- Email: [email protected]
HbA1c for the Diagnosis of Diabetes Mellitus in a Developing Country. A Position Article
Archives of Medical Research 41 (2010) 302e308
OPINION
HbA1c for the Diagnosis of Diabetes Mellitus in a Developing Country. A Position Article Francisco J. Gomez-Perez, Carlos A. Aguilar-Salinas, Paloma Almeda-Valdes, Daniel Cuevas-Ramos, Israel Lerman Garber, and Juan A. Rull Departamento de Endocrinologı´a y Metabolismo, Instituto Nacional de Ciencias Me´dicas y Nutricion Salvador Zubiran, Mexico, D.F., Mexico Received for publication April 16, 2010; accepted May 24, 2010 (ARCMED-D-10-00181).
An Expert Committee of the American Diabetes Association and the European Association for the Study of Diabetes recommended a move to the use of HbA1c level to diagnose diabetes mellitus. Diagnosis should be made if the A1c level is $6.5%. HbA1c provides a reliable measure of chronic glycemia, correlates well with the risk of long-term diabetes complications and technical limitations for standardization have been overcome in laboratories of the U.S. and Europe. The objective of this paper is to analyze critically the advantages and disadvantages of the use of HbA1c as a diagnostic method of diabetes in a developing country. The lack of a universal threshold for the diagnosis of diabetes, the cost of the test and the absence of the standardization network in the majority of the countries are major arguments for not including HbA1c as diagnostic criteria of diabetes. HbA1c diagnostic criteria has a low sensitivity. As a result, there is a lack of agreement between the HbA1c criteria with the other diagnostic methods that lead into significant variations in the number of affected cases. In addition, sensitivity and specificity vary among ethnic groups. No study has compared the diagnostic properties of the HbA1c in Latin America. In conclusion, the logistic limitations that exist in a large proportion of developing countries and the unsolved uncertainties that exist for the definition of the A1c criterion are strong arguments against the inclusion of HbA1c among the diagnostic criteria of diabetes. Ó 2010 IMSS. Published by Elsevier Inc. Key Words: HbA1c, Diabetes, Oral glucose tolerance test, Diagnosis.
Introduction Recently, an Expert Committee of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) led by David Nathan (1) recommended a move to the use of HbA1c level to diagnose diabetes mellitus. In 2010, the ADA adopted the proposal and it became part of the diagnosis criteria of diabetes in the 2010 Clinical Practice Recommendations (2). The document proposed that the diagnosis of diabetes should be made with an A1c level $6.5%. Diagnosis should be confirmed with a repeat A1c test unless clinical symptoms
Address reprint requests to: Francisco J. Gomez-Perez, MD, Departamento de Endocrinologia y Metabolismo, Instituto Nacional de Ciencias Me´dicas y Nutricio´n Salvador Zubira´n, Vasco de Quiroga #15, Tlalpan, 14000, Me´xico D.F., Me´xico; Phone: (þ52) (55) 5513-3891; FAX: (þ52) (55) 5513-0002; E-mail: [email protected]
and high glucose levels O200 mg/dL are present. The expert panel recommends not mixing different methods to diagnose diabetes because of the lack of agreement between them. In addition, the Committee initially concluded that an A1c level $6% but !6.5% should be considered at the highest risk for progression to diabetes (Table 1). However, A1c limits were modified in the 2010 Clinical Practice Recommendations, establishing the lower limit to 5.7%. Several days later, the Endocrine Society and the American Association of Clinical Endocrinologists endorsed this recommendation, but some limitations of the diagnostic test were highlighted in both reports. Before these statements, the only acceptable tests for the diagnosis of diabetes in non-pregnant adults were plasma glucose levels ($126 mg/dL fasting, $200 mg/dL randomly obtained with symptoms of diabetes, and $200 mg/dL after an oral glucose tolerance test) (3). The ADA committee states that HbA1c provides a reliable measure of chronic
0188-4409/$ - see front matter. Copyright Ó 2010 IMSS. Published by Elsevier Inc. doi: 10.1016/j.arcmed.2010.05.007
HbA1c and Diabetes in Developing Countries Table 1. Criteria for the diagnosis of diabetes (ADA) 2010 Any of the following criteria: 1. A1c $6.5%. The test should be performed in a laboratory using a method that is NGSP certified and standardized to the DCCT assay 2. FPG $126 mg/dL (7.0 mmol/l). Fasting is defined as no caloric intake for at least 8 h* 3. 2-h plasma glucose $200 mg/dL (11.1 mmol/l) during an OGTT. The test should be performed as described by the WHO using a load containing the equivalent of 75 g anhydrous glucose dissolved in water* 4. In a patient with classic symptoms of hyperglycemia or hyperglycemic crisis, a random plasma glucose $200 mg/dL (11.1 mmol/l) ADA, American Diabetes Association; DCCT, Diabetes Control and Complications Trial; NGSP, National Glycohemoglobin Standardization Program; OGTT, oral glucose tolerance test; WHO, World Health Organization. *In the absence of unequivocal hyperglycemia, criteria 1e3 should be confirmed by repeated testing.
glycemia, correlates well with the risk of long-term diabetes complications and technical limitations for standardization have been overcome in laboratories of the U.S. and Europe. In addition, the use of A1c for screening and diagnosis of patients with diabetes may offer some advantages for patients and caregivers. It does not require sampling patients after an overnight fast or 2 h after the administration of oral glucose. Although the ADA is recommending the use of this test as an alternative to the previously used measures of fasting plasma glucose and 2-h oral glucose tolerance test, it is assumed that A1C will replace these other tests in most cases because of its ease of use. A proposal for a major shift from the present criteria to diagnose diabetes frequently leads to controversy, especially if the supporting evidence is in a developing stage. The objective of this paper is to analyze critically the advantages and disadvantages of the use of HbA1c as a diagnostic method of diabetes in a developing country. Evidence that Favors the Use of HbA1c in the Diagnosis of Diabetes Evidence for the use of A1c for diagnostic purposes is largely based upon cross-sectional data showing that the prevalence of the microvascular complications of diabetes (retinopathy) increases in non-diabetic patients in direct proportion with the A1c concentration. The committee selected evidence from Egypt, the Pimas and the National Health and Nutrition Examination Survey (NHANES) III survey. In these surveys, the highest two or three deciles of fasting glycemia, 2 h post-challenge glucose concentration and HbA1c levels had a remarkably greater prevalence of retinopathy compared to the rest of the individuals (1). In addition, they have access to preliminary data of the DETECT collaborative study that shows that the prevalence of retinopathy (graded by fundus photog-
303
raphy) significantly increased in cases with HbA1c $6.5% (4). The evidence considered by the panel was in accordance with a systematic review of nine studies done in Asian or European populations that compared HbA1c vs. fasting plasma glucose (FPG) for the diagnosis of diabetes using the WHO criteria as the gold standard (5). Area under the ROC curve was similar for both methods. On the other hand, Sabanayagam and coworkers showed that HbA1c was useful for the detection of several diabetes-related chronic complications in a population-based study done in Singapore (6). The areas under the ROC curve were 0.754 for any degree of retinopathy, 0.899 for mild retinopathy, 0.904 for moderate retinopathy, 0.615 for chronic kidney disease, 0.673 for micro-/macroalbuminuria and 0.57 for neuropathy. HbA1c thresholds ranged from 6.6e7.0%. In addition, a recent report (7) derived from the ARIC study (Atherosclerotic Risk in Communities) found that higher HbA1c levels, even in the nondiabetic range, are associated with an increased risk of cardiovascular disease. The relationship has a J-shape and it is stronger compared to that found for fasting glycemia. Similar conclusions were published by the DECODE Study Group (8). Advantages of A1c testing compared with FPG and 2HPG for the diagnosis of diabetes according to the International Expert Committee report are as follows. First, a strong effort has been made to achieve worldwide standardization of HbA1c measurements. The HbA1c test should be performed using a method that is certified by the National Glycohemoglobin Standardization Program (NGSP) and standardized or traceable to the Diabetes Control and Complications Trial assay. This effort has made possible that HbA1c can be properly measured in the majority of U.S. laboratories (9). However, not every method fulfills the required precision. Lenters-Westra and Slingerland (10) report on the imprecision, bias, and total error of eight portable points of care (POC) HbA1c analyzers. The study used well-defined protocols and appropriate secondary reference measurement procedures. Two of the eight manufacturers withdrew from the study after initial unpromising results with their POC methods. Only two met the !0.85% error criterion of the National Glycohemoglobin Standardization Program. These results show that HbA1c estimated with POC analyzers are not suitable for diagnosis of diabetes. Second, the test has a low interindividual variability. The coefficient of variation (CV) for the mean within-person variation of HbA1c derived from six studies is approximately 3.5%. The variation is smaller for that reported for FPG (12e15%) (11). In addition, HbA1c has a low biological variability and less pre-analytic instability. Its results are unaffected by acute (e.g., stress or illness-related) perturbations in glucose levels (12). Also, it is not affected by variations caused by administering the same amount of glucose load to
304
Gomez-Perez et al./ Archives of Medical Research 41 (2010) 302e308
individuals with different body sizes as happens with the 2-h post-challenge glucose levels (13). Its small biological variability is in contrast with the significant variation of the 2-h postchallenge glucose levels. Finally, there is no need for fasting or timed samples, which are major limitations for the proper diagnosis based on FPG or the oral glucose tolerance test. Arguments Against the Use of HbA1c as a Tool for the Diagnosis of Diabetes Local and general issues overshadow the above-described advantages of HbA1c in populations other than Europe and the U.S. Lack of a universal threshold for the diagnosis of diabetes, absence of the standardization network in the majority of the countries and uncertanties in the diagnosis resulting from the use of nonstandardized methods are major arguments for not including HbA1c as a diagnostic criteria of diabetes in the developing world. In addition, biological factors limit its use in a small but significant proportion of cases in all populations. Issues Limited to Developing Countries Absence of a national program of standardization and the resulting uncertanties. The NGSP is a successful program organized by the American Diabetes Association (ADA), European Association for the Study of Diabetes (EASD) and the International Diabetes Federation (IDF) (9). Its implementation has made possible that the majority of the laboratories in the U.S. have a standardized A1c method. Data from a 2003 survey on glycohemoglobin done by the College of American Pathologist indicated that $98% of participating laboratories use NGSP certified methods and report results as HbA1c or HbA1c equivalents (14). There are similar standardization programs in Sweden and Japan. National standardization programs including all different designated comparision methods (i.e., ionexchange chromatography, immunoassays, affinity chromatography and others) are complimentary actions that are required in every country to become comparable with the results of the programs above described. These programs require a clear definition of the analyte based on its molecular structure, a primary reference material containing the analyte in a pure form, a validated reference method that specifically measures the analyte in human samples (usually a HPLC method), and a global network of reference laboratories that guarantees that the reference method is performed with the necessary analytical quality and is capable of assigning reliable values to matrix-based secondary reference materials and calibrators. In Mexico, as in other developing countries, lack of a national Glycohemoglobin Standardization Program precludes the use of this method as a diagnostic tool. Seven commercial kits are used for HbA1c measurements in Mexico. However, only a few laboratories (|50) have an
HPLC method in place. This is considered the reference method. Hence, the necessity of a standardization network is inevitable; it will allow the proper adjustment of the results provided by the non-reference methods. This issue is highlighted by a report by Garcı´a-Alcala´ et al. (15) showing that variations in HbA1c results related to the laboratory method utilized are large enough to lead to significant changes in the management of patients with diabetes. Despite the existence of some institutional programs (i.e., PACAL, ACP) that provide external quality control to the participating laboratories, many private and public clinics are not part of these inititatives. Such logistic problems are common in many developing countries. Unless goverments and medical or clinical chemistry societies work together to build a national Glycohemoglobin Standardization Program, clinical decision making will have a high degree of uncertainty generated by these methodological problems. HbA1c Has a Higher Cost Compared to the FPG or the OGTT. HbA1c is too expensive to be used as a diagnostic tool. The net cost of the A1c test is 13.6 times higher compared to a plasma glucose measurement. The cost is still higher than the OGTT (including the cost of the glucose load and the five plasma glucose measurements). In addition, it is highly probable that at least a FPG measurement will be required in a significant proportion of cases due to the low specificity of the A1c criterion. Thus, inclusion of the A1c among diabetes diagnostic criteria will increase the economic burden to our health system. Increased prevalence of biological conditions that alter the diagnostic performance of HbA1c anemia caused by iron deficiency or by hemoglobin variants is more common in developing countries (i.e., Africa, the Caribbean and Latin American nations) than in the U.S. or Europe (16). Issues Applicable to All Populations Problems for the selection of the A1c threshold: methodological and ethnic issues. The study by Sabanayagam and coworkers (6) shows that the relationship between HbA1c and the prevalence of moderate retinopathy is a continuous phenomenon. No inflection points are evident. As a result, the selection of the threshold is an arbitrary decision and adjustments by country or ethnic group could be required. Despite that, the ADA committee report proposes that an HbA1c $6.5% should be used for diagnosis. This decision was based on data of the DETECT-2 study (4) showing that moderate retinopathy is virtually nonexistent below this concentration. Receiver operating characteristic (ROC) curve analysis of the same data indicated that the optimal cut-off point for detecting at least moderate retinopathy was an A1c of 6.5%. This threshold coincides with the concentration that is 3 standard deviations (SD) above the
HbA1c and Diabetes in Developing Countries
mean A1c concentration found in the NHANES III survey (5.17%, SD 0.45%). The authors state that ‘‘in selecting a diagnostic A1C level $6.5%, the International Expert Committee balanced the stigma and costs of mistakenly identifying individuals as diabetic against the minimal clinical consequences of delaying the diagnosis in someone with an A1C level $6.5%.’’ However, this decision is in opposition to other reports showing that the A1c threshold with the best diagnostic performance is !6.5%. Among these are a systematic review of nine studies (5) and several cross-sectional population-based surveys (16e18). For example, Bennett and coworkers (5) proposed that the limit should be established between 6.0 and 6.2% after reviewing the results of nine studies. Also, Buell and coworkers (17) searched for the optimal A1c level in the 1999e2004 NHANES data. Diagnosis of diabetes was considered established if FPG was 126 mg/dL or greater. Using a ROC analysis, they found that HbA1c of 5.8% or greater is the point that yielded the highest sum of sensitivity (86%) and specificity (92%). These results are in contrast with the significantly lower sensitivity of the 6.5% value (42.8 and 44.3% observed in the NHANES-II and -III surveys, respectively). The stricter threshold results in a slightly higher specificity (99.6%). Recently, Ravi Kumar and coworkers found that an A1c of 6.1% has the best diagnostic performance in an Indian community-based study. The 6.5% value has a sensitivity and specificity of 65 and 88%, respectively (18). Using a cut-off point of 6.1%, the specificity decreased to 81% but the sensitivity increased to 81%. Thus, the conscientious decision to have HbA1c diagnostic criteria with a lower than optimal sensitivity has led to controversy (19,20). A limited sensitivity has been identified for HbA1c criteria compared to the other FPG and OGTT based criteria. In the Rancho Bernardo Study (n 5 2107 adults with known diabetes) (21), the A1c cut-off point of 6.5% had a sensitivity of 44% and a specificity of 79% and an area under ROC curve of 0.65. As in the previously referred reports (18e20), the A1c cut-off point of 6.15% yielded the highest combination of sensitivity (63%) and specificity (60%). However, lowering the threshold does not solve the diagnostic inaccuracy because despite the modification, A1c-based diagnosis would miss one-third of those with type 2 diabetes by ADA criteria and misclassify one-third of those without it. The agreement with type 2 diabetes diagnosis was low (kappa coefficient !0.119) regardless of which criteria were selected to establish the presence of diabetes. Lack of agreement between the HbA1c criteria with the other options results in significant variations in the number of affected cases. For example, in the NHANES 2003e2006 (22), 5.4% of the population aged O20 years was classified as having undiagnosed diabetes based on at least one of these criteria. All three criteria simultaneously classified only 23% of undiagnosed cases. The A1c criterion diagnosed the smallest percent of the population
305
(30% of the undiagnosed diabetic group). In contrast, the 2-h plasma glucose test diagnosed 90% of those with undiagnosed diabetes. Close to 1 out of 5 undiagnosed cases was diagnosed by both FPG and 2-h plasma glucose but not by A1c. As a result there is a remarkable discrepancy in the cases considered as affected using U.S. nationwide data. The HbA1c criterion may result in only 19.7 million U.S. adults identified as having diabetes, as opposed to 21.5 million by fasting plasma glucose and 26.5 million by glucose tolerance test. This discrepancy is undesirable for patients, physicians and the global health system. Lack of agreement among methods may have a more critical effect in the evaluation of at-risk populations compared to the results derived from the population-based surveys above described. Lu and coworkers (23) compared the diagnostic proficiency of HbA1c in a clinic-derived cohort and a population-based study (AusDiab). HbA1c criteria diagnosed only 58% of the patients with diabetes in the clinicbased group. In contrast, it was useful to detect 91.1% of diabetic subjects in the population-based survey. Even more, the majority of cases with either impaired glucose tolerance or impaired fasting glucose were missed by the HbA1c-based approach in both populations. Thus, the low sensitivity of the HbA1c criterion may leave undetected a significant proportion of cases seeking attention because they are considered at risk for having diabetes. In addition, the sensitivity and specificity varies among ethnic groups (24). Six studies from Denmark, UK, Kenya, Australia, Greenland and India were analyzed to measure the impact of using the HbA1c criterion among diabetic cases diagnosed using an OGTT. In this report (25), diabetes prevalence was lower using HbA1c as compared against the OGTT criterion in the majority of the studies. In a subsidiary analysis of the southeastern Asian and Black minorities compared with the Caucasian studied in England, this indicates that the magnitude of the discrepancies varies among ethnic groups. Prevalence was increased by 63% with the use of the HbA1c definition in the Danish study but 82% decreased in the Australian study. Also, Anand (26) stratified their findings according to the three ethnic groups included in their study population, East Asian, Chinese and European, and reported ethnic variations in sensitivity and specificity for both HbA1c and FPG tests. In addition, several studies have shown race-related differences in HbA1c curve distribution. Data derived from the NHANES-III survey showed higher values for non-Hispanic blacks and Mexican-Americans than for nonHispanic whites (27). Herman (28) compared HbA1c levels in population-based studies of 11 countries. A1C was higher in Hispanics (9.4 1.4%), Asians (9.2 1.4%) and patients from other racial/ethnic groups (9.7 1.5%) compared with Caucasians (8.9 1.2%) after adjusting for factors affecting glycemia. However, this finding has not been consistent in every study. Thus, different sensitivity and specificity for HbA1c occurs among ethnic groups, which may be related to genetic differences in the concentration of hemoglobin,
306
Gomez-Perez et al./ Archives of Medical Research 41 (2010) 302e308
rates of glycation and the lifespan or amount of red blood cells. No study has compared the diagnostic properties of the HbA1c in populations from Latin America. Regrettably, all available methods for the diagnosis of diabetes have limitations. One example of this with respect to the OGTT-based approach is the significant variability of the 2-h post-challenge glucose levels and the lack of replication of the OGTT result. More practical and specific approaches will be welcome in the future. However, the above-described limitations of the HbA1c criterion precludes its use in many regions of the world. Abnormal hemoglobins. Hemoglobin A1c should not be used in some hemoglobin traits such as HbS, HbC, HbF and HbE (29). Many assay methods can correct for the presence of the most common hemoglobin traits and there are affinity assays unaffected by hemoglobin traits. Nevertheless, the trait should be diagnosed and the adequate A1c methods should be available. Anemia and Iron Deficiency Any condition that shortens erythrocyte survival (i.e., hemolytic anemia, spherocytosis, pregnancy) will proportionally decrease HbA1c because hemoglobin in younger red cells has less time of exposure to ambient glycemia (30). Also, bleeding and the resulting increased reticulocyte production will decrease the half-life of erythrocytes and thereby will lower HbA1C. In contrast, any factor that prolongs the erythrocyte half-life will increase the level of HbA1C (i.e., splenectomy, aplastic anemia). Iron deficiency can lead to rises in HbA1c up to 2% that can be reversed with iron supplementation. Aging A consistent increment in A1c with age was found in a crosssectional analysis of both the Framingham Offspring Study (FOS) and the NHANES study 2001e2004 in nondiabetic populations (31). The longitudinal analysis of FOS confirmed an increase in A1c with aging. A 0.03-point increment per year was found in subjects without abnormalities in glucose metabolism. More studies are needed to establish whether age-specific diagnosis criteria are required. Renal failure Glycated hemoglobin levels can be affected by several factors that are specific for chronic kidney disease (32). Increased blood urea nitrogen levels can cause the formation of carbamylated hemoglobin, which cannot be distinguished from glycated hemoglobin by using electrical charge-based assays. It has been estimated that an increase in time-averaged blood urea nitrogen concentration from 3.57e24.9 mol/l will result in an increment in glycated hemoglobin of 1.26%. This effect is much smaller compared with the impact of an increase in
Table 2. Categories of increased risk for diabetes* (ADA 2010) FPG 100 mg/dL (5.6 mmol/l) to 125 mg/dL (6.9 mmol/l) (IFG) 2-FPG in the 75-g OGTT, 140 mg/dL (7.8 mmol/l) to 199 mg/dL (11 mmol/l) (IGT) A1C 5.7e6.4% *For all three tests, risk is continuous, extending below the lower limit of the range and becoming disproportionately greater at higher ends of the range.
blood glucose concentration of similar proportion. Overestimation of glycated hemoglobin can be overcome by using assays that do not depend on electrical charge or assays specific for HbA1c, which are less prone to interference from carbamylated hemoglobin. Other factors that modify HbA1c. Lower values of HbA1c have been reported in patients with HIV infection. In contrast, severe hypertriglyceridemia, hyperbilirubinemia, chronic alcoholism, chronic ingestion of opioid drugs and chronic use of high doses of salicylates are potential causes of higher HbA1c concentrations. Use of antioxidants may alter HbA1c results (33). HbA1c and Increased Risk for Diabetes The Expert Committee proposed that HbA1c values between 6.0 and 6.4% identify subjects at-risk for having diabetes. This category could be an equivalent for the categories enclosed in prediabetes (i.e., impaired fasting glucose and impaired glucose tolerance) (Table 2). Selection of the threshold was based on the HbA1c concentration distribution found in the NHANES surveys. An A1c value of 6.0% corresponds to two SDs above the population mean. However, this recommendation was modified in the 2010 Clinical Practice Recommendations in which the lower limit was decreased to 5.7%. This decision was based on a ROC curve analysis of the NHANES 2005e2006 data in which an A1c value of 5.7% had the best combination of sensitivity (39%) and specificity (91%) to identify cases of impaired fasting glucose. The same strategy has been used in the past by the ADA when the fasting plasma glucose was incorporated as a potential alternative method for the OGTT. The lower limit of the criterion was changed from 6.1 to 5.6 mmol/l in order to diagnose the same number of cases using the OGTT and the FGP. However, this strategy was partially successful because of the lack of agreement among the diagnostic categories (34). It is very likely that the change in the A1c criterion will cause the same diagnostic disagreements. The use of HbA1c as a diagnostic test in ‘‘prediabetes’’ is debated. The agreement between prediabetes categories and an abnormal A1c (5.7e6.4%) has not been included in the majority of the surveys. The low sensitivity (39%) of the A1c criterion is a strong argument against its use in clinical practice because a highly sensitive test is
HbA1c and Diabetes in Developing Countries
required to screen cases at risk for diabetes. In addition, the cost of the test makes A1c testing an unfeasible option for screening diabetes. The HbA1c criterion (5.7e6.4%) to detect at-risk cases will identify 12% of the U.S. adult population. This number is significantly smaller compared to that diagnosed using the current FPG criteria (25%). Finally, there are several studies supporting the concept that the combined use of HbA1c and FPG has a better predictive value compared to FPG to estimate the likelihood of diabetes in high-risk subjects. However, this approach implies a significantly higher cost (35). This approach was not considered by the ADA Committee. In conclusion, a proposal for a major shift from the present criteria to diagnose diabetes frequently leads to controversy especially if the evidence that support it is not strong enough. High A1c levels are clearly associated with incident microvascular complications in cohorts comprised of previously diagnosed cases with diabetes. The available evidence supports the role of the A1c test as a marker of glycemic control and as a prognostic factor for future complications. However, it is insufficient to recommend it as the method of choice for diagnosis. In addition, the limited access to standardized laboratory A1c methods makes unfeasible the inclusion of the A1c criterion among the diagnostic criteria of diabetes in a large proportion of the world. On the other hand, the selection of a diagnostic threshold is usually made by means of the ROC approach followed by its validation in the target population (using likelihood ratios). Such evidence is missing for Latin American and African countries. Even less information was presented to support the 5.7% threshold. The lack of agreement between the glycemia-based diagnosis and the A1c may lead to a series of conflicting reports regarding various diagnostic criteria-related issues (i.e., prevalence of the disease) that may result in confusion among patients and health providers (34). This change could be in opposition to the principles behind the 1997 ADA recommendations (i.e., to obtain an earlier diagnosis of diabetes and glucose intolerance). Finally, it will increase the A1c tests and related costs. Prospective evaluations are required to assess the clinical and economic consequences that may result from the modification of the current diabetes diagnostic criteria. In the meantime, we should continue using plasma glucose levels and the OGTT for the diagnosis of diabetes. References 1. The International Expert Committee Report on the Role of the A1C Assay in the diagnosis of Diabetes. Diabetes Care 2009;32:1327e1334. 2. American Diabetes Association. Diagnosis Criteria for Diabetes Mellitus. Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 2010;33(suppl 1):S62eS69. 3. American Diabetes Association. Screening for type 2 diabetes (Clinical Practice Recommendations 2004: Position Statement). Diabetes Care 2004;27:S11eS14.
307
4. Vistisen D, Colagiuri S, Borch-Johnsen K. The DETECT-2 Collaboration. Bimodal distribution of glucose is not universally useful for diagnosing diabetes. Diabetes Care 2009;32:397e403. 5. Bennett AL, Guop M, Dharmage SC. HbA1C as a screening tool for detection of type 2 diabetes: a systematic review. Diabet Med 2007;24: 333e343. 6. Sabanayagam C, Liew G, Tai ES, et al. Relationship between glycated haemoglobin and microvascular complications: is there a natural cutoff point for the diagnosis of diabetes? Diabetologia 2009;52: 1279e1289. 7. Selvin E, Steffes MW, Zhu H, et al. Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults. N Engl J Med 2010; 362:800e811. 8. The DECODE Study Group. Is the current definition for diabetes relevant to mortality risk from all causes and cardiovascular and non-cardiovascular diseases. Diabetes Care 2003;26:688e696. 9. Goldstein DE, Little RR, Lorenz RA, et al. Global harmonization of hemoglobin A1c. Clin Chem 2005;51:681e683. 10. Lenters-Westra E, Slingerland RJ. Six out of eight hemoglobin A1c point-of care instruments do not meet the generally accepted analytical performance criteria. Clin Chem 2010;56:44e52. 11. Amquist H, Wallensteen M, Jeppsson JO. Standardization of long-term glucose measurements established. Lakartidningen 1997;50:4789e4790. 12. Peters AL, Davidson MB, Schigrer DL, et al. A clinical approach for the diagnosis of diabetes mellitus. An analysis using glycosylated hemoglobin levels. JAMA 1996;276:1246e1252. 13. Faerch A, Borch-Johnsen K, Vaag A, et al. Sex differences in glucose levels: a consequence of physiology or methodological convenience? The Inter99 study. Diabetologia 2010;53:858e865. 14. Shima K, Endo J, Oimoni Y, et al. A report of the committee on an inter-laboratory standardization of HbA1C determination. The Japan Diabetes Society. J Jpn Diabetes Soc 1994;37:855e864. 15. Garcı´a-Alcala´ H, Ruiz-Argu¨elles A, Cedillo-Carvallo B. Effect of the method to measure levels of glycated hemoglobin on individual clinical decisions comparison of an immunoassay with high-performance liquid chromatography. Am J Clin Pathol 2009;132:332e335. 16. Ezenwaka C, Jones-LeCointe A, Nwagbara E, et al. Anaemia and kidney dysfunction in Caribbean type 2 diabetic patients. Cardiovasc Diabetol 2008;7:25e31. 17. Buell C, Kermah D, Davidson MB. Utility of A1C for diabetes screening in the 1999 2004 NHANES population. Diabetes Care 2007;30:2233e2235. 18. Ravi Kumar P, Bhansali A, Ravikiran M, et al. Utility of glycated hemoglobin in diagnosing type 2 diabetes mellitus: a communitybased study. J Clin Endocrinol Metab. Published ahead of print April 6, 2010 (doi:10.1210/jc.2009e2433). 19. Saudek CD, Herman WH, Sacks DB, et al. A new look at screening and diagnosing diabetes mellitus. J Clin Endocrinol Metab 2008;93: 2447e2453. 20. Herman WH, Fajans SS. Hemoglobin A1c for the diagnosis of diabetes: practical considerations. Pol Arch Med Wewn 2010;120:37e41. 21. Kramer CK, Araneta MR, Barrett-Connor E. A1C and diabetes diagnosis: The Rancho Bernardo Study. Diabetes Care 2010;33:101e103. 22. Carson A, Reynolds K, Fonseca V, et al. Comparison of A1C and fasting glucose criteria to diagnose diabetes among U.S. adults. Diabetes Care 2010;33:95e97. 23. Lu X, Walker K, O’Dea K, et al. A1C for screening and diagnosis of type 2 diabetes in routine clinical practice. Diabetes Care 2010;33:817e819. 24. Herman WH, Dungan KM, Wolffenbuttel BHR, et al. Racial and ethnic differences in mean glucose, hemoglobin A1c, and 1,5 anhydroglucitol in over 2000 patients with type 2 diabetes. J Clin Endocrinol Metab 2009;94:1689e1694. 25. Christensen DL, Witte DR, Kduka L, et al. Moving to an A1c-based diagnosis of diabetes has a different impact on prevalence in different ethnic groups. Diabetes Care 2010;33:580e582.
308
Gomez-Perez et al./ Archives of Medical Research 41 (2010) 302e308
26. Anand SS, Razak F, Vuksan V, et al. Diagnostic strategies to detect glucose intolerance in a multiethnic population. Diabetes Care 2003; 26:290e229. 27. Cowie C, Rust K, Byrd-Holt D, et al. Prevalence of diabetes and high risk for diabetes using A1C criteria in the U.S. population in 1988e2006. Diabetes Care 2010;33:562e568. 28. Herman WH, Ma Y, Uwaifo G, et al. Diabetes Prevention Program Research Group Differences in A1C by race and ethnicity among patients with impaired glucose tolerance in the Diabetes Prevention Program. Diabetes Care 2007;30:2453e2457. 29. Kim C, McKeever K, Herman WH, et al. Association between iron deficiency spherocytosis and HbA1C levels among adults without diabetes in the National Health and Nutrition Examination Survey, 1999e2006. Diabetes Care 2010;33:780e785. 30. El-Agouza I, Abu SA, Syrdah M. The effect of iron-deficiency anaemia on the glycosylated hemoglobin subtypes: possible consequences for clinical diagnosis. Clin Lab Haematol 2002;24:285e289.
31. Pani LN, Korenda L, Meigs JM, et al. Effect of aging on A1C levels in individuals without diabetes. Diabetes Care 2008;31:1991e1996. 32. Krovesdy CP, Sharma K, Kalantar-Zadeh K. Glycemic control in diabetic CKD patients: where do we stand? Am J Kidney Dis 2008; 52:766e777. 33. Gomez-Perez FJ, Valles-Sanchez VE, Lopez-Alvarenga JC, et al. Vitamin E modifies neither fructosamine nor HbA1c levels in poorly controlled diabetes. Rev Invest Clin 1996;48:421e424. 34. Gomez-Perez FJ, Aguilar-Salinas CA, Lo´pez-Alvarenga JC, et al. Lack of agreement between the World Health Organization category of impaired glucose tolerance and the American Diabetes Association category of impaired fasting glucose. Diabetes Care 1998;21: 1886e1888. 35. Ko GTC, Chan JCN, Yeung VTF, et al. Combined use of fasting plasma glucose concentration and HbA1C or fructosamine predicts the likelihood of having diabetes in high risk subjects. Diabetes Care 1998;21:1221e1225.
OPINION
HbA1c for the Diagnosis of Diabetes Mellitus in a Developing Country. A Position Article Francisco J. Gomez-Perez, Carlos A. Aguilar-Salinas, Paloma Almeda-Valdes, Daniel Cuevas-Ramos, Israel Lerman Garber, and Juan A. Rull Departamento de Endocrinologı´a y Metabolismo, Instituto Nacional de Ciencias Me´dicas y Nutricion Salvador Zubiran, Mexico, D.F., Mexico Received for publication April 16, 2010; accepted May 24, 2010 (ARCMED-D-10-00181).
An Expert Committee of the American Diabetes Association and the European Association for the Study of Diabetes recommended a move to the use of HbA1c level to diagnose diabetes mellitus. Diagnosis should be made if the A1c level is $6.5%. HbA1c provides a reliable measure of chronic glycemia, correlates well with the risk of long-term diabetes complications and technical limitations for standardization have been overcome in laboratories of the U.S. and Europe. The objective of this paper is to analyze critically the advantages and disadvantages of the use of HbA1c as a diagnostic method of diabetes in a developing country. The lack of a universal threshold for the diagnosis of diabetes, the cost of the test and the absence of the standardization network in the majority of the countries are major arguments for not including HbA1c as diagnostic criteria of diabetes. HbA1c diagnostic criteria has a low sensitivity. As a result, there is a lack of agreement between the HbA1c criteria with the other diagnostic methods that lead into significant variations in the number of affected cases. In addition, sensitivity and specificity vary among ethnic groups. No study has compared the diagnostic properties of the HbA1c in Latin America. In conclusion, the logistic limitations that exist in a large proportion of developing countries and the unsolved uncertainties that exist for the definition of the A1c criterion are strong arguments against the inclusion of HbA1c among the diagnostic criteria of diabetes. Ó 2010 IMSS. Published by Elsevier Inc. Key Words: HbA1c, Diabetes, Oral glucose tolerance test, Diagnosis.
Introduction Recently, an Expert Committee of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) led by David Nathan (1) recommended a move to the use of HbA1c level to diagnose diabetes mellitus. In 2010, the ADA adopted the proposal and it became part of the diagnosis criteria of diabetes in the 2010 Clinical Practice Recommendations (2). The document proposed that the diagnosis of diabetes should be made with an A1c level $6.5%. Diagnosis should be confirmed with a repeat A1c test unless clinical symptoms
Address reprint requests to: Francisco J. Gomez-Perez, MD, Departamento de Endocrinologia y Metabolismo, Instituto Nacional de Ciencias Me´dicas y Nutricio´n Salvador Zubira´n, Vasco de Quiroga #15, Tlalpan, 14000, Me´xico D.F., Me´xico; Phone: (þ52) (55) 5513-3891; FAX: (þ52) (55) 5513-0002; E-mail: [email protected]
and high glucose levels O200 mg/dL are present. The expert panel recommends not mixing different methods to diagnose diabetes because of the lack of agreement between them. In addition, the Committee initially concluded that an A1c level $6% but !6.5% should be considered at the highest risk for progression to diabetes (Table 1). However, A1c limits were modified in the 2010 Clinical Practice Recommendations, establishing the lower limit to 5.7%. Several days later, the Endocrine Society and the American Association of Clinical Endocrinologists endorsed this recommendation, but some limitations of the diagnostic test were highlighted in both reports. Before these statements, the only acceptable tests for the diagnosis of diabetes in non-pregnant adults were plasma glucose levels ($126 mg/dL fasting, $200 mg/dL randomly obtained with symptoms of diabetes, and $200 mg/dL after an oral glucose tolerance test) (3). The ADA committee states that HbA1c provides a reliable measure of chronic
0188-4409/$ - see front matter. Copyright Ó 2010 IMSS. Published by Elsevier Inc. doi: 10.1016/j.arcmed.2010.05.007
HbA1c and Diabetes in Developing Countries Table 1. Criteria for the diagnosis of diabetes (ADA) 2010 Any of the following criteria: 1. A1c $6.5%. The test should be performed in a laboratory using a method that is NGSP certified and standardized to the DCCT assay 2. FPG $126 mg/dL (7.0 mmol/l). Fasting is defined as no caloric intake for at least 8 h* 3. 2-h plasma glucose $200 mg/dL (11.1 mmol/l) during an OGTT. The test should be performed as described by the WHO using a load containing the equivalent of 75 g anhydrous glucose dissolved in water* 4. In a patient with classic symptoms of hyperglycemia or hyperglycemic crisis, a random plasma glucose $200 mg/dL (11.1 mmol/l) ADA, American Diabetes Association; DCCT, Diabetes Control and Complications Trial; NGSP, National Glycohemoglobin Standardization Program; OGTT, oral glucose tolerance test; WHO, World Health Organization. *In the absence of unequivocal hyperglycemia, criteria 1e3 should be confirmed by repeated testing.
glycemia, correlates well with the risk of long-term diabetes complications and technical limitations for standardization have been overcome in laboratories of the U.S. and Europe. In addition, the use of A1c for screening and diagnosis of patients with diabetes may offer some advantages for patients and caregivers. It does not require sampling patients after an overnight fast or 2 h after the administration of oral glucose. Although the ADA is recommending the use of this test as an alternative to the previously used measures of fasting plasma glucose and 2-h oral glucose tolerance test, it is assumed that A1C will replace these other tests in most cases because of its ease of use. A proposal for a major shift from the present criteria to diagnose diabetes frequently leads to controversy, especially if the supporting evidence is in a developing stage. The objective of this paper is to analyze critically the advantages and disadvantages of the use of HbA1c as a diagnostic method of diabetes in a developing country. Evidence that Favors the Use of HbA1c in the Diagnosis of Diabetes Evidence for the use of A1c for diagnostic purposes is largely based upon cross-sectional data showing that the prevalence of the microvascular complications of diabetes (retinopathy) increases in non-diabetic patients in direct proportion with the A1c concentration. The committee selected evidence from Egypt, the Pimas and the National Health and Nutrition Examination Survey (NHANES) III survey. In these surveys, the highest two or three deciles of fasting glycemia, 2 h post-challenge glucose concentration and HbA1c levels had a remarkably greater prevalence of retinopathy compared to the rest of the individuals (1). In addition, they have access to preliminary data of the DETECT collaborative study that shows that the prevalence of retinopathy (graded by fundus photog-
303
raphy) significantly increased in cases with HbA1c $6.5% (4). The evidence considered by the panel was in accordance with a systematic review of nine studies done in Asian or European populations that compared HbA1c vs. fasting plasma glucose (FPG) for the diagnosis of diabetes using the WHO criteria as the gold standard (5). Area under the ROC curve was similar for both methods. On the other hand, Sabanayagam and coworkers showed that HbA1c was useful for the detection of several diabetes-related chronic complications in a population-based study done in Singapore (6). The areas under the ROC curve were 0.754 for any degree of retinopathy, 0.899 for mild retinopathy, 0.904 for moderate retinopathy, 0.615 for chronic kidney disease, 0.673 for micro-/macroalbuminuria and 0.57 for neuropathy. HbA1c thresholds ranged from 6.6e7.0%. In addition, a recent report (7) derived from the ARIC study (Atherosclerotic Risk in Communities) found that higher HbA1c levels, even in the nondiabetic range, are associated with an increased risk of cardiovascular disease. The relationship has a J-shape and it is stronger compared to that found for fasting glycemia. Similar conclusions were published by the DECODE Study Group (8). Advantages of A1c testing compared with FPG and 2HPG for the diagnosis of diabetes according to the International Expert Committee report are as follows. First, a strong effort has been made to achieve worldwide standardization of HbA1c measurements. The HbA1c test should be performed using a method that is certified by the National Glycohemoglobin Standardization Program (NGSP) and standardized or traceable to the Diabetes Control and Complications Trial assay. This effort has made possible that HbA1c can be properly measured in the majority of U.S. laboratories (9). However, not every method fulfills the required precision. Lenters-Westra and Slingerland (10) report on the imprecision, bias, and total error of eight portable points of care (POC) HbA1c analyzers. The study used well-defined protocols and appropriate secondary reference measurement procedures. Two of the eight manufacturers withdrew from the study after initial unpromising results with their POC methods. Only two met the !0.85% error criterion of the National Glycohemoglobin Standardization Program. These results show that HbA1c estimated with POC analyzers are not suitable for diagnosis of diabetes. Second, the test has a low interindividual variability. The coefficient of variation (CV) for the mean within-person variation of HbA1c derived from six studies is approximately 3.5%. The variation is smaller for that reported for FPG (12e15%) (11). In addition, HbA1c has a low biological variability and less pre-analytic instability. Its results are unaffected by acute (e.g., stress or illness-related) perturbations in glucose levels (12). Also, it is not affected by variations caused by administering the same amount of glucose load to
304
Gomez-Perez et al./ Archives of Medical Research 41 (2010) 302e308
individuals with different body sizes as happens with the 2-h post-challenge glucose levels (13). Its small biological variability is in contrast with the significant variation of the 2-h postchallenge glucose levels. Finally, there is no need for fasting or timed samples, which are major limitations for the proper diagnosis based on FPG or the oral glucose tolerance test. Arguments Against the Use of HbA1c as a Tool for the Diagnosis of Diabetes Local and general issues overshadow the above-described advantages of HbA1c in populations other than Europe and the U.S. Lack of a universal threshold for the diagnosis of diabetes, absence of the standardization network in the majority of the countries and uncertanties in the diagnosis resulting from the use of nonstandardized methods are major arguments for not including HbA1c as a diagnostic criteria of diabetes in the developing world. In addition, biological factors limit its use in a small but significant proportion of cases in all populations. Issues Limited to Developing Countries Absence of a national program of standardization and the resulting uncertanties. The NGSP is a successful program organized by the American Diabetes Association (ADA), European Association for the Study of Diabetes (EASD) and the International Diabetes Federation (IDF) (9). Its implementation has made possible that the majority of the laboratories in the U.S. have a standardized A1c method. Data from a 2003 survey on glycohemoglobin done by the College of American Pathologist indicated that $98% of participating laboratories use NGSP certified methods and report results as HbA1c or HbA1c equivalents (14). There are similar standardization programs in Sweden and Japan. National standardization programs including all different designated comparision methods (i.e., ionexchange chromatography, immunoassays, affinity chromatography and others) are complimentary actions that are required in every country to become comparable with the results of the programs above described. These programs require a clear definition of the analyte based on its molecular structure, a primary reference material containing the analyte in a pure form, a validated reference method that specifically measures the analyte in human samples (usually a HPLC method), and a global network of reference laboratories that guarantees that the reference method is performed with the necessary analytical quality and is capable of assigning reliable values to matrix-based secondary reference materials and calibrators. In Mexico, as in other developing countries, lack of a national Glycohemoglobin Standardization Program precludes the use of this method as a diagnostic tool. Seven commercial kits are used for HbA1c measurements in Mexico. However, only a few laboratories (|50) have an
HPLC method in place. This is considered the reference method. Hence, the necessity of a standardization network is inevitable; it will allow the proper adjustment of the results provided by the non-reference methods. This issue is highlighted by a report by Garcı´a-Alcala´ et al. (15) showing that variations in HbA1c results related to the laboratory method utilized are large enough to lead to significant changes in the management of patients with diabetes. Despite the existence of some institutional programs (i.e., PACAL, ACP) that provide external quality control to the participating laboratories, many private and public clinics are not part of these inititatives. Such logistic problems are common in many developing countries. Unless goverments and medical or clinical chemistry societies work together to build a national Glycohemoglobin Standardization Program, clinical decision making will have a high degree of uncertainty generated by these methodological problems. HbA1c Has a Higher Cost Compared to the FPG or the OGTT. HbA1c is too expensive to be used as a diagnostic tool. The net cost of the A1c test is 13.6 times higher compared to a plasma glucose measurement. The cost is still higher than the OGTT (including the cost of the glucose load and the five plasma glucose measurements). In addition, it is highly probable that at least a FPG measurement will be required in a significant proportion of cases due to the low specificity of the A1c criterion. Thus, inclusion of the A1c among diabetes diagnostic criteria will increase the economic burden to our health system. Increased prevalence of biological conditions that alter the diagnostic performance of HbA1c anemia caused by iron deficiency or by hemoglobin variants is more common in developing countries (i.e., Africa, the Caribbean and Latin American nations) than in the U.S. or Europe (16). Issues Applicable to All Populations Problems for the selection of the A1c threshold: methodological and ethnic issues. The study by Sabanayagam and coworkers (6) shows that the relationship between HbA1c and the prevalence of moderate retinopathy is a continuous phenomenon. No inflection points are evident. As a result, the selection of the threshold is an arbitrary decision and adjustments by country or ethnic group could be required. Despite that, the ADA committee report proposes that an HbA1c $6.5% should be used for diagnosis. This decision was based on data of the DETECT-2 study (4) showing that moderate retinopathy is virtually nonexistent below this concentration. Receiver operating characteristic (ROC) curve analysis of the same data indicated that the optimal cut-off point for detecting at least moderate retinopathy was an A1c of 6.5%. This threshold coincides with the concentration that is 3 standard deviations (SD) above the
HbA1c and Diabetes in Developing Countries
mean A1c concentration found in the NHANES III survey (5.17%, SD 0.45%). The authors state that ‘‘in selecting a diagnostic A1C level $6.5%, the International Expert Committee balanced the stigma and costs of mistakenly identifying individuals as diabetic against the minimal clinical consequences of delaying the diagnosis in someone with an A1C level $6.5%.’’ However, this decision is in opposition to other reports showing that the A1c threshold with the best diagnostic performance is !6.5%. Among these are a systematic review of nine studies (5) and several cross-sectional population-based surveys (16e18). For example, Bennett and coworkers (5) proposed that the limit should be established between 6.0 and 6.2% after reviewing the results of nine studies. Also, Buell and coworkers (17) searched for the optimal A1c level in the 1999e2004 NHANES data. Diagnosis of diabetes was considered established if FPG was 126 mg/dL or greater. Using a ROC analysis, they found that HbA1c of 5.8% or greater is the point that yielded the highest sum of sensitivity (86%) and specificity (92%). These results are in contrast with the significantly lower sensitivity of the 6.5% value (42.8 and 44.3% observed in the NHANES-II and -III surveys, respectively). The stricter threshold results in a slightly higher specificity (99.6%). Recently, Ravi Kumar and coworkers found that an A1c of 6.1% has the best diagnostic performance in an Indian community-based study. The 6.5% value has a sensitivity and specificity of 65 and 88%, respectively (18). Using a cut-off point of 6.1%, the specificity decreased to 81% but the sensitivity increased to 81%. Thus, the conscientious decision to have HbA1c diagnostic criteria with a lower than optimal sensitivity has led to controversy (19,20). A limited sensitivity has been identified for HbA1c criteria compared to the other FPG and OGTT based criteria. In the Rancho Bernardo Study (n 5 2107 adults with known diabetes) (21), the A1c cut-off point of 6.5% had a sensitivity of 44% and a specificity of 79% and an area under ROC curve of 0.65. As in the previously referred reports (18e20), the A1c cut-off point of 6.15% yielded the highest combination of sensitivity (63%) and specificity (60%). However, lowering the threshold does not solve the diagnostic inaccuracy because despite the modification, A1c-based diagnosis would miss one-third of those with type 2 diabetes by ADA criteria and misclassify one-third of those without it. The agreement with type 2 diabetes diagnosis was low (kappa coefficient !0.119) regardless of which criteria were selected to establish the presence of diabetes. Lack of agreement between the HbA1c criteria with the other options results in significant variations in the number of affected cases. For example, in the NHANES 2003e2006 (22), 5.4% of the population aged O20 years was classified as having undiagnosed diabetes based on at least one of these criteria. All three criteria simultaneously classified only 23% of undiagnosed cases. The A1c criterion diagnosed the smallest percent of the population
305
(30% of the undiagnosed diabetic group). In contrast, the 2-h plasma glucose test diagnosed 90% of those with undiagnosed diabetes. Close to 1 out of 5 undiagnosed cases was diagnosed by both FPG and 2-h plasma glucose but not by A1c. As a result there is a remarkable discrepancy in the cases considered as affected using U.S. nationwide data. The HbA1c criterion may result in only 19.7 million U.S. adults identified as having diabetes, as opposed to 21.5 million by fasting plasma glucose and 26.5 million by glucose tolerance test. This discrepancy is undesirable for patients, physicians and the global health system. Lack of agreement among methods may have a more critical effect in the evaluation of at-risk populations compared to the results derived from the population-based surveys above described. Lu and coworkers (23) compared the diagnostic proficiency of HbA1c in a clinic-derived cohort and a population-based study (AusDiab). HbA1c criteria diagnosed only 58% of the patients with diabetes in the clinicbased group. In contrast, it was useful to detect 91.1% of diabetic subjects in the population-based survey. Even more, the majority of cases with either impaired glucose tolerance or impaired fasting glucose were missed by the HbA1c-based approach in both populations. Thus, the low sensitivity of the HbA1c criterion may leave undetected a significant proportion of cases seeking attention because they are considered at risk for having diabetes. In addition, the sensitivity and specificity varies among ethnic groups (24). Six studies from Denmark, UK, Kenya, Australia, Greenland and India were analyzed to measure the impact of using the HbA1c criterion among diabetic cases diagnosed using an OGTT. In this report (25), diabetes prevalence was lower using HbA1c as compared against the OGTT criterion in the majority of the studies. In a subsidiary analysis of the southeastern Asian and Black minorities compared with the Caucasian studied in England, this indicates that the magnitude of the discrepancies varies among ethnic groups. Prevalence was increased by 63% with the use of the HbA1c definition in the Danish study but 82% decreased in the Australian study. Also, Anand (26) stratified their findings according to the three ethnic groups included in their study population, East Asian, Chinese and European, and reported ethnic variations in sensitivity and specificity for both HbA1c and FPG tests. In addition, several studies have shown race-related differences in HbA1c curve distribution. Data derived from the NHANES-III survey showed higher values for non-Hispanic blacks and Mexican-Americans than for nonHispanic whites (27). Herman (28) compared HbA1c levels in population-based studies of 11 countries. A1C was higher in Hispanics (9.4 1.4%), Asians (9.2 1.4%) and patients from other racial/ethnic groups (9.7 1.5%) compared with Caucasians (8.9 1.2%) after adjusting for factors affecting glycemia. However, this finding has not been consistent in every study. Thus, different sensitivity and specificity for HbA1c occurs among ethnic groups, which may be related to genetic differences in the concentration of hemoglobin,
306
Gomez-Perez et al./ Archives of Medical Research 41 (2010) 302e308
rates of glycation and the lifespan or amount of red blood cells. No study has compared the diagnostic properties of the HbA1c in populations from Latin America. Regrettably, all available methods for the diagnosis of diabetes have limitations. One example of this with respect to the OGTT-based approach is the significant variability of the 2-h post-challenge glucose levels and the lack of replication of the OGTT result. More practical and specific approaches will be welcome in the future. However, the above-described limitations of the HbA1c criterion precludes its use in many regions of the world. Abnormal hemoglobins. Hemoglobin A1c should not be used in some hemoglobin traits such as HbS, HbC, HbF and HbE (29). Many assay methods can correct for the presence of the most common hemoglobin traits and there are affinity assays unaffected by hemoglobin traits. Nevertheless, the trait should be diagnosed and the adequate A1c methods should be available. Anemia and Iron Deficiency Any condition that shortens erythrocyte survival (i.e., hemolytic anemia, spherocytosis, pregnancy) will proportionally decrease HbA1c because hemoglobin in younger red cells has less time of exposure to ambient glycemia (30). Also, bleeding and the resulting increased reticulocyte production will decrease the half-life of erythrocytes and thereby will lower HbA1C. In contrast, any factor that prolongs the erythrocyte half-life will increase the level of HbA1C (i.e., splenectomy, aplastic anemia). Iron deficiency can lead to rises in HbA1c up to 2% that can be reversed with iron supplementation. Aging A consistent increment in A1c with age was found in a crosssectional analysis of both the Framingham Offspring Study (FOS) and the NHANES study 2001e2004 in nondiabetic populations (31). The longitudinal analysis of FOS confirmed an increase in A1c with aging. A 0.03-point increment per year was found in subjects without abnormalities in glucose metabolism. More studies are needed to establish whether age-specific diagnosis criteria are required. Renal failure Glycated hemoglobin levels can be affected by several factors that are specific for chronic kidney disease (32). Increased blood urea nitrogen levels can cause the formation of carbamylated hemoglobin, which cannot be distinguished from glycated hemoglobin by using electrical charge-based assays. It has been estimated that an increase in time-averaged blood urea nitrogen concentration from 3.57e24.9 mol/l will result in an increment in glycated hemoglobin of 1.26%. This effect is much smaller compared with the impact of an increase in
Table 2. Categories of increased risk for diabetes* (ADA 2010) FPG 100 mg/dL (5.6 mmol/l) to 125 mg/dL (6.9 mmol/l) (IFG) 2-FPG in the 75-g OGTT, 140 mg/dL (7.8 mmol/l) to 199 mg/dL (11 mmol/l) (IGT) A1C 5.7e6.4% *For all three tests, risk is continuous, extending below the lower limit of the range and becoming disproportionately greater at higher ends of the range.
blood glucose concentration of similar proportion. Overestimation of glycated hemoglobin can be overcome by using assays that do not depend on electrical charge or assays specific for HbA1c, which are less prone to interference from carbamylated hemoglobin. Other factors that modify HbA1c. Lower values of HbA1c have been reported in patients with HIV infection. In contrast, severe hypertriglyceridemia, hyperbilirubinemia, chronic alcoholism, chronic ingestion of opioid drugs and chronic use of high doses of salicylates are potential causes of higher HbA1c concentrations. Use of antioxidants may alter HbA1c results (33). HbA1c and Increased Risk for Diabetes The Expert Committee proposed that HbA1c values between 6.0 and 6.4% identify subjects at-risk for having diabetes. This category could be an equivalent for the categories enclosed in prediabetes (i.e., impaired fasting glucose and impaired glucose tolerance) (Table 2). Selection of the threshold was based on the HbA1c concentration distribution found in the NHANES surveys. An A1c value of 6.0% corresponds to two SDs above the population mean. However, this recommendation was modified in the 2010 Clinical Practice Recommendations in which the lower limit was decreased to 5.7%. This decision was based on a ROC curve analysis of the NHANES 2005e2006 data in which an A1c value of 5.7% had the best combination of sensitivity (39%) and specificity (91%) to identify cases of impaired fasting glucose. The same strategy has been used in the past by the ADA when the fasting plasma glucose was incorporated as a potential alternative method for the OGTT. The lower limit of the criterion was changed from 6.1 to 5.6 mmol/l in order to diagnose the same number of cases using the OGTT and the FGP. However, this strategy was partially successful because of the lack of agreement among the diagnostic categories (34). It is very likely that the change in the A1c criterion will cause the same diagnostic disagreements. The use of HbA1c as a diagnostic test in ‘‘prediabetes’’ is debated. The agreement between prediabetes categories and an abnormal A1c (5.7e6.4%) has not been included in the majority of the surveys. The low sensitivity (39%) of the A1c criterion is a strong argument against its use in clinical practice because a highly sensitive test is
HbA1c and Diabetes in Developing Countries
required to screen cases at risk for diabetes. In addition, the cost of the test makes A1c testing an unfeasible option for screening diabetes. The HbA1c criterion (5.7e6.4%) to detect at-risk cases will identify 12% of the U.S. adult population. This number is significantly smaller compared to that diagnosed using the current FPG criteria (25%). Finally, there are several studies supporting the concept that the combined use of HbA1c and FPG has a better predictive value compared to FPG to estimate the likelihood of diabetes in high-risk subjects. However, this approach implies a significantly higher cost (35). This approach was not considered by the ADA Committee. In conclusion, a proposal for a major shift from the present criteria to diagnose diabetes frequently leads to controversy especially if the evidence that support it is not strong enough. High A1c levels are clearly associated with incident microvascular complications in cohorts comprised of previously diagnosed cases with diabetes. The available evidence supports the role of the A1c test as a marker of glycemic control and as a prognostic factor for future complications. However, it is insufficient to recommend it as the method of choice for diagnosis. In addition, the limited access to standardized laboratory A1c methods makes unfeasible the inclusion of the A1c criterion among the diagnostic criteria of diabetes in a large proportion of the world. On the other hand, the selection of a diagnostic threshold is usually made by means of the ROC approach followed by its validation in the target population (using likelihood ratios). Such evidence is missing for Latin American and African countries. Even less information was presented to support the 5.7% threshold. The lack of agreement between the glycemia-based diagnosis and the A1c may lead to a series of conflicting reports regarding various diagnostic criteria-related issues (i.e., prevalence of the disease) that may result in confusion among patients and health providers (34). This change could be in opposition to the principles behind the 1997 ADA recommendations (i.e., to obtain an earlier diagnosis of diabetes and glucose intolerance). Finally, it will increase the A1c tests and related costs. Prospective evaluations are required to assess the clinical and economic consequences that may result from the modification of the current diabetes diagnostic criteria. In the meantime, we should continue using plasma glucose levels and the OGTT for the diagnosis of diabetes. References 1. The International Expert Committee Report on the Role of the A1C Assay in the diagnosis of Diabetes. Diabetes Care 2009;32:1327e1334. 2. American Diabetes Association. Diagnosis Criteria for Diabetes Mellitus. Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 2010;33(suppl 1):S62eS69. 3. American Diabetes Association. Screening for type 2 diabetes (Clinical Practice Recommendations 2004: Position Statement). Diabetes Care 2004;27:S11eS14.
307
4. Vistisen D, Colagiuri S, Borch-Johnsen K. The DETECT-2 Collaboration. Bimodal distribution of glucose is not universally useful for diagnosing diabetes. Diabetes Care 2009;32:397e403. 5. Bennett AL, Guop M, Dharmage SC. HbA1C as a screening tool for detection of type 2 diabetes: a systematic review. Diabet Med 2007;24: 333e343. 6. Sabanayagam C, Liew G, Tai ES, et al. Relationship between glycated haemoglobin and microvascular complications: is there a natural cutoff point for the diagnosis of diabetes? Diabetologia 2009;52: 1279e1289. 7. Selvin E, Steffes MW, Zhu H, et al. Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults. N Engl J Med 2010; 362:800e811. 8. The DECODE Study Group. Is the current definition for diabetes relevant to mortality risk from all causes and cardiovascular and non-cardiovascular diseases. Diabetes Care 2003;26:688e696. 9. Goldstein DE, Little RR, Lorenz RA, et al. Global harmonization of hemoglobin A1c. Clin Chem 2005;51:681e683. 10. Lenters-Westra E, Slingerland RJ. Six out of eight hemoglobin A1c point-of care instruments do not meet the generally accepted analytical performance criteria. Clin Chem 2010;56:44e52. 11. Amquist H, Wallensteen M, Jeppsson JO. Standardization of long-term glucose measurements established. Lakartidningen 1997;50:4789e4790. 12. Peters AL, Davidson MB, Schigrer DL, et al. A clinical approach for the diagnosis of diabetes mellitus. An analysis using glycosylated hemoglobin levels. JAMA 1996;276:1246e1252. 13. Faerch A, Borch-Johnsen K, Vaag A, et al. Sex differences in glucose levels: a consequence of physiology or methodological convenience? The Inter99 study. Diabetologia 2010;53:858e865. 14. Shima K, Endo J, Oimoni Y, et al. A report of the committee on an inter-laboratory standardization of HbA1C determination. The Japan Diabetes Society. J Jpn Diabetes Soc 1994;37:855e864. 15. Garcı´a-Alcala´ H, Ruiz-Argu¨elles A, Cedillo-Carvallo B. Effect of the method to measure levels of glycated hemoglobin on individual clinical decisions comparison of an immunoassay with high-performance liquid chromatography. Am J Clin Pathol 2009;132:332e335. 16. Ezenwaka C, Jones-LeCointe A, Nwagbara E, et al. Anaemia and kidney dysfunction in Caribbean type 2 diabetic patients. Cardiovasc Diabetol 2008;7:25e31. 17. Buell C, Kermah D, Davidson MB. Utility of A1C for diabetes screening in the 1999 2004 NHANES population. Diabetes Care 2007;30:2233e2235. 18. Ravi Kumar P, Bhansali A, Ravikiran M, et al. Utility of glycated hemoglobin in diagnosing type 2 diabetes mellitus: a communitybased study. J Clin Endocrinol Metab. Published ahead of print April 6, 2010 (doi:10.1210/jc.2009e2433). 19. Saudek CD, Herman WH, Sacks DB, et al. A new look at screening and diagnosing diabetes mellitus. J Clin Endocrinol Metab 2008;93: 2447e2453. 20. Herman WH, Fajans SS. Hemoglobin A1c for the diagnosis of diabetes: practical considerations. Pol Arch Med Wewn 2010;120:37e41. 21. Kramer CK, Araneta MR, Barrett-Connor E. A1C and diabetes diagnosis: The Rancho Bernardo Study. Diabetes Care 2010;33:101e103. 22. Carson A, Reynolds K, Fonseca V, et al. Comparison of A1C and fasting glucose criteria to diagnose diabetes among U.S. adults. Diabetes Care 2010;33:95e97. 23. Lu X, Walker K, O’Dea K, et al. A1C for screening and diagnosis of type 2 diabetes in routine clinical practice. Diabetes Care 2010;33:817e819. 24. Herman WH, Dungan KM, Wolffenbuttel BHR, et al. Racial and ethnic differences in mean glucose, hemoglobin A1c, and 1,5 anhydroglucitol in over 2000 patients with type 2 diabetes. J Clin Endocrinol Metab 2009;94:1689e1694. 25. Christensen DL, Witte DR, Kduka L, et al. Moving to an A1c-based diagnosis of diabetes has a different impact on prevalence in different ethnic groups. Diabetes Care 2010;33:580e582.
308
Gomez-Perez et al./ Archives of Medical Research 41 (2010) 302e308
26. Anand SS, Razak F, Vuksan V, et al. Diagnostic strategies to detect glucose intolerance in a multiethnic population. Diabetes Care 2003; 26:290e229. 27. Cowie C, Rust K, Byrd-Holt D, et al. Prevalence of diabetes and high risk for diabetes using A1C criteria in the U.S. population in 1988e2006. Diabetes Care 2010;33:562e568. 28. Herman WH, Ma Y, Uwaifo G, et al. Diabetes Prevention Program Research Group Differences in A1C by race and ethnicity among patients with impaired glucose tolerance in the Diabetes Prevention Program. Diabetes Care 2007;30:2453e2457. 29. Kim C, McKeever K, Herman WH, et al. Association between iron deficiency spherocytosis and HbA1C levels among adults without diabetes in the National Health and Nutrition Examination Survey, 1999e2006. Diabetes Care 2010;33:780e785. 30. El-Agouza I, Abu SA, Syrdah M. The effect of iron-deficiency anaemia on the glycosylated hemoglobin subtypes: possible consequences for clinical diagnosis. Clin Lab Haematol 2002;24:285e289.
31. Pani LN, Korenda L, Meigs JM, et al. Effect of aging on A1C levels in individuals without diabetes. Diabetes Care 2008;31:1991e1996. 32. Krovesdy CP, Sharma K, Kalantar-Zadeh K. Glycemic control in diabetic CKD patients: where do we stand? Am J Kidney Dis 2008; 52:766e777. 33. Gomez-Perez FJ, Valles-Sanchez VE, Lopez-Alvarenga JC, et al. Vitamin E modifies neither fructosamine nor HbA1c levels in poorly controlled diabetes. Rev Invest Clin 1996;48:421e424. 34. Gomez-Perez FJ, Aguilar-Salinas CA, Lo´pez-Alvarenga JC, et al. Lack of agreement between the World Health Organization category of impaired glucose tolerance and the American Diabetes Association category of impaired fasting glucose. Diabetes Care 1998;21: 1886e1888. 35. Ko GTC, Chan JCN, Yeung VTF, et al. Combined use of fasting plasma glucose concentration and HbA1C or fructosamine predicts the likelihood of having diabetes in high risk subjects. Diabetes Care 1998;21:1221e1225.