Interpreting HbA1c in Presence of Deficiency Anemias

Abstract

HbA1c is used extensively for the diagnosis and management of diabetes mellitus. It constitutes 80% of glycated HbA1(Glycated haemoglobin(GHb)A), and depends upon blood glucose and RBC life span. RBC life span varies with anemia, leading to a consequent alteration in the HbA1c value irrespective of the circulating blood glucose concentration. But to the best of our knowledge no Hb cut offs have been derived for appropriate interpretation of HbA1c. The prevalence of anemia in Indian population is nearly 40% as per its definition by WHO—Hb < 12 g/dL in females and < 13 g/dL in males—with most cases attributable to nutritional deficiencies. Hence, we aimed to identify Hb cut-off for accurate interpretation of HbA_1c in presence of deficiency anemias. Partial correlation between random blood glucose (RBG) and HbA_1c was studied in 1312 subjects, 470 of whom had deficiency-related anemia]. The data was adjusted for age, sex and Hb. Partial correlation between RBG and HbA1c was highly significant (p < 0.0001) till Hb of 8.1 gm/dL. Significance reduced to p = 0.003 and p = 0.006 as the cut off of Hb reduced to 7.1 gm/dL and 5.0 gm/dL, respectively, but was not lost. Hence, caution in interpretation of HbA_1c is not required till an Hb of 5 g/dL.

Introduction

Glycation is a non-enzymatic reaction between reducing sugars and protein-bound amines [1]. All proteins, including haemoglobin, are subject to glycation, which is the process of attachment of a molecule of glucose to the amino acid chain of the protein. As per the IFCC working group, glycated Hb is defined as the haemoglobin that is irreversibly glycated at one or both N-terminal valines. This process is called the Maillard reaction. The initial   phase of Maillard reaction is reversible and results in the formation of a Schiff base product. This is followed by an essentially irreversible phase of a stable ketoamine linkage, called Amadori rearrangement. Once this phase is complete, the glucose remains attached to the Hb molecule for its lifetime [2, 3].

Circulating haemoglobin in a normal adult is comprised of 95–97% of HbA (adult type 1), 2–3% of HbA₂ (adult type 2) and 0.5–2% of HbF (fetal type). Chromatographic analysis of HbA identifies several minor fractions named HbA_1a, HbA_1b and HbA_1c, which together comprise HbA₁ and are also known as glycated hemoglobins or fast hemoglobins as they migrate earlier than HbA0 (the non-glycated Hb) in an electrical field. HbA1c constitutes 80% of the HbA1—the glycated Hb—and has been ascertained to be directly proportional to mean blood glucose levels. Hence, amongst the glycated proteins, HbA1c is preferred for the diagnosis of diabetes mellitus (DM) and for assessment of glycemic control in management of DM [3, 4].

The major complications of diabetes mellitus (DM) are often mediated by the pathological properties of glycated proteins as well as their advanced glycation endproducts(AGEs) which result from further transformation (oxidation, dehydration, polymerization) of glycated proteins [5]. The mainstay of treatment of DM is, therefore, towards achievement of normoglycemia so that excessive glycation of proteins does not occur.

Amongst the glycated proteins, glycated Hb has been shown to have the best correlation with circulating glucose concentrations, and the estimation of HbA_1C has enabled easier diagnosis and better management of diabetes mellitus. In fact, it is the gold standard for diagnosis, long-term management, risk stratification and prevention of complications [6].

The process of glycation of Hb may be altered in several situations for e.g. in the presence of anemia, hemoglobinopathies, uremia, pregnancy, end-stage renal disease, etc [7]. This study has focussed on the effect of anemia on HbA_1c, as data from a WHO report on the global prevalence of anemia has ascertained that in India this prevalence is nearly 40%, the major cause being nutritional deficiencies[8, 9] This makes interpretation of HbA1c very challenging in the Indian perspective. Currently, there is no yardstick to quantitate this alteration. Hence, in the current retrospective study, we have directed our attention to the interpretation of HbA1c in the presence of deficiency anemias.

Materials and Methods

All the patients’ samples received for estimation of random blood glucose, HbA1c and complete blood count (in sodium fluoride/EDTA containing evacuated tubes) from 15th March, 2018 to 15th April, 2018 were included in this study. Subjects with age < 18 years and with co-morbidities like renal and hepatic diseases or complicated diabetes were excluded from the study. Presence of haemolytic anemias or anemia due to acute blood loss was considered an exclusion criteria. The number of subjects included was 1312.

Study population was categorized into five groups on the basis of Hb levels: group 1–males(M) with Hb ≥ 13 g/dL and females(F) with Hb ≥ 12 g/dL; group 2—males with Hb 10—12.9 g/dL and females with Hb 10–11.9 g/dL; group 3—subjects with Hb 8.1–9.9 g/dL (both male and female), group 4—subjects with Hb 7.1–8 g/dL (both male and female), and group 5—subjects with Hb 5–7 gm/dL (both male and female). For males, the haemoglobin cut off for anaemia was 13 g/dL and for females the cut off was 12 g/dL (as per WHO criteria [10]).

The plasma RBG level was estimated by hexokinase method (Beckman DXC800). HbA1c was estimated by ion-exchange HPLC (Biorad Variant Turbo) and haemoglobin was estimated by photometric method on the Sysmax XN3000. All analyses were performed on fully automated platforms.

Data was tabulated and statistical analysis was performed. Correlation was derived between RBG and HbA1c (adjusted for age, sex, and Hb levels) in groups based on haemoglobin levels (groups 1 to 5).

All statistical analysis were performed with the programme statistical package for Social Sciences version 20.0 (SPSS Inc., Chicago, Illinois). Correlation significance was estimated by Pearson partial correlation (adjusting for age, sex and Hb) and a p value < 0.05 was accepted as statistically significant.

Results

Subjects

Out of 1312 subjects, 1038 (85.7%) were in the age group of 31–70 years with a mean age of 50.93 years. There were 570 (47.1%) females and 641 (52.9%) males. 479 of the enrolled subjects had anemia. None of the subjects had haemolytic anemia or anemia due to acute blood loss. Hence, by exclusion, all the anemic subjects had deficiency anemia.

Correlation of RBG with HbA1c

Partial correlation between RBG and HbA1c (adjusting for age, sex and Hb) was derived. For this the subjects were distributed in 5 groups on basis of Hb, as described above and shown in Table 1. The correlation between RBG and HbA1c was highly significant(r = 0.770; p < 0.00001) in group 1 which included subjects with normal Hb (Hb ≥ 13 g/dL [M] and ≥ 12 g/dL [F]). There was decrease in correlation ‘r’ value with decrease in haemoglobin cut off values. Significance of correlation was not lost till Hb is 5gm/dL. (The lowest Hb observed in our subjects was 5.3 g/dL.)

Table 1.

Correlation of HbA1c with RBG at different cut-offs of Haemoglobin

Variables	Subjects with Hb ≥ 13 g/dL (M) ≥ 12 g/dL (F) (group 1) (n = 833)	Subjects with Hb 10–12.9 g/dL (M) 10–11.9 g/dL (F) (group 2) (n = 364)	Subjects with Hb 8.1–9.9 g/dL (group3) (n = 75)	Subjects with Hb 7.1–8 g/dL (group 4) (n = 22)	Subjects with Hb 5–7 g/dL (group 5) (n = 18)
Mean Hb (in g/dL)	13.96 ± 0.05	11.42 ± 0.04	9.15 ± 0.06	7.53 ± 0.06	6.24 ± 0.21
Mean HbA1c (%)	6.38 ± 0.06	6.60 ± 0.08	6.63 ± 0.17	6.23 ± 0.35	6.67 ± 0.53
Mean RBG (in mg/dL)	162.08 ± 2.90	170.61 ± 4.35	173.49 ± 8.67	138.91 ± 14.59	197.67 ± 20.74
Correlation ‘r’ (between HbA1c and RBG)	0.770	0.677	0.651	0.608	0.634
Significance ‘p’	< 0.00001	< 0.0001	< 0.0001	0.003	0.006

There is decrease in significance of correlation between RBG and HbA1c (adjusted for Hb) with decrease in Hb but significance is not lost till Hb of 5 g/dL

Discussion

HbA_1c and Diabetes Mellitus

Diabetes management revolves around tight glycaemic control as evidenced through HbA_1c estimations. The basis for this was the Diabetes Control and Complications Trial (1983 to 1993), where 1441 type 1 diabetics were followed up to evaluate the long-term benefits of good control of diabetes either by standard treatment or by intensive treatment. It revealed that initial intensive treatment of type 1 diabetes towards maintaining an HbA1c as close as 6% benefitted the patient with a 50% lower risk of renal disease, 60% lower risk of neuropathy and 76% lower risk of retinopathy (micro-vascular diseases). The sequel to this study—the Epidemiology of Diabetes Interventions and Complications (EDIC) study evidenced a similar end-point HbA1c in both groups (standard treatment group and intensive treatment group) of patients [6]. Thus, HbA1c was established as an excellent basis for the assessment of diabetic control. It, therefore, becomes necessary that the interpretation of HbA1c should be accurate and confounders should be addressed adequately.

Anaemia and HbA1c

Though the life span of a red blood cell (RBC) is 120 days (4 months), since their formation and lyses is a dynamic process, the percentage of Hb which is glycated is a representation of 3 months average circulating glucose concentrations [2]. Hence, any change in RBC life span would change the interpretation of HbA1c. For example, in case of microcytic/macrocytic anaemia, where the Hb is low but the molecule is exposed to the intracellular glucose for longer periods due to the delayed turnover of RBCs, the process of glycation is affected and interpretation is altered [7]. Haemolytic anaemia or acute/chronic blood loss or splenomegaly, on the other hand, are associated with decreased RBC survival and, hence, a lower HbA1c [5].

Cohen et al. [11] observed that variation in RBC survival was enough to cause clinically important differences in HbA1c for a given mean blood glucose level. Jandric Balen et al. [12] concluded that in diabetic population with haemolytic anaemia, HbA1c is a very poor marker of overall glycemia. English et al. conducted an extensive review (January 1990 to May 2014 from 24 worldwide studies), based upon a systematic electronic database search on research articles regarding the effects of anaemia and abnormalities of erythrocyte indices on HbA1c. They summarised that abnormalities of erythrocyte indices are considerable confounders in the analysis of HbA1c [13]. Ford et al. collected the data from NHANES (National Health And Nutrition Examination Survey) and examined the association between anaemia and HbA1c. The authors concluded that haemoglobin positively correlated with HbA1c, such that there was a need for caution while diagnosing diabetes and prediabetes in people with high or low haemoglobin when HbA1c level was close to the 6.5% or 5.7% cut-offs [14]. Balasubramanian et al. took 50 non-diabetic iron deficiency anaemia (IDA) patients and 50 healthy age matched participants. Haemoglobin, packed cell volume, mean corpuscular volume, mean corpuscular haemoglobin, ferritin, fasting plasma glucose and HbA1c were analysed to determine the effects of IDA on HbA1c levels in non-diabetics. The authors observed that iron deficiency had a substantial effect on HbA1c and proposed that it was the consequent changes in the quaternary structure of haemoglobin that lead to increased glycation [15].

In haemolytic anaemia and individuals with recent significant blood loss, RBC life span is shortened as a result of intravascular and extravascular haemolysis. This leads to an increased turnover of RBC, resulting in a higher fraction of young erythrocytes, which are exposed to the circulating glucose for a shorter period. This results in a false low value of HbA1c [12]. But in IDA, due to decreased RBC turnover, the proportion of old erythrocytes is high in the circulation, resulting in longer exposure to circulating glucose and, therefore, falsely elevated HbA1c [16].

However, none of these studies addressed the need of the clinician—a cut-off of Hb below which HbA1c interpretation was affected significantly. Though a generalisation may be difficult, a cut-off could provide some guidance to the diabetologist. This becomes all the more relevant in our country (India) where the prevalence of anemia is nearly 40% (as per its definition by WHO—Hb < 12 g/dL in females and < 13 g/dL in males) [8]. The National Family Health Survey in 2015–2016 reported a prevalence of deficiency anemia of 59% in children below 3 years, 53% in women and 23% in men [17]. Most of these are attributed to nutritional deficiency states especially iron deficiency, making anemia the major cause of disability in India [14]. Recently, we analysed retrospective data (as yet unpublished) from 9006 subjects with anemia; iron deficiency was found in 66.7%, folate deficiency in 2.9% and vitamin B₁₂ deficiency in 44.1% of these subjects. This indicates that deficiency anemias are still the predominant type of anemias in India.

Hence, we aimed to identify Hb cut-off for accurate interpretation of HbA_1c in the  presence of deficiency anemias and our data showed that till  a  Hb of 5 g/dL, interpretation of HbA1c is not affected.

Conclusion

Interpretation of HbA1c in the presence of anemia is known to be altered. To the best of our knowledge, however, there are no guidelines as to what level of Hb alters its interpretation significantly. We have tried to address this issue in our study.

It is well-documented that HbA1c must be interpreted with caution in  the presence of anemia. Our data indicates that in the presence of deficiency-related anemias, caution in HbA1c interpretation is not required up to Hb of 5 g/dL.

However, more studies need to be conducted that take into consideration the alterations in glycation of Hb in the presence of haemolytic anemias, acute blood loss and other causes of changes in RBC indices.

Strengths

We included a statistically significant number of subjects (n = 1312) and have excluded subjects with haemolytic anemias and acute blood loss, as well as those with co-morbidities including diabetic complications. All biochemical analyses were done on freshly drawn blood specimen by globally accepted standard automated methods. Lastly, during statistical analysis, the correlation of RBG with HbA1c was assessed after adjusting for variations due to age, sex and Hb.

Limitations

We have measured RBG, not fasting and postprandial glucose levels, nor mean glucose. In addition, our data only includes subjects with deficiency anemias; none of our subjects were diagnosed with acute blood loss or hemolysis. Also, we did not take into consideration the RBC indices.

Abbreviations

Hb

Hemoglobin

HbA1

Adult haemoglobin type 1

HbA2

Adult haemoglobin type 2

HbF

Fetal hemoglobin

HbA_1c

Glycated haemoglobin

RBG

Random blood glucose

IFCC

International Federation of Clinical Chemistry

HPLC

High performance liquid chromatography

EDTA

Ethylene diamine tetraacetic acid

SPSS

Statistical Package for social sciences

VW

Variant window

BRI

Biological reference interval

EDIC

Epidemiology of Diabetes Interventions and Complications

NHANES

National Health And Nutrition Examination Survey

IDA

Iron deficiency anemia

Funding

No author has received any funding for any aspect of this study.

Compliance with Ethical Standards

Conflict of interest

None of the authors have any conflict of interest.

Ethical approval

All procedures performed in this study were in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was obtained from all individual participants included in the study.