tests, and ser-
we were interested in evaluating the incidence of anemia ial complete blood cell counts. On the basis of these stud-
in young, otherwise healthy persons with newly diagnosed ies, 45 of the 83 patients were found to have a readily iden-
diabetes and in ascertaining the mechanism of its tifiable cause of their anemia, including 13 patients with
development. iron deficiency anemia, 13 with β-thalassemia minor, 7
with G6PD deficiency, 6 with hemoglobin AS, 2 with
PATIENTS AND METHODS hemoglobin AC, and 2 with hemoglobin SC. One patient
had both hemoglobin AS and G6PD deficiency. One
The data for this report were compiled from a review patient from Thailand had hemoglobin EE, which is com-
of medical records of patients with new-onset diabetes and mon in Asia. In addition, 10 patients with mild anemia did
unexplained anemia admitted during the 10-year period not undergo hematologic evaluation at diagnosis and thus
from 1989 to 1999 to the University Diabetes Treatment were not included in the statistical analysis. Two of these
Center at Parkland Memorial Hospital in Dallas, Texas. patients were diagnosed years later with prostate and
The patients were identified by the presence of anemia, as endometrial cancer, respectively.
defined by a hemoglobin concentration of less than 12.5 Thus, our study population included 28 patients with
g/dL in women and less than 14 g/dL in men, at the time new-onset diabetes who had an unexplained nor-
of admission to the hospital before intravenous adminis- mochromic normocytic anemia at the time of diagnosis.
tration of any fluids. Patients with a prior history of dia- Of these 28 patients, 12 had no follow-up laboratory stud-
betes or an obvious cause of anemia, such as renal insuffi- ies and could not be included in the statistical analysis.
ciency (serum creatinine level in excess of 1.5 mg/dL), The other 16 patients had an initial workup and had a fol-
infection, pancreatitis, chronic hepatitis, alcohol abuse, low-up evaluation including blood cell counts and glycat-
hemolysis, or acute or chronic blood loss, were excluded ed hemoglobin (HbA1c). Therefore, our report consists of
from this study. Patients who received intravenously data from 16 otherwise seemingly healthy patients with
administered fluids before any laboratory measurements new-onset diabetes who had a normochromic normocytic
were excluded from analysis to avoid confusion with dilu- anemia at the time of diagnosis. The 16 patients (8 men
tion anemia. Hematologic studies and workup for the ane- and 8 women) had a mean age of 33 ± 10 years. They had
mia were done at the time of diagnosis. Although the com- no other acute or chronic systemic illness or history of
ponents of the evaluation were variable, all study patients cancer to explain their anemia. Eight patients were African
underwent sufficient assessment to rule out the presence American, six were Hispanic, one was Asian Indian, and
of hemoglobinopathy, blood loss, acute hemolysis, or iron one was non-Hispanic white. At the time of initial diagno-
deficiency. The calculations of the corrected reticulocyte sis, eight patients with type 1 diabetes had diabetic
count were done by using the reticulocyte production ketoacidosis, and eight patients had hyperglycemia with-
index. This index corrects the reticulocyte count on the out ketoacidosis. The clinical characteristics at the time of
basis of the patient’s hematocrit and also the maturation diagnosis are summarized in Table 1.
index, which accounts for the early release of immature At the time of diagnosis of diabetes in the 16 study
reticulocytes (“shift reticulocytes”) from the bone marrow patients, the mean HbA1c was 15.5 ± 3.4%, the mean
depending on the degree of anemia. A reticulocyte pro- hemoglobin level was 12.5 ± 0.6 g/dL, and the mean
duction index of less than 2 was considered reflective of a hematocrit was 36.2 ± 2% (mean corpuscular volume,
hypoproliferative marrow. None of the patients had ane- 86.4 ± 4 fL; range, 81 to 91.2). The corrected mean retic-
mia that was sufficiently severe to warrant bone marrow ulocyte production index of all patients at diagnosis was
aspiration. For statistical analysis of all variables, 0.89 ± 0.3 (median, 0.86; range, 0.38 to 1.47). No statisti-
Student’s t test for nominal values was used (Glantz’s cally significant difference was found between men (0.92
Statistical Software, version 3.02, McGraw-Hill, 1992). ± 0.32) and women (0.85 ± 0.3) (difference of 0.07; P =
The values are reported as the mean ± standard deviation; 0.66). All patients were treated with insulin in the hospital
most results were not normally distributed. and after dismissal. At dismissal from the hospital, the
mean hemoglobin concentration was 12.4 ± 1.4 g/dL, and
RESULTS the mean hematocrit was 36.1 ± 3.9%.
After a mean follow-up of 10.8 ± 17 months (median,
We reviewed the medical records of more than 1,500 6; range, 2 to 48), insulin treatment resulted in a decline in
patients who were admitted to the University Diabetes HbA1c to 7.7 ± 1.7% (P<0.001; confidence interval [CI],
Treatment Center at Parkland Memorial Hospital in 5.7 to 9.8). The mean hemoglobin level increased to 14.3
Dallas, Texas, between 1989 and 1999 for management of ± 0.9 g/dL (P<0.001 versus baseline; CI, 1.22 to 2.38), and
new-onset diabetes. Of these patients, 83 were found to the mean hematocrit increased to 42.1 ± 1.9% (P<0.001
fulfill the initial criteria for anemia. Thus, the incidence of versus baseline; CI, 3.59 to 7.04) (Fig. 1). Men and
anemia in patients with new-onset diabetes was 5.5%. The women had equal improvement in hematologic variables
workup of patients with anemia usually included a reticu- after improvement in glycemic control. In women, the
locyte count, corrected reticulocyte count, serum iron, mean hemoglobin concentration was 12.1 ± 0.37 g/dL at
278 Diabetes and Anemia, Endocr Pract. 2002;8(No. 4)
Table 1
Clinical Characteristics of Patients With
Anemia Associated With New-Onset Diabetes*†
Age (yr) … 33 ± 10
Weight (kg) … 78.7 ± 29
Glucose (mg/dL) 70-110 498 ± 98
HbA1c (%) 3.8-5.6 15.5 ± 3.4
RBC (×106/µL) 3.8-5.5 4.29 ± 0.33
Hemoglobin (g/dL) 12.5-14 12.5 ± 0.6
Hematocrit (%) 37-46 36.2 ± 2
MCV (fL) 80-101 86.4 ± 4
MCHC (g/dL) 31-35 34.5 ± 0.77
RDW (%) 11.5-14 12.6 ± 0.95
Iron (µg/dL) 48-180 71.6 ± 25
Ferritin (ng/mL) 7-282 248 ± 150
TIBC (µg/dL) 262-474 287 ± 61
Reticulocyte count (%) 0.8-2.5 1.61 ± 0.57
Hemoglobin A2 (%) 0.0-3.5 2.7 ± 0.5
Creatinine (mg/dL) 0.6-1.2 0.8 ± 0.3
MCH (pg) 27-33 29.5 ± 1.74
Fig. 1. Effects of insulin treatment on hemoglobin concentration Fig. 2. Correlation between changes in glycated hemoglobin
and glycated hemoglobin (HbA1c) values in patients with (HbA1c) values and hemoglobin concentration at follow-up of
anemia associated with new-onset diabetes. patients with new-onset diabetes and unexplained anemia.
chronic illness to explain their anemia. None had evidence levels of oxidized glutathione (12,13), a shorter half-life
of hemolysis. After workup of the anemia, the findings (14), a decreased NADPH/NADP ratio (15), and an
were consistent with so-called anemia of inflammation increased cellular fragility (16). Oxidative stress has also
(formerly known as anemia of chronic disease) (9,10), been shown to cause mitochondrial DNA damage to skele-
even though these patients were otherwise apparently tal myocytes in patients with diabetes (17). This DNA
healthy. The most interesting aspect of this anemia seems damage causes apoptosis involving various cell lines
to be the notable improvement in the hemoglobin and including neurons (18) and perhaps bone marrow precur-
hematocrit levels once glycemic control is improved. This sors, leading to decreased production of erythrocytes or
result suggests that hyperglycemia has a key role in the decreased synthesis of hemoglobin.
pathogenesis of this phenomenon. This association has not The NADPH in erythrocytes originates from the reac-
been described before, and there is limited literature avail- tions derived from the hexose-monophosphate shunt
able to explain this phenomenon. We propose that hyper- (HMPS). Hyperglycemia will theoretically increase the
glycemia has an important pathophysiologic role in the flux of glucose through this pathway and should protect
development of this anemia. erythrocytes from oxidative stress. Why then did hyper-
Possible explanations for our findings are the key role glycemia not protect erythrocytes or its precursors from
that hyperglycemia and oxidative stress have in the oxidative damage in these patients? Why do not all
destruction of mature erythrocytes or the decrease in pro- patients with diabetes have anemia?
duction of erythrocytes by bone marrow precursors (or The administration of glucose has been shown to
both). Hyperglycemia causes oxidative stress by different increase markers of oxidative stress in erythrocyte mem-
mechanisms, including the increased flux of glucose branes in patients with diabetes (19) and those without dia-
through the polyol pathway (Fig. 3). The polyol pathway betes (20). High glucose levels have also been shown to
is believed to be the major contributor of oxidative stress inhibit G6PD in vitro (21) and in experimental diabetes
in patients with diabetes (11). Increased flux of glucose (22). The inhibition of production of G6PD will reduce the
through the polyol pathway causes greater consumption of flux of glucose through the HMPS and can result in
the reduced form of nicotinamide adenine dinucleotide decreased production of NADPH. Therefore, it is possible
phosphate (NADPH) due to increased production of sor- that glucose is being used preferentially in the polyol and
bitol. This augmented utilization of NADPH can cause a glycolytic pathway rather than through the HMPS. The
deficiency of this cofactor for other enzymes that use preferential shunting of glucose through the polyol
NADPH. One important enzyme that uses NADPH as a pathway will make erythrocytes prone to oxidative
cofactor is glutathione reductase. Glutathione reductase is damage and fragility because of peroxidation of cell mem-
responsible for the reduction of oxidized glutathione, brane lipids. Another theoretical mechanism to explain
which is one of the major antioxidants in cells, including this phenomenon is that this group of patients may have a
erythrocytes. Reduced glutathione is a scavenger for free gene polymorphism that causes some of the enzymes in
oxygen radicals and peroxides that can damage cell mem- the HMPS or G6PD to be less effective in producing
brane lipids. The decreased production of glutathione NADPH.
causes an increase in free oxygen radicals that damage cell Therefore, we believe that severe persistent hyper-
membrane lipids in erythrocytes, an action that causes glycemia is sufficient to cause abnormalities in the
erythrocytes to be fragile and prone to lysis. Erythrocytes production of mature erythrocytes and the fragility of
of patients with diabetes have been shown to have high mature erythrocytes. Once the metabolic derangements
280 Diabetes and Anemia, Endocr Pract. 2002;8(No. 4)
and severe hyperglycemia have been corrected, the pro- cance because of the small sample size. Finally, other
duction of mature and new erythrocytes improves with an factors that could influence erythrocyte production were
increase in hemoglobin and hematocrit. Peterson et al (23) not directly measured. Because most of the patients had
reported improvements in erythrocyte half-life in patients normal reticulocyte counts, peripheral smears, and serum
with a long history of diabetes with achievement of inten- creatinine levels, erythropoietin levels, haptoglobin levels,
sive glycemic control. There was no mention of improve- and bone marrow aspiration were not done. Perhaps these
ments in hemoglobin concentration with improved studies should be done in future investigations in an effort
glycemic control. to understand this phenomenon better.
Another potential explanation is that all these patients
had α-thalassemia with only one α-chain locus affected CONCLUSION
and that hyperglycemia somehow affected the synthesis of
the other α-chains. This possibility seems unlikely New-onset diabetes can be associated with a mild nor-
because all the patients had improvements in their hemo- mochromic normocytic anemia. The anemia is not
globin concentration after insulin treatment, although most explainable by usual causes, such as infection, pancreati-
patients were not euglycemic at follow-up. Published tis, or blood loss, and the hemoglobin concentration nor-
reports have described resistance to the action of erythro- malizes with improvement in glycemic control. The cause
poietin in erythrocyte precursors in the bone marrow of anemia may be direct “glucose toxicity” to erythrocyte
(24,25) or low production of erythropoietin (26) in precursors in the bone marrow or to mature erythrocytes,
patients with long-standing diabetes. There are no such perhaps because of oxidative stress. Further studies are
reports in patients with new-onset diabetes. Because these needed to evaluate these findings.
patients had normal renal function, plasma erythropoietin
was not measured. Perhaps acute hyperglycemia as is seen ACKNOWLEDGMENT
in new-onset diabetes can inhibit the production of eryth-
ropoietin; such a process would explain the low corrected We thank the staff of the University Diabetes
reticulocyte index. Treatment Center and the Medical Records Department at
Certain limitations that are common to retrospective Parkland Memorial Hospital for helping with data collec-
studies warrant mention. In this retrospective study, con- tion. We also thank Beverly Adams-Huet, the General
siderable variation existed from the time of initial sam- Clinical Research Unit biostatistician, for assistance with
pling to the time of follow-up. Moreover, 12 patients with statistical analyses and Dan Chan for collection of data
unexplained anemia were untraceable at follow-up, and and review of medical records. Drs. Piñero-Piloña and
their inclusion might have altered the statistical analysis. Litonjua were supported in part by NIDDK Training Grant
Even though we observed a tendency toward a correlation T-32-DK070307-22. Beverly Adams-Huet was supported
between glycemic control and improvement in hemoglo- in part by the General Clinical Research Unit Grant
bin level, this correlation did not reach statistical signifi- M01RR00633.
Diabetes and Anemia, Endocr Pract. 2002;8(No. 4) 281